Energy efficiency-task driven 6g passive iot terminal autonomous collaborative management method

Abstract

The present application relates to the field of 6G passive Internet of Things and low-power energy management, aiming at the core defects of traditional passive terminal energy flow and task flow fragmentation, low operation reliability and poor multi-terminal collaborative efficiency, a kind of energy efficiency-task driven 6G passive Internet of Things terminal autonomous collaborative management method is proposed.The present application constructs the complete data flow closed loop of energy collection-management-task adaptation-reliable execution-data transmission-multi-terminal collaboration, through independent maximum power tracking of heterogeneous multi-source energy, energy gradient driven micro task segmentation, checkpoint mechanism of energy risk and task income joint constraint, mixed communication adaptive switching and incomplete information distributed local signaling collaboration, the intermittent reliable operation of passive terminal is realized.The present application significantly improves energy utilization efficiency and global sensing accuracy, greatly reduces system signaling overhead, and is suitable for large-scale passive Internet of Things deployment scenarios such as industrial monitoring and smart agriculture.

Description

Technical Field

[0001] This invention relates to the fields of 6G communication, passive Internet of Things (IoT) and low-power energy management, and particularly to an energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals. Background Technology

[0002] In passive IoT applications for 6G communication, traditional energy management and task execution methods have many limitations. Early IoT devices were mostly based on the assumption of continuous and stable power supply, relying on a single energy source or simple threshold control, making it difficult to adapt to heterogeneous and complex dynamic energy scenarios in the environment. For example, classic energy harvesting and terminal computing algorithms are prone to frequent power outages when there are rapid and random fluctuations in solar, wind, or radio frequency energy in the environment. This is because they lack effective adaptation to the low-loss aggregation of heterogeneous micro-energy and the intermittent power supply characteristics, resulting in a significant decrease in the reliability of task execution progress and state persistence.

[0003] Traditional passive communication technologies, such as fixed-mode backscattering or active transmission, suffer from severely limited communication performance in complex environments with scarce energy or poor channel conditions. Furthermore, when multi-terminal systems conduct collaborative sensing over wide areas, the resource scheduling and wake-up mechanisms between terminals are relatively simple, lacking a deep understanding and peak-shifting integration of node geographical locations, energy levels, and charging / discharging cycles. This fails to adequately compensate for the insufficient sensing continuity of a single terminal and easily leads to sampling gaps in spatiotemporal joint sensing.

[0004] Meanwhile, existing methods for handling heterogeneous multi-source energy capture and communication tasks are relatively simplistic, making it difficult to achieve efficient dynamic adaptive processing of complex energy flows and sensing information. This fails to meet the demands of modern 6G intelligent systems for efficient and continuous energy utilization and seamless collaborative sensing in complex dynamic environments. Furthermore, while some complex intelligent dynamic scheduling methods can improve policy adaptation capabilities in complex environments, they generally suffer from problems such as uninterpretable black-box models, high complexity in online updates and inference, and high energy consumption at the edge. Additionally, collaborative allocation methods that rely entirely on centralized solutions from edge controllers require continuous collection of complete network-wide state information, which introduces additional signaling overhead and scheduling latency under conditions of channel constraints, intermittent terminal power-on, or large-scale deployment. Therefore, low-complexity interpretable models for passive terminal constraints and distributed local coordination mechanisms under incomplete information are still needed. Summary of the Invention

[0005] To address the aforementioned issues, this invention aims to propose an energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals. This method constructs a complete data flow closed loop encompassing energy harvesting, management, task adaptation, reliable execution, data transmission, and multi-terminal collaboration. Through heterogeneous multi-source energy independent maximum power tracking, energy gradient-driven micro-task segmentation, a checkpoint mechanism with joint constraints of energy risk and task benefits, hybrid communication adaptive switching, and distributed local signaling collaboration based on incomplete information, the method achieves intermittent reliable operation of passive terminals.

[0006] To achieve the above objectives, the technical solution of the present invention is implemented as follows: An energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals includes the following steps: Step 1, Independent Maximum Power Tracking Acquisition and Time-Frequency Domain Mapping of Heterogeneous Multi-Source Energy: To address the random fluctuation characteristics of solar, micro-wind, and radio frequency energy in the environment, independent energy capture is performed using a variable step-size maximum power tracking algorithm and a nonlinear impedance matching algorithm, respectively; a multi-source energy characteristic model is established, and the instantaneous available power of the three heterogeneous input sources is decoupled and time-aligned to generate and output a stable instantaneous power input sequence; Step 2, Multi-Input Single-Output Converter Energy Adaptive Flow Control Fusion and Nonlinear Storage: The built-in energy routing control logic is used to monitor the impedance mismatch status of each energy channel in real time; based on the decoupled and aligned instantaneous power input sequence output in Step 1, the duty cycle of each channel of the converter is dynamically adjusted, and the three heterogeneous DC micro-energy is aggregated with low loss and stored in the supercapacitor bank; at the same time, the effective injected energy is calculated in real time, and seven core energy state parameters are continuously monitored and generated: remaining energy, average energy harvesting power, energy harvesting power change rate, static leakage power, energy release gradient, predicted remaining available time, and energy level. The energy level is the ratio of remaining energy to the rated energy storage capacity of the supercapacitor bank. Step 3, Intermittent decoupling and segmentation of microtasks based on energy gradient and semantic entropy: Introduce a semantic communication model to model the macroscopic perception and communication tasks as a directed acyclic task flow graph; Based on the energy release gradient of the energy storage element generated in Step 2 and the predicted remaining available time, find the optimal cut set in the task flow graph, decouple the task logic into multiple microtask nodes whose energy consumption matches the current energy and whose state can be independently persisted, and synchronously generate five microtask attribute parameters for each microtask: atomic operation energy consumption, atomic operation duration, checkpoint backup energy consumption, semantic value, and task urgency. Step 4, Real-time checkpoint calculation and intermittent operation reliability assurance based on joint constraints of energy risk and task benefits: Using the remaining energy, average energy harvesting power, energy harvesting power change rate, and static leakage power generated in Step 2, the energy consumption of micro-task atomic operations, atomic operation duration, checkpoint backup energy consumption, semantic value, and task urgency generated in Step 3, as well as the non-volatile storage write wear and power failure restart risk obtained from real-time terminal monitoring, as input parameters, an explicit interpretable mathematical model is used to calculate the checkpoint trigger threshold and backup execution indication; during the execution of micro-tasks, when the remaining energy reaches the trigger threshold and it is predicted that the next atomic operation cannot be completed safely, micro-state backup is forcibly executed to ensure that execution continues from the breakpoint after a power failure restart, and finally generates a reliable execution task result; Step 5, Hybrid active communication and backscatter adaptive information transmission: Establish a joint evaluation function of energy level and channel state; Based on the energy level generated in Step 2 and the real-time channel quality obtained by the terminal in real time monitoring, dynamically switch the communication mode: When energy is sufficient, adopt the active communication mode to improve the transmission rate; when energy is scarce or there is a strong carrier in the environment, adopt the backscatter communication mode to reduce energy consumption, transmit the reliable execution results generated in Step 4, and update the local information freshness loss data after the transmission is completed. Step 6, Distributed Local Signaling Coordination and Optimization for Incomplete Information in Spatiotemporal Joint Coverage: This includes optimizing the low-frequency broadcast coverage requirements of the regional edge controller, the upper bound of the neighborhood coordination capacity, the wake-up decision threshold, and the total number of local iteration rounds. Each passive terminal generates local wake-up decisions and task allocation shares based solely on its own local state and the lightweight signaling of neighboring terminals. The local state includes the energy level generated in Step 2, the information freshness loss updated in Step 5, and the terminal's own coverage capability status. Through local congestion price updates, marginal utility calculations, and energy-task share feasibility projection, a near-optimal scheduling result is obtained, achieving seamless spatiotemporal joint perception of multi-terminal staggered wake-up and the target area.

