An ultra-low power consumption container positioning device and a control method thereof

Abstract

The application relates to the technical field of Internet of Things positioning and low-power networking, and discloses a super-low-power container positioning device and a control method thereof. The method comprises the following steps: collecting multi-modal sensor data to identify the physical stacking state and the motion state of a container, and constructing a group three-dimensional space topology; a center node is elected according to the topology and equivalent available power, and the power supply of a satellite positioning module and a long-distance communication module is cut off from the hardware power supply layer by a slave node; the center node and the slave node adaptively adjust near-field communication parameters according to the metal environment characteristics of the container to perform data interaction; the center node dynamically evaluates a rotation period according to the equivalent available power and a predicted power consumption model, broadcasts a rotation warning when the power is insufficient to support the next period, and dynamically transfers roles in the group. The application significantly reduces the overall energy consumption of the equipment and prolongs the endurance time of the equipment under the premise of ensuring the continuity of positioning data.

Description

Technical Field

[0001] This invention relates to the field of IoT positioning and low-power networking technology, and in particular to an ultra-low power container positioning device and its control method. Background Technology

[0002] As a core carrier of global freight transport, the traceability of container locations during storage, sea freight, and rail and road transshipment is crucial for logistics scheduling, asset management, and safety supervision. To meet the need for real-time container location monitoring, current technologies typically involve installing positioning devices on the container body, equipped with satellite positioning modules (GPS / BeiDou, etc.) and long-distance communication modules (such as NB-IoT, GPRS, 4G Cat.1, etc.). The positioning device periodically activates the satellite positioning module to acquire location information and uploads this information to a cloud server via the long-distance communication module, thereby enabling remote tracking.

[0003] However, the following problems mainly exist in the actual use cases of container positioning: First, containers are usually stacked vertically in the yard in four to six layers. The electromagnetic environment between the steel containers is similar to a "metal canyon". Satellite signals and long-distance wireless signals are strongly blocked and multipath interference inside the stack. Positioning devices located inside or at the bottom of the stack cannot independently complete satellite positioning and data transmission. Secondly, the complete business cycle of containers from loading, transportation to stacking often lasts for months or even years. The positioning device relies on built-in batteries for power and is difficult to replace frequently, which imposes extremely strict constraints on power consumption. Third, the frequency of container position changes varies significantly under different motion states such as static stacking, sea vibration, road transport, and rail transport. If a unified positioning and reporting cycle is adopted, it will result in a large amount of ineffective energy consumption. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides an ultra-low power container positioning device and its control method.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: According to a first aspect of the present invention, a control method for an ultra-low power container positioning device is provided, comprising: acquiring data from a multimodal sensor built into the positioning device, identifying the physical stacking state and motion state of the container, and constructing a three-dimensional spatial topology relationship among multiple positioning devices within the same physical area; selecting at least one central node among multiple positioning devices within the communication range, and the remaining positioning devices as slave nodes, based on the three-dimensional spatial topology relationship and the equivalent available power of each positioning device; controlling the central node to activate the satellite positioning module, and controlling the slave nodes to disconnect the power supply of the satellite positioning module and the long-distance communication module from the hardware power supply level; establishing a near-field communication connection between the central node and the slave nodes, and adaptively adjusting the near-field communication parameters for data interaction based on the characteristics of the container's metal environment; dynamically evaluating the supported rotation cycle based on its own equivalent available power and a predicted power consumption model, and broadcasting a rotation warning to the slave nodes when predicting that its own equivalent available power is insufficient to support the next rotation cycle, so as to dynamically transfer the role of the central node within the group.

[0006] According to a second aspect of the present invention, an ultra-low power container positioning device is provided, comprising: a satellite positioning module, a long-distance communication module, a near-field communication module, a multimodal sensor, a microcontroller unit, and a power management unit; the microcontroller unit includes a topology construction module, an election module, an energy management module, and a communication adaptive module; the multimodal sensor includes a barometer and an inertial measurement unit, used to collect data to identify the physical stacking state and motion state of the container; the topology construction module is used to construct a three-dimensional spatial topology relationship of multiple positioning devices within the same physical area based on the data from the multimodal sensor; the election module is used to elect at least one central node among the multiple positioning devices within the communication range, with the remaining positioning devices acting as slave nodes, based on the three-dimensional spatial topology relationship and the equivalent available power of each positioning device; the energy management module is used to control the central node to activate the satellite positioning module and control the slave nodes to disconnect the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level; the communication adaptive module... The adaptation module is used to establish near-field communication connections between the central node and the slave nodes, and adaptively adjusts near-field communication parameters for data interaction based on the characteristics of the container metal environment. The energy management module is also used to dynamically assess the supported rotation cycle based on the central node's own equivalent available power and predicted power consumption model. When it is predicted that its own equivalent available power is insufficient to support the next rotation cycle, it broadcasts a rotation warning to the slave nodes to dynamically transfer the central node's role within the group. The power management unit includes a dual-channel power switch circuit and a bus switch. The dual-channel power switch circuit is connected in series between the battery power supply and the power input terminals of the satellite positioning module and the long-distance communication module, respectively, and is controlled by the switch control signal output by the microcontroller unit. The bus switch is connected between the data signal pins of the satellite positioning module, the long-distance communication module, and the microcontroller unit, and is driven by the same switch control signal as the dual-channel power switch circuit. In the off state, it simultaneously cuts off the power supply circuit and blocks the data signal path.

[0007] The technical solution provided by this invention has the following beneficial technical effects compared with the prior art: (1) Achieve group spatial perception without external dependence in metal-obstructed environments. By collecting data from the multimodal sensors built into the positioning devices, the physical stacking state and motion state of the containers are identified, and the three-dimensional spatial topology relationship of multiple positioning devices in the same physical area is constructed. In the stacking environment where satellite signals are severely obstructed by steel containers, the present invention can autonomously restore the relative spatial position and group motion state of each device without relying on external infrastructure, providing accurate spatial and state basis for subsequent node election and power consumption scheduling, and overcoming the defects of isolated devices and lack of collaborative benchmarks in the prior art.

[0008] (2) Eliminate energy redundancy caused by repeated positioning and uploading within the group. Based on the three-dimensional spatial topology and the equivalent available power of each device, a central node is elected and the rest are slave nodes. High-power responsibilities such as satellite positioning and long-distance data uploading are concentrated on a few nodes. The same stack only needs one positioning and one upload path to meet the position awareness needs of all devices. The overall energy consumption of the group no longer increases exponentially with the number of devices. The equivalent available power participates in the election to avoid low-power devices taking on the central role and failing prematurely.

[0009] (3) Achieve near-zero power standby at the hardware level for slave nodes. Control the slave nodes to cut off the power supply to the satellite positioning module and the long-distance communication module at the hardware power supply level. Unlike the software hibernation mode, where the module still maintains static standby current and there is backflow leakage at the interface, this invention physically disconnects the power supply circuit, and the static leakage is completely eliminated. The long-term standby power consumption of the slave nodes is reduced to the microampere level. For container positioning scenarios that are statically stacked or in service on a time scale of months or years, the endurance improvement is particularly significant.

[0010] (4) Resolve the contradiction between communication reliability and energy consumption in a metallic environment. The central node establishes a near-field communication connection with the slave nodes and adaptively adjusts the communication parameters according to the characteristics of the container metallic environment. This ensures that the link remains reliable when the obstruction is severe, reduces the extra power consumption of repeated retransmissions, and transmits efficiently when the obstruction is reduced. At the same time, the data from the slave nodes is aggregated to the central node via the near-field link for unified external interaction, avoiding the need for each device to independently activate the long-distance communication module.

[0011] (5) The switching of the central node is changed from passive remediation to active and orderly handover. The central node dynamically evaluates the rotation cycle it can support based on its own equivalent available power and predicted power consumption model. When the prediction is insufficient to support the next cycle, it broadcasts a rotation warning to the slave nodes and dynamically transfers the role, avoiding the positioning gap and data disconnection caused by switching after the power is exhausted or communication fails in the existing technology. The central role is rotated and distributed within the group, so that the power consumption of each device tends to be balanced. The system life is determined by the group as a whole rather than a single node, which significantly extends the service life. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of an embodiment of the control method for an ultra-low power container positioning device in this invention. Figure 2 This is a schematic diagram of the hierarchical sleep state machine of the ultra-low power container positioning device in an embodiment of the present invention; Figure 3 This is a flowchart of the fault takeover and data compensation of the ultra-low power container positioning device in an embodiment of the present invention; Figure 4 This is a schematic diagram of one embodiment of the ultra-low power container positioning device in this invention. Detailed Implementation

[0013] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0014] The ultra-low power container positioning device described in various embodiments of the present invention includes the following hardware components: (1) Microcontroller unit (MCU): a low-power microcontroller with deep sleep mode is selected, and the deep sleep current is in the microampere level; the microcontroller unit integrates the main control chip; (2) Satellite positioning module, supporting GPS / BeiDou dual-mode or multi-mode reception, with a typical operating current of about 35 mA during the acquisition phase, a typical hot start positioning time of less than 5 seconds, and a typical cold start positioning time of more than 30 seconds; (3) Long-distance communication module, supporting NB-IoT or LTE Cat.1 cellular communication; (4) Near-field communication module, supporting low-power Bluetooth or IEEE 802.15.4 short-range communication; (5) Multimodal sensors, including barometers and inertial measurement units (including triaxial accelerometers and triaxial gyroscopes, supporting motion wake-up interruption); (6) Power management unit, including dual power switch circuit (composed of P-channel MOSFET load switch or integrated load switch chip, controlled by MCU general input / output pins) and bus switch.

[0015] In the following embodiments, all positioning devices within the same stack constitute a group; the nodes elected to assume the central role within the group are collectively referred to as central nodes. In some embodiments, the central nodes are further subdivided into edge communication nodes, area central nodes, and positioning master nodes; nodes other than the central nodes are collectively referred to as slave nodes.

[0016] Example 1 Traditional container positioning devices operate in a standalone autonomous mode: each device independently activates its satellite positioning module for location and its long-distance communication module for data reporting. In multi-layered stacking scenarios, devices located inside the stack and at the bottom are obstructed by the steel containers above and around them, resulting in a decrease in the signal-to-noise ratio of satellite signals. The positioning module needs to significantly extend its acquisition time to complete the positioning or may even fail to locate the device. In metal multipath environments, the long-distance communication module's transmission power increases, retransmissions increase, and power consumption further deteriorates. At the same time, the uniform power consumption of all devices leads to the simultaneous depletion of power for all devices within the same stack, and the group's endurance is determined by the weakest single node, making it impossible to utilize the group's power redundancy. This embodiment, in a container stacking group composed of a large number of positioning devices, reduces the overall power consumption of the group by more than an order of magnitude by sensing the physical stacking shape, assigning power according to roles, implementing hardware-level power-off and dynamic rotation, while maintaining the group's positioning and data reporting capabilities. Furthermore, it makes the group's endurance determined by the total power of the group.

[0017] like Figure 1 As shown, the control method for the ultra-low power container positioning device in this embodiment includes: Step 101: Collect data from the multimodal sensors built into the positioning device, identify the physical stacking state and motion state of the containers, and construct the three-dimensional spatial topology relationship of multiple positioning devices in the same physical area; The multimodal sensor includes at least a barometer and an inertial measurement unit (IMU). The barometer senses the vertical position of the device in the atmosphere: since atmospheric pressure decreases approximately linearly with altitude, there are discernible differences in atmospheric pressure measured by devices on different stacked layers within the same stack, allowing for the calculation of the relative layer number of each device in the stack. The IMU senses the motion and vibration characteristics of the device: by performing frequency domain characteristic analysis such as Fast Fourier Transform on the triaxial acceleration signals, it distinguishes between static stacking, road or rail transport, and marine oscillation states based on the distribution of vibration energy in the spectrum. After state identification, each positioning device periodically broadcasts a message containing its own device identifier, relative layer number, motion status, and equivalent available power via its near-field communication module. Within the communication range, these devices receive each other's broadcast messages, gradually forming a three-dimensional spatial topology describing the relative spatial position and status of each device within the same physical area. This topology is stored in the memory of each device in the form of a topology table. Each entry in the topology table includes at least the fields of device identifier, relative layer number, near-field link signal strength, motion status, and equivalent available power.

[0018] The relative altitude is calculated using the barometric altitude formula: ; In the formula, The height of the positioning device relative to the reference plane, in meters; Temperature at the reference plane, in Kelvin; The vertical lapse rate of atmospheric temperature is taken as 0.0065 Kelvin per meter under standard atmospheric conditions. The barometer measures atmospheric pressure, with the unit being Pascals (Pa). Atmospheric pressure at the reference plane, in Pascals; Let be the gas constant of dry air, taken as 287.05 joules per kilogram per Kelvin; The acceleration due to gravity is taken as 9.80665 meters per second squared.

[0019] The number of stacking layers is calculated based on the relative height using the following formula: ; In the formula, For the first The stacking layer number of the positioning device starts from zero; For the first The height of the platform is calculated based on the air pressure height formula; The reference height of the storage yard is determined by statistical analysis of the air pressure measurements of the equipment located at the bottom of the stack. The effective stacking height for a single layer of containers is the sum of the height of a single container and the stacking gap between containers. For a standard container, it is taken as 2.65 meters; "round" indicates rounding to the nearest integer.

[0020] For example, an export storage yard has a six-layer stack with a total of 210 positioning devices (equipment identification numbers D001 to D210). The barometer on device D108 measures an atmospheric pressure of 100856.23 Pa, while the ground reference device in the storage yard measures an atmospheric pressure of 100962.5 Pa. Substituting these values ​​into the pressure-height formula, the height difference is approximately 8.86 meters (for simplified estimation, the pressure difference formula can be used directly). ,in The air density is taken as 1.225 kg per cubic meter. It is the acceleration due to gravity. (This is the pressure difference), and then according to the formula for the number of layers, 8.86 / 2.65=3.34, rounded to get... That is, device D108 is located on the fourth layer from the ground. If the triaxial acceleration signal collected by the inertial measurement unit of device D056 within the half-hour observation window is concentrated in the 0.1 to 0.5 Hz frequency band and exhibits periodic oscillation characteristics after frequency domain analysis, it is determined to be in a marine oscillation state; if the vibration energy of device D172 is concentrated in the 2 to 20 Hz vehicle body bump frequency band and the 20 to 200 Hz road surface high-frequency excitation frequency band, it is determined to be in a road transport state; if the acceleration amplitude of the other devices in devices D001 to D210 is lower than the silence threshold of 0.03g for a long time, it is determined to be in a static stacking state. Each device sends a broadcast message through the near-field communication module at a broadcast interval of once every 10 seconds. For example, the broadcast message content of device D108 is {Device ID D108, Layer No. 3, Movement Status: Static Stacking, Equivalent Available Power 7200 mAh, Link Signal Strength -52 dBmW}. After receiving the messages from each other, the topology table construction of all 210 entries is completed within 40 seconds.

[0021] Step 102: Based on the three-dimensional spatial topology and the equivalent available power of each positioning device, at least one central node is selected from among the multiple positioning devices within the communication range, and the remaining positioning devices are slave nodes. The overall score for the election is a weighted sum of the normalized values ​​of equivalent available power, satellite visibility, and communication signal quality. The equivalent available power is evaluated according to a set method, reflecting the battery's true discharge capacity under current temperature, self-discharge, and aging conditions. The scoring formula is: ; In the formula, Scoring the election of the central node for the i-th positioning device; The equivalent available power of the i-th device is expressed in milliampere-hours. This refers to the battery's nominal capacity, measured in milliampere-hours (mAh). For the first The satellite visibility quantification index of the device can be calculated based on the layer number of the device in the topology table and whether it is blocked by the upper-layer nodes. The maximum value is taken for the top layer unblocked nodes. This is a reference value for satellite visibility. The received signal strength of the link between the i-th device and the outside of the stack is expressed in decibels and milliwatts. This is a reference value for the received signal strength. , , The weighting coefficients are and satisfy the following conditions: .

[0022] Each device exchanges scores within the election window, and at least one device with the highest score is elected as the central node, with the remaining devices serving as slave nodes. When there are multiple central nodes, they can be further subdivided into edge communication nodes, area-based central nodes, and positioning master nodes, as described in subsequent embodiments.

[0023] For example, take , , , milliampere hour, , Decibels and milliwatts. Device D210 is located in an unobstructed position on the top layer (sixth layer), with an equivalent available power of 7600 mAh and a link signal strength of -45 dBmW to the outside of the stack. Device D042 is located on the second floor, obscured by the box above it on the fourth floor, and has an equivalent available power. milliampere-hours, signal strength -75 dBmW, ,but After a group-wide scoring exchange, device D210 was selected as the central node with the highest score, and the remaining 209 devices became slave nodes.

[0024] Step 103: Control the central node to turn on the satellite positioning module, and control the slave node to cut off the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level; The central node's main control chip outputs a first turn-on control signal, enabling the dual-power switching circuit to connect the battery power supply to the power supply circuits of the satellite positioning module and the long-distance communication module, thus giving the central node satellite positioning and long-distance communication capabilities. The slave node's main control chip outputs a turn-off control signal, physically disconnecting the power supply circuits of the satellite positioning module and the long-distance communication module from the battery power supply via the dual-power switching circuit. Simultaneously, it disconnects the data signal path between these modules and the main control chip. The slave node only retains power to the near-field communication module, the multi-modal sensor, and the main control chip's low-power standby circuit. The fundamental difference between hardware power-off and software hibernation is that in software hibernation mode, the module's internal registers and power domain remain in place, resulting in leakage current ranging from microamps to tens of microamps. For example, the typical current in software hibernation for a satellite positioning module is 15 microamps, and the typical current in power-saving mode for a cellular module is 10 microamps. However, after the hardware power supply is cut off, the module's power domain is completely de-energized, and the leakage current approaches zero.

[0025] For example, within 100 milliseconds after selection and confirmation, the master control chip of the central node D210 outputs a high-level first conduction control signal through the general-purpose input / output pin, turning on the MOSFET load switch, powering on the satellite positioning module and performing warm-start positioning, and powering on the long-distance communication module and registering with the cellular network. The master control chip of the slave node D042 outputs a low-level shutdown control signal, turning off both MOSFET load switches, completely de-energizing the satellite positioning module and the long-distance communication module. A multimeter measurement shows the voltage at the module's power input pin to be zero volts. Based on 209 slave nodes, compared to the software hibernation scheme, eliminating only the leakage current of the satellite positioning module during hibernation saves approximately 75 mA / h per day (209 × 24 × 0.015).

[0026] Step 104: The central node establishes a near-field communication connection with the slave node, and adaptively adjusts the near-field communication parameters according to the characteristics of the container metal environment to exchange data; The central node and each slave node establish a connection through a near-field communication module. During the connection establishment process, they exchange probe messages. The round-trip delay of the probe messages and the received signal strength are used to assess the attenuation of the link due to metal blockage. Based on this, the central node selects and switches between the first communication parameter group (low rate, high power, short packet length, multiple transmission) and the second communication parameter group (high rate, low power, long packet length, fewer retransmissions). The data exchange content includes at least the status data reported by the slave nodes and the group location sent by the central node.

