Low-power mesh networking system and method for taming crystal oscillator of intermittent satellite time service

Abstract

This invention discloses a low-power mesh networking system and method for intermittent satellite timing disciplined crystal oscillators. Each routing node in the system is equipped with a GNSS positioning and timing unit and a local crystal oscillator timing unit. The GNSS unit operates intermittently in a sleep-timed wake-up mode. After waking up, it captures the standard 1PPS pulse and UTC timestamp from the satellite and then goes into sleep mode. The routing node has a built-in crystal oscillator discipline correction module. Each time a 1PPS pulse is successfully acquired, the module captures the latched count value of the local crystal oscillator's free-running counter at the rising edge of the pulse. It calculates the phase deviation between this latched value and the theoretically expected count value. The instantaneous frequency deviation (ppm) of the local crystal oscillator is calculated from the phase deviation over multiple consecutive cycles, and the corresponding compensation amount is written to the frequency compensation register, allowing the crystal oscillator to run freely at the compensated frequency during subsequent GNSS sleep periods. Simultaneously, a dual-table architecture of a backbone routing table and a neighbor forwarding table is used to isolate long- and short-range services. Furthermore, a routing node silent listening-coordinator invitation network access protocol is used instead of active flooding scanning.

Description

Technical Field

[0001] This invention belongs to the technical field of wireless communication and the Internet of Things, and relates to a low-power Mesh networking system and method with intermittent satellite timing disciplined crystal oscillators. Background Technology

[0002] Wireless mesh self-organizing network technology is widely used in scenarios such as municipal lighting, industrial monitoring, field geological and hydrological monitoring, and smart agriculture. In these scenarios, a large number of nodes rely on batteries or solar energy storage for power, requiring that the overall power consumption be minimized while maintaining network topology reachability and transmission timing consistency.

[0003] The existing technology has the following interrelated technical defects: First, local clock drift causes timing discrepancies. Most low-power Mesh nodes maintain their local time reference using passive quartz crystals or ordinary RTCs. The crystal frequency is affected by temperature gradients, aging, and power supply voltage fluctuations, resulting in a drift of ±10ppm to ±30ppm. Over long-term operation, the accumulated error can reach hundreds of milliseconds or even several seconds, directly leading to two types of consequences: (1) the sleep-wake window based on time slot alignment is misaligned, the node's receiving window and sending window no longer coincide, and the packet loss rate increases; (2) the time-dependent route keep-alive timeout judgment is inaccurate, the link is misjudged and thus triggers unnecessary network-wide route reconstruction, further increasing power consumption and air interface overhead.

[0004] Second, there is the power consumption contradiction when GNSS acquires absolute time. The RF front-end and baseband power consumption of GNSS / GPS / BeiDou modules in continuous operation is typically in the range of 20mA to 45mA. If battery nodes are required to maintain GNSS power 24 / 7, even with a 5000mAh battery pack, the battery life will be compressed to several weeks or even days, completely failing to meet the engineering goal of over one year of battery life for unattended field nodes. Some existing solutions attempt to wake up GNSS at longer intervals (e.g., 30 minutes to several hours) to read the current UTC timestamp, and then directly overwrite the local software counter with this timestamp. This "read time overwrite counter" approach only corrects the absolute time deviation but does not perform any closed-loop correction at the crystal oscillator physical frequency level. The local counter continues to drift freely at the uncompensated original crystal oscillator frequency during subsequent sleep periods. Its long-term error growth rate is not fundamentally different from the undisciplined state; it is merely periodically "hard-jumped" to the correct scale. This jump operation, however, disrupts the consistency of continuous timing between adjacent nodes, which is detrimental to protocols that rely on accumulated time slots.

[0005] Third, there is the problem of error accumulation in multi-hop relative clock synchronization chains. Another approach uses parent-tracking to transmit clock offset corrections hop-by-hop, with child nodes adjusting their local counters based on the offset values ​​broadcast by their parent nodes. However, this type of approach is essentially relative synchronization; if the root node itself lacks a high-precision absolute reference, the entire network will still drift. Even if the root node is connected to GNSS, only one or a few nodes in the entire network have an absolute reference; the remaining nodes rely on multi-hop propagation of offsets. This results in slow re-convergence during topology changes, and each hop introduces additional measurement noise, making it difficult to guarantee clock quality for long-distance edge nodes.