[0007] Furthermore, in step 1, a comprehensive power capture model is established to describe the energy convergence relationship of multiple heterogeneous energy sources in the time domain. This model simultaneously considers the nonlinear physical characteristics of the three types of energy sources: The solar energy section introduces a temperature power decay coefficient to characterize the negative feedback effect of the difference between the actual temperature of the photovoltaic cell and the standard reference temperature on the conversion efficiency. The wind energy project uses a wind energy utilization coefficient that depends on the tip speed ratio and the blade pitch angle to describe the hydrodynamic power extraction law of micro wind energy. The radio frequency energy term uses a logistic mapping function to simulate the nonlinear startup characteristics and maximum efficiency limit of the rectifier diode under small signal input.

[0008] Furthermore, in step 1, a dynamic reflection coefficient evaluation equation is established to address the impedance mismatch problem in RF energy acquisition, enabling online adjustment of the impedance matching network. The reflection loss is quantified by calculating the voltage mapping relationship between the load impedance and the antenna complex conjugate impedance, guiding the switching of the varactor diode array and maintaining high sensitivity of the rectifier front end when the input power fluctuates significantly.

[0009] Furthermore, in step 2: by calculating the switching conduction loss and nonlinear conversion efficiency of the multi-input converter, a charge inflow net value model is established to calculate the effective injected energy, and the inductor parasitic resistance loss and power switch conversion heat dissipation are deducted. Construct a lower limit evaluation equation for supercapacitor capacity to ensure that the system can safely complete the extreme operation closed loop, including task execution, state backup and communication switching, when there is no external energy injection. The supercapacitor bank adopts an asymmetric micro supercapacitor architecture, with a transition metal oxide as the positive electrode, a carbon-based nanomaterial with high specific surface area as the negative electrode, and a high-impedance solid electrolyte. At the same time, a charge state evolution equation for the supercapacitor is established to describe the remaining available energy state of the device in an intermittent power supply environment in real time, and the survival probability of the terminal in the next task time slot is calculated based on the remaining available energy state.

[0010] Furthermore, in step 3: the micro-task segmentation aims to maximize the semantic information energy efficiency ratio. A time evolution penalty term is introduced into the objective function to describe the timeliness loss caused by the task being forced to suspend due to insufficient energy. A dynamic sleep compensation mechanism based on device material characteristics and leakage loss is established. By dynamically raising the wake-up level, the supercapacitor is forced to operate in the medium-low voltage stable range where leakage current is minimized.

[0011] Furthermore, in step 4, the specific process for calculating the checkpoint trigger threshold using an explicitly interpretable mathematical model includes: First, the explainable safety energy margin is calculated based on the energy income, leakage loss and operating expenses in the next atomic operation cycle. Secondly, based on the difference between the current energy level and the safe energy margin, the failure risk of the next atomic operation is calculated; Then, a comprehensive cost function is established that includes power outage failure risk, backup energy consumption, storage wear, semantic value, and task urgency. Finally, the checkpoint trigger threshold is constructed by combining the safety energy margin and the comprehensive cost function. The safety of the remaining energy after deducting the energy consumption of the next atomic operation is used to determine whether to trigger backup. Simultaneously, a discretized decision mapping table is generated based on the current average energy harvesting power of the passive terminal and the urgency of the task. The terminal directly obtains the checkpoint trigger threshold through a lightweight index function.

[0012] Furthermore, in step 4, the processor classifies and addresses the data based on the semantic information content of the microtask node and the terminal survival probability: high-value key breakpoint data with semantic information content greater than a preset semantic threshold are written into the protected sector of the ferroelectric memory, while non-key intermediate variables with semantic information content less than or equal to the preset semantic threshold are retained in static random access memory and selectively discarded when power is lost.

[0013] Furthermore, in step 5: the communication mode switching judgment criterion considers both the energy margin threshold and the instantaneous channel capacity. When the energy is sufficient, the active communication transmit power is calculated, and when the energy is insufficient, the backscatter receive power is calculated based on the radar equation. Before adaptive switching, physical layer identity authentication logic based on energy capture waveform characteristics is introduced. The instantaneous voltage fluctuation ripple and ambient light intensity pulse sequence of the RF rectification front end are extracted. Dynamic identity tags are constructed using physical non-cloning functions. Illegal energy injection attacks are identified by verifying the consistency of the statistical distribution of the tags.

[0014] Furthermore, in step 5, in active communication mode, an adaptive modulation order selection equation based on semantic importance is established. By balancing effective transmission rate, successful reception probability and single-bit energy efficiency, the constellation mapping is dynamically adjusted according to the current signal-to-noise ratio and energy constraints to maximize the semantic entropy of transmission.