[0027] For example, if the round-trip latency of the probe message between the central node D210 and the underlying slave node D017 is 18 milliseconds, the received signal strength is -87 dBmW, and the metal obstruction attenuation assessment value exceeds the first attenuation threshold, D210 will adjust the link configuration towards D017 to the first communication parameter group: the physical layer rate is changed from... The transmission power was reduced from megabits per second to 250 kilobits per second. The decibel and milliwatt were increased to The single packet length was shortened from 255 bytes to 64 bytes, and the maximum number of retransmissions was increased from 2 to 5. The received signal strength between D210 and its adjacent top-level slave node D195 was -52 dBmW, below the second attenuation threshold, and configured with the second communication parameter group: data rate 2 Mbps, transmit power -4 dBmW, single packet length 255 bytes, and 1 retransmission. During data exchange, D017 successfully reported a status frame using the first communication parameter group, and D195 successfully reported a status frame using the second communication parameter group.

[0028] Step 105: The central node dynamically assesses the number of rotation cycles it can support based on its own equivalent available power and predicted power consumption model. When it predicts that its own equivalent available power is insufficient to support the next rotation cycle, it broadcasts a rotation warning to the slave nodes so as to dynamically transfer the role of the central node within the group.

[0029] The predicted power consumption model is constructed based on the actual power consumption statistics of the central node performing various tasks such as satellite positioning, long-distance communication, and near-field networking during historical cycles. The total power consumption required for the next rotation cycle is the sum of the power consumption components of each task. ; In the formula, The total power consumption required to complete one rotation cycle is expressed in milliampere-hours; k is the number of task types within one rotation cycle; The operating current for the k-th type of task is expressed in milliamperes. This represents the cumulative power-on duration for the k-th type of task within a cycle, expressed in hours.

[0030] The central node performs a judgment at the end of each rotation cycle: ; In the formula, To rotate the warning signs, take This indicates that a rotation needs to be initiated; The equivalent available power of the central node itself; Predict total power consumption for the next power cycle; For safety margin, use 5% of the battery's nominal capacity.

[0031] For example, the predicted power consumption model statistics for the central node D210 are as follows: During each rotation cycle (seven days), the satellite positioning task is powered on for a cumulative total of 8 hours with an average current of 35 mA; the long-distance communication task is powered on for a cumulative total of 3 hours with an average current of 90 mA; the near-field networking task is powered on for a cumulative total of 14 hours with an average current of 8 mA; and the MCU is active for 168 hours with an average current of 3 mA. Substituting these values ​​into the total power consumption formula yields E. cycle =(35×8+90×3+8×14+3×168) / 1000=1.166 ampere-hours, or 1166 milliampere-hours. When the equivalent usable capacity of D210 drops to... When the current is 1580-1166=414 mA, it is lower than the safety margin of 425 mA, so D210 is set. The system broadcasts a rotation warning message to all slave nodes, and slave node D203 adjusts the response based on the equivalent available power. The milliampere-hour and spatial topology location (top layer, distance from D210 and two boxes) are determined to be the successor node. After D210 hands over the role of the central node to D203, it cuts off the power supply to the satellite positioning module and long-distance communication module of this node and switches to the slave node low power mode. D203 turns on the satellite positioning module to take over the group positioning and reporting functions.

[0032] This embodiment, through role division, ensures that only a very small number of nodes in the stacking group undertake high-power tasks such as satellite positioning and long-distance communication, while the hardware of the remaining nodes is powered off. The daily power consumption of the group is reduced from approximately 210 × 100 = 21,000 mAh in the traditional scheme to approximately 60 + 40 + 208 × 1.3 ≈ 370 mAh (approximately 60 mAh for central node positioning and reporting, approximately 40 mAh for auxiliary communication, and approximately 1.3 mAh for each slave node), a power consumption reduction of approximately 98%. Through dynamic rotation driven by equivalent available power, the power within the group is consumed evenly, and the group's endurance approaches the total equivalent power of the group divided by the group's daily power consumption, approximately 210 × 8500 × 0.85 / 370 ≈ 4100 days, close to the battery's storage life. This completely changes the situation in the traditional scheme where the group's endurance is limited by the weakest single node and fails completely after about one year.

[0033] Example 2 In container stacking scenarios, positioning devices cannot directly determine their own stacking layer or transportation status, leading to a lack of spatial and state-based basis for subsequent role selection: if the layer number cannot be determined, the positioning master node with optimal satellite visibility at the top layer cannot be selected; if the movement state cannot be determined, it is impossible to distinguish between stationary stacking (which can be located at low frequency over long periods) and transportation in progress (which requires higher frequency positioning). Existing locators rely on satellite positioning results to infer the state, which falls into the logical contradiction of "first consuming power to locate and then determining whether positioning is needed." This embodiment, without activating the satellite positioning module, utilizes a low-power multimodal sensor to complete stacking layer identification and movement state identification with extremely low power consumption, and constructs a group's 3D topology at low cost.

[0034] In this embodiment, step 101, "collecting data from the multimodal sensors built into the positioning device, identifying the physical stacking and movement states of the containers, and constructing a three-dimensional spatial topological relationship between multiple positioning devices within the same physical area," further includes: Step 201: Collect atmospheric pressure data using the barometer built into the positioning device, and calculate the relative layer number of the positioning device in the container stack; The barometer collects raw atmospheric pressure data at fixed sampling intervals. To suppress pressure disturbances caused by gusts and mechanical operations in the storage yard, the raw data is first filtered using a moving average before being used for layer number calculation. The selection of the filter window length and sampling interval must balance response speed and stability. Since the atmospheric pressure difference corresponding to a single standard container layer is approximately 30 Pa (approximately 12 Pa per meter of height, with an effective layer height of 2.65 meters), while the barometer resolution is on the order of 0.1 Pa and the relative accuracy is on the order of ±6 Pa, the inter-layer pressure difference is significantly greater than the measurement error, ensuring reliable resolution margin for layer number determination. The formula for calculating the relative layer number is: ; In the formula, Let i be the stack layer number of the i-th positioning device; The reference atmospheric pressure for the stacked ground is determined by statistically analyzing the maximum value of the air pressure data broadcast by each device in the group (the lower the altitude, the higher the air pressure, and the lowest device has the highest air pressure). The measured atmospheric pressure after filtering by the i-th device is expressed in Pascals. Let the air density be denoted as . kilograms per cubic meter; g is the acceleration due to gravity, taken as 9.8 meters per second squared; The effective stacking height for a single-layer container is set at 2.65 meters.

[0035] The filtering formula is: ; In the formula, This is the filtered air pressure value after the nth sampling. For the first The original sampled value of the secondary air pressure; α is the filter coefficient, with a range of values. In this embodiment, the value is 0.2.

[0036] For example, device D061 samples the air pressure every 30 seconds. Ten consecutive raw samples are 100901.3, 100902.1, 100900.8, 100901.5, 100903.0, 100901.1, 100900.2, 100901.9, 100902.4, and 100901.0 Pa. After first-order inertial filtering, the pressure converges to approximately 100901.5 Pa. The ground reference air pressure P broadcast by the group... base =100962.5 Pa (obtained from statistics of the bottom devices D001 to D035), the pressure difference is 61 Pa. Substituting into the layer number formula, we get 1.91, which is rounded to 2. That is, D061 is located in the third layer from the ground. The filtering process effectively suppressed the instantaneous air pressure disturbance of about ±10 Pa caused by the operation of the gantry crane in the yard: a disturbance caused the original sample to jump to 100912 Pa, but after filtering, it was only 100903.5 Pa. The layer number calculation result remained stable at 2, and no layer number jump occurred.

[0037] Step 202: Collect vibration signals through the inertial measurement unit built into the positioning device, and perform frequency domain feature analysis on the vibration signals; The inertial measurement unit (IMU) acquires triaxial acceleration signals at a high sampling rate. The main control chip performs a fast Fourier transform on the acceleration amplitude sequence of a fixed duration to obtain the spectrum and calculates the energy proportion of each predetermined frequency band. For example... Figure 2As shown, the judgment rules are as follows: If periodic low-frequency oscillation vibrations of a specific frequency are detected (the energy proportion in the 0.05 to 0.5 Hz frequency band exceeds the low-frequency threshold and exhibits periodicity), it is judged as a marine oscillation state; if high-frequency vibration characteristics are detected (the energy proportion in the 2 to 200 Hz frequency band exceeds the high-frequency threshold), it is judged as a road or rail transport state; if the vibration signal amplitude is lower than the silence threshold, it is judged as a static stacking state. The formula for calculating the frequency band energy proportion is: ; In the formula, For the predetermined frequency band The proportion of energy; For the frequency spectrum of the acceleration signal The amplitude at that point; , To analyze the lower and upper limits of the frequency band, we take... Hertz and hertz.

[0038] For example, if device D118 is in ocean transport, and its inertial measurement unit collects vertical acceleration at a sampling rate of 100 Hz, performing a fast Fourier transform on the 1024-point sequence, the energy proportion η in the 0.05 to 0.5 Hz frequency band is calculated to be 0.71, and the spectrum shows a significant peak at 0.18 Hz, with the waveform exhibiting a regular oscillation with a period of approximately 5.5 seconds, satisfying the condition of periodic low-frequency oscillation, it is determined to be in a marine oscillation state; if device D061 is in road transport, the spectrum calculation shows that the energy proportion in the 2 to 20 Hz frequency band is 0.52, and the energy proportion in the 20 to 200 Hz frequency band is 0.31, totaling 0.83, exceeding the high-frequency threshold of 0.6, it is determined to be in a road or rail transport state; if device D061 returns to the storage yard and the acceleration amplitude is below the 0.03g silence threshold for ten consecutive minutes, it is determined to be in a stationary storage state. Throughout the entire identification process, the satellite positioning module remained powered off, and the total power consumption of the inertial measurement unit and signal processing was less than 1% of the power consumption of a single satellite positioning operation.

[0039] Step 203: Each positioning device broadcasts its relative layer number, motion status, and device identifier through the near-field communication module to construct a three-dimensional topology map containing the relative spatial positions of each node.

[0040] Each positioning device transmits broadcast messages carrying its own information at preset broadcast intervals on a low-power broadcast channel, while simultaneously listening for broadcast messages from its neighbors. Each device maintains a local topology table, indexed by device identifiers and using relative layer number and near-field link signal strength as spatial information carriers. The relationship between near-field link signal strength and free-space distance is as follows: ; In the formula, Received signal strength, measured in decibels and milliwatts; For reference distance (Take the received signal strength at a distance of one meter); The path loss index is 2.5 to 4 for multipath environments of metal stacks; d is the distance between the transmitting and receiving nodes. The masking fading term follows a normal distribution with zero mean and a standard deviation σ of 4 to 10 dB.

[0041] Based on the layer number and signal strength, the topology map uses the "layer" as the vertical dimension and the horizontal proximity derived from the signal strength as the horizontal dimension to form a three-dimensional spatial topological relationship of the group.

[0042] For example, all 210 devices broadcast at ten-second intervals. An example broadcast message would be {D061, floor 2, stationary stacked state, power 7100 mAh}. Device D061 detects over one hundred broadcasts from its neighbors D060 (floor 2, RSSI = -48 dB / mW), D062 (floor 2, RSSI = -51 dB / mW), D025 (floor 1, RSSI = -70 dB / mW), and D097 (floor 3, RSSI = -73 dB / mW) within sixty seconds. Using the signal strength formula, for D060, -48 = -40 - 10 × 3 × log₂ 10 (d / 1), solving for d, we get d≈1.7 meters, meaning D060 and D061 are horizontally adjacent to each other by about one container. Each device completes the topology table within two minutes and then summarizes it to form a three-dimensional topology map describing six layers, with dozens of containers in each layer. The relative positional relationship between the top-level unobstructed nodes (such as D210 and D203) and the bottom-level nodes (such as D001 to D035) is clear at a glance, providing an accurate basis for the subsequent selection of the positioning master node with the best satellite visibility.

[0043] This embodiment completes stacking layer identification and motion state identification without activating the satellite positioning module. The daily power consumption of a single device for state perception is only about 0.6 mAh (24 × 0.01 mAh for continuous barometer sampling, 24 × 0.02 mAh for intermittent inertial measurement unit sampling, and 24 × 0.02 mAh for filtering and frequency domain calculation), which is less than 1% of the traditional "positioning first, then state determination" scheme. The construction of the three-dimensional topology map provides a spatial data foundation for subsequent functions such as role election, multi-hop relay path planning, and stacking split detection. Moreover, all communication uses low-power near-field broadcasting, avoiding the consumption of long-distance communication power for topology acquisition.

[0044] Example 3 In Example 1, the central node plays a single role, with a single device simultaneously handling three high-power functions: satellite positioning, long-distance communication, and near-field convergence. This approach has two drawbacks: First, the overlapping of these three functions causes the central node's power consumption to be significantly higher than the group average, resulting in frequent rotations and high management overhead. Second, the location with optimal satellite visibility (top layer of the stack) and the location with optimal cellular signal quality (outermost layer of the stack) typically do not coincide, making it impossible for a single central node to simultaneously handle both communication conditions, leading to long satellite positioning times or numerous cellular retransmissions. This example subdivides the central node's functions into three categories: edge communication nodes, area-based central nodes, and positioning master nodes. This allows high-power modules to operate at their respective locations and layers with optimal signal conditions, further reducing group power consumption and improving positioning and communication success rates.

[0045] In this embodiment, step 102, "selecting at least one central node from among multiple positioning devices within the communication range based on the three-dimensional spatial topology and the equivalent available power of each positioning device," further includes: Step 301: In the three-dimensional spatial topology, select the node located on the outermost side of the stack with the best long-distance communication signal quality as the edge communication node, and only the edge communication node will enable the long-distance communication module to communicate with the cloud server. The selection criteria for edge communication nodes are based on the quantification value of long-distance communication signal quality and the external location conditions. Since the cellular signal of the outermost device in the stack is least obstructed by metal, edge communication nodes are selected from the set of nodes located on the outer side of the stack (horizontally at the stack boundary layer) in the topology diagram. The election score is primarily based on the detection results of long-distance communication signal quality, supplemented by equivalent available power. The detection method is as follows: the candidate node momentarily activates its long-distance communication module to perform signal quality detection (measuring the reference signal received power or signal field strength), records the result, and immediately shuts it down. The detection consumes approximately a few milliampere-hours of power. The signal quality scoring formula is: ; In the formula, For the first Scoring of edge communication nodes of the device; For the first The received power of the cellular reference signal detected by the device, measured in decibels and milliwatts; , These represent the minimum and maximum acceptable reference signal received power, respectively, taken as -110 dBmW and -60 dBmW; This is the equivalent usable power. This refers to the battery's nominal capacity. , The weighting coefficients are 0.7 and 0.3.

[0046] For example, in the six-layer stack, candidate devices D041 to D070 located on the south facade sequentially perform cellular signal detection, and D058 detects... decibels per milliwatt, equivalent available power Milliampere-hours, rating D058 was selected as the edge communication node due to its high ranking among the candidates. D058 activated its long-range communication module, registered with the network, and established a connection with the cloud server. The long-range communication modules of the remaining 209 devices in the group remained powered off. During the detection phase, each candidate device only activated its communication module. The measurement is performed in seconds, consuming approximately [amount missing] power. The total power consumption for the full candidate set detection is less than 60 mAh, which is far lower than the scheme of having each device maintain its own cellular connection.

[0047] Step 302: Elect a planar central node in each layer to collect data reported by the nodes in this layer through near-field communication. The planar central node transmits the data of this layer to the edge communication node through near-field communication multi-hop relay. The planar center node is elected within this layer. The election objective is to minimize the sum of the total near-field communication power consumption from each slave node to the planar center node and the multi-hop relay power consumption from the planar center node to edge communication nodes. The communication cost within this layer is measured by the sum of the signal strengths from each slave node to the candidate node. A higher sum of signal strengths indicates a more centrally located candidate node within this layer, resulting in better link quality and fewer retransmissions. The planar center node scoring formula is as follows: ; In the formula, For the first The platform device is used as the central node of the local surface for scoring; For the first The set of all nodes in the layer where the device is located; For this layer Taiwan node to the first The strength of the near-field received signal at the station node; , As the normalization boundary for near-field signal intensity, take decibels and milliwatts decibels and milliwatts; The weighting factor for electricity consumption is taken as follows: The meanings of the remaining parameters are the same as before.

[0048] The planar central node transmits the data of this layer to the edge communication node via multi-hop relays. The relay path selects the cross-layer link based on the inter-layer signal strength in the topology table.

[0049] For example, in the fourth layer, there are 35 devices. Candidate device D127's normalized sum of the signal strength of the remaining 34 nodes in this layer is 28.6. Adding the power consumption factor of 5 × 0.87 = 4.35, the total score is 32.95, which is the highest in this layer. D127 is selected as the top-level node of the fourth layer. Each slave node in the fourth layer reports status frames to D127 at an hourly aggregation cycle. D127 aggregates and compresses the data of this layer, and then transmits it to D058 via two hops through the relay path D127 (fourth layer) → D093 (third layer, cross-layer signal strength -66 dBmW) → D058 (edge ​​communication node, signal strength -59 dBmW). D058 then packages and compresses the data and uploads it to the cloud server. Compared with the single-center scheme in Example 1, the reporting communication distance of the 34 slave nodes in this layer is shortened from a maximum of 400 meters to less than 50 meters per hop, and the total power consumption of near-field communication in this layer is reduced by about 60%.

[0050] Step 303: Elect the node located at the top of the stack, with optimal satellite visibility and equivalent available power greater than the preset power threshold as the positioning master node. Only the positioning master node activates the satellite positioning module to obtain the group location and distributes the location information to each area center node via near-field communication.

[0051] The positioning master node is elected from the top-level node set. Satellite visibility is determined based on the topology map (nodes on the top level with no overhead obstructions have the best visibility; if a node on the top level is partially obstructed by a taller device, its visibility is reduced accordingly). Furthermore, the equivalent available power must be greater than a preset power threshold (60% of the battery's nominal capacity) to ensure the positioning master node has sufficient power reserves to support longer rotation cycles. The positioning master node scoring formula is as follows: ; In the formula, For the first Positioning master node scoring of the platform device; For the first Satellite visibility quantification value of the device, with the top layer unobstructed. , affected by partial occlusion For non-top floors, the discount is reduced for each floor that is obscured. ; , For the weighting coefficients, take... and .

[0052] The positioning master node is the only node in the group with its satellite positioning module enabled. The calculated position represents the position of the entire stack. The position information is transmitted via near-field communication along the "positioning master node". Each planar center node This layer distributes resources layer by layer from the path of the node.

[0053] For example, among the candidate devices in the sixth layer of the top layer, D210 has no obstruction and high power. milliampere hour, D194 has no obstructions but has sufficient battery power. milliampere-hours (below the threshold) (milliampere-hours) were excluded from the selection; D206 is located directly below the gantry crane track and is partially obstructed. , D210 was selected as the master positioning node and activated its satellite positioning module to perform positioning. The measured positioning success rate was 98%, while the bottom-level stacking device D001 achieved only a 31% success rate and an average acquisition time of 48 seconds during the same period. The group position calculated by D210 (East longitude)... latitude, north latitude After being labeled with a credibility tag, the data is distributed to six planar central nodes via near-field communication. Each planar central node forwards the data within its own layer. Within approximately 20 seconds, all 210 devices in the group simultaneously store the group's location. The entire process consumes only the satellite positioning power of the D210 device.