[0006] Fourth, the simplistic routing architecture leads to unnecessary wake-ups and air interface congestion. Existing Mesh routing systems mostly use a unified routing table to handle all services. Long-distance backbone forwarding frames and near-field short-distance control frames share the same routing criterion and the same transmission queue. This causes short near-field frames and long-distance multi-hop frames to compete for the same transmission window, or to frequently wake up the routing calculation logic under a low duty cycle system, increasing the time of ineffective activity and clock uncertainty.

[0007] Fifth, proactive flooding during the network access phase increases power consumption and introduces patent feature risks. Many commercial solutions actively scan the channel and send flooding probe / beacon frames to discover the network after the node is powered on. This not only increases the peak power consumption during the network access phase, but its combination of features, namely "proactive frequency scanning + dual-threshold RSSI parent selection + flooding neighbor discovery," overlaps with the technical means in many published patent documents, posing a risk of design circumvention when exercising rights in the future.

[0008] Therefore, a technical solution is needed that can form a complete low-power Mesh networking system that does not rely on constant GNSS power supply, multi-hop relative clock chains, or active flooding scanning. Summary of the Invention

[0009] The purpose of this invention is to provide a low-power Mesh networking system and method with intermittent satellite timing disciplined crystal oscillators, which solves the problems of severe clock drift, excessive power consumption of GNSS constant power supply, accumulation of multi-hop relative synchronization error, unnecessary wake-up due to single routing architecture, and power consumption increase and feature overlap caused by network flooding in the prior art. This invention suppresses the average power consumption of GNSS to the microampere level while ensuring the consistency of the timing of the entire network, which significantly extends the battery node endurance.

[0010] The technical solution adopted in this invention is as follows: A low-power mesh networking system with intermittent satellite timing disciplined crystal oscillators includes a coordinator and a plurality of routing nodes. The coordinator and each routing node are respectively equipped with a radio frequency transceiver unit, a local crystal oscillator timing unit and a GNSS positioning and timing unit. The radio frequency transceiver unit adopts industrial unlicensed low-power radio frequency, supports at least one unlicensed ISM frequency band, and has anti-interference radio frequency transceiver capability.

[0011] The GNSS positioning and timing unit is configured to work intermittently in a sleep-timed wake-up mode: during non-wake-up periods, the power supply to the radio frequency front-end is turned off while only the wake-up timer is kept running; during wake-up periods, the unit is powered on to capture satellite signals and outputs at least one standard 1PPS pulse and its corresponding UTC timestamp, and then re-enters sleep mode.

[0012] The local crystal timing unit contains a free-running counter (typically implemented by the SoC's GPT / timer array) that uses the local crystal as its clock source.

[0013] Its core lies in the fact that the routing node also has a crystal oscillator discipline correction module, which is configured as follows: Each time the GNSS successfully wakes up and outputs a valid 1PPS pulse, the rising edge of the 1PPS pulse triggers hardware capture, latching the current value CNT of the free-running counter. cap ; Calculate CNT cap The theoretical expected count value CNT corresponding to 1PPS expected Phase deviation ΔN = CNT cap -CNT expected The instantaneous frequency deviation (ppm) of the local crystal oscillator is calculated from the ΔN value of at least two consecutive 1PPS cycles (K≥4 in this example); and the compensation amount corresponding to the ppm is written into the frequency compensation register—the control quantity output by the register acts on at least one of the following: For subsequent counting cycles, insert or subtract a sub-count every N cycles (fractional tick injection / deletion), or Adjusting the digital control word (load-capacitance trimming) of the programmable load capacitor array connected in parallel with the crystal oscillator, or Adjust the fractional division control parameters of the internal clock divider of the RF SoC; This allows the free-running counter to run freely at a compensated, approximate nominal frequency during subsequent GNSS dormancy periods, rather than drifting at the original, uncompensated crystal oscillator frequency.