[0015] Furthermore, in step 6, the distributed local signaling coordination optimization specifically includes: Using the spatiotemporal joint penalty cost as the global evaluation criterion, we comprehensively consider changes in energy deficit, loss of information freshness, gain of effective coverage and interference from neighborhood conflicts. A virtual queue evolution equation for the energy deficit of passive terminals is established to replace the full constraint of centralized physical energy storage to support low-complexity local judgment; the regional edge controller predicts the available energy harvesting power of downstream terminals based on the correlation of environmental energy flow between adjacent terminals. Construct a blind zone urgency function for coverage units to quantify the urgency of perceived coverage holes; Each terminal only exchanges compressed local signaling vectors with its neighbor set, and generates wake-up decisions by updating local congestion prices and calculating local marginal utility; Allocate task shares based on effective coverage gain and blind zone urgency, and perform energy and task share feasibility projection. When the preset total number of local iterations is reached or the task allocation change in two consecutive rounds is less than the convergence threshold, a near-optimal wake-up vector and task allocation matrix are output. The error between the output and the optimal solution in the complete information centralized mode satisfies the preset error bound constraint, and the regional edge controller performs low-frequency correction to ensure global consistency.

[0016] Beneficial Effects: This invention addresses common industry bottlenecks in 6G passive IoT terminals, such as intermittent energy supply, extremely limited computing resources, and poor scalability for large-scale deployments. It constructs a closed-loop autonomous management system deeply coupled with "energy flow, task flow, and collaborative flow." Through the fusion of heterogeneous multi-source energy independent maximum power point tracking and adaptive flow control, it achieves efficient capture and aggregation of environmental energy, significantly improving overall energy utilization efficiency. By employing energy gradient-driven semantic micro-task segmentation and an explicit, explainable checkpoint mechanism with joint constraints of energy risk and task reward, it achieves dynamic and precise matching between energy supply and task requirements, effectively solving the reliability problem of task execution under intermittent power supply in passive terminals. Through a hybrid communication mode adaptive switching technology jointly driven by energy level and channel quality, it reduces communication energy consumption while ensuring transmission performance. Through a distributed local signaling collaborative optimization mechanism under incomplete information, it significantly reduces system signaling overhead and scheduling latency, significantly improving the scalability of large-scale multi-terminal deployments. Simultaneously, it enhances system security through lightweight physical layer authentication technology based on environmental energy capture waveform fingerprints. Furthermore, this invention is compatible with existing communication standards and mainstream hardware platforms, and can be widely adapted to various complex application environments. Attached Figure Description

[0017] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a complete data flow closed-loop flowchart of the energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals of the present invention; Figure 2 This is a simulation comparison diagram of the energy storage state of the present invention, showing the difference in energy storage state between the present invention and the traditional fixed strategy under a 24-hour dynamic environment; Figure 3 The image shows a physical diagram of the multi-source energy convergence experimental device of this invention, illustrating the basic hardware implementation. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0019] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Example 1 This embodiment provides an energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals. The overall process is as follows: Figure 1 As shown, it strictly follows the closed-loop data flow logic of "energy harvesting → energy management → task adaptation → task assurance → data transmission → multi-terminal collaboration", specifically including the following steps: Step 1: Independent Maximum Power Point Acquisition and Time-Frequency Domain Mapping of Heterogeneous Multi-Source Energy This step is the energy source for the entire system, responsible for capturing three types of heterogeneous energy from the environment: solar energy, micro-wind energy, and radio frequency energy, and converting them into usable direct current (DC) electricity. To address the different physical characteristics of these three energy sources, a differentiated independent maximum power point tracking (MPPT) strategy is employed to avoid mutual interference between the different energy sources. For solar and wind power, a perturbation observation method combined with a variable step size adjustment mechanism is adopted: when the power change rate is greater than a preset threshold (e.g., 5% / s), the step size is increased to quickly track the maximum power point; when the power change rate is less than the preset threshold, the step size is decreased to improve tracking accuracy, thus avoiding the problem of oscillation near the maximum power point in the traditional fixed step size method.

[0021] For radio frequency energy, due to its extremely weak signal strength and violent fluctuations, a nonlinear impedance matching network is used. By adjusting the bias voltage of the varactor diode array in real time, dynamic matching between the load impedance and the antenna impedance is achieved, thereby maximizing the rectification efficiency.

[0022] A comprehensive power capture model is established to describe the energy convergence relationship of various heterogeneous energy sources in the time domain. The specific calculation formula is as follows:

[0023] in, This refers to the continuous physical sampling moments during the heterogeneous multi-source energy harvesting process. Let be the total capture power at time t. For photovoltaic reference conversion efficiency, The effective area of ​​the photovoltaic panel. Instantaneous solar irradiance, The power attenuation coefficient is the temperature-dependent factor. This refers to the actual physical temperature of the photovoltaic cell. For standard reference temperature, air density, For the area swept by the wind turbine, For instantaneous wind speed, For depends on the tip speed ratio With pitch angle Wind energy utilization coefficient, The instantaneous radio frequency power received at the antenna end. This represents the maximum efficiency limit of the rectifier circuit. For impedance matching slope parameters, This is the sensitivity threshold for starting the rectifier circuit.

[0024] This formula accurately characterizes the instantaneous total power input of the terminal in a complex environment through a nonlinear superposition model. The solar energy term considers the negative feedback effect of temperature on physical efficiency, the wind energy term describes the hydrodynamic power extraction characteristics, and the radio frequency term uses a logistic mapping to simulate the nonlinear startup behavior of the rectifier diode under small signal conditions.

[0025] To address the impedance mismatch problem in the radio frequency energy acquisition process, an online adjustment of the impedance matching network is achieved by establishing a dynamic reflection coefficient evaluation equation. The logic follows:

[0026] This formula calculates the load impedance. Antenna complex conjugate impedance The voltage mapping relationship between them is used to quantify reflection loss and guide the switching of the varactor diode array, thereby maintaining high sensitivity of the rectifier front end when the input power fluctuates significantly. The reflection coefficient, This is the rectified DC feedback voltage.

[0027] This step outputs the three instantaneous available power sequences after decoupling and alignment, which serve as the input for energy fusion in step 2.

[0028] Step 2: Energy Adaptive Flow Control Fusion and Nonlinear Storage of Multi-Input Single-Output Converter This step is responsible for the low-loss aggregation and stable storage of the three heterogeneous DC micro-energy sources collected in step 1, while simultaneously calculating the precise energy state of the terminal in real time to provide a reliable energy boundary for subsequent task scheduling. This embodiment employs a three-input single-output Boost converter, corresponding to the three energy acquisition channels: solar, wind, and radio frequency. Each channel is independently controlled, preventing interference between different energy sources.