[0054] This embodiment employs a division of labor among three types of central nodes, enabling the satellite positioning module to operate on the top layer of the stack with optimal satellite visibility, the long-distance communication module to operate on the outer facade of the stack with optimal cellular signal, and the near-field convergence task to be distributed among the layered central nodes. This increases the positioning success rate from 31% for devices inside the stack to 98%, and reduces the cellular retransmission rate from over 40% to below 5%. A single node no longer has three high-power functions stacked on top of each other, and the power consumption of each role node is balanced. The rotation cycle is extended from 7 days in Embodiment 1 to approximately 21 days, and the rotation management overhead (handover messages, topology synchronization) is reduced by about two-thirds. The intra-layer short-distance convergence of the layered central nodes reduces the average transmission power of near-field communication of slave nodes by about six decibels, and further reduces the power consumption of near-field communication of the entire group by about 50%.

[0055] Example 4 Existing low-power designs typically employ software hibernation or register-level shutdown to turn off the satellite positioning module and long-distance communication module. In software hibernation, the module's power domain remains active, resulting in continuous leakage current ranging from microamps to tens of microamps. Furthermore, even when the module power is partially shut down, the module's signal pins (universal asynchronous transceiver pins, serial peripheral interface pins) remain connected to the main control chip. When in a high-impedance or unpredictable state, this creates additional leakage paths to ground and unexpected module wake-ups due to signal crosstalk. Considering two hundred slave nodes and a service life of several years, the cumulative leakage current can reach tens of thousands of milliamp-hours per unit, becoming a hidden bottleneck for achieving ultra-long battery life. This embodiment completely cuts off the power and signal loops between the satellite positioning module and the long-distance communication module at the hardware power supply level, reducing the continuous leakage current of the slave nodes to near zero.

[0056] In this embodiment, step 103, "controlling the central node to activate the satellite positioning module and controlling the slave node to disconnect the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level," further includes: Step 401: Set up a dual-power switch circuit in each positioning device, which is connected to the satellite positioning module and the long-distance communication module respectively. The dual-power switch circuit is controlled by the switch control signal output by the main control chip. In the off state, it simultaneously cuts off the power supply circuit and blocks the signal circuit between the module signal pin and the main control chip. The dual-power switching circuit consists of two identical switching branches: the first branch is connected in series between the battery power supply and the power input terminal of the satellite positioning module, and the second branch is connected in series between the battery power supply and the power input terminal of the long-distance communication module. Each switching branch includes a P-channel MOSFET load switch (or an integrated load switch chip) and a gate drive circuit. The switching state is determined by the switch control signal output from the general-purpose input / output pins of the main control chip. Signal loop blocking is achieved through analog switches or bus switch chips: the data pins and serial peripheral interface pins of the general-purpose asynchronous transceivers of the satellite positioning module and the long-distance communication module are connected to the corresponding pins of the main control chip via a bus switch chip. The enable terminal of the bus switch chip and the load switch are driven by the same switch control signal. When the power supply loop is disconnected, the signal pins are in a high-impedance isolation state. The formulas for calculating the MOSFET on-resistance and voltage drop are as follows: ; In the formula, This refers to the drain-source on-state voltage drop of the MOSFET, measured in volts. This is the load current, measured in amperes. This is the on-resistance of the MOSFET, measured in ohms.

[0057] The turn-off leakage current in the off state is: ; In the formula, This represents the total leakage current after the node is turned off. For load switch shutdown leakage current, the typical value is less than microampere; For high-impedance leakage current of bus switches, the typical value is less than microampere.

[0058] For example, within a certain node, the satellite positioning module branch uses an on-resistance circuit. Ohm's load switch, satellite positioning module operating current milliampere, conduction voltage drop The millivolts have a negligible impact on the module's power supply; the peak transmit current of the long-distance communication module is... milliampere voltage drop The voltage is in millivolts, still within the module's allowable voltage tolerance. When the switch control signal is at the off level, both load switches are open, the bus switch is at high impedance, and the measured voltage at the power input terminal of the satellite positioning module is... The voltage at the power input terminal of the long-distance communication module is 100 kV. The impedance of the universal asynchronous transceiver pin to ground is greater than 10 volts. Ohm, measured leakage current from node turn-off is microampere.

[0059] Step 402: The main control chip of the central node outputs a first conduction control signal to control the dual-channel power switch circuit to connect the power supply circuit of the battery power supply to the power supply circuit of the satellite positioning module and the long-distance communication module, so that the central node has satellite positioning and long-distance communication capabilities. After the central node determines its role, the main control chip outputs the first conduction control signal through the control pin corresponding to that role. For the positioning master node, the branch containing the satellite positioning module is activated; for the edge communication node, the branch containing the long-distance communication module is activated; for the central node that performs both functions, both branches are activated. After activation, the main control chip first delays until the module power supply stabilizes, then releases the bus switch to connect the signal path, and subsequently sends an initialization command sequence to the module. The empirical formula for the power supply stabilization delay is: ; In the formula, The waiting time from powering on the module to receiving an acceptable instruction; The power-on time constant of the module power supply is determined by the module input capacitance and the switch conduction resistance. Battery voltage; This is the normal operating voltage threshold for the module; This refers to the module firmware self-booting time, a typical value. to millisecond.

[0060] For example, after the positioning master node D210 is selected, the main control chip outputs a high-level first conduction control signal to the load switch gate drive circuit of the satellite positioning module branch, the MOSFET is turned on, and the module power supply voltage rises to [value] within 150 milliseconds. Volts (battery voltage) Voltage drop after conduction (millivolts); after a 300-millisecond delay, the main control chip enables the bus switch, connecting the transmit and receive pins of the satellite positioning module's universal asynchronous transceiver to the main control chip, and the main control chip transmits at a baud rate of... Upon receiving the initialization command, the module responds with a ready response and then performs a warm-start positioning. The first positioning takes 4.2 seconds to output a valid positioning result.

[0061] Step 403: The master control chip of the slave node outputs a shutdown control signal to control the dual power switch circuit to physically disconnect the power supply circuit of the satellite positioning module and the long-distance communication module from the battery power supply, and simultaneously disconnect the data signal path between the satellite positioning module and the long-distance communication module and the master control chip, so that the slave node only retains the power supply of the near-field communication module, the multi-modal sensor and the low-power standby circuit of the master control chip.

[0062] After the slave node is assigned its group role, the master control chip outputs a shutdown control signal: First, the master control chip sends a software shutdown command to the module (causing the module to write necessary state data to its internal non-volatile memory; for example, the satellite positioning module writes valid ephemeris data to the battery backup domain or flash memory to preserve conditions for subsequent hot starts). After waiting for the module to confirm the shutdown, the bus switch is disconnected to make the signal path high-impedance. Finally, the two load switches are turned off to physically disconnect the power supply circuit. The shutdown sequence follows the order of "signal first, then power" to avoid damage to the module due to hot-plugging effects. The reserved power supply domain after the slave node is shut down includes only: the near-field communication module (controlled by an independent third switch, activated within the wake-up window), the multimodal sensor (intermittent sampling power supply for the barometer and inertial measurement unit), and the master control chip's low-power standby circuit (real-time clock and random access memory retention domain). The residual power consumption of the slave node is: ; In the formula, Average power consumption of slave nodes; Battery voltage; To determine the low-power standby current of the main control chip, take microampere; To sample the operating current of the sensor, take milliampere; The sensor duty cycle is calculated at one second per minute. ; For the transmit and receive current of the near-field communication module, take milliampere; For near-field communication duty cycle, wake-up time is calculated per hour. Instant retrieval ; To cut off leakage current in a dual-circuit switch, take microampere.

[0063] For example, the shutdown sequence executed by node D042 is as follows: the master control chip sends a software shutdown command to the satellite positioning module, the module replies with confirmation and saves the ephemeris data to its internal flash memory; the bus switch is set to high impedance; the load switch is turned off; subsequently, the measured overall sleep current of the slave node is 2.1 microamps. Substituting these values ​​into the power consumption formula, the average power consumption is calculated as follows: With 97 microwatts, the average daily power consumption is 1.3 mAh. The annual power consumption of the slave node is about 475 mAh, and the cumulative power consumption over eight years is about 3800 mAh, which only accounts for 45% of the nominal battery capacity, leaving ample margin for ultra-long service life. However, if a software hibernation mode is adopted (15 microamps for satellite positioning module hibernation, 10 microamps for cellular module power saving mode, and 5 microamps for signal pin leakage), the annual power consumption of the slave node due to leakage alone reaches 30×24×365 / 1000≈263 mAh, and the cumulative power consumption over eight years is 2100 mAh, accounting for 25% of the nominal capacity, and is accompanied by the risk of unexpected module wake-up.

[0064] This embodiment simultaneously cuts off the power supply circuit and the signal circuit at the hardware power supply level, reducing the continuous leakage current of the high-power module of the slave node from about 30 microamps in the software hibernation scheme to 0.12 microamps, a reduction of more than 99%; the high-impedance isolation of the signal pins eliminates the unexpected wake-up of the module and the leakage path to ground caused by level crosstalk; the software shutdown command is executed first to retain ephemeris data for the satellite positioning module, so that when the slave node takes over the central node in subsequent role rotation, it can perform a warm start positioning instead of a cold start, shortening the first positioning time and further saving positioning power.

[0065] Example 5 Near-field communication links in container stacks operate in an extremely complex metallic propagation environment: the steel container reflects, scatters, and absorbs radio waves, resulting in deep fading between nodes inside the stack, while links between adjacent nodes outside the stack may have good signal due to superimposed reflections from the metal surface. If fixed near-field communication parameters are used, two types of failures occur: with conservative parameters (low data rate, high power), the air interface transmission time on good links is prolonged, and transmission power is wasted; with aggressive parameters (high data rate, low power), the bit error rate of deep-fading links skyrockets, and retransmission storms exponentially increase power consumption and data latency. This embodiment quantifies the degree of link attenuation due to blockage based on the characteristics of the container's metallic environment and adaptively switches between two sets of communication parameters, achieving a dynamic optimization of near-field communication between connectivity and power consumption.

[0066] In this embodiment, step 104, "the central node establishes a near-field communication connection with the slave node and adaptively adjusts the near-field communication parameters for data interaction based on the characteristics of the container's metal environment," further includes: Step 501: During the process of establishing a near-field communication connection, the central node and the slave node exchange probe messages with each other, and evaluate the degree of metal blockage attenuation of the current link based on the round-trip delay of the probe messages and the received signal strength. During the connection establishment phase, the central node and slave nodes perform link probing: the initiator sends a probe message carrying the transmission time, and the receiver immediately replies with an acknowledgment message carrying the reception and transmission times. The initiator calculates the round-trip time and acknowledgment signal strength based on the acknowledgment message. The degree of attenuation due to metallic obstruction is quantified by the excess of link loss relative to free space loss. The formula for estimating total link attenuation is: ; In the formula, This represents the link path attenuation value, in decibels (dB). The signal strength corresponding to the transmission power, expressed in decibels and milliwatts; The received signal strength is measured by the receiver and is expressed in decibels and milliwatts.

[0067] The attenuation caused by metal obstruction is defined as the difference between the measured path attenuation and the free space attenuation at the same distance. ; In the formula, The degree of attenuation due to metal shielding is expressed in decibels. The transmit / receive distance is estimated based on the round-trip delay. ,in To determine the propagation speed of electromagnetic waves meters per second To detect the round-trip time of messages, the unit is seconds; The carrier frequency is measured in Hertz (Hz). It is the speed of light in a vacuum.

[0068] For example, edge communication node D058 establishes a connection with the fourth-layer slave node D131, and D058 uses... The D131 transmitted a probe message with a power of decibels and milliwatts. The response signal strength of the D131 was [value missing]. Decibels and milliwatts, path attenuation Decibels; round-trip time detection seconds, estimated distance Meters (This estimate in multipath environments includes propagation path broadening and is only used for attenuation assessment); carrier frequency gigahertz, free space decay decibels, indicating the degree of attenuation due to metal shielding. Decibels. Another link: between D058 and the underlying slave node D023, received signal strength. decibels per milliwatt, path attenuation Decibels, degree of attenuation due to occlusion The decibel level indicates that D023 is located deep within the stack and is shielded by multiple layers of steel boxes.

[0069] Step 502: When the metal blockage attenuation exceeds the first attenuation threshold, the central node adjusts the near-field communication parameters to the first communication parameter group, which includes reducing the communication rate, increasing the transmission power, shortening the single packet data length, and increasing the number of retransmissions. First attenuation threshold Based on the receiver sensitivity margin setting: when obstruction attenuates the signal strength to near the receiver sensitivity (typically...) When the threshold value (dB / mW) is reached, the link is at the edge of reliability and needs to be switched to anti-fading parameter configuration. The threshold setting formula is: ; In the formula, The first attenuation threshold is expressed in decibels. To determine the receiving sensitivity of the near-field communication module, take... decibels and milliwatts; For reliability margin, take Decibels. Transmission power. decibels milliwatt-hours decibels, i.e., path attenuation exceeding decibels (equivalent shading attenuation exceeding approximately) Switching between decibel levels.

[0070] The parameters of the first communication parameter group are: the physical layer communication rate is determined by... Megabits per second dropped Kilobits per second (improving receiver sensitivity at low rates by approximately) (decibels); transmission power is determined by The decibel and milliwatt were increased to decibels and milliwatts; single packet data length is from bytes shortened to Bytes (shorter packets have a higher single-packet survival probability in deep fading channels); the upper limit of retransmission times is increased from... The number increased to Next. The formula for estimating the survival probability of short packets is: ; In the formula, This represents the probability of a single packet being correctly received. The channel bit error rate; This is the length of a single packet, in bytes.

[0071] For example, the bit error rate of deep fading links hour, Byte long packet survival probability Average retransmission required It takes two tries to succeed; switch to After the short packet Average retransmission Success was achieved on the first attempt. After switching the link configuration of D058 to D023 with the first communication parameter group, combined with rate and power adjustments, the link completed one operation. The total power consumption time for byte status frame transmission plus retransmission power consumption is reduced by approximately 55% compared to the fixed aggressive parameter scheme.

[0072] Step 503: When the metal shielding attenuation is lower than the second attenuation threshold, the central node adjusts the near-field communication parameters to the second communication parameter group, which includes increasing the communication rate, reducing the transmission power, increasing the single packet data length, and reducing the number of retransmissions. Second attenuation threshold Below the first attenuation threshold, a hysteresis interval is set between the two to prevent parameter jitter. The second communication parameter group takes the following values: the physical layer communication rate is increased to... Megabits per second; transmit power reduced to decibels and milliwatts; single packet data length increased to bytes; the maximum number of retransmissions is reduced to Secondly, high speeds under good link conditions reduce air interface occupancy time to half of its original value, low power reduces transmit energy consumption by approximately 75%, and long packets reduce protocol overhead. The transmit energy consumption calculation formula is as follows: ; In the formula, The energy consumption per packet is expressed in joules (or milliampere-hours). Battery voltage; The operating current for transmission is measured in milliamperes (mA). This refers to the length of a single packet, in bytes. This is the physical layer rate, measured in bits per second.

[0073] For example, the degree of occlusion attenuation between D058 and its adjacent slave node D060 on the same layer is: decibels, below the second attenuation threshold Decibels, link configuration as the second communication parameter group: transmit current milliampere, voltage Fu, Bao Chang byte, rate Megabits per second, single packet transmission power consumption Joules, fold milliampere-hours; compared to the first parameter group ( kilobits per second Decibel-milliwatt emission current single-pack energy consumption (mA) At milliampere-hours, the voltage drops by approximately 93%. The link successfully transmits status frames on the first attempt, requiring no retransmission.

[0074] Step 504: The central node continuously calculates the bit error rate and retransmission rate during the data interaction process, and dynamically switches between the first communication parameter group and the second communication parameter group based on the statistical results.

[0075] The switching decision is based on two indicators: the bit error rate and the retransmission rate within the sliding statistical window. When the retransmission rate within the window exceeds the upper retransmission threshold... (Pick (or the bit error rate exceeds the upper threshold) (Pick If the current communication parameter group is not in the first communication parameter group, then switch to the first communication parameter group; when the retransmission rate within the window is lower than the retransmission lower threshold. (Pick And the bit error rate is lower than the lower bit error threshold. (Pick If the attenuation due to obstruction is below the second attenuation threshold, then the system switches to the second communication parameter group. The hysteresis interval between the upper and lower thresholds prevents repeated parameter switching when link quality fluctuates near the threshold. The sliding window retransmission rate is calculated using the following formula: ; In the formula, For the first Retransmission rate for each statistical window; This represents the total number of retransmitted data packets within the window. This represents the total number of data packets sent within the window (including retransmitted packets).

[0076] For example, links D058 to D131 initially operate according to the second communication parameter group. Within a certain statistical window (50 packets), a total of 83 packets are sent (including 33 retransmissions). The retransmission rate is... Approaching the upper limit; the retransmission rate for the next window increased due to two additional rows of containers being added to the yard, causing obstruction. Exceed D058 switches the link to the first communication parameter group, and the window retransmission rate drops back to [value missing]. After two statistical windows, the removal of yard obstructions and the degree of obstruction attenuation decreased to decibels, retransmission rate If the value falls below the lower threshold, the link switches back to the second communication parameter group. The link parameters switch a total of three times throughout the day, saving approximately 41% of near-field communication power consumption compared to the fixed first parameter group scheme, and reducing power consumption for failed retransmissions by approximately 68% and eliminating status frame loss compared to the fixed second parameter group scheme.

[0077] This embodiment employs a parameter adaptive mechanism based on the quantitative evaluation of metal obstruction attenuation, enabling reliable connectivity for deep fading links (retransmission rate reduced from over 50% to below 8%, and state frame loss rate approaching zero), while achieving extremely low power consumption for good links (power consumption per packet reduced by approximately 93%). The dual-threshold hysteresis switching strategy avoids oscillation overhead caused by parameter jitter. The total power consumption of full-stack near-field communication is reduced by approximately 55% compared to the fixed conservative parameter scheme, and by approximately 35% compared to the fixed aggressive parameter scheme while ensuring reliability. The latency and integrity of group state data are simultaneously guaranteed.

[0078] Example 6 When the central node role is held by a fixed device for an extended period, the device's power is depleted much earlier than that of other nodes in the group. Once the power is exhausted and there is no takeover mechanism, the group's positioning and data reporting capabilities are immediately lost. If a simple timed rotation (forced rotation at fixed intervals) is adopted, the rotation is too frequent and wastes power in long-term scenarios where the group is statically stacked for months. In high-power scenarios where the group is in transit, the rotation is too late, and the central node may be deeply depleted midway through the cycle. This embodiment constructs a predictive power consumption model based on historical actual power consumption statistics. The central node dynamically assesses the rotation cycle it can support based on its equivalent available power. When the power is insufficient to support the next cycle, it accurately triggers a rotation warning and completes the role handover, so that the group's power consumption is balanced and the handover overhead is minimized.

[0079] In this embodiment, step 105, "the central node dynamically evaluates the number of rotation cycles it can support based on its own equivalent available power and predicted power consumption model, and broadcasts a rotation warning to the slave nodes when it predicts that its own equivalent available power is insufficient to support the next rotation cycle, so as to dynamically transfer the role of the central node within the group," further includes: Step 601: Construct a predicted power consumption model, which is based on the actual power consumption statistics of the central node performing various tasks such as satellite positioning, long-distance communication and near-field networking within the historical period; The central node records the actual power consumption for each task using coulomb counting during each rotation cycle. The main control chip has a built-in coulomb counting sampling circuit (or calculates based on the task's power-on duration and nominal current), and accumulates the power consumption for satellite positioning, long-distance communication, near-field networking, and main control processing tasks separately according to task type, forming a historical sequence of task power consumption. The predicted power consumption model uses exponential moving average to smooth the trend of power consumption for each task. The prediction formula is as follows: ; In the formula, For the first The first cycle Predicted power consumption for this type of task, in milliampere-hours; For the first The first cycle Measured power consumption for this type of task, in milliampere-hours; For the first The first cycle Historical predicted power consumption values ​​for this type of task; For the smoothing coefficient, take... It balances the response speed to recent changes in operating conditions (such as the increase in positioning frequency due to the yard entering a transport state) with the ability to suppress random fluctuations.