[0014] The system further runs a dual-routing table architecture: A backbone routing table (BRT) is generated by the coordinator through a set of route descriptors or by the coordinator. It describes the end-to-end forwarding path that spans multiple hops and is used for long-distance service frames (such as monitoring data reporting, downlink configuration, and alarm feedback). A proximity forwarding table (NFT) is autonomously constructed by the node locally based on the detected neighbor link quality. The maximum number of hops for each entry is no more than 5 hops. It is used for near-field control frames and local linkage frames (such as chain-like segment lighting and local alarm flashing). The BRT and NFT entry aging strategies and transmission queues are isolated from each other to prevent short-range high-frequency frames from crowding out long-range backbone bandwidth, and vice versa.

[0015] During the network entry phase, after powering on, the routing node tunes its radio frequency to the network entry channel with a fixed number and enters a pure receive state for silent listening. It does not send active scanning probe frames or beacon frames on this channel. Only when it receives an invitation frame broadcast by the coordinator on the network entry channel containing valid access credentials (whitelist information / key verification information) will it send a network entry request frame. The coordinator will then perform whitelist verification or key authentication to decide whether to allocate short address / time slot parameters and allow the node to join.

[0016] Both the routing nodes and the coordinator are configured with security engines, which perform integrity verification and encryption protection on network access credentials and service frames.

[0017] Furthermore, let F be the nominal number of counts per second for the free-running counter. nom The latch count value when the k-th 1PPS pulse arrives is CNT. cap，k The theoretical expected count value is CNT. expected，k Then the phase deviation ΔN k =CNT cap，k -CNT expected，k ; The crystal oscillator discipline correction module consists of K consecutive effective 1PPS pulses from ΔN1 to ΔN K According to Δppm=[(ΔN K -ΔN1) / (K-1)] / F nom ×10 6 Calculate the instantaneous frequency deviation, where K is the number of samples of consecutive effective 1PPS pulses and K≥4; After the compensation amount is written to the frequency compensation register, it is applied to the free-running counter in at least one of the following ways: per N trim Insert or subtract a subcount every counting cycle; or Adjust the digital control word of the programmable load capacitor array connected in parallel with the local crystal oscillator; or Adjust the fractional division parameters of the clock divider.

[0018] Furthermore, the crystal oscillator discipline correction module also maintains a discipline quality index: when the variance of the phase deviation ΔN of K consecutive valid 1PPS pulses is lower than the preset variance threshold and the number of valid GNSS satellites is not less than 4, the clock state of the routing node is marked as disciplined, and the disciplinary local clock drives the timing reference of the Mesh network; when the GNSS wake-up fails to acquire a valid 1PPS, the routing node continues to run using the compensation amount most recently written to the frequency compensation register and marks the clock state as free-running drift state.

[0019] Furthermore, each neighbor entry in the proximity forwarding table includes the neighbor node identifier, received signal strength, and last update time. Neighbor entries that are not refreshed within the preset neighbor timeout period are eliminated. In chain-like or near-one-dimensional deployment scenarios, routing nodes calculate the trunk axis projection position relationship based on their own GPS coordinates obtained from GNSS and the GPS coordinates of neighbor entries, distinguishing neighbors into upstream and downstream directions. Near-field control frames are only forwarded to neighbors determined to be downstream, and the default maximum forwarding depth can be configured to 1 to 5 hops.

[0020] Furthermore, the backbone routing table is centrally constructed by the coordinator: the coordinator collects the hop count, parent node and link quality indicators of each routing node layer by layer through control frames, generates a reverse forwarding tree with the coordinator as the root, and unicasts the backbone routing table entries of each routing node to the corresponding routing node through the distribution frames; the update trigger condition of the backbone routing table is a partial or full refresh triggered by a topology change event, rather than a periodic full network flood.

[0021] Furthermore, the channel number of the network access channel remains fixed during system operation; the coordinator periodically broadcasts an invitation frame containing the network identifier, the current working channel number, and a temporary random number on the network access channel; when constructing a network access request frame during silent listening, the routing node performs a hash operation on the device's unique identifier and encapsulates it with an encrypted verification field protected by a preset network key, and sends the network access request frame within the sending window indicated by the invitation frame; if the routing node does not receive any invitation frame after the network access listening period expires, the routing node shuts down the RF front-end, enters deep sleep, and delays retrying with a wake-up timer, supporting a network access cache with a maximum delay of 30 days.