[0029] By calculating the switching conduction loss and nonlinear conversion efficiency of the multi-input converter, the energy loss process from the acquisition end to the storage end is quantified, and the formula for calculating the effective injected energy is as follows:

[0030] in, To effectively inject energy, the starting time of the integral window for assessment, The length of the integration window is... The continuous integration time variable is located within the integration window. The instantaneous integrated total power captured in step 1 at continuous integration time... The function value is obtained below. The effective energy actually injected into the supercapacitor bank within the integration window. To be related to the duty cycle matrix The relevant dynamic power conversion efficiency function, This represents the effective value of the inductor current for the corresponding channel. The on-resistance of the power switch transistor. The operating frequency of the converter. and These represent the instantaneous input voltage and current for each channel. The physical time required for a power transistor to complete a single switching operation.

[0031] The physical essence of this formula lies in establishing a net value model for charge storage, which provides a realistic energy boundary constraint for the allocation of subsequent intermittent calculation tasks by deducting inductor parasitic resistance losses and power switch conversion heat dissipation.

[0032] This step continuously monitors and generates the following seven core energy state parameters: Remaining energy: Calculated by monitoring the terminal voltage of the supercapacitor bank and combining it with the capacitance.

[0033] Average power harvesting: The average value of the instantaneous power sequence input in step 1 is obtained within a sliding time window.

[0034] Energy power change rate: obtained by first-order difference calculation of the instantaneous power sequence input in step 1.

[0035] Static leakage power: Calculated by monitoring the voltage drop rate of the supercapacitor bank during system sleep and combining it with the capacitance.

[0036] Energy release gradient: calculated based on the discharge characteristic curve of the supercapacitor and the current remaining energy.

[0037] Predicted remaining available time: Based on the current remaining energy, average energy harvesting power, and static leakage power, it is predicted through the energy balance equation.

[0038] Energy level: The ratio of remaining energy to the rated energy storage capacity of the supercapacitor bank, used to normalize and characterize the degree of charging.

[0039] To address the energy adaptive flow control fusion and the intermittent decoupling and segmentation of microtasks, a lower limit evaluation equation for the supercapacitor capacity is constructed to ensure that the system can safely complete the extreme operation closed loop, including task execution, state backup, and adaptive communication, even without external energy injection. The minimum required capacity of this supercapacitor bank is calculated as follows:

[0040] in, This represents the minimum required capacity for a supercapacitor bank. The energy limit required to execute a maximum energy-consuming microtask node. The energy consumption required to trigger an immediate checkpoint and write field data to non-volatile storage media. The maximum transmission power consumption for performing a single active communication or backscatter mode switch, This is the evaluation value for the longest single operation cycle under extremely low energy conditions. This refers to the instantaneous leakage power of the supercapacitor bank during this operating cycle. The physical efficiency of charge-discharge conversion of the capacitor dielectric. This refers to the rated storage voltage limit of the output of the multi-input single-output converter. The minimum physical dead zone voltage required to maintain normal operation of the computing processing unit and radio frequency module.

[0041] In this embodiment, the supercapacitor bank adopts an asymmetric micro-supercapacitor architecture. It utilizes transition metal oxides at the positive electrode and high-specific-surface-area carbon-based nanomaterials at the negative electrode to ensure high power density and overall energy storage capacity for burst communication, while employing a high-impedance solid-state electrolyte to suppress static leakage current. Under this hardware architecture, the system extracts the natural voltage drop curve to fit the equivalent parallel leakage resistance and performs dynamic sleep compensation threshold calculation during deep sleep.

[0042] in, The dynamic sleep compensation threshold. The system's preset basic energy wake-up threshold, The leakage current penalty weighting coefficient is determined based on the characteristics of the selected capacitor solid-state material. The duration of deep sleep predicted by the system. This is the equivalent parallel leakage resistance of the supercapacitor bank obtained through real-time fitting. This logic dynamically raises the wake-up level, forcing the capacitor components to operate in the low-to-medium voltage stable range where leakage current is minimized, thereby maximizing the overall leakage efficiency and energy retention time of the device during intermittent power outages.

[0043] Meanwhile, by establishing the charge state evolution equation of the supercapacitor, the remaining available energy state of the device under intermittent power supply environment is described in real time, and the survival probability of the terminal to maintain normal operation in the next task time slot is further calculated based on the remaining available energy state:

[0044] in, This is an index for discrete scheduling time slots during the microtask execution phase. For the first The remaining available energy of the supercapacitor bank at the start of each microtask scheduling slot. The remaining available energy of the supercapacitor bank at the start of the next microtask scheduling slot. For the first The energy injected into the database within each micro-task scheduling slot, determined based on the aforementioned effective energy injection model. Let be the total number of microtask nodes to be scheduled, and j be the index of the microtask node. For the j-th microtask node at the th The execution decision variables within each scheduling time slot, and This indicates that the j-th microtask node is being executed. This indicates that the j-th microtask node will not be executed; The computational energy required to execute the j-th microtask node; The energy consumption required to perform state backup of the j-th microtask node; The energy consumption for data communication transmission corresponding to the j-th microtask node; This refers to the static leakage current loss generated by the supercapacitor bank and terminal circuits during the current task scheduling time slot.

[0045] This step outputs seven core energy state parameters that are updated in real time, serving as the core inputs for task splitting in step 3 and checkpoint calculation in step 4.

[0046] Step 3: Intermittent decoupling and segmentation of microtasks based on energy gradient and semantic entropy This step maps energy flow to task flow, solving the core problem of "determining how much work to do based on current power levels." A semantic communication model is introduced to model the macroscopic perception and communication tasks as a directed acyclic task flow graph, where each node represents an atomic operation or a set of closely related operations, and edges represent dependencies between operations.

[0047] The microtask segmentation logic is achieved by maximizing the energy efficiency ratio of semantic information. Its core lies in balancing perception accuracy and energy expenditure from an information theory perspective. The selection of the optimal task subset S follows the following optimization objective:

[0048] This logic introduces a time evolution penalty term. This describes the timeliness loss caused by tasks being forced to suspend due to insufficient energy, ensuring that the algorithm prioritizes removing nodes with high computational cost and low semantic value when energy is extremely scarce. For nodes The amount of semantic information, This is the semantic fidelity weighting coefficient. To implement energy efficiency expenditures, Energy consumption for state backup As a time penalty weight, This represents the waiting time for a node in the execution queue.