[0080] The total power consumption forecast for the next replacement cycle is the sum of the forecasts for each task: ; In the formula, This is the predicted total power consumption for the next replacement cycle; The total number of task types is taken in this embodiment. These correspond to satellite positioning, long-distance communication, near-field networking, and main control processing, respectively.

[0081] For example, the power consumption statistics for the four historical cycles of the positioning master node D210 are as follows: satellite positioning mission , , , Milliampere-hours (mAh) are used for long-distance communication tasks (indirectly generated via edge communication nodes). , , , milliampere-hours, near-field networking tasks , , , Milliampere-hour, main control processing task , , , Milliampere-hours (mAh). Taking a satellite positioning mission as an example, let's assume the initial prediction value is taken as the measured value of the first cycle. Milliampere-hours, and so on: ; ; ; milliampere-hours. Similarly, recursively, we obtain... , , At milliampere time, then Milliampere-hour.

[0082] Step 602: The central node calculates the total power consumption required for the next rotation cycle based on the predicted power consumption model. When the difference between its own equivalent available power and the total power consumption is lower than the safety margin, it generates a rotation warning and broadcasts it to the slave node. The central node performs a judgment at the evaluation moment at the end of each rotation cycle: ; In the formula, The unit is milliampere-hours (mAh) to indicate the margin for rotational decision-making. The current equivalent available power of the central node; Predict total power consumption for the next replacement cycle; For safety margin, use 5% of the battery's nominal capacity. Milliampere-hours (mAh) are used to absorb prediction bias, double-running power consumption during handover, and bursty tasks. When A rotation warning is generated periodically. The rotation warning message carries: warning reason code, current rotation cycle count, remaining equivalent available power, predicted total power consumption, and a list of suggested successor candidates. The safety margin can be represented as follows: ; In the formula, To predict the amount of deviation absorption, take 10% of the predicted value; For the dual power consumption during role handover, take milliampere-hours; Reserve capacity for unforeseen tasks, take Milliampere-hour.

[0083] For example, at the evaluation time at the end of the 21st rotation cycle, the equivalent available power of the location master node D210 is... milliampere-hours, predicted total power consumption milliampere-hours, substituting into the equation, yields... When the mAh is less than zero, D210 generates a cycle warning and broadcasts an alert message to the group: {Warning reason: insufficient power, cycle count: 21, remaining power:} Milliampere-hours (mAh) of predicted power consumption: Milliampere-hours (mAh), candidate list: D203, D206, D209. If D210 adopts a timed rotation strategy, it will rotate every 21 days regardless of the remaining battery level. Unnecessary handover is triggered at milliampere-hours, and the power consumption and management message overhead during the handover period are approximately [amount missing]. Milliampere-hours are a waste; while in high-power transportation scenarios, they may run out of power midway through the cycle. The dynamic evaluation mechanism implemented in this paper avoids both of these types of losses.

[0084] Step 603: Determine the successor node based on the equivalent available power and spatial topology of each slave node in the group. After the original central node hands over its role, cut off the power supply to the satellite positioning module and long-distance communication module of this node.

[0085] The successor node is determined based on a succession suitability score, which considers both equivalent available power and spatial topology location: the successor node for the positioning master node should be located on the top layer of the stack (maintaining satellite visibility advantage) and have sufficient power; the successor node for the edge communication node should be located on the outer facade of the stack; and the successor node for the area center node should be located in the center of its layer. The succession suitability score formula is as follows: ; In the formula, For the first The succession suitability score for the platform from the node; For the first The equivalent available power of each slave node; This refers to the battery's nominal capacity. For the first The matching degree between the platform node and the target role's position conditions (top-level node takes...). Fifth layer (Lower layers decrease in number) Take the matching reference value ; and For the weighting coefficients, take... and .

[0086] After the original central node completes the role handover, it immediately cuts off the power supply to the satellite positioning module and long-distance communication module of this node, and switches to the low-power mode of the slave node. This causes the old node with the power outage to stop all high-power tasks, and the remaining power is used to maintain low-power sensing and data snapshot caching.

[0087] For example, after the rotation warning broadcast, the candidate slave node reports its power and location information: D203 (top-level, milliampere hour, )score D206 (Top Floor) milliampere hour, )score D191 (fifth floor) milliampere hour, )score D203 was selected as the successor node with the highest score. D210 sent a role handover message to D203. After D203 confirmed the handover, it activated its satellite positioning module to take over as the master positioning node. D210 then cut off power to its own satellite positioning module and long-distance communication module, using the remaining power... The mAh / m³ power is switched to slave node mode, based on the average daily power consumption of the slave node. Even with a power consumption of 1000 mAh, it can still maintain low-power sensing operation for more than three and a half years. Within the group, the role of the location master node is successively assumed by D210, D203, D206, D209, and D212 in 21 cycles. The power consumption of each node is evenly distributed, and the group's battery life is no longer limited by a single node.

[0088] The predicted power consumption model in this embodiment is based on historical measured power consumption statistics rather than theoretical nominal current. The prediction error is less than 8% according to actual measurements. The rotation triggering point is accurate, avoiding premature consumption (avoiding unnecessary handover overhead) and delayed consumption (avoiding power loss failure in the middle of the cycle). Dynamic rotation makes the power consumption of each node in the group balanced, and the group's battery life is increased from about one year in the traditional single-node solution to several years. The hardware is immediately powered off after the role is handed over, so that the remaining power of the old node that is about to run out of power is converted into low-power sensing use, maximizing the overall energy utilization of the group.

[0089] Example 7 Near-field communication (NFC) has a communication range (tens to hundreds of meters) significantly larger than the physical scale of a single stack. Adjacent stacks within a storage yard, and between the storage yard and roads, are communicatively reachable. Without physical proximity authentication, the positioning devices of adjacent stacks will receive broadcast messages from each other and may be included in the same topology, leading to: interference or even election of the central node by external stack nodes; data aggregation mixed with irrelevant stack data; and role handover messages being received by irrelevant nodes. Traditional proximity determination based on signal strength is unreliable in multipath environments with metal structures. This embodiment utilizes the mechanical coupling and transmission characteristics of the steel enclosure to external mechanical excitation, using common vibration event fingerprints to achieve stack-level physical proximity authentication, ensuring the physical purity of the group topology.

[0090] In this embodiment, after constructing a three-dimensional topology map containing the relative spatial positions of each node, the method further includes physical proximity authentication for the positioning devices participating in the group networking: Step 701: Within a preset common observation window, each positioning device records at least one event fingerprint of a common vibration event generated by the same external excitation through mechanical coupling transmission via the steel box using the inertial measurement unit built into the positioning device. The event fingerprint includes the event start time, peak intensity, duration, and spectral energy distribution. The preset public observation window is a fixed duration during the group networking phase (e.g., ten consecutive minutes after network initiation). Within this window, the storage yard typically experiences sufficient external mechanical excitation (movement of loading / unloading machinery, operation of lifting equipment, vibration from adjacent operations). Any excitation is transmitted through the ground and the container to various parts of the stack. Devices within the same stack, due to mechanical coupling, sense vibration events with highly similar waveform characteristics. The inertial measurement unit records vibration events in an event-triggered manner: when the acceleration amplitude exceeds the event trigger threshold (take...). High-speed sampling (sampling rate) is initiated when Hertz), continuous The event ends when the amplitude value falls below the silence threshold. The event fingerprint extraction formula is as follows: the event start time is the time of the first sampling point exceeding the threshold; the peak intensity is: ; In the formula, Peak acceleration of the event, in units of ; for The magnitude of the triaxial composite acceleration at time t; The duration of the event.

[0091] The duration is the difference between the end and start times of the event; the spectral energy distribution is the result of performing a Fast Fourier Transform on the event acceleration sequence, divided into four pre-divided frequency bands ( to , to , to , to Hertzian energy percentage vector: ; In the formula, This is the spectral energy distribution vector; For the first Energy percentage of each frequency band; For the first One frequency band; This represents the spectral amplitude.

[0092] For example, within the network's public observation window, a gantry crane moving and operating beside stack A generates an excitation. Device D103 within stack A records the event fingerprint: the local time within the start time window. Peak intensity Duration Second-band energy distribution Fingerprint recorded by device D047 inside stack A: start time Peak ,continued Seconds, distribution The adjacent stack B is approximately sixty meters from stack A, and its device D305 records an independent excitation from another machine: the initial time... Peak ,continued Seconds, distribution There is no correspondence between the event fingerprint of stack A and the event fingerprint of stack A in either the time domain or the frequency domain.

[0093] Step 702: Each pair of positioning devices exchanges their recorded event fingerprints through a near-field communication module, and calculates the event fingerprint matching degree for each pair. When the number of mutually matching event fingerprints reaches a preset authentication threshold, it is determined that the two positioning devices are physically adjacent and belong to the same stack. Event fingerprint matching accuracy is measured using a combination of time and frequency domain metrics. The formula for calculating the matching accuracy of a single fingerprint pair is: ; In the formula, For a single pair of event fingerprint matching scores, the value range is... ; The difference in the start times of the events recorded by the two devices, in seconds; Let be the time tolerance factor, and take . Second; , The peak intensity recorded by the two devices; For peak intensity tolerance, take ; , The duration recorded by both devices; For duration tolerance, take Second; , The vectors representing the spectral energy distribution recorded by the two devices; It is a norm; to For the weight of each item, take , , , A single-pair match exceeds the single-pair threshold (take...). This is recorded as a matching event.

[0094] For example, the fingerprint matching degree between devices D103 and D047 for the same gantry crane excitation: Seconds, first item Peak difference Second item Time difference Seconds, third item Energy distribution norm distance Item 4 ;total Record one mutual match. Stack A within the observation window experienced four identifiable common stimuli, with D103 and D047 matching four times, reaching the preset authentication threshold (taking...). (Times), it was determined that D103 and D047 are physically adjacent and belong to the same stack. Between D103 of stack A and D305 of stack B, the matching degree of the four fingerprints within the window is lower than (The starting time difference is on the order of minutes), and the number of mutual matches is... The authentication threshold was not met, and the device was determined to be not physically nearby.

[0095] Step 703: Mark the positioning devices that pass the physical proximity determination as members of the same group, and assign a group access credential to each group member. The group access credential is used for voting qualification verification during the central node election, near-field communication network access verification, and role handover message legality verification. Positioning devices that fail the physical proximity determination are refused to join the group, do not participate in the central node election, data aggregation, and role rotation of the group, and the node record of the positioning device is removed from the three-dimensional topology map.

[0096] Group access credentials are derived from the group key. Credential generation involves each authenticated device using the group identifier, the set of authenticated node device identifiers, and the network setup time as input. A key derivation function is then used to generate the group key, and finally, a member credential is generated for each member. Credential verification uses a message authentication code mechanism, with the following formula: ; In the formula, For message authentication code labels; For group keys; The payload content of the message being verified (election voting message, network access request message, or role handover message); MAC is the message authentication code calculation. The receiver recalculates the tag using the locally held group key and compares it with the tag carried in the message; if they match, the verification passes.

[0097] For example, all 210 devices in stack A pass pairwise authentication (100% authentication success rate, with 42 node pairs achieving three matches from the same three strong stimuli plus one weak stimuli). Each device receives a 16-byte group access credential. For instance, D103's credential is a valid combination of node identifier D103, group identifier GRP-0007, and derived key. During subsequent central node election voting, D203's voting message includes a message authentication code tag. The vote counting device D210 verifies the tag's consistency using the group key, confirming the vote's validity. Although device D305 in stack B is within near-field communication range and receives the election message, its response message is discarded by D210 because it fails credential verification. Simultaneously, when constructing the topology graph, D210 removes node records such as D305 and D306, which were within signal reach but failed authentication, from the topology table. The final topology table contains only 210 clean entries from stack A, and data aggregation and role rotation are performed within this clean member set.

[0098] Physical proximity authentication based on common vibration event fingerprints distinguishes between communication reachability and physical co-location. The authentication basis (similarity of vibration waveforms transmitted through mechanical coupling of the steel enclosure) cannot be forged by devices outside the communication range, and security stems from the physical channel itself. The purity of the group topology is guaranteed, and the election of the central node, data aggregation, and role rotation are free from interference from neighboring nodes, significantly improving the correctness of the election results and the accuracy of group data attribution. The authentication process reuses the existing hardware of the inertial measurement unit and near-field communication module, with additional power consumption being only a small amount of power consumed by high-speed event sampling (approximately [amount missing] per device window). (milliampere-hours), with extremely low cost and power consumption.

[0099] Example 8 Traditional simple voltage lookup table methods for assessing remaining battery capacity suffer from three systematic biases: First, lithium thionyl chloride batteries have an extremely flat discharge plateau, with the voltage change rate during the mid-discharge phase only on the order of about one-thousandth of a volt per hour, and the ohmic voltage drop caused by changes in load current is much greater than the voltage difference caused by differences in capacity. Second, the battery's discharge capacity is significantly reduced at low temperatures; at -20 degrees Celsius, the usable capacity is only about 60% of the nominal value. Third, container positioning devices operate in a static stacked state for months to years, continuously consuming power through self-discharge, and the battery exhibits voltage relaxation recovery after long-term low-current discharge. Distorted capacity assessment directly leads to errors in role election and rotation decisions. This embodiment provides an equivalent usable capacity assessment method that integrates temperature reduction, self-discharge loss, and recyclable compensation during dormancy, providing a true measure of dischargeable capacity for group decision-making.

[0100] In this embodiment, the equivalent available power is determined by evaluation in the following manner: (1) Collect the real-time voltage and real-time temperature of the batteries of each positioning device, and determine the nominal remaining power according to the pre-calibrated voltage-capacity correspondence; Each positioning device acquires the battery terminal voltage through a built-in voltage acquisition circuit (via a voltage divider network and an analog-to-digital converter), and acquires the real-time temperature through a temperature sensor mounted on the battery surface. To eliminate the contamination of voltage measurement by the ohmic voltage drop caused by the load current, voltage sampling is scheduled during the rest period (the rest window after near-field communication is off, the module is powered off, and the main control chip has completed sampling preparation), and multiple samples are taken consecutively to obtain the median value. The nominal remaining capacity is determined based on a pre-calibrated voltage-capacity correspondence. The calibration method is as follows: in the laboratory, the same batch of batteries are subjected to a calibrated load (… A full discharge test was performed using a constant current (mA), and the curves corresponding to the voltage and the cumulative discharged charge during the discharge process were recorded. These curves were then discretized into a lookup table. Linear interpolation was used for the table lookup. ; In the formula, Nominal remaining power, in milliampere-hours; Battery voltage collected during the resting window, in volts; , To find the bounding box in the table Two voltage calibration points; , This represents the remaining power corresponding to the two designated points.

[0101] For example, device D126 continuously collected battery voltages eleven times during the resting window, with the median value being... Volts, battery temperature Degrees Celsius. Lookup table Fu Ying milliampere-hours Fu Ying milliampere-hours, linear interpolation Milliampere-hour.

[0102] (2) Based on the real-time temperature, the nominal remaining power is reduced by low-temperature capacity to obtain temperature-corrected power; The discharge capacity of a battery decreases with decreasing temperature due to a decrease in electrolyte conductivity and an increase in electrochemical polarization. The temperature derating factor is obtained from the battery manufacturer's technical specifications and actual calibration, and the derating model is as follows: ; In the formula, This is the temperature reduction factor; This is the real-time battery temperature, in degrees Celsius. To reduce the initial reference temperature, take Celsius; To reduce the slope, lithium-ion batteries in Actual measurements taken in the range of minus 20 degrees Celsius Each degree Celsius decrease (i.e., a reduction of 1.2% for every degree Celsius decrease).

[0103] Temperature correction charge is: In the formula, Temperature-corrected charge, in milliampere-hours; This represents the nominal remaining battery power.

[0104] For example, the temperature of the D126 battery in device D. Celsius, below the reference temperature Celsius, reduction factor Temperature-corrected power Milliampere-hours. It is evident that without temperature reduction, the D126's capacity would be overestimated by approximately [missing value]. The milliampere-hours are overestimated by 50.5% (in low-temperature winter scenarios, this deviation is enough to cause D126 to be mistakenly selected as the successor node and fail due to power depletion midway).

[0105] (3) Calculate the self-discharge loss power based on the historical self-discharge records of each positioning device and the duration of the current static stacking state, deduct the self-discharge loss power from the temperature correction power, and add the recoverable power compensation after hibernation to obtain the equivalent usable power.

[0106] The self-discharge loss is calculated based on the device's historical self-discharge records (monthly self-discharge rate records established at the factory and continuously updated over time) and the duration of static stacking. ; In the formula, This represents the amount of electricity lost due to self-discharge, measured in milliampere-hours (mAh). This refers to the battery's nominal capacity. The monthly self-discharge rate is used for lithium-ion batteries. to This embodiment takes data based on the device's historical records. ; This represents the duration of the current static stacking state, measured in days, and is obtained by accumulating the static stacking state time from the motion state determination module.

[0107] The recoverable charge compensation after dormancy stems from the voltage relaxation characteristics of lithium-ion batteries: after prolonged low-current discharge, internal concentration polarization accumulates, resulting in a lower apparent voltage and a conservative nominal charge level obtained from a table. Upon resting, the polarization dissipates, and the discharge capacity partially recovers. The recoverable compensation is estimated as follows: ; In the formula, This is compensation for recyclable energy after hibernation, measured in milliampere-hours (mAh). This is the recyclability factor, which is related to the duration of rest. Tiantori , to Tiantori ,Exceed Tiantori ; To determine the maximum compensation, take Milliampere-hour.

[0108] The final equivalent usable power is: In the formula, This is the equivalent usable power. Temperature-corrected charge; The power loss is due to self-discharge. This is to compensate for recyclable electricity.

[0109] For example, device D126 is currently in a static stacking state and has been there for a period of time. Daily, monthly self-discharge rate ,but milliampere-hours; resting time exceeding Day, recyclability coefficient taken , milliampere-hours; equivalent usable power Milliampere-hours. Comparison of triple evaluation results: Simple table lookup method. milliampere-hours, temperature reduction only milliampere-hours, complete equivalent evaluation Milliampere-hours. Group decision-making adopts... mAh: In the succession node selection, the D126 was no longer selected as a top-level successor candidate due to insufficient battery power, while in summer scenarios (battery temperature) Celsius The equivalent charge of D126 is approximately Only batteries with a capacity of 1000 mAh or less are eligible for selection. Laboratory comparative testing (low-temperature actual discharge verification on 20 batteries from the same batch) showed that the equivalent evaluation method had a capacity estimation error of ±6%, a significant improvement over the simple lookup table method which had an error of over ±40% in low-temperature, long-term static scenarios.

[0110] In this embodiment, the equivalent available power assessment integrates four mechanisms: resting voltage lookup, low-temperature capacity reduction, self-discharge loss deduction, and hibernation recyclable compensation. This reduces the power assessment error in low-temperature long-term static scenarios from more than ±40% to ±6%, providing a reliable power benchmark for role election, rotation cycle assessment, and successor node selection. It eliminates the failure of central nodes due to power overestimation and unnecessary rotation due to power underestimation from the source, ensuring the long-term stable operation of the group.