[0022] Furthermore, the radio frequency transceiver unit is compatible with at least one of the SUB-1G band and the 2.4GHz ISM band; the system divides the network access channel into a fixed channel and a dynamically selected working channel; the network access channel always falls on a predefined fixed numbered channel, and the working channel is selected by the coordinator from the available channels in the current frequency band through energy detection ED; when the coordinator detects that the interference energy of the current working channel exceeds the limit, it triggers channel migration, and the migration notice is broadcast on the new channel. Nodes that have not joined the network can still be discovered because the network access channel number is fixed.

[0023] Furthermore, the security engine is a symmetric encryption engine, specifically an AES-128 encryption engine. The credential field of the network access request frame and the payload of the service frame are protected by AES-128 operations. The coordinator is configured with two levels of permission keys: operation and maintenance and administrator. The keys are periodically updated and distributed to each routing node.

[0024] A low-power mesh networking method for intermittent satellite timing discipline crystal oscillators includes the following steps: S1. After the routing node is powered on, it will adjust the radio frequency transceiver unit to the network access channel with a fixed number and enter a pure receive state to silently listen to the network access invitation frame broadcast by the coordinator. It will not send active scanning probe frames or beacon frames on the network access channel. S2. After receiving an invitation frame containing valid access credentials, the routing node sends a network access request within the sending window indicated by the invitation frame. The coordinator performs whitelist verification or key authentication. If successful, the routing node is allowed to join the network and the working channel parameters are notified. The S3 GNSS positioning and timing unit operates intermittently in a sleep-timed wake-up mode. During the wake-up period, it captures the standard 1PPS pulse and UTC timestamp of the satellite and then goes back to sleep. Each time it is successfully woken up, the crystal oscillator discipline correction module captures the latched count value of the local crystal oscillator free-running counter at the rising edge of the 1PPS pulse, calculates the phase deviation and estimates the instantaneous frequency deviation of the local crystal oscillator, and writes the corresponding compensation amount into the frequency compensation register so that the crystal oscillator runs at the compensated frequency during the sleep period. S4. The network operates a backbone routing table and a neighbor forwarding table. The backbone routing table is centrally built and distributed by the coordinator to describe end-to-end forwarding paths that span multiple hops. The neighbor forwarding table is built locally by the routing nodes based on the neighbors they hear, and the maximum number of hops is no more than 5. The aging policies and sending queues of the two types of routing tables are isolated from each other.

[0025] Further, in step S3, the compensation amount is applied to the local crystal oscillator timing unit in at least one of the following ways: per N trim Insert or subtract a sub-count in each counting cycle, adjust the digital control word of the programmable load capacitor array connected in parallel with the local crystal oscillator; when consecutive GNSS wake-ups fail to acquire a valid 1PPS, the routing node continues to operate with the most recent compensation amount and maintains near-field communication with a widened keep-alive protection band until the next wake-up is successful and it is re-tamed.

[0026] This invention constructs a crystal oscillator discipline closed loop of "1PPS phase deviation → ppm quantization → frequency compensation register feedback", enabling each distributed battery node to operate at the frequency after being pulled and corrected during the dormant period without maintaining constant GNSS power supply. This is equivalent to each node having its own intermittent miniature GPSDO. Its discipline mechanism is different from the shallow use of "reading UTC time overwrite counter" (the latter does not correct the crystal oscillator frequency, so the long-term drift rate remains unchanged), and also different from "multi-hop parent-child deviation transmission chain" (the latter is relatively synchronized, the error accumulates with the number of hops, and the root drift is still unconstrained). Experiments and theoretical calculations show that an undisciplined ±20ppm crystal oscillator can produce a maximum cumulative deviation of about ±36ms in a 30-minute no-GNSS window. After being disciplined and pulled to ±2ppm, the cumulative deviation in the same window is reduced to within about ±3.6ms, which is sufficient to maintain no missed reception in the dormant window based on coarse time slot alignment.