[0049] Based on the energy release gradient of the energy storage element generated in step 2 and the predicted remaining available time, the optimal cut set is found in the task flow graph to decouple the task logic into a series of energy-consumption-matched and independently persistent micro-task nodes. The energy consumption of each micro-task node should be less than 80% of the current remaining available energy to ensure that the execution and state backup of the micro-task can be completed even in the worst case.

[0050] This step simultaneously generates the following five attribute parameters for each microtask: Atomic operation energy consumption: obtained by statistically modeling the historical execution energy consumption of each atomic operation in the task flow graph. Atomic operation duration: Calculated by statistical modeling of the historical execution time of each atomic operation in the task flow graph.

[0051] Checkpoint backup energy consumption: calculated based on the amount of state data that needs to be backed up for this microtask node, combined with the unit write energy consumption of the ferroelectric memory.

[0052] Semantic value: Based on the semantic communication model, the importance of the output information of each micro-task node is quantitatively evaluated, and the value range is [0,1].

[0053] Task urgency: Calculated based on the microtask node's position in the task flow graph, its deadline, and dependencies. This step outputs the sequence of split microtask nodes and their five attribute parameters, which serve as input for the checkpoint calculation in step 4.

[0054] Step 4: Real-time checkpoint calculation and intermittent operation reliability assurance based on joint constraints of energy risk and task benefit This step provides reliability assurance for microtask execution, addressing the issue of "what to do if the power goes out halfway through." An explicit, interpretable mathematical model scheme employing joint constraints of energy risk and task reward predicts the checkpoint trigger threshold, allowing the terminal to make decisions during the time slot. The seven core energy state parameters generated in step 2 of the internal acquisition process, the five microtask attribute parameters generated in step 3, and the non-volatile memory write wear obtained from real-time monitoring by the terminal are all included. and the risk factor of power outage restart It calculates the checkpoint trigger threshold and backup execution indication based on the explicit interpretable energy security margin, power outage risk function and comprehensive cost function.

[0055] Among them, non-volatile memory write wear The power failure restart risk coefficient is obtained by real-time counting of write and erase counts from the ferroelectric memory controller. It is obtained by the energy management module based on historical energy fluctuation data and the current energy status in real time.

[0056] First, based on the terminal's energy income, leakage loss, and operating expenses in the next atomic operation cycle, the explainable safety energy margin is calculated:

[0057] in, This represents the safety energy margin that terminal u needs to reserve to avoid power outages, complete backups, and offset negative energy fluctuations when executing microtask r.

[0058] Secondly, based on the difference between the current energy level and the safe energy margin, the risk of failure in the next atomic operation is calculated:

[0059] in, This refers to the probability that the terminal will survive to complete the minimum necessary operations within the next microtask scheduling slot without experiencing unexpected power loss. The steepness coefficient is used to map the survival probability during the microtask execution phase. The energy level at the start of the next microtask scheduling slot. The minimum energy level threshold required to maintain a minimum operational closed loop during the microtask execution phase.

[0060] Then, establish the checkpoint backup comprehensive cost function:

[0061] in, This represents the comprehensive cost of checkpoint triggering. The first term characterizes the risk of power outage failure, the second term characterizes backup energy consumption, the third term characterizes storage media write wear, and the fourth and fifth terms characterize the positive demand for early backups from high semantic value tasks and high urgency tasks, respectively. , , , , All are non-negative weighting coefficients.

[0062] Furthermore, an interpretable checkpoint trigger threshold is constructed based on the safety energy margin and the comprehensive cost function:

[0063] in, This represents the checkpoint trigger threshold output by the mathematical model. This indicates the upper limit of energy storage for terminal u. This indicates the minimum operating energy required for terminal u. , , , All are non-negative threshold adjustment coefficients.

[0064] Finally, based on the remaining energy minus the energy consumption of the next atomic operation, a safety assessment is made to determine whether to trigger a backup:

[0065] in, This indicates that a checkpoint backup is triggered and the microtask context data is written to non-volatile storage media. This indicates that the current microtask will continue to run without triggering a backup.

[0066] Based on the current average energy harvesting power of the passive terminal and the task urgency, a discretized decision mapping table is extracted and sent to the passive terminal. The passive terminal directly obtains the checkpoint trigger threshold through a lightweight index function.

[0067] in Threshold triggered at checkpoints, For decision mapping table, This represents the remaining available energy at the current moment. For energy quantization step size, A Boolean variable representing the sign of the rate of change of energy capture power.

[0068] The processor calculates the node semantic information based on step 3. Survival probability Classify and address the data. Greater than the preset semantic threshold High-value critical breakpoint data is written into the protected sector of the ferroelectric memory. Less than or equal to a preset semantic threshold Non-critical intermediate variables are retained in static random access memory and selectively discarded upon power failure.

[0069] This step outputs the results of a reliably executed task, which serves as the input for communication transmission in step 5.

[0070] Step 5: Hybrid Active Communication and Backscatter Adaptive Information Transmission This step is responsible for transmitting the results of the reliably executed task in step 4 to the edge controller, solving the problem of "transmitting data with minimal power". A joint evaluation function for energy level and channel state is established to dynamically switch communication modes. The adaptive switching logic achieves smooth and seamless switching between the two communication modes by jointly evaluating the energy margin threshold and instantaneous channel capacity. The switching judgment criteria are expressed as follows: Active transmission power calculation is performed when the level threshold condition is met (energy level is greater than a preset threshold).

[0071] Otherwise, perform backscatter signal received power calculation.

[0072] This logic introduces an energy level. As a safety boundary, it ensures that active communication does not cause instantaneous system crash, and uses radar equations to describe the feasibility of backscatter communication links under extremely low power consumption; among which, This serves as an index of the communication decision rounds during the communication mode switching phase. For the first Transmit power of active communication mode in each communication decision round The remaining available energy of the supercapacitor bank detected at the start of the communication decision round. To ensure that the terminal can maintain a minimum operating closed loop even after performing active communication, a survival energy threshold must be reserved. The duration of a single active communication. For gateway receive power, For ambient carrier transmit power, For antenna gain, For the scattering cross-section, For wavelength, and Physical distance To modulate the difference in complex reflection coefficients; Before the adaptive information transmission mode switching, physical layer authentication logic based on energy capture waveform characteristics is introduced, and the passive terminal extracts the instantaneous voltage fluctuation ripple from the RF rectification front end. With ambient light intensity pulse sequence Utilizing physically non-clonable functions to construct dynamic identity tags

[0073] The regional collaborative scheduling edge gateway identifies illegal energy injection attacks by verifying the consistency of the statistical distribution of the tag. For dynamic identity tags, This refers to instantaneous voltage fluctuation ripple. It is an ambient light intensity pulse sequence. For authentication timestamps; In active communication mode, an adaptive modulation order selection equation based on semantic importance is established:

[0074] This formula weighs the effective transmission rate. Successful reception probability and single-bit energy efficiency This enables the terminal to adjust according to the current signal-to-noise ratio. and energy limitation Dynamically adjust the constellation mapping to maximize the semantic entropy of the transmission; among which, This is the optimal modulation order.