[0111] Example 9 The mismatch between the positioning trigger cycle and the actual motion state is a direct cause of power waste or data loss in traditional solutions: In a static stacked state, the container position remains unchanged for months, making high-frequency positioning a pure waste of power; in road or rail transport states, the container position changes rapidly, and long-cycle positioning results in sparse trajectory points and loss of critical path information; in a maritime oscillating state, ships move at a constant speed along their routes, and satellite signals are partially blocked by the ship's structure, placing the positioning frequency requirement and reliability requirement between the two. Traditional single positioning cycles cannot simultaneously meet the needs of all three states. This embodiment uses a positioning master node to adaptively determine the positioning trigger cycle based on the group's motion state and performs quality screening on the raw satellite observation data, balancing low power consumption with complete and reliable positioning data.

[0112] In this embodiment, step 303 above, "the positioning master node activates the satellite positioning module to obtain the group location and distributes the location information to each area center node via near-field communication," further includes: Step 901: The positioning master node determines the positioning triggering period according to the current motion state. In the static stacking state, the first long period is used to trigger satellite positioning. In the road or rail transportation state, the second period, which is shorter than the first long period, is used to trigger satellite positioning. In the sea oscillation state, the third period, which is between the first long period and the second period, is used to trigger satellite positioning. The selection of the positioning trigger cycle is based on the principle of matching "power consumption per unit time for positioning with data freshness requirements": In a static stacked state, the position remains unchanged, requiring only low-frequency confirmation of no position drift (to prevent undetected lifting) and updating the ephemeris. The first long cycle is taken as... Hours; under the condition of continuous change of location in road or rail transport, in order to completely reconstruct the transport trajectory, the second period is taken. Minutes; Under maritime oscillations, ships move hundreds of nautical miles per day but their trajectories are smooth; the third period is taken. The relationship between power consumption of the positioning task within the cycle and the cycle length is as follows: (hours) ; In the formula, The average power consumption of the positioning task (in milliampere-hours per hour); Power consumption for a single positioning operation, typical value for warm start. An Shi Ji Typical cold start value (mAh, including module power-on stabilization and calculation) milliampere-hours; The trigger period is set in hours.

[0113] For example, if the master node D210 is in a static stacked state, then... Location tracking triggered by an hour; power consumption per warm start. Average power consumption per milliampere-hour The average daily power consumption for positioning is approximately 100 mAh per hour. Milliampere-hours. Once the motion state recognition determines that the group has transitioned to a road transport state, the D210 will switch the trigger cycle to the second cycle. Minutes, average daily power consumption for positioning is At milliampere-hours, a complete road transport route can be depicted with trajectory points spaced ten minutes apart; once the transition to a marine oscillation state is determined, the system switches to the third cycle. Hourly average daily power consumption for positioning Milliampere-hours (mAh) are used to record the ship's route at a location point every six hours, which, combined with subsequent trajectory calculations, is sufficient to reconstruct the complete route. If a traditional fixed ten-minute cycle method is used, the daily power consumption in a static stacking state is... Milliampere-hours, consumed in one year The mAh capacity accounts for 30% of the battery capacity; the state-adaptive solution implemented in this case consumes only 1% of the positioning power of the traditional solution during the 95% static stacking period.

[0114] Step 902: The positioning master node continuously collects multiple sets of raw satellite observation data at the trigger time and filters them, removing observation data with signal quality below the quality threshold before calculating the group position; In a metal stacked environment, satellite signals are diffracted and reflected by the edges of the enclosure before entering the receiving antenna. This multipath effect causes significant errors in the pseudorange observations of some channels. The positioning master node continuously acquires data at the trigger moment. Next (take) (Several) satellite raw observation data, for each observation, the carrier-to-noise ratio of each channel is recorded, and the selection rule is: the carrier-to-noise ratio of a single channel is lower than the channel quality threshold (taken as follows). Observation channels with a frequency (dB / Hertz) were directly eliminated; those that passed the channel screening were... For each observation, the mean and dispersion of the carrier-to-noise ratio (CNR) of each channel are calculated. The observation with the smallest dispersion and the highest mean CNR is retained for the solution. The lower limit for the number of channels used in the solution after screening is [value missing]. (Minimum number of observations for 3D positioning and clock error calculation). The solution uses least squares, and the weighted formula for the residuals of the observation equations in the position calculation is: ; In the formula, This is the position and clock error correction vector obtained from the solution; The observation matrix is ​​composed of the cosines of the line-of-sight directions of each channel; The weighted matrix has its diagonal elements as a function of the carrier-to-noise ratio for each channel. ; This is the pseudorange observation residual vector, where each element represents the difference between the observed pseudorange and the predicted pseudorange.

[0115] For example, the positioning master node D210 continuously collects five observations at a certain trigger time, among which the third observation receives a total of [number missing] observations. One channel: The carrier-to-noise ratio of each channel is between to decibels, Carrier-to-noise ratio of each channel and decibels (below the threshold) (removed) Carrier-to-noise ratio of each channel dB / Hertz (reserved); the average channel carrier-to-noise ratio of the five observations was highest in the third observation. (dB / Hertz) and has the smallest dispersion (standard deviation) (dB / Hertz), using the third observation. Each channel is subjected to weighted least squares solution, and the solution result is east longitude. latitude, north latitude The horizontal positioning error is determined by differential reference comparison. Meters; if the full-observation solution containing low carrier-to-noise ratio channels is used directly without screening, the error reaches [missing information]. Meters, and occasional cycle slips in low-quality channels can cause solution failures and trigger reacquisition, increasing the power consumption of a single positioning operation to [missing information]. Above milliampere-hours.

[0116] Step 903: After the location master node adds a location confidence mark and a collection time mark to the calculated group location, it distributes the data to each area center node in sequence via near-field communication. Each area center node forwards the data within its own layer, so that each slave node can synchronously store the group location.

[0117] Location confidence markers are generated based on the calculated quality indices, and the confidence quantification formula is as follows: ; In the formula, For location confidence, the range of values ​​is... ; This represents the number of channels that participate in the solution after filtering. Number of visible satellites; This is the geometric accuracy attenuation factor; As the geometric precision factor threshold, take ; The average carrier-to-noise ratio of the participating channels; For the carrier-to-noise ratio threshold, take decibels Hertz; , , As the weight, take , , Credibility is no less than Marked as highly reliable to Marked as medium confidence, below If marked as low trust, the low trust location will not be distributed to the group but will be retested by the location master node at the next trigger time.

[0118] The structure of the distributed message is {group location latitude and longitude, location confidence marker, acquisition time marker (the time of solution completion represented by the group's common time base), and location master node identifier}.

[0119] For example, the solution above Number of participating channels and visible satellites , Average carrier-to-noise ratio decibels (Hz), credibility Marked as high confidence. D210 will use {East Longitude} degrees, North latitude High degree of reliability, high reliability, and constant speed. In seconds, D210 is distributed sequentially to six planar center nodes along the near-field path (the distribution order follows the topology from top to bottom and from near to far). Each planar center node forwards the message within its own layer via broadcast. Seconds later, all 210 devices in the group write the group's location record to their local memory; each slave node determines the data freshness based on the acquisition time marker, and the cloud server downgrades the weight of low-reliability historical data based on the reliability marker. The entire distribution process consumes only near-field communication power (approximately [amount missing] on the D210 side). milliampere-hours, approximately at each planar central node milliampere-hours, each slave node approximately (milliampere-hours), without involving any power consumption for satellite positioning of slave nodes.

[0120] In this embodiment, state-adaptive positioning triggering reduces positioning power consumption during the static stacking period (approximately 95% of the service life) to about 1% of the traditional fixed-cycle scheme. The trajectory point density during transport still meets the requirements for complete characterization at ten-minute intervals. During the sea period, a six-hour interval is used in conjunction with trajectory extrapolation to restore the route. The total positioning power consumption is reduced by approximately 93% compared to the fixed ten-minute cycle scheme. Observation quality screening reduces positioning errors in metal multipath environments from... meter-level compression Meter-level accuracy and the recapture rate for failed calculations have been reduced from 12% to below 1%; the credibility markers and acquisition time markers make the quality of group location data traceable, improving the reliability of cloud trajectory reconstruction.

[0121] Example 10 If over 200 devices in a group communicate directly with the cloud server, the number of cellular connections and air interface traffic will be enormous. Simply aggregating all status data and uploading it as is results in lengthy reporting messages, high cellular traffic costs, and high power consumption. If cloud configuration commands need to be sent to each device individually, the time and power consumption of the transmission channel are also considerable. To address this technical problem, this embodiment constructs a three-tiered data interaction system from nodes to area central nodes, from area central nodes to edge communication nodes, and from edge communication nodes to the cloud server. Through deduplication, sorting, aggregation, lossless compression, and downlink layer-by-layer transmission, cellular traffic and transmission time are minimized, reducing the total power consumption and operating costs of group communication.

[0122] (1) The data interaction between the slave node and the surface center node includes: each slave node encapsulates its own status data collected by the near-field communication module into a status reporting frame according to a preset aggregation period and uploads it to the surface center node of its own area. The status reporting frame includes device identifier, relative layer number, motion status, equivalent available power and near-field link quality index.

[0123] From the node, according to the preset aggregation period (take... Every minute, the device encapsulates its own status data into a fixed-length status reporting frame. The frame structure is: device identifier ( bytes), relative number of layers ( byte), motion status code ( Bytes, static heap state Highway or rail transport , sea oscillation state ), equivalent available power ( Bytes, quantized in tens of milliampere-hours), near-field link quality metrics ( Bytes, reporting the received signal strength and repeatability of the data from the slave node to the local surface center node, and the checksum ( (bytes), total frame length Bytes. Quantized reporting of status codes enables incremental identification of bit changes in group status by the cloud.

[0124] For example, a frame reported by node D131 at the end of a certain aggregation cycle, encapsulating the following information: device identifier D131, layer number. Motion status codes (Static stacking state) Equivalent available power The milliampere-hour is quantified as Link quality indicators (received signal strength) Decibel-milliwatt quantization value, window retransmission rate Quantization value and cyclic redundancy check code, frame length Bytes. Within the common wake-up window of this layer, D131 sends the frame to the local layer's topology center node D127 using the first communication parameter group (D131 is deep in the stack). A single report consumes approximately [amount missing]. Milliampere-hours. The 35 slave nodes at the fourth layer reported a total of [number] times within one aggregation cycle. byte.

[0125] (2) The data interaction between the surface center node and the edge communication node includes: the surface center node receives the status reporting frames of each slave node in the current layer within a preset aggregation window, sorts the status reporting frames according to the relative layer number and device identifier, aggregates them into data packets of the current layer, performs lossless compression on the data packets of the current layer, and uploads them to the edge communication node together with the location information distributed by the positioning master node.

[0126] The center node of the plane is in the aggregation window (take) Within a given second (covering the off-peak reporting period for each slave node in this layer), status reporting frames are received and processed in three ways: deduplication (removing duplicate frames based on device identifier and frame sequence number, handling duplicate reception caused by near-field broadcast); sorting (arranging frames in ascending order by device identifier to maximize the correlation of field content between adjacent frames and improve compression ratio); and aggregation (concatenating the sorted frames in order, adding a layer number header and a frame count header for this layer to form the data packet for this layer). Lossless compression uses a dictionary-based compression algorithm (such as LZ4 or a simplified LZW), and the compressed size is: ; In the formula, This represents the number of bytes in the compressed data packet. This represents the number of bytes in the current layer's data packets before compression. To achieve the compression ratio, the status data shows continuous device identifiers, highly repetitive status codes (identical across all layers in the static stacking state), and gradual changes in electrical charge values. The measured compression ratio is approximately [missing information]. .

[0127] For example, the fourth-layer center node D127 receives the data of this layer within the aggregation window. Each status reporting frame is deduplicated to remove duplicates received due to broadcast errors. Frame, Remaining Frame (deduplication targets duplicate frames, actual valid nodes) One, repeating frame Each received (Frames), assembled in order of device identifiers D101 to D135, with additional... Byte header, current layer data packet Bytes; Perform LZ4 compression, compression ratio After compression Bytes. D127 will send the compressed package along with the location information distributed by the master node ( Bytes) merged into uplink data The data was uploaded to edge communication node D058 via a two-hop relay. The relay link used the second communication parameter group (inter-layer link quality was good). The upload process D127 consumed approximately [amount missing] power. Milliampere-hour.

[0128] (3) The data interaction between the edge communication node and the cloud server includes: the edge communication node merges the local data packets transmitted by each surface center node with the location information distributed by the positioning master node into a group data packet, adds a group identifier and a rotation cycle count, and uploads it to the cloud server through the long-distance communication module, and receives the downlink configuration instructions returned by the cloud server, and sends them down to each slave node layer by layer through near-field communication.

[0129] Edge communication nodes aggregate six layers of compressed data packets and location information to form group data packets: the packet header contains a group identifier ( byte), rotation cycle count ( byte), data timestamp ( bytes) and number of layers ( (Bytes), the message body consists of compressed packets at each layer and location information. The message is reported via the long-distance communication module using NB-IoT non-frequency radio. The power consumption per report and the transmission duration are as follows: ; In the formula, Power consumption per cellular report; For the module standby current, take milliampere; For registration and attachment duration, take Second; For the emission current, take milliampere; The duration of a single launch is taken as... Second; To calculate the average number of retransmissions, edge communication nodes are located in areas with good signal strength. .

[0130] Downlink configuration commands are sent from the cloud to the edge communication node, and then transmitted via near-field communication along the edge communication node. planar center node This layer sends out the message layer by layer from the node path, and uses intra-layer broadcasting to avoid unicasting to each node.

[0131] For example, edge communication node D058 aggregates data from six layers (approximately per layer). (bytes) and location information Bytes, total length of group data packets Bytes, uploaded via NB-IoT, power consumption mAh, cellular data rate Bytes; compared to the traditional solution where each device reports independently (each message contains approximately [number] bytes of location data); byte, Taiwan total byte, The total number of cellular transmissions is approximately (mAh), cellular traffic decreased by approximately 95%, and cluster cellular transmission power consumption decreased by approximately 99.94%. The cloud returned the downlink configuration command {aggregation period adjusted to...} Minutes later, the candidate sequence of the master node is updated. D058 sends this instruction down layer by layer via intra-layer broadcast. The six-layer topological center node forwards the broadcast within its own layer, and all slave nodes approximately... The system receives instructions and updates local configurations within seconds. The entire downlink process is near-field communication, consuming no cellular resources of any slave node. The rotation cycle count field allows the cloud to fully track the group's 21-round role rotation history, enabling remote auditing of the group's power consumption balance.

[0132] The three-level data interaction system in this embodiment reduces the number of cellular transmissions per cycle from 210 to 1, and the cellular traffic from approximately [missing information] per cycle. Bytes compressed to approximately Byte (down about 95%), group cellular transmission power consumption down from about Compressed to approximately 1000 mAh The power consumption of near-field aggregated data decreased by approximately 62% due to deduplication, sorting, aggregation, and lossless compression, resulting in a corresponding decrease in relay power consumption. Downlink commands are broadcast layer by layer in the near field, and configuration changes take effect across the entire network in approximately 25 seconds with zero cellular overhead. Group identifiers and rotation cycle counts provide complete data support for tracking group status and rotation history in the cloud.

[0133] Example 11 The role rotation process itself carries reliability risks: if the handover fails midway, the handover message may be damaged or lost in the metal multipath channel, or the succeeding node may be unable to respond due to its own malfunction, leaving the group in a masterless state: the original central node has already entered the degrade preparation process, the succeeding node has not yet completed its functional takeover, and the group's satellite positioning and data reporting are completely interrupted during this period. Furthermore, if the succeeding node does not possess the complete three-dimensional spatial topology, device identifiers at each layer, and near-field communication parameters, it must re-execute topology discovery and link-by-link parameter negotiation, which takes several minutes and consumes considerable power. To address this technical problem, this embodiment provides a complete role handover process and a failure retry mechanism to ensure the atomicity, seamlessness, and high success rate of the central node's role transfer.

[0134] In this embodiment, the dynamic transfer of the central node role within the group includes: Step 1101: The original central node sends a role handover message to the slave node that has been determined as the successor node. The role handover message includes the three-dimensional spatial topology, the device identifiers of each node in this layer, near-field communication parameters, and the rotation cycle count. The role handover message consists of the following fields: message type and version field, rotation cycle count field (recording the current rotation sequence number for cloud tracking and successor node continuation calculation), compressed 3D spatial topology table (each entry includes device identifier, layer number, link signal strength, and equivalent available power; 210 entries compressed to approximately 1.2 kilobytes), near-field communication parameter table (recording the communication parameter group number currently used by each active link in the group, including link-by-link assignment of the first and second communication parameter groups), original central node identifier and new role effective sequence number field, cyclic redundancy check field, and message authentication code tag field based on the group key. The receiving end performs dual verification of integrity and legality for the handover message. ; In the formula, The value is a verification result flag; 1 indicates that the message is complete and valid, and 0 indicates that the verification failed. The cyclic redundancy check value carried in the message; This is the checksum value recalculated by the receiving end for the message payload; The message authentication code label carried in the message; This is a message authentication code tag recalculated by the receiving end using the group key on the message payload. Only after successful verification does the confirmation process begin, preventing message transmission loss in metal multipath channels and preventing neighboring nodes within the communication range but without physical proximity authentication from forging handover messages. Due to near-field single-packet length limitations, long messages are transmitted in fragments. ; In the formula, This refers to the number of fragments. This represents the total number of bytes in the compressed message. This represents the upper limit of the single packet data length for the current link communication parameter group; The fragment header overhead for each fragment (including the fragment sequence number and total fragment number field, taken as six bytes); This is a rounding operation. The loss of any fragment triggers a retransmission of that fragment, with a maximum of three retransmissions.

[0135] For example, at the end of the 21st rotation cycle, the original positioning master node D210 (equivalent available power 1720 mAh, rotation triggered according to the evaluation mechanism of Example 6) sends a role handover message to the successor node D203. The message is compressed to 1240 bytes. D210 and D203 are on the top layer of the stack, the link is blocked and attenuated by 3dB, and the second communication parameter group (255 bytes per packet) is used. Five chips; the transmission of five chips, along with the acknowledgment and retransmission mechanism, is completed within approximately 45 milliseconds, with verification at the D203 end. The message is complete and legitimate.

[0136] Step 1102: After verifying the integrity of the role handover message, the succeeding node replies with a confirmation message and re-establishes near-field communication connections with each slave node according to the received near-field communication parameters; The confirmation message includes the successor node's device identifier, the receive confirmation bitmap for all segments, the successor node's current equivalent available power and readiness flag, and an appended message authentication code tag based on group access credentials. The successor node then broadcasts a new central node beacon frame to the group according to the near-field communication parameter table carried in the handover message: the beacon frame announces the new central node identifier, role activation sequence number, and the assignment of communication parameter groups for each link; upon receiving the beacon frame, each slave node directly updates its local link configuration according to the parameter table, without needing to re-execute probe message exchange and parameter negotiation for each link (in the initial network setup, single-link probe negotiation takes approximately 200 milliseconds, and the total overhead and power consumption of renegotiation for over 200 links are considerable). The composition of the network-wide connection reconstruction latency: ; In the formula, The delay is due to the completion of the network-wide connection reconstruction. The transmission delay of the new central node beacon frame; The number of stacking levels; For single-layer signaling forwarding delay; This refers to the aggregation delay of the last-layer slave node response.