[0075] This step outputs data on the freshness loss of the transmitted information, which serves as input for step 6, multi-terminal collaboration.

[0076] Step 6: Coordination and Optimization of Distributed Local Signaling for Incomplete Information in Spatiotemporal Joint Coverage This step upgrades the system from "single-terminal independent operation" to "multi-terminal collaborative sensing," solving the problem of "how a bunch of devices can cooperate seamlessly." Based on the area-level coordination parameters of the area edge controller's low-frequency broadcast and the local signaling interaction of each passive terminal's neighborhood, a distributed local signaling coordination optimization process is invoked. The area edge controller's low-frequency broadcast coverage task requirements, the upper bound of the neighborhood coordination capacity, the wake-up decision threshold, and the total number of local iteration rounds are considered. Each passive terminal generates its local wake-up decision and task allocation share solely based on its own energy level generated in step 2, the information freshness loss updated in step 5, its own coverage capability status, neighboring coverage holes, neighborhood interference, and neighboring terminal signaling.

[0077] First, the spatiotemporal joint penalty cost is used as the global evaluation criterion for edge multi-terminal collaborative scheduling, and its calculation formula is as follows:

[0078]

[0079] in, Add a joint spatiotemporal penalty cost to distributed cooperative scheduling. The decision matrix for waking up passive terminals. Assign matrices to tasks. For the set of passive terminals participating in coordinated scheduling, For the target coverage unit set, For passive terminal indexing, For covering cell indexes, In order to cover the coverage unit The set of candidate passive terminals, Passive terminal The change in energy deficit For coverage unit Information freshness loss, Passive terminal Assigned to coverage unit Task share, Passive terminal For coverage unit Effective coverage gain, Passive terminal Awakening decision variables and Indicates waking up, Indicates sleep or low-power listening. Passive terminal The cost of neighborhood conflict interference, , , and All are non-negative weighting coefficients; Secondly, the regional edge controller predicts the available energy harvesting power of downstream terminals based on the correlation of environmental energy flow between adjacent passive terminals. The calculation formula is as follows:

[0080] in, Passive terminal In scheduling time slots Predicted energy extraction power, For edge collaborative scheduling time slots, Passive terminal The neighbor terminal index, Passive terminal The neighborhood group, For neighboring terminals For passive terminals Environmental energy flow related weights, For neighboring terminals With passive terminals Geometric distance between them The spatial attenuation coefficient of environmental energy flow. For neighboring terminals In scheduling time slots Real-time energy acquisition power; Then, a virtual queue evolution equation for the energy deficit of passive terminals is established to replace the full constraint of centralized physical energy storage and support low-complexity local judgment. The calculation formula is as follows: .

[0081] in, Passive terminal In scheduling time slots Energy deficit queue Passive terminal The energy deficit queue in the next scheduling time slot Passive terminal In scheduling time slots Awakening decision, Passive terminal Energy consumption required to perform a single collaborative sensing task The time interval between adjacent scheduling slots; To address coverage gaps, a blind zone urgency function for coverage units is constructed, and its calculation formula is as follows:

[0082] in, For coverage unit In scheduling time slots The urgency of blind spots This represents the spatial blind zone attenuation coefficient. For coverage unit Task importance weighting For coverage unit Distance to the nearest effectively covered location; In local iteration rounds Inside, passive terminal Gathering only from neighbors The formula for calculating the compressed local signaling vector is as follows:

[0083] in, Passive terminal In local iteration rounds The local signaling vector sent. For local iteration rounds, The normalized energy level This represents the normalized neighborhood coverage gain. This represents a loss in information freshness after normalization. This represents the normalized energy deficit state. The normalized proximity cover hole urgency, Passive terminal Local congestion price, Passive terminal In local iteration rounds Awakening decision, Indicates vector transpose; Passive terminal The local marginal utility is calculated based on the local state and neighbor signaling, using the following formula:

[0084] in, Passive terminal In local iteration rounds The local marginal utility This is the normalized estimate of neighborhood conflict interference. , , , and All are non-negative marginal utility weights; passive terminals The local congestion price is updated based on the wake-up status of neighboring terminals, and the calculation formula is as follows:

[0085] in, Passive terminal The local congestion price in the next local iteration round For the first Local iteration step size, Passive terminal With neighboring terminals The local constraint coupling coefficient between them For neighboring terminals In local iteration rounds Awakening decision, Passive terminal Upper bound of the tolerable neighborhood coordination capacity Indicates nonnegative projection; Passive terminal The wake-up decision is generated based on the result of subtracting the neighborhood congestion cost from the local marginal utility, and the calculation formula is as follows:

[0086] in, Passive terminal The wake-up decision in the next local iteration round For neighboring terminals In local iteration rounds Local congestion price, For neighboring terminals For passive terminals The local constraint coupling coefficient, Passive terminal The wake-up determination threshold; After the wake-up decision is determined, the passive terminal For coverage unit The task allocation share is calculated based on the effective coverage gain and the urgency of blind spots, and the calculation formula is as follows:

[0087] in, Passive terminal Assign to the covering unit in the next local iteration round Task share, Passive terminal In local iteration rounds For coverage unit Effective coverage gain, This is the adjustment coefficient for the urgency of the blind spot. For coverage unit In local iteration rounds The urgency of blind spots Candidate terminal For coverage unit Effective coverage gain, Candidate terminal The wake-up decision in the next local iteration round To prevent positive numbers with a denominator of zero; Coverage unit The actual workload is determined by the task requirements and the task allocation share, and the calculation formula is as follows:

[0088] in, Passive terminal The actual coverage unit undertaken in the next local iteration round workload For coverage unit Task requirements; To ensure that the distributed suboptimal result is executable under the condition of energy-constrained passive terminal, the constraint formula for the feasibility projection of task execution energy and share is as follows:

[0089] in, To execute the coverage unit Energy consumption required per unit task Passive terminal In local iteration rounds Available energy level; When the preset total number of local iterations is reached Or the change in task allocation for two consecutive rounds is less than the convergence threshold. At that time, output the passive terminal wake-up vector, task allocation matrix, and area coverage completion rate, and make the first... The cost of round-distributed iteration and the optimal cost of centralized full-information iteration satisfy the following approximate optimal error bound:

[0090] in, For the first The spatiotemporal joint penalty cost obtained from round-distributed iteration, The optimal cost for centralized complete information. To preset the total number of local iterations, This is the upper bound of the initial local decision-making disagreement. An upper bound for error introduced for incomplete observations. An upper bound on the error introduced for local signaling quantization, delay, or packet loss. An upper bound for the error introduced for the feasibility projection of energy and task share.

[0091] The distributed local signaling coordination and optimization process sequentially executes the following steps: the regional edge controller initializes the neighbor set and broadcasts low-frequency parameters; passive terminals collect local states; passive terminals exchange local signaling vectors; passive terminals update local congestion prices; passive terminals generate wake-up decisions; passive terminals calculate coverage unit task shares; passive terminals perform energy and share feasibility projection; and the regional edge controller performs low-frequency correction and outputs near-optimal wake-up and task allocation results. Thus, without requiring continuous collection of complete state information across the entire network, it achieves staggered wake-up, relay sensing, and seamless collaborative coverage for multiple edge terminals.

[0092] Figure 2 This is a simulation comparison of the energy storage state of the energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to the present invention. The figure uses a 24-hour dynamic environmental energy change as the simulation period to compare the changes in the terminal's energy storage state under the baseline of the complete strategy of the present invention and the traditional fixed strategy. The strategy of the present invention, through the collaborative capture and adaptive convergence of heterogeneous energy sources such as solar, micro-wind, and radio frequency energy, combined with semantic micro-task scheduling, energy risk constraint checkpoint management, and active communication / backscatter communication switching mechanisms, enables the terminal to maintain a high level of remaining usable energy even during energy fluctuations and periods of weak energy harvesting. In contrast, the traditional fixed strategy experiences a rapid decline in energy storage state during periods of low light or insufficient energy, easily leading to undervoltage shutdown. The simulation results demonstrate that the present invention can improve the continuous energy utilization capability and task operation reliability of passive IoT terminals under intermittent power supply conditions.

[0093] Figure 3 This is a physical diagram of a multi-source energy aggregation experimental device, illustrating the hardware foundation for the heterogeneous environment energy harvesting, aggregation, and energy storage management process of this invention. The diagram shows a photovoltaic energy harvesting module, a micro-wind energy harvesting module, a multi-source energy aggregation unit, an energy management unit, and an energy storage module. The multi-source energy aggregation unit rectifies, conditions, and aggregates weak electrical energy from different sources with varying output characteristics. The energy management unit regulates the voltage, controls charging and discharging, and manages the energy storage after aggregation. This device provides a stable energy input foundation for passive terminals to perform subsequent sensing, computing, state backup, and communication transmission tasks. In further implementations, a radio frequency energy harvesting branch can also be connected to the multi-source energy aggregation unit, thus forming an experimental verification platform corresponding to the heterogeneous multi-source energy harvesting and autonomous energy management method of this invention.

[0094] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals, characterized in that, Includes the following steps: Step 1, Independent Maximum Power Tracking Acquisition and Time-Frequency Domain Mapping of Heterogeneous Multi-Source Energy: To address the random fluctuation characteristics of solar, micro-wind, and radio frequency energy in the environment, independent energy capture is performed using a variable step-size maximum power tracking algorithm and a nonlinear impedance matching algorithm, respectively; a multi-source energy characteristic model is established, and the instantaneous available power of the three heterogeneous input sources is decoupled and time-aligned to generate and output a stable instantaneous power input sequence; Step 2, Multi-Input Single-Output Converter Energy Adaptive Flow Control Fusion and Nonlinear Storage: The built-in energy routing control logic is used to monitor the impedance mismatch status of each energy channel in real time; based on the decoupled and aligned instantaneous power input sequence output in Step 1, the duty cycle of each channel of the converter is dynamically adjusted, and the three heterogeneous DC micro-energy is aggregated with low loss and stored in the supercapacitor bank; at the same time, the effective injected energy is calculated in real time, and seven core energy state parameters are continuously monitored and generated: remaining energy, average energy harvesting power, energy harvesting power change rate, static leakage power, energy release gradient, predicted remaining available time, and energy level. The energy level is the ratio of remaining energy to the rated energy storage capacity of the supercapacitor bank. Step 3, Intermittent decoupling and segmentation of microtasks based on energy gradient and semantic entropy: Introduce a semantic communication model to model the macroscopic perception and communication tasks as a directed acyclic task flow graph; Based on the energy release gradient of the energy storage element generated in Step 2 and the predicted remaining available time, find the optimal cut set in the task flow graph, decouple the task logic into multiple microtask nodes whose energy consumption matches the current energy and whose state can be independently persisted, and synchronously generate five microtask attribute parameters for each microtask: atomic operation energy consumption, atomic operation duration, checkpoint backup energy consumption, semantic value, and task urgency. Step 4, Real-time checkpoint calculation and intermittent operation reliability assurance based on joint constraints of energy risk and task benefits: Using the remaining energy, average energy harvesting power, energy harvesting power change rate, and static leakage power generated in Step 2, the energy consumption of micro-task atomic operations, atomic operation duration, checkpoint backup energy consumption, semantic value, and task urgency generated in Step 3, as well as the non-volatile storage write wear and power failure restart risk obtained from real-time monitoring of the terminal as input parameters, an explicit interpretable mathematical model is used to calculate the checkpoint trigger threshold and backup execution indication. During the execution of a microtask, when the remaining energy reaches the trigger threshold and it is predicted that the next atomic operation cannot be completed safely, a microstate backup is forcibly executed to ensure that execution continues from the breakpoint after a power outage and restart, ultimately generating a reliable task result. Step 5, Hybrid active communication and backscatter adaptive information transmission: Establish a joint evaluation function of energy level and channel state; Based on the energy level generated in Step 2 and the real-time channel quality obtained by the terminal in real time monitoring, dynamically switch the communication mode: When energy is sufficient, adopt the active communication mode to improve the transmission rate; when energy is scarce or there is a strong carrier in the environment, adopt the backscatter communication mode to reduce energy consumption, transmit the reliable execution results generated in Step 4, and update the local information freshness loss data after the transmission is completed. Step 6, Distributed Local Signaling Coordination and Optimization for Incomplete Information in Spatiotemporal Joint Coverage: This includes optimizing the low-frequency broadcast coverage requirements of the regional edge controller, the upper bound of the neighborhood coordination capacity, the wake-up decision threshold, and the total number of local iteration rounds. Each passive terminal generates local wake-up decisions and task allocation shares based solely on its own local state and the lightweight signaling of neighboring terminals. The local state includes the energy level generated in Step 2, the information freshness loss updated in Step 5, and the terminal's own coverage capability status. Through local congestion price updates, marginal utility calculations, and energy-task share feasibility projection, a near-optimal scheduling result is obtained, achieving seamless spatiotemporal joint perception of multi-terminal staggered wake-up and the target area.

2. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 1, a comprehensive power capture model is established to describe the energy convergence relationship of various heterogeneous energy sources in the time domain. This model simultaneously considers the nonlinear physical characteristics of the three types of energy sources: The solar energy section introduces a temperature power decay coefficient to characterize the negative feedback effect of the difference between the actual temperature of the photovoltaic cell and the standard reference temperature on the conversion efficiency. The wind energy project uses a wind energy utilization coefficient that depends on the tip speed ratio and the blade pitch angle to describe the hydrodynamic power extraction law of micro wind energy. The radio frequency energy term uses a logistic mapping function to simulate the nonlinear startup characteristics and maximum efficiency limit of the rectifier diode under small signal input.

3. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 1, a dynamic reflection coefficient evaluation equation is established to address the impedance mismatch problem in RF energy acquisition, enabling online adjustment of the impedance matching network. The reflection loss is quantified by calculating the voltage mapping relationship between the load impedance and the antenna complex conjugate impedance, guiding the switching of the varactor diode array and maintaining high sensitivity of the rectifier front end when the input power fluctuates significantly.

4. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 2: By calculating the switching conduction loss and nonlinear conversion efficiency of the multi-input converter, a charge inflow net value model is established to calculate the effective injected energy, and the inductor parasitic resistance loss and power switch conversion heat dissipation are deducted. Construct a lower limit evaluation equation for supercapacitor capacity to ensure that the system can safely complete the extreme operation closed loop, including task execution, state backup and communication switching, when there is no external energy injection. The supercapacitor bank adopts an asymmetric micro supercapacitor architecture, with a transition metal oxide as the positive electrode, a carbon-based nanomaterial with high specific surface area as the negative electrode, and a high-impedance solid electrolyte. At the same time, a charge state evolution equation for the supercapacitor is established to describe the remaining available energy state of the device in an intermittent power supply environment in real time, and the survival probability of the terminal in the next task time slot is calculated based on the remaining available energy state.

5. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 3: the micro-task segmentation aims to maximize the energy efficiency ratio of semantic information. A time evolution penalty term is introduced into the objective function to describe the timeliness loss caused by the task being forced to suspend due to insufficient energy. A dynamic sleep compensation mechanism based on device material characteristics and leakage loss is established. By dynamically raising the wake-up level, the supercapacitor is forced to operate in the medium-low voltage stable range where leakage current is minimized.

6. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 4, the specific process for calculating the checkpoint trigger threshold using an explicit interpretable mathematical model includes: First, the explainable safety energy margin is calculated based on the energy income, leakage loss and operating expenses in the next atomic operation cycle. Secondly, based on the difference between the current energy level and the safe energy margin, the failure risk of the next atomic operation is calculated; Then, a comprehensive cost function is established that includes power outage failure risk, backup energy consumption, storage wear, semantic value, and task urgency. Finally, the checkpoint trigger threshold is constructed by combining the safety energy margin and the comprehensive cost function. The safety of the remaining energy after deducting the energy consumption of the next atomic operation is used to determine whether to trigger backup. Simultaneously, a discretized decision mapping table is generated based on the current average energy harvesting power of the passive terminal and the urgency of the task. The terminal directly obtains the checkpoint trigger threshold through a lightweight index function.

7. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 4, the processor classifies and addresses the data based on the semantic information content of the microtask node and the terminal survival probability: high-value key breakpoint data with semantic information content greater than a preset semantic threshold are written into the protected sector of the ferroelectric memory, while non-key intermediate variables with semantic information content less than or equal to the preset semantic threshold are retained in static random access memory and selectively discarded when power is lost.

8. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 5: the communication mode switching judgment criterion considers both the energy margin threshold and the instantaneous channel capacity. When the energy is sufficient, the active communication transmit power is calculated, and when the energy is insufficient, the backscatter receive power is calculated based on the radar equation. Before adaptive switching, physical layer identity authentication logic based on energy capture waveform characteristics is introduced. The instantaneous voltage fluctuation ripple and ambient light intensity pulse sequence of the RF rectification front end are extracted. Dynamic identity tags are constructed using physical non-cloning functions. Illegal energy injection attacks are identified by verifying the consistency of the statistical distribution of the tags.

9. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, In step 5, under active communication mode, an adaptive modulation order selection equation based on semantic importance is established. By balancing effective transmission rate, successful reception probability and single-bit energy efficiency, the constellation mapping is dynamically adjusted according to the current signal-to-noise ratio and energy constraints to maximize the semantic entropy of transmission.

10. The energy efficiency-task-driven autonomous collaborative management method for 6G passive IoT terminals according to claim 1, characterized in that, Step 6, the distributed local signaling coordination optimization specifically includes: Using the spatiotemporal joint penalty cost as the global evaluation criterion, we comprehensively consider changes in energy deficit, loss of information freshness, gain of effective coverage and interference from neighborhood conflicts. A virtual queue evolution equation for the energy deficit of passive terminals is established to replace the full constraint of centralized physical energy storage to support low-complexity local judgment; the regional edge controller predicts the available energy harvesting power of downstream terminals based on the correlation of environmental energy flow between adjacent terminals. Construct a blind zone urgency function for coverage units to quantify the urgency of perceived coverage holes; Each terminal only exchanges compressed local signaling vectors with its neighbor set, and generates wake-up decisions by updating local congestion prices and calculating local marginal utility; Allocate task shares based on effective coverage gain and blind zone urgency, and perform energy and task share feasibility projection. When the preset total number of local iterations is reached or the task allocation change in two consecutive rounds is less than the convergence threshold, a near-optimal wake-up vector and task allocation matrix are output. The error between the output and the optimal solution in the complete information centralized mode satisfies the preset error bound constraint, and the regional edge controller performs low-frequency correction to ensure global consistency.

Images