[0137] For example, D203 completes verification and replies with an acknowledgment message (including a 5-fragment acknowledgment bitmap and a ready flag) within 0.6 seconds after initiating the handover; then it broadcasts the new master node beacon frame, which is forwarded at this layer by the six-layer topological central node. Each slave node directly updates its configuration according to the parameter table in the beacon frame (where D131 maintains the first communication parameter group on the deep fading link and D060 maintains the second communication parameter group on the good link); Second, , Second, The reconnection of all 209 slave nodes in the network was completed in about 9 seconds, which is about 79% shorter than re-performing the full-link probe negotiation (estimated to be about 42 seconds). The power consumption of probe message transmission and reception without the negotiation process is about 2.3 mAh.

[0138] Step 1103: After receiving the confirmation message, the original central node broadcasts a role change notification, downgrades itself to a slave node, cuts off the power supply to the satellite positioning module and long-distance communication module of this node from the hardware power supply level, and enters low-power standby. The role change notification is broadcast across the entire network, containing the original central node identifier, the new central node identifier, the role activation sequence number, and the activation time. All devices within the group (including edge communication nodes and each area central node) update the role field in their local topology table and the destination address for subsequent status data interactions accordingly, ensuring strict synchronization between the data stream destination address switch and the new role activation. The degradation sequence of the original central node follows the order specified in Example 4: first, a software shutdown command is sent to the satellite positioning module to write the valid ephemeris into its internal non-volatile memory (to reserve conditions for a warm-start when the node takes over again in the future); then, the bus switch is disconnected to isolate the signal pins with high impedance; finally, the dual-path load switch is turned off to physically disconnect the power supply circuit, subsequently entering low-power standby (standby current 2.1 microamps). The duration of residual power support after degradation is as follows: ; In the formula, The duration of slave node operation that can be supported by the residual equivalent power of the degraded node; The remaining equivalent available power of the original central node after the role handover is completed; The average current in slave node mode is approximately 0.054 mA (calculated based on a daily power consumption of 1.3 mA / h). Overlapping overhead due to the coexistence of the old and new central nodes during the handover period: / 3600; In the formula, This refers to the additional power consumption generated by the original central node continuing to operate in central mode during the handover period; The average operating current of the node during the handover period (including near-field transmission and reception and main control processing, taken as 15 mA). This is the total time from the start of the handover to the completion of the power outage at the original central node.

[0139] For example, upon receiving the confirmation message, D210 immediately broadcasts a role change notification {Original Center: D210, New Center: D203, Effective Sequence Number: 22}. Each device in the group updates its role fields within 3 seconds, and the destination address for subsequent status data switches from D210 to D203. D210 executes a degradation sequence: the satellite positioning module software shuts down and saves the ephemeris for approximately 2 seconds, signal isolation and power cut-off for approximately 0.5 seconds, and then enters a 2.1 microamp low-power standby. Total handover time. Approximately 15 seconds (message transmission 0.05 seconds, acknowledgment processing 0.6 seconds, beacon broadcast and network-wide reconstruction 9 seconds, role change notification 3 seconds, degradation sequence 2.5 seconds), overlapping overhead. The power consumption per milliampere-hour (mAh) is negligible compared to the power consumption per single switching cycle (approximately 1371 mAh). The remaining equivalent usable power of the D210 is approximately 1719 mAh, which, theoretically, can support [the following] based on slave node mode. Hour The actual lifespan of 3.6 years is constrained by battery self-discharge and storage life, which is sufficient to support the battery until it is listed as a successor candidate again or until the end of its service life.

[0140] Step 1104: If the original central node does not receive the confirmation message within the preset handover time limit, it will reselect the slave node with the second best equivalent available power as the successor node and re-initiate the central node role handover.

[0141] Preset handover time limit The entire process of fragmented transmission, retransmission, and acknowledgment return is covered, with a 30-second timeframe (covering 5 fragments multiplied by 3 retransmissions plus processing delay). If a complete acknowledgment message is not received within this timeframe (missing acknowledgment, verification failure, or no response from the successor node), the handover is considered a failure. The retry mechanism is as follows: the original central node pre-sorts candidate slave nodes based on the succession suitability score in Example 6. After a failure, it initiates handover to the next best candidate in sequence, with the number of retries limited to the length of the candidate sequence. If all candidates are exhausted and the handover still fails, the original central node broadcasts an emergency status message, and the group enters the fault takeover process. The cumulative handover success probability considering retries is: ; In the formula, This represents the cumulative probability of a successful handover. This represents the number of available candidate nodes. To the first The single success probability of a candidate initiating a handover (including link errors and candidate node failures, the measured average is approximately 0.97).

[0142] For example, in a simulated handover exercise, the primary successor node D203 experienced a main control circuit failure due to water damage from the previous day's heavy rain. D210 did not receive an acknowledgment message within the 30-second time limit. Based on the candidate sequence (D203 score 0.918, D206 score 0.841, D209 score 0.826), D210 switched to the secondary node D206 (equivalent available power 5800 mAh) and re-initiated the handover. The second node's segmented transmission and acknowledgment were normal, and the handover was completed within 28 seconds. Based on a single success probability of 0.97 and 3 candidate nodes, the cumulative success probability... That is, the probability of failure is only about 0.000027%.

[0143] The handover process in this embodiment ensures the atomicity of role transfer: the successor node is confirmed to be ready first, and the original central node is downgraded and powered off afterward. There is a valid central node at any time in the entire network, completely eliminating the masterless window. The handover message fully carries the three-dimensional spatial topology relationship, device identifiers at each layer, near-field communication parameters, and rotation cycle count. The successor node is exempt from re-performing topology discovery and per-link parameter negotiation. The network reconstruction latency is reduced from about 42 seconds to about 9 seconds, and negotiation power consumption is reduced by about 80%. The failure retry mechanism increases the cumulative success probability of handover from 97% for a single attempt to over 99.99%. After the original central node is downgraded, it is immediately powered off from the hardware power supply level and retains ephemeris data. The remaining power is transferred to low-power sensing purposes and reserves the warm-start condition for the node to take over again in the future.

[0144] Example 12 In areas without cellular network coverage, such as existing ocean voyages and mountain railways, traditional positioning devices suffer from two major power consumption drawbacks: First, the device cannot accurately determine its location without coverage, repeatedly attempting cellular network registration and searching. Each failed attempt takes an average of 60 seconds and draws an average current of 80 mA. Assuming an attempt every 30 minutes, this amounts to approximately 896 mAh over a 14-day voyage, representing 10.5% of the battery's nominal capacity. Second, maintaining high-frequency satellite positioning to ensure the continuity of the transport trajectory can lead to sparse trajectory points after power-saving frequency reduction, making it impossible to reconstruct the route. To address these technical issues, this embodiment keeps the long-distance communication module powered off in areas without coverage, uses low-frequency satellite positioning to obtain discrete anchor points, and uses inertial measurement unit trajectory calculation to fill in the trajectories between anchor points and stitch them together to form a continuous transport trajectory, thus simultaneously meeting the requirements of low power consumption and trajectory integrity.

[0145] In this embodiment, the control method for the ultra-low power container positioning device further includes: the positioning master node performing low-power trajectory calculation, as detailed below: (1) When the edge communication node determines that it is in an area without cellular network coverage, keep the long-distance communication module powered off and only turn on the satellite positioning module of the positioning master node at low frequency to obtain discrete location points; The coverage failure determination adopts a periodic probing and consecutive failure counting mechanism: edge communication nodes briefly activate the long-distance communication module to perform registration attempts according to the probing period (a simplified attachment attempt with early termination is used when the single probing time is limited to 40 seconds, the average current is 60 mA, and the power consumption is about 0.67 mA). If the number of consecutive failures reaches the determination threshold, it is determined that there is no coverage; in order to prevent false judgments caused by registration jitter in weak signal edge areas, the threshold is selected based on the false judgment probability constraint: ; In the formula, The probability of being in a coverage area but being mistakenly identified as having no coverage; The probability of a single detection registration failure within the coverage area (approximately 0.2 in the measured weak signal edge area). This is the threshold for determining consecutive failures. hour The false positive rate is only 0.16%. After determining that there is no coverage, the long-distance communication module remains powered off, and the detection cycle is extended from 30 minutes to 4 hours. The positioning master node activates the satellite positioning module to obtain discrete location points according to the low-frequency cycle (6 hours for the marine oscillation state, consistent with the third cycle of Example 9), which are used as anchor points for track estimation.

[0146] For example, a container ship sails from Shanghai Port to Los Angeles Port, a journey of approximately fourteen days. Edge communication node D058 performs cellular registration probes every 30 minutes after departure. The first four probes (within two hours of departure) fail consecutively (the ship has already left the shore-based coverage area). Based on the false alarm probability constraint, it determines that the ship is in an area without cellular network coverage, immediately shuts down the long-distance communication module, and extends the probe cycle to 4 hours. Subsequently, positioning master node D210 performs a hot-start satellite positioning every 6 hours (e.g., using ephemeris and rough position distributed within the group), obtaining a total of 56 discrete location anchors throughout the process. The total power consumption of satellite positioning is [not specified]. mAh. Compared to traditional standalone locators that repeatedly attempt cellular registration (every 30 minutes, 60 seconds per attempt, average 80 mA), cellular search power consumption over fourteen days is approximately 896 mA; the cellular power consumption in this embodiment is only the detection overhead after the relaxed period. At milliampere-hours, the decrease is approximately 94%.

[0147] (2) Between adjacent discrete locations, the positioning master node calculates the trajectory based on the attitude and acceleration data collected by the inertial measurement unit; The trajectory estimation employs a simplified strapdown mechanism combining attitude quaternion updates, force coordinate transformation, and velocity increment integration. Furthermore, quasi-uniform velocity constraints and online estimation of equivalent velocity errors within segments are introduced to address the low-dynamic characteristics of ship motion. Attitude quaternion updates: ; In the formula, for The attitude quaternion at time (describing the orientation of the vehicle coordinate system relative to the navigation coordinate system); This refers to quaternion multiplication operations; The pose increment quaternion for a single update step; , , The value is the angular velocity measurement of the three axes of the gyroscope, in radians per second. The attitude update step size is set to 0.1 seconds. The velocity increment is obtained by integrating the force of the navigation frame after subtracting gravity: ; In the formula, Since the last anchor point At that time The velocity increment vector, in meters per second; The direction cosine matrix is ​​calculated from the attitude quaternion (to realize the transformation from the vehicle coordinate system to the navigation coordinate system); This is the triaxial specific force measurement vector of the accelerometer, with units of meters per second squared. Let be the gravity vector of the navigation system, and take... The speed limit is meters per second squared. The quasi-uniform speed constraint is as follows: the ship maintains a quasi-uniform straight-line speed for most of the time within the anchor point interval. If the residual amplitude after subtracting gravity from the specific force is below the significance threshold (0.002 meters per second squared), it is considered noise and set to zero. The speed increment is only integrated when a significant residual (real speed change or steering maneuver) lasting longer than five seconds is detected. This constraint prevents accelerometer zero bias below the threshold from being accumulated during integration. The equivalent speed error within each segment (including the equivalent effects of maneuver capture deviation and sub-threshold residual zero bias) is estimated online at each anchor point based on the closure error. ; In the formula, For the first The equivalent velocity error estimation vector after segment correction, in meters per second; This is the estimated value from the previous section; For the first Satellite positioning vector of each anchor point; To calculate the position vector at the time of arrival at the same anchor point for the flight track; The anchor point interval is set to 21600 seconds. The corrected estimated position for the next segment is adjusted by subtracting the equivalent velocity error. The growth model of the estimation error within the anchor point interval is as follows: ; In the formula, This serves as the upper bound for calculating the positional error within the anchor point interval; The equivalent velocity error growth rate after convergence of quasi-uniform velocity constraint and online error estimation is expressed in meters per second. This represents the anchor point interval duration.

[0148] For example, the inertial measurement unit of D210 (the positioning master node, installed on the top layer of the stacked container, moving with the ship as a whole) collects gyroscope and acceleration data at 10 Hz; within the first anchor point interval, due to the lack of correction for equivalent velocity error, the closure error of the track to the second anchor point is 2.4 kilometers (equivalent to equivalent velocity error). The error in meters per second (originating from measurement deviations of two maneuvers not precisely captured) was corrected according to the online estimation formula. The second segment's closure error decreased to 710 meters, and subsequent segments stabilized between 450 and 700 meters. The equivalent velocity error converged to the order of 0.03 meters per second. Substituting this into the error growth model yielded... The measurement accuracy is consistent with the deviation magnitude within the measured range. The low duty cycle sampling configuration of the inertial measurement unit (gyroscope sampling for 1 second every 10 seconds, accelerometer sampling at 10 Hz) has an average current of approximately 100 microamps, and consumes approximately 34 milliamp-hours of power over a 14-day flight.

[0149] (3) The splicing forms a continuous transportation trajectory.

[0150] The closure error between adjacent anchor points (the difference between the anchor point's location and the calculated arrival location) is eliminated using a piecewise linear distribution method, forming a continuous trajectory: ; In the formula, For splicing The trajectory position vector at any given moment; Obtained from the trajectory calculation Time-position vector; , The time for two adjacent anchor points; For the first The satellite positioning vector of each anchor point. This formula linearly distributes the closure error over time to each point within the segment, ensuring that the trajectory at the anchor point strictly matches the satellite positioning result and that the transition within the segment is smooth. The maximum positional deviation within the segment after splicing is approximately half of the closure error.

[0151] For example, the 14-day voyage is divided into 55 estimated segments by 56 anchor points, and each segment is spliced ​​together. Taking the segment from anchor point 9 to 10 as an example, the closure difference is 680 meters, and the maximum deviation within the spliced ​​segment is 340 meters. After all 55 segments are spliced ​​together, the continuous trajectory is compared with the shipping company's automatic identification system route. The average lateral deviation is about 520 meters, and the maximum deviation is 1.71 kilometers, which meets the requirements for container route monitoring and arrival time estimation (the total voyage is about 18,500 kilometers, and the deviation within the segment is negligible relative to the voyage). The location data and the status data of each time slice during the period of no coverage are cached together in the local snapshot (during the loading period of the ocean-going scenario, the snapshot retention period is pre-configured to 20 days via downlink configuration command), and will be uploaded after the network is restored. Total power consumption: satellite positioning 2.7 mAh, inertial measurement and calculation 34 mAh, cellular detection 56 mAh, totaling approximately 93 mAh; compared with the traditional solution (cellular search 896 mAh plus satellite positioning with a 10-minute cycle of approximately 99 mAh, totaling approximately 995 mAh), the power consumption is reduced by approximately 91%.

[0152] In this embodiment, the power consumption for cellular search in uncovered areas is reduced from approximately 896 mAh to approximately 56 mAh for detection, a decrease of approximately 94%. Quasi-uniform velocity constraints and online estimation of equivalent velocity errors within segments reduce the calculated closure error between anchor points from 2.4 km in the first segment to within 700 meters. After splicing, the deviation within the segment does not exceed 350 meters. Continuous trajectory reconstruction is achieved at a six-hour anchor point interval, with an average deviation of approximately 520 meters between the trajectory and the automatic identification system's route. The total power consumption is reduced by approximately 91% compared to traditional solutions, and the trajectory continuity is not lost due to the reduced positioning frequency. The use of snapshot caching and compensation upload mechanisms ensures zero data loss during periods of no coverage.

[0153] Example 13 The group data link suffers from a structural single point of vulnerability: in the unidirectional uplink where the status data of the slave nodes is aggregated by the area-based central node and uploaded by the edge communication nodes, failure of any link (central node failure, near-field link interruption, or continuous periods without cellular coverage) will result in the permanent loss of data for the corresponding period in the cloud. Traditional solutions use a report-and-discard approach for slave nodes, leaving no local copies and no means of recovery after failure. This embodiment, with extremely low storage and power consumption, retains data snapshots in the local flash memory of the slave nodes on a time-slice basis, providing a data foundation for fault takeover and data compensation, while ensuring that flash memory writing does not damage the ultra-low power consumption characteristics of the slave nodes and that the wear and tear life far exceeds the service life of the device.

[0154] In this embodiment, the control method for the ultra-low power container positioning device further includes: distributed data caching from the nodes, as detailed below: (1) While each slave node reports data to its respective area center node, it also saves the reported data as a data snapshot in the local flash memory according to time slices. The data snapshots save the data reported in the most recent N time slices. The time slices are aligned with the preset aggregation period (one time slice per hour). The time slice sequence number is uniformly numbered according to the group's common time base, ensuring that time slices across nodes can be aligned and merged during fault recovery. Snapshot content includes: a copy of the status reporting frame within the time slice (device identifier, relative layer number, motion status, equivalent available power, near-field link quality indicators), event records (motion status transition time, common vibration event fingerprint summary, power jump record, timestamp and magnitude of motion segment events in transport state), and a snapshot header (time slice sequence number, number of records) and a cyclic redundancy check field. The entire snapshot is appended with a message authentication code tag based on the group key, allowing the new central node to verify the authenticity and ownership of the snapshot during recovery, preventing mistransmitted data from neighboring stacks or links from being mixed in. After the snapshot is written, a readback verification is performed. Physical pages that fail the readback verification are marked as bad blocks and replaced with spare blocks. Total storage space in the snapshot area: ; In the formula, This represents the total size of the snapshot area, in bytes. The number of time slices to retain; This represents the number of bytes in a single time-slice snapshot. The loop-overwrite write address is: ; In the formula, For the first The starting address for writing a time-slice snapshot; This is the base address of the snapshot area; For modulo operation; the first The time slice will automatically overwrite the first one. The storage location of each time slice enables rolling retention of the most recent one. A time slice. The value can be adjusted via downlink configuration commands to adapt to the recovery window requirements of different scenarios. Flash memory wear and tear calculation: ; In the formula, This refers to the number of days the flash memory can last before it wears out. This represents the maximum number of erase / write cycles per sector (typical value of 100,000 cycles according to flash memory datasheets). The number of bytes in the sector (4096 bytes). This represents the average number of bytes written per day. Snapshot writes are scheduled to be completed within the same common wake-up window as reporting (written immediately after reporting), incurring no additional wake-up overhead; power consumption per snapshot write: ; In the formula, Power consumption for a single snapshot write; Battery voltage; The flash memory write current is set to 3 mA. This is the time taken for a single write operation (compensated for by page programming rate).

[0155] For example, from the default configuration of node D131 (Retaining the most recent 240 hours, or ten days) Bytes (status frame 10 bytes, event log 30 bytes, header and checksum 8 bytes); Total snapshot area This represents 2.3% of the 4 megabits (512 kilobytes) of serial flash memory. The address to be written for the 731st time slice (corresponding to 07:00 on day 31) is... Automatically overwrites old data from the 491st time slice. Wear accounting: 4096 bytes per sector accommodates 85 snapshots ( (bytes), meaning a sector erase / write cycle is completed every 85 hours. 10,000 hours, approximately 971 years, far exceeding the device's service life. Write time and power consumption: 48-byte page programming and erase / write balancing overhead is approximately 8 milliseconds. At 1.152 kilobytes per day, the power consumption is less than 0.001 mAh, which is less than one-thousandth of the daily power consumption of the slave node (1.3 mAh). Scenario Verification 1: During a six-hour period of railway transportation through mountainous tunnels with no cellular coverage, the six time-slice data reported by each slave node could not be uploaded to the cloud, but all were retained in local flash memory as snapshots. After the network was restored, the data was returned via the recovery mechanism, resulting in zero data loss for the six hours. Scenario Verification 2: In a known ocean shipping scenario, during loading at port, the cloud-based downlink configuration command will... Adjusted to 480 (20-day recovery window, covering 14-day flight time plus margin), snapshot area usage increased to approximately 23 kilobytes (4.5% of flash memory), wear rate remains unchanged (write frequency and...). (Irrelevant), data compensation during network recovery after the flight ends fully covers the entire journey.

[0156] In this embodiment, the distributed data cache provides local redundancy to the group data link: if any uplink link fails within the default 10-day (20-day in the ocean scenario) time window, it can be recovered from the snapshots of each slave node, reducing the risk of data loss from a structural single point of failure to failure only outside the snapshot window; the storage overhead is only 2.3% of the flash memory capacity (4.5% in the ocean configuration), the daily power consumption for writing is less than 0.001 mAh, and the wear life is about 970 years, which does not constitute any power consumption or lifespan burden; the snapshot is attached with group credential authentication and a common time base time slice number, providing authenticity and alignment guarantee for cross-node fusion splicing; the write reuses the existing wake-up window, without generating additional wake-up overhead, and the ultra-low power consumption characteristics of the slave node are not affected.

[0157] Example 14 A single point of failure at the central node will cause the group to lose its location and time reference. If the failure coincides with a period of no cellular coverage (such as the failure of the main node during railway transportation in mountainous areas), both location data and status data will be lost, which traditional solutions cannot recover. This embodiment provides a complete fault takeover and data compensation closed loop, from failure detection, group reselection, snapshot recovery, trajectory refitting to a one-time upload after network recovery, eliminating the permanent data loss caused by the superposition of single point failure at the central node and coverage blind spots.

[0158] like Figure 3 As shown, in this embodiment, the control method for the ultra-low power container positioning device further includes: the central node performing fault takeover and data compensation, as detailed below: (1) When a slave node fails to receive a handshake signal from the central node for a preset number of consecutive times, it is determined that the central node has failed, and it leaves the current group and re-initiates the group networking election or temporarily switches to the independent working mode. The master node broadcasts a handshake frame (including a rotation cycle count and the central node identifier) ​​within each common wake-up window. This handshake frame is forwarded by the central node at this layer, covering all slave nodes. Each slave node performs a continuous loss count of the handshake frames; if the continuous loss count reaches a threshold, the central node is considered to have failed. Failure detection latency limit: ; In the formula, This is the upper limit for failure detection delay; The threshold for determining continuous data loss is set to 3. The common wake-up window period is 1 hour. The probability of false positives due to link packet loss: ; In the formula, This is the probability that a normal central node will be mistakenly identified as a failure due to packet loss in the link. The single-window handshake frame loss rate is approximately 0.1 in actual measurements for deep fading links, which is reduced to below 0.02 through window alignment mechanisms. Upon failure determination, the handling branch is selected based on near-field topology integrity: if the group topology remains intact (all area center nodes are in place and responding normally), a new group network election is initiated; if the near-field link is also damaged (e.g., a partial collapse of the stack causing intra-layer link breakage), the affected node temporarily switches to independent working mode, autonomously activating its satellite positioning module and long-distance communication module for low-frequency positioning and reporting in independent mode until it is re-accepted by the group.

[0159] For example, in another railway transportation scenario, the positioning master node D210 fails completely at 02:00 AM due to a short circuit caused by condensation in the main control circuit. The handshake frame is broadcast normally in the 01:00 window, but is missing in the 02:00, 03:00, and 04:00 windows. Each slave node completes three consecutive loss counts in the 04:00 window, determining that D210 has failed and the detection delay limit is reached. The actual detection delay is between 2 and 3 hours; according to Calculate the probability of misjudgment. The risk of misjudgment is negligible. At the time of failure determination, the group's near-field topology is complete (all six layered central nodes respond normally), and the slave nodes re-initiate the group network election branch.

[0160] (2) After the newly elected central node takes office, it broadcasts a data recovery request to each slave node through near-field communication; The re-election process reuses the scoring and election mechanism of Embodiments 1 and 3. Failure judgment triggers an emergency extension window (extended to 120 seconds) to complete the scoring exchange and election. The first action of the new central node after taking office is to broadcast a data recovery request frame. The request frame content includes the recovery start time (determined by the most recent successful cloud upload time provided by the edge communication node), the recovery end time (current time), the time slice length parameter, and the group credential authentication tag. The request frame is forwarded by the planar central node at this layer, covering all slave nodes.

[0161] For example, the failure determination within the 04:00 window triggers an emergency extension window, and the candidate nodes exchange scores. D203 (top-level, equivalent available power 7100 mAh, score 0.918) is elected as the new positioning master node at 04:00:40. After being elected, D203 immediately starts its own node's satellite positioning module to perform a hot start positioning (using existing positioning auxiliary data in the group, the first positioning takes 4.1 seconds). Then it broadcasts a data recovery request frame {recovery interval: 22:00 to 04:10 of the previous day, time slice length: 60 minutes}. The most recent successful upload time reported by the edge communication node D058 is 22:00 of the previous day (after which the train entered the mountainous area with no coverage, and the cell lost power). Therefore, the snapshot recovery interval is a total of seven whole time slices. The request frame is forwarded through six layers, covering all slave nodes in the group in about 25 seconds.

[0162] (3) Each slave node sends a snapshot of the data in its local flash memory to the new central node. The new central node merges and splices the data for the missing time period, and combines its own satellite positioning data and inertial measurement unit data to refit the transportation trajectory. Snapshot backhaul is performed in batches and at off-peak times according to device identifiers (approximately 70 devices are backhauled per common wake-up window to avoid near-field channel collisions). Snapshots are aggregated and compressed by the area-based central node and then relayed to the new central node via multiple hops. Fusion and splicing processing includes: deduplication using (device identifier, time slice number) tuples, sorting by time slice number in ascending order, and cross-validation of data within the same layer and slice (the layer number and motion status field of multiple devices on the same layer are cross-verified, and abnormal records that fail to read the verification are removed). Measurement of data recovery rate: ; In the formula, For data recovery rate; The actual number of device time-slice snapshots recovered; To recover the total number of device time slices that should exist within the interval (the product of the number of devices and the number of time slices), trajectory refitting uses two types of anchor points as boundaries: the pre-failure anchor point (the group position last distributed before the original central node failed) and the initial positioning anchor point after the new central node is selected; the trajectory between anchor points is estimated from the motion segment events (timestamps and magnitudes of significant acceleration and turning events) continuously recorded by the inertial measurement unit of the new central node in transport mode, and a segmented splicing formula is used to eliminate closure errors. For position points that simultaneously have satellite positioning references and estimated references, inverse variance weighted fusion is performed: ; In the formula, The merged position vector; The standard deviation of the satellite positioning anchor point (take five meters); The standard deviation for estimating the location (the failed blind estimation segment is taken as being on the order of 300 meters). and These are the satellite positioning reference position vector and the calculated reference position vector, respectively.

[0163] For example, 206 out of 209 slave nodes returned complete snapshots, 3 nodes failed read verification due to individual physical pages in the flash memory being at the edge of the data retention period (each node missing approximately 4 time slices), and 1 node (D209 in the same row as D210, affected by the same moisture erosion event) did not respond; the recovery interval should exist. Each device time slice, actual recovery Data recovery rate during snapshot phase %. Trajectory refitting: Using the 01:00 anchor point (the last group position distributed before D210 failed, near 118.32 degrees east longitude and 30.45 degrees north latitude) and the 04:05 anchor point (the initial positioning result of D203) as boundaries, the 3-hour blind-pushing trajectory was calculated based on the motion segment events (02:40 train deceleration and cornering event, 03:50 train acceleration event) archived by D203. This was compared with the actual train route geographical data, with a maximum lateral deviation of 1.9 kilometers. The group position and transportation trajectory were 100% complete within the recovery interval. After D203 was selected, the 10-minute positioning cycle for highway or railway transportation was restored according to Example 9. Between 04:05 and 08:00, 23 new positioning anchor points were added, and the trajectory continuity was fully restored.

[0164] (4) When the cellular network signal is detected to be restored, the new central node will package the spliced ​​and completed group data and upload it to the cloud server at once.

[0165] Network recovery detection is performed by the edge communication node according to a relaxed cycle (4 hours, following the detection mechanism of Example 12). Upon successful registration, the new central node is notified via near-field communication. The new central node merges the snapshot recovery data, the data obtained through normal aggregation after the recovery interval, the refit trajectory, and the fault event report (original central node identifier, failure determination time, detection delay, new central node identifier, and data recovery statistics) into a single group data packet. Exceptionally, it activates its long-distance communication module once to perform a one-time upload. Power is immediately cut off after upload, and the routine reporting function is returned to the edge communication node. One-time upload duration: ; In the formula, The duration of a single upload; This refers to the number of bytes in the packaged message. The effective uplink rate for cellular data is 20 kilobits per second (measured in practice). The advantage of a one-time upload over time-slice retransmission is that all data can be transmitted in a single network attach, avoiding the repeated attaching and multi-packet protocol overhead of time-slice retransmission.

[0166] For example, at 08:00, edge communication node D058 successfully registered its 4-hour cycle detection (the train left the mountainous area), and notified D203 via near-field communication; D203 packaged the full group data for 10 time slices [from 22:00 to 08:00 the previous day]: 1451 device time slices recovered from the snapshot and 2078 device time slices that converged normally after the recovery period, totaling approximately 99.7 kilobytes, which, after lossless compression (compression rate 0.62), is approximately 38 kilobytes; the weighted fitted trajectory (55 trajectory points and motion segment event records, approximately 4 kilobytes) and the fault event report, approximately 1 kilobyte, totaling approximately 43 kilobytes; uploaded in one go. Each second, the combined power consumption for launch and attachment is approximately 1.7 mAh. The cloud ultimately obtains complete data for this 10-hour period: including the fault compensation phase, the total data recovery rate is... % (the missing portion is concentrated in 10 time slices of the failed device D209 and some time slices of 3 flash edge devices), the group location and transport trajectory are 100% complete; D209 is marked as a suspected fault node and added to the cloud maintenance list. If time-slice retransmission is used (10 time slices are attached and uploaded separately), the attachment overhead alone will reach The increased bandwidth and multi-packet protocol overhead increase the total traffic by about 30%, and also significantly increase latency.

[0167] In this embodiment, the fault takeover closed loop ensures that a single point of failure of the central node no longer leads to the loss of group functionality: the detection latency is no more than 3 hours, the false judgment probability is less than 0.08%, and the new central node completes the election and initial positioning within 120 seconds after the failure determination; the superimposed data compensation mechanism for the coverage period enables the data recovery rate of 99% for the 10-hour double missing period, and the group location and transportation trajectory are 100% complete, with the lateral deviation of the trajectory in the blind push segment not exceeding 1.9 kilometers; one-time packaged upload saves about 30% of the protocol overhead compared to piece-by-piece re-upload and avoids repeated attachment; the credential authentication of the snapshot and the public time base number ensure the authenticity of the recovered data and cross-node fusionability, and the independent working mode branch of the edge device ensures that monitoring is not interrupted in the near-field link damage scenario.

[0168] Example 15 Group location only represents the overall centroid of the stack and cannot answer the individual location of a single container. If container-level location is required in the cloud, traditional solutions can only position each container independently, returning to the old high-power approach. A more serious issue is the dynamic change of the stack: after a container is lifted off the stack (split), firstly, the group topology table and location mapping table retain records of the device that has left, forming "ghost nodes" that contaminate data aggregation, power assessment, and rotation calculations; secondly, the device on the lifted container is in a hardware-off state, neither locating nor reporting, monitoring is immediately interrupted and no one is aware of it; thirdly, the group lacks a means to detect the splitting event. To solve this technical problem, this embodiment achieves zero-power restoration of individual positions based on relative position offset, detects stack splitting events based on lifting vibration characteristics and the drop in air pressure at the same level, automatically shrinks the group topology and marks out-of-group devices through member counting, and allows out-of-group devices to autonomously switch between independent tracking and re-entry into the group, forming a complete adaptive closed loop for dynamic changes in the stack.

[0169] In this embodiment, the control method for the ultra-low power container positioning device further includes: Step 1501: The positioning master node determines the relative position offset of each positioning device relative to itself based on the three-dimensional spatial topology relationship, superimposes the relative position offset with the calculated group position, restores the individual position of each container in the stack, and stores the individual position as a location mapping table associated with the equipment identifier. In the three-dimensional spatial topology, the relative position offset vector of each device consists of two components: vertical and horizontal. The vertical component is determined by the difference in layer number and the effective height of a single layer, while the horizontal component is determined by the intra-layer relative proximity relationship derived from the mutual signal strength measurements between nodes in the same layer in the topology table (the distance between each pair of nodes in the same layer is inverted using the signal strength ranging formula, and then triangulated using intra-layer anchor nodes). The formula for superimposing individual positions is: ; In the formula, For the first The positioning device corresponds to the individual position vector of the container (represented in northeast-central coordinates); The group position vector obtained by solving for the location of the master node; The transformation matrix from the local coordinate system of the stack to the northeast-sky coordinate system is determined by the heading angle of the most recent transport state and the orientation angle of the stack placement (the attitude record at the end of the most recent transport is used in the static stacking state). For the first The relative position offset vector of the device relative to the master node, in meters. The position mapping table is indexed by the device identifier, and the table entries contain {device identifier, individual position, layer number, slot}. The individual position is stored in association with the device identifier.

[0170] For example, locate the master node D210 (top level, sixth level, level number). The solution group's location is east longitude. latitude, north latitude Degree; from node D131 (fourth layer, layer number) Relative position offset Meter (vertical component) (meters, horizontal components are obtained from the fourth layer of inner triangulation); transformation matrix Take the heading angle of the nearest transport state The rotation matrix corresponding to the degree, after superposition, the individual position of container D131 is the group position. Accuracy assessment of latitude and longitude coordinates after meter offset projection: vertical component error Meter, horizontal component error meter, group positioning error Meters, the composite individual position error is approximately Meters, meeting the requirements of box-level management. D210 calculates and stores the location mapping table for each of the 210 devices in the group, with a total of 210 records, which are refreshed with each group location calculation.

[0171] Step 1502: When the positioning master node detects a vibration event that matches the characteristics of a lifting event, and detects a drop in atmospheric pressure at this layer exceeding the lifting air pressure change threshold within a set time period before and after the vibration event, it determines that a stacking splitting event has occurred. The characteristic vibration of a lifting event, distinguishing it from transportation vibration and general public vibration events, follows a "shock-decrease-quiet" time sequence: peak values ​​occur at the moment the lifting device touches the box and during tensioning. to The impact peak lasts less than 0.5 seconds, followed by a 2-4 second damped oscillation as the container leaves the stack. It then enters a state of silence as the container detaches from the stack's mechanical coupling. This timing pattern is sensed and identified by the positioning master node's own inertial measurement unit via the stack's steel structure mechanical coupling. The physical mechanism of the air pressure criterion is as follows: as the container rises, the atmospheric pressure decreases corresponding to the height of ascent. ; In the formula, The change in air pressure on the device being lifted is expressed in Pascals. Set the filter air pressure for the time period preceding the vibration event; Set the filter air pressure for the duration following a vibration event; Let the air density be denoted as . kilograms per cubic meter; It is the acceleration due to gravity; The lifting height is expressed in meters. The lifting air pressure change threshold is taken as... Pa (corresponding to an increase of approximately 3.3 meters or more, significantly different from air pressure fluctuations and interlayer pressure differences of approximately 30 Pa). When the device detects that its own air pressure is continuously decreasing at a rate exceeding -2 Pa per second for more than 3 seconds, it immediately broadcasts an abnormal air pressure report via near-field broadcast in an event-triggered manner (carrying the equipment identifier, layer number, air pressure change, and detection time, without waiting for the aggregation cycle); when the positioning master node detects lifting characteristic vibrations and receives a report within a set duration window of 60 seconds before and after the vibration event that the air pressure drop of its own layer (the layer identified by the layer number carried in the report, i.e., the layer where the container was lifted) exceeds the lifting air pressure change threshold, both criteria are met simultaneously, and a stacking splitting event is determined to have occurred. The two criteria mutually suppress false detections: although the transmitted vibrations from the loading and unloading of adjacent stacks can be perceived as vibration events by the master node, there is no air pressure drop report for this layer; sudden changes in temperature or gradual changes in air pressure caused by gusts can produce air pressure fluctuations, but there are no lifting characteristic vibrations; only actual lifting of this stack and this layer simultaneously meets both criteria.

[0172] For example, at 09:15:22 on a certain day during the 18th rotation cycle, the gantry crane spreader lifted a container on the sixth layer of the top stack (the container where device D182 is located); the D210's own inertial measurement unit sensed the impact peak. (Continued for 0.3 seconds), followed by a damped oscillation of 2.7 seconds, and then silence, matching the "impact-decrease-silence" pattern of the lifting event; D182, lifted off the container, rose 11 meters with the container, and the air pressure... Pa drops to Pa, Pa exceeded At 09:15:31, device D182 broadcast an abnormal air pressure report via event triggering. {Device ID D182, Floor Number} air pressure change At 09:15:31, D210 received a report of a drop in air pressure on the sixth floor (floor 6) nine seconds after the characteristic vibration of the lifting operation occurred. Since both the vibration and air pressure criteria were met within a 60-second window, a stacking dismantling event was determined to have occurred at 09:15:31. In contrast, in scenario 1, the transmitted vibration from the adjacent gantry crane operation at 09:40 was detected by D210, but no abnormal air pressure report was received within the 60-second window, so dismantling was not determined. In scenario 2, the air pressure changed slowly by approximately ten Pa per hour due to a sudden rise in afternoon temperature, without the accompanying characteristic vibration of the lifting operation, so dismantling was not determined.

[0173] Step 1503: After determining that a stacking splitting event has occurred, the positioning master node initiates a group member count via near-field communication. Positioning devices that respond within the set response time limit are retained in the original group. Positioning devices that fail to respond within the set response time limit and whose relative position in the location mapping table is offset from the layer that was removed are marked as out-of-group devices. The corresponding records are deleted from the location mapping table. The last individual position of the out-of-group device and the splitting event time are reported to the cloud server via the edge communication node. The member count process is as follows: the location master node broadcasts a count request frame (carrying a 30-second response time limit and a count cycle identifier), which is forwarded by each area center node at its own layer; each device sends back a count response frame within the time limit (carrying a device identifier and a group access credential authentication tag); the location master node summarizes the response list, and the response rate and judgment rules are as follows: ; ; In the first formula, To check the response rate; The number of devices that complete the response within the time limit; The total number of records in the location mapping table. In the second formula, a non-responding device must simultaneously meet the condition that its relative position offset in the location mapping table corresponds to the removed layer (the layer number and slot fall within the layer and adjacent slot identified in the split event pressure report) to be marked as an outlier device. Devices that do not respond but are not located in the removed layer (e.g., devices with transient link failures) are not classified as outliers and are handled as link failures rather than being removed from the list. After being classified as an outlier, the positioning master node deletes the corresponding record from the location mapping table and reports the last individual position of the outlier device (the last record value before deletion in the mapping table) and the split event time to the cloud server via the edge communication node.

[0174] For example, after D210 determines the split at 09:15:31, it immediately initiates a group member count: the count request is forwarded through the six-layer top-level central node, and 209 devices complete the response within 30 seconds (response rate). D182 did not respond (it has been lifted off the ground with the container and transported outside the yard, out of the reliable range of near-field communication, and the metal structure being transported is blocking the link); the record layer number of D182 in the location mapping table The slot corresponds to the sixth layer slot that was removed. If both conditions are met—no response and offset corresponding to the removed layer—it is marked as an outlier device. D210 deletes record D182 from the mapping table (entry number drops from 210 to 209), and sets the last individual position of {D182: East Longitude}. latitude, north latitude The value recorded in the mapping table before splitting, and the splitting event time of 09:15:31, were uploaded to the cloud server via an event message through edge communication node D058. This allowed the cargo owner and yard management to know in real time that the container had been removed from stack GRP-0007 at 09:15:31 and its exact location before removal. After inventory, the group continued to operate at a scale of 209 units, with power assessment, data aggregation, and rotation candidate sequences all based on the shrunk, clean member set.

[0175] Step 1504: After a positioning device marked as an out-of-group device has not been included in the original group for several consecutive counting cycles, it removes its group membership and enters independent working mode. It autonomously activates its satellite positioning module and long-distance communication module to independently track its own position at a positioning and reporting frequency higher than that of the group working mode, until it is accepted as a member of a new group and then enters the slave node low-power working mode again.

[0176] The criterion for determining whether an out-of-group device is included is that it has not been included in the count for consecutive inventory cycles: During the initial lifting phase, the out-of-group device intermittently receives inventory broadcasts or handshake signals from the original group (accessible in the near field when suspended above the yard), and compares its response list with the broadcast to confirm that it has not been included; once the number of consecutive inventory cycles without inclusion reaches a threshold, its group membership is revoked. ; In the formula, Counting for consecutive inventory periods that were not included; To determine the threshold, take (The inventory cycle is 1 hour, and the threshold covers the 3-hour hoisting and transfer period to prevent misjudgment of being out of the group due to transient link blockage). After switching to independent working mode, the device autonomously activates its satellite positioning module and long-distance communication module (the main control chip outputs a conduction control signal, and the dual power switch circuit connects the two power supply circuits), and independently tracks at a frequency higher than that of the group working mode: a positioning cycle of 10 minutes and a reporting cycle of 30 minutes (compared to the zero positioning of the slave node and indirect reporting via the aggregation channel in the group working mode). The path to being accepted by a new group is as follows: after the device lands on the new stack, the new stack group initiates networking, the device participates in the physical proximity authentication of the new group, obtains the new group admission certificate after passing the authentication, and assumes a role in the new group role election according to its own location and power (in most cases, a slave node). Then, the satellite positioning module and long-distance communication module are disconnected from the hardware power supply level, and the device re-enters the low-power working mode of the slave node.

[0177] For example, D182 was transported to another section of the yard along with the container. Between 09:20 and 12:20, it intermittently received inventory broadcasts from the original group GRP-0007. Its own identifier was not found in the response list for three consecutive inventory cycles. Upon reaching the threshold, at 12:20, GRP-0007 member status was removed, local group credentials were deleted, and the system transitioned to independent working mode: the satellite positioning module was activated (10-minute cycle, using pre-hospitalization location data for a warm start, initial positioning takes 5 seconds) and the long-distance communication module (30-minute reporting cycle). Independent mode consumes approximately 9 mAh per day (high power consumption during the transition period). At 15:40 that day, the container containing D182 was placed in a new stack (new group GRP-0032 network). At 15:45, the event fingerprint matching degree between D182 and the new stacking device for the same container drop excitation within the public observation window was determined. After matching three times within the window to reach the authentication threshold, D182 passes the physical proximity authentication and obtains the GRP-0032 group access credential; in the new group role election, D182's corrected equivalent available power... The device operates at a capacity of 1000 mAh, but is located in the second layer (not the top layer), so the selection criteria are not met, and it becomes a slave node. The D182 master control chip outputs a shutdown control signal, and the hardware cuts off the power supply to the satellite positioning and long-distance communication modules, and switches back to the slave node low-power operating mode with an average daily power consumption of about 1.3 mAh (the independent mode lasts only 3 hours and 20 minutes, consuming about 1.3 mAh, and then seamlessly returns to the ultra-low power group operation). The cloud monitors the D182's trajectory throughout: the last individual position before being lifted (reported by GRP-0007), the trajectory at ten-minute intervals during the independent mode (reported independently), and the group channel reported by GRP-0032 after joining the group.

[0178] In this embodiment, individual position restoration achieves box-level position output with zero additional positioning power consumption (only reusing topology offset and group position superposition), with an accuracy of approximately [missing information]. The system obtains 210 individual location records in the cloud without requiring any additional positioning devices; the dual criteria of lifting vibration characteristics and local air pressure drop mutually suppress false detections of splitting events (neither adjacent stack vibration nor slow air pressure changes trigger false detections), and missed detections are protected by event-triggered air pressure broadcasting; member counting automatically shrinks the group topology, eliminating the pollution of ghost nodes on power assessment, data aggregation, and rotation calculations; independent tracking of out-of-group devices ensures uninterrupted monitoring of split containers, with seamless 10-minute interval trajectory connection between the last individual location before lifting and the group channel reporting after joining, ensuring zero breakpoints in the entire trajectory; the independent mode is only a short-term transition (3 hours and 20 minutes in this example, consuming 1.3 mAh of power), and automatically returns to the ultra-low power consumption state of the node hardware being powered off after rejoining the group, ensuring both power consumption control and monitoring continuity for the group and individuals under frequent stacking and dismantling operations.

[0179] Example 16 The control method for the ultra-low power container positioning device in the embodiments of the present invention has been described above. The ultra-low power container positioning device in the embodiments of the present invention will be described below. Please refer to [link / reference]. Figure 4 One embodiment of the ultra-low power container positioning device of the present invention includes: The satellite positioning module 10 consists of a multi-mode satellite positioning receiver chip (supporting both BeiDou and GPS dual-mode reception), a dielectric-coupled stacked ceramic antenna (mounted on the top surface of the device and arranged upwards through the container mounting holes), an RF pre-filtering circuit, and a temperature-compensated crystal reference. The satellite positioning module 10 is used to receive BeiDou / GPS dual-mode satellite signals, calculate the group position, and is the sole source of positioning data within the group.

[0180] The long-range communication module 20 consists of a narrowband cellular communication chip, a user identification card slot, a power amplifier, a cellular band antenna, and an energy storage capacitor network. The long-range communication module 20 is used to communicate with the cloud server via the NB-IoT cellular network and is the group's only external cellular interface.

[0181] The near-field communication module 30 consists of a low-power wireless transceiver chip (supporting Bluetooth Low Energy or IEEE 802.15.4 physical layer) and a printed antenna. The near-field communication module 30 is used to handle all data exchanges within the group.

[0182] The multimodal sensor 40 includes a barometer and an inertial measurement unit: the barometer is a MEMS digital barometer, mounted on a static pressure port on the device housing; the inertial measurement unit uses a six-axis micromechanical sensor (a three-axis accelerometer and a three-axis gyroscope). The barometer is used to collect atmospheric pressure data to calculate the relative number of stacked layers and to detect sudden drops in air pressure during lifting; the inertial measurement unit is used to determine the motion state through frequency domain analysis of vibration signals and to extract fingerprints of common vibration events.

[0183] The microcontroller unit 50 is a low-power microcontroller that includes a topology building module, an election module, an energy management module, and a communication adaptive module, carrying all the core logic of group collaborative control.

[0184] In this embodiment, the topology construction module is used to construct a three-dimensional spatial topology relationship between multiple positioning devices within the same physical area based on data from multimodal sensors; the election module is used to elect at least one central node from among the multiple positioning devices within the communication range, based on the three-dimensional spatial topology relationship and the equivalent available power of each positioning device, with the remaining positioning devices acting as slave nodes; the energy management module is used to control the central node to activate the satellite positioning module and control the slave nodes to disconnect the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level; the communication adaptation module is used to establish a near-field communication connection between the central node and the slave nodes, and adaptively adjust the near-field communication parameters for data interaction based on the characteristics of the container metal environment. The energy management module is also used to dynamically evaluate the supported rotation cycle based on the central node's own equivalent available power and the predicted power consumption model, and broadcast a rotation warning to the slave nodes when it predicts that its own equivalent available power is insufficient to support the next rotation cycle, so as to dynamically transfer the role of the central node within the group.

[0185] The power management unit 60 includes a dual-channel power switch circuit and a bus switch. The power management unit 60 is used to manage the power supply to each module within the positioning device. The positioning device is powered by a battery.

[0186] In this embodiment, the power management unit 60 includes a dual-channel power switch circuit and a bus switch. The dual-channel power switch circuit is connected in series between the battery power supply and the power input terminals of the satellite positioning module 10 and the long-distance communication module 20, respectively, and is controlled by the switch control signal output by the microcontroller unit 50. The bus switch is connected between the data signal pins of the satellite positioning module 10, the long-distance communication module 20, and the microcontroller unit 50, and is driven by the same switch control signal as the dual-channel power switch circuit. In the off state, it simultaneously cuts off the power supply circuit and blocks the data signal path.

[0187] Since the embodiments of the device part correspond to the embodiments of the above method, the description of the ultra-low power container positioning device provided by the present invention should refer to the above method embodiments. The present invention will not be described again here, but it has the same beneficial effects as the control method of the ultra-low power container positioning device described above.

[0188] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A control method for an ultra-low power container positioning device, characterized in that, include: Data from the multimodal sensors built into the positioning device is collected to identify the physical stacking and movement states of the containers, and to construct a three-dimensional spatial topology relationship between multiple positioning devices within the same physical area. Based on the three-dimensional spatial topology and the equivalent available power of each positioning device, at least one central node is selected from among the multiple positioning devices within the communication range, and the remaining positioning devices are selected as slave nodes. Control the central node to turn on the satellite positioning module, and control the slave node to cut off the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level; The central node establishes a near-field communication connection with the slave node and adaptively adjusts the near-field communication parameters to exchange data based on the characteristics of the container's metal environment. The central node dynamically assesses the number of rotation cycles it can support based on its own equivalent available power and predicted power consumption model. When it predicts that its own equivalent available power is insufficient to support the next rotation cycle, it broadcasts a rotation warning to the slave nodes in order to dynamically transfer the central node role within the group.

2. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, The data acquisition and positioning device, equipped with a multimodal sensor, identifies the physical stacking and movement states of the containers and constructs a three-dimensional spatial topology relationship among multiple positioning devices within the same physical area, including: Atmospheric pressure data is collected by the barometer built into the positioning device, and the relative layer number of the positioning device in the container stack is calculated. The positioning device collects vibration signals through its built-in inertial measurement unit and performs frequency domain feature analysis on the vibration signals. The frequency domain feature analysis includes: if periodic low-frequency swaying vibrations of a specific frequency are detected, it is determined to be a marine oscillation state; if high-frequency vibration features are detected, it is determined to be a road or rail transport state; if the vibration signal is below the silence threshold, it is determined to be a static stacking state. Each positioning device broadcasts its relative layer number, motion status, and device identifier through a near-field communication module to construct a three-dimensional topology map containing the relative spatial positions of each node.

3. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, The central node includes an edge communication node, a planar central node, and a positioning master node; The step of selecting at least one central node from multiple positioning devices within the communication range, based on the three-dimensional spatial topology and the equivalent available power of each positioning device, with the remaining positioning devices acting as slave nodes, includes: In the three-dimensional spatial topology, the node located on the outermost side of the stack with the best long-distance communication signal quality is selected as the edge communication node, and only the edge communication node enables the long-distance communication module to communicate with the cloud server. In each layer, a planar central node is elected to collect the data reported by the nodes in this layer through near-field communication. The planar central node transmits the data of this layer to the edge communication node through near-field communication multi-hop relay. The edge communication node then packages and compresses the data and uploads it to the cloud server. The node located at the top of the stack, with optimal satellite visibility and equivalent available power greater than a preset power threshold is selected as the positioning master node. Only the positioning master node activates the satellite positioning module to obtain the group location and distributes the location information to each area center node via near-field communication.

4. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, The establishment of a near-field communication connection between the central node and the slave node, and the adaptive adjustment of near-field communication parameters based on the characteristics of the container's metal environment for data interaction, includes: During the process of establishing a near-field communication connection, the central node and the slave node exchange probe messages with each other, and evaluate the degree of metal blockage attenuation of the current link based on the round-trip delay of the probe messages and the received signal strength. When the metal shielding attenuation exceeds the first attenuation threshold, the central node adjusts the near-field communication parameters to the first communication parameter group, which includes reducing the communication rate, increasing the transmission power, shortening the single packet data length, and increasing the number of retransmissions. When the metal shielding attenuation is lower than the second attenuation threshold, the central node adjusts the near-field communication parameters to the second communication parameter group, which includes increasing the communication rate, reducing the transmission power, increasing the single packet data length, and reducing the number of retransmissions. The central node continuously calculates the bit error rate and retransmission rate during data interaction, and dynamically switches between the first communication parameter group and the second communication parameter group based on the statistical results.

5. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, The central node dynamically assesses the number of rotation cycles it can support based on its own equivalent available power and a predicted power consumption model. When it predicts that its equivalent available power is insufficient to support the next rotation cycle, it broadcasts a rotation warning to the slave nodes. The dynamic transfer of the central node's role within the group includes: A predictive power consumption model is constructed based on the actual power consumption statistics of the central node performing various tasks such as satellite positioning, long-distance communication and near-field networking within the historical period. The central node calculates the total power consumption required to complete the next rotation cycle based on the predicted power consumption model. When the difference between its own equivalent available power and the total power consumption is lower than the safety margin, it generates a rotation warning and broadcasts it to the slave node. The successor node is determined based on the equivalent available power and spatial topology of each slave node in the group. After the original central node hands over its role, the power supply to the satellite positioning module and long-distance communication module of this node is cut off.

6. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, After constructing the three-dimensional topology map containing the relative spatial positions of each node, the method also includes physical proximity authentication of the positioning devices participating in the group networking. The physical proximity authentication of the positioning devices participating in the group networking includes: Within a preset common observation window, each positioning device records at least one common vibration event generated by the same external excitation through mechanical coupling transmission via a steel box using its built-in inertial measurement unit. The event fingerprint includes the event start time, peak intensity, duration, and spectral energy distribution. Each pair of positioning devices exchanges their recorded event fingerprints through a near-field communication module and calculates the matching degree of event fingerprints pair by pair. When the number of matching event fingerprints reaches a preset authentication threshold, it is determined that the two positioning devices are physically adjacent and belong to the same stack. Positioning devices determined by physical proximity are marked as members of the same group, and a group access credential is assigned to each group member. The group access credential is used for voting qualification verification during the election of the central node, near-field communication network access verification, and legality verification of role handover messages. Location devices that fail the physical proximity determination are refused entry into the group, do not participate in the election of the group's central node, data aggregation, and role rotation, and their node records are removed from the three-dimensional topology map.

7. The control method for the ultra-low power container positioning device according to claim 1, characterized in that, The equivalent available power is determined in the following way: Collect the real-time voltage and temperature of the batteries of each positioning device, and determine the nominal remaining power according to the pre-calibrated voltage-capacity correspondence. Based on the real-time temperature, the nominal remaining power is reduced by the low-temperature capacity to obtain the temperature-corrected power. The self-discharge loss is calculated based on the historical self-discharge records of each positioning device and the duration of the current static stacking state. The self-discharge loss is then deducted from the temperature-corrected power and the recoverable power compensation after hibernation is added to obtain the equivalent usable power.

8. The control method for the ultra-low power container positioning device according to claim 3, characterized in that, The data interaction between the slave node and the planar center node includes: each slave node encapsulates its own status data collected by the near-field communication module into a status reporting frame according to a preset aggregation period and uploads it to the planar center node of its own layer. The status reporting frame includes device identifier, relative layer number, motion status, equivalent available power and near-field link quality index. The data interaction between the planar center node and the edge communication node includes: the planar center node receives status reporting frames from each slave node in its own layer within a preset aggregation window, sorts the status reporting frames according to the relative layer number and device identifier after deduplication, aggregates them into data packets for its own layer, performs lossless compression on the data packets for its own layer, and uploads them to the edge communication node together with the location information distributed by the positioning master node; The data interaction between the edge communication node and the cloud server includes: the edge communication node merging the data packets of this layer transmitted by each area center node with the location information distributed by the positioning master node into a group data packet, adding a group identifier and a rotation cycle count, and then uploading it to the cloud server via the long-distance communication module, and receiving the downlink configuration instructions returned by the cloud server, and then sending them down to each slave node layer by layer via near-field communication.

9. The control method for the ultra-low power container positioning device according to claim 3, characterized in that, The control method for the ultra-low power container positioning device also includes: The positioning master node determines the relative position offset of each positioning device relative to itself based on the three-dimensional spatial topology relationship, superimposes the relative position offset with the calculated group position, restores the individual position of each container in the stack, and stores the individual position as a position mapping table associated with the equipment identifier. When the positioning master node detects a vibration event that matches the characteristics of a lifting event, and detects a drop in atmospheric pressure at the current layer exceeding the lifting pressure change threshold within a set time period before and after the vibration event, it determines that a stacking splitting event has occurred. After determining that a stacking splitting event has occurred, the positioning master node initiates a group member count via near-field communication. Positioning devices that respond within the set response time limit are retained in the original group. Positioning devices that fail to respond beyond the set response time limit and whose relative position in the location mapping table is offset from the layer that was removed are marked as out-of-group devices. The corresponding records are deleted from the location mapping table. The last individual position of the out-of-group device and the splitting event time are reported to the cloud server via the edge communication node. If a positioning device marked as an out-of-group device fails to be included in the original group for several consecutive counting cycles, it will remove its group membership and switch to independent working mode. It will autonomously activate its satellite positioning module and long-distance communication module to independently track its own location at a higher positioning and reporting frequency than in the group working mode, until it is accepted as a member of a new group and then switches back to the slave node low-power working mode.

10. A low-power container positioning device, characterized in that, include: The system includes a satellite positioning module, a long-distance communication module, a near-field communication module, a multimodal sensor, a microcontroller unit, and a power management unit. The microcontroller unit includes a topology building module, an election module, an energy management module, and a communication adaptive module. The multimodal sensor includes a barometer and an inertial measurement unit, used to collect data to identify the physical stacking and motion states of the containers. The topology construction module is used to construct a three-dimensional spatial topology relationship between multiple positioning devices within the same physical area based on the data from the multimodal sensor; the election module is used to elect at least one central node among the multiple positioning devices within the communication range based on the three-dimensional spatial topology relationship and the equivalent available power of each positioning device, with the remaining positioning devices serving as slave nodes. The energy management module is used to control the central node to turn on the satellite positioning module and to control the slave node to cut off the power supply to the satellite positioning module and the long-distance communication module from the hardware power supply level. The adaptive communication module is used to establish a near-field communication connection between the central node and the slave node, and adaptively adjust the near-field communication parameters for data interaction according to the characteristics of the container metal environment; the energy management module is also used to dynamically evaluate the available rotation cycle based on the central node's own equivalent available power and predicted power consumption model, and broadcast a rotation warning to the slave node when it is predicted that its own equivalent available power is insufficient to support the next rotation cycle, so as to dynamically transfer the role of the central node within the group; The power management unit includes a dual-channel power switch circuit and a bus switch. The dual-channel power switch circuit is connected in series between the battery power supply and the power input terminals of the satellite positioning module and the long-distance communication module, respectively, and is controlled by the switch control signal output by the microcontroller unit. The bus switch is connected between the data signal pins of the satellite positioning module, the long-distance communication module and the microcontroller unit, and is driven by the same switch control signal as the dual-channel power switch circuit. In the off state, it simultaneously cuts off the power supply circuit and blocks the data signal path.

Images