[0027] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. A low-power Mesh networking system and method for intermittent satellite timing discipline crystal oscillators, which constructs a crystal oscillator discipline closed loop of "1PPS phase deviation → ppm quantization → frequency compensation register feedback", so that each distributed battery node can still operate at the frequency after being pulled and corrected during the dormant period without maintaining constant GNSS power supply, which is equivalent to each node having its own intermittent miniature GPSDO.

[0028] 2. In this invention, the tamed clock provides a distributed absolute reference for the Mesh timing plane. Mechanisms such as route keep-alive timeout, sleep window alignment, directional time slot advancement, and micro-time slot OTA thus achieve long-term stability and reduce routing oscillations and additional power consumption caused by accidental link disconnection.

[0029] 3. In this invention, dual routing tables isolate and decouple the scheduling of long and short-distance services: BRT is updated by coordinator events (non-periodic flooding) for multi-hop backbone, and NFT is maintained by local passive listening for near-field services with ≤5 hops. The queues of the two are isolated, which reduces unnecessary wake-up times and the probability of air interface conflicts from the architecture level.

[0030] 4. In this invention, silent listening + fixed network access channel + invitation system completely eliminates power-on active frequency scanning and flood detection, significantly reduces peak power consumption during network access, and the network access behavior characteristics are clearly distinguished from the publicly disclosed patented technology routes of "active scanning + dual-threshold RSSI parent selection + flood neighbor discovery".

[0031] 5. In this invention, while reducing the average power consumption of GNSS to the microampere level, the distributed nodes maintain a clock accuracy sufficient to support the timing consistency of the Mesh over a long period of time, which significantly extends the battery life of field battery nodes and improves the robustness of the network. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments, experimental examples, and comparative examples will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is an architecture diagram of the system of the present invention; Figure 2 This is an architecture diagram of the crystal oscillator discipline correction module in the system of this invention; Figure 3 This is an architecture diagram of the proximity forwarding table in the system of this invention; Figure 4 This is an architecture diagram of routing nodes and neighbor forwarding tables in the system of this invention; Figure 5 This is an architecture diagram of the backbone routing table and coordinator in the system of this invention; Figure 6 This is an architecture diagram of the network access phase in the system of this invention; Figure 7 This is an architecture diagram of the radio frequency transceiver unit in the system of this invention; Figure 8 This is an architecture diagram of the routing nodes and coordinator in the system of this invention; Figure 9 This is a flowchart of the method of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings, embodiments, experimental examples, and comparative examples. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described embodiments are merely some embodiments of the invention, and not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0034] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0035] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0036] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0037] I. Implementation Examples Example 1: Chain deployment of municipal streetlights / highway This invention discloses a low-power mesh networking system and method for intermittent satellite timing discipline crystal oscillators, such as... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown, the specific implementation method of this embodiment is as follows: taking smart street lights deployed along one or both sides of the highway as a typical application scenario, the node spacing is about 20m to 50m, the total length can reach several kilometers to tens of kilometers, the nodes are powered by solar energy + storage battery or disposable lithium-ion battery, and are required to operate without maintenance for ≥1 year after single installation.

[0038] Hardware configuration: No specific chip model required. Table 1 lists the portable modules included in the routing node hardware platform. The coordinator shares the same hardware architecture as the routing node, but is equipped with larger non-volatile storage (for centralized maintenance of the entire network backbone routing table) and optional Ethernet / 4G backhaul interfaces. The coordinator can remain in GNSS sleep mode if it has constant power; if the coordinator is also battery powered, it will execute the intermittent wake-up discipline logic.

[0039] 2. GNSS intermittent wake-up and crystal oscillator discipline closed loop 2.1 Wake-up Scheduling The routing node records the timestamp T of the last successful taming. last With the number of consecutive failures F cnt Set the wake-up cycle T.wake_intvl (In this example, the factory default value is 1800s = 30min, which can be configured to 900s to 3600s via downlink commands).

[0040] Upon the timer's arrival, the SoC pulls the GPIO to enable power to the GNSS module, and the GNSS internally performs a hot / warm start capture. The capture timeout T is set. time out (Example time: 120s): If successful locking and continuous valid 1PPS output before timeout, proceed to the discipline sampling phase; otherwise, wake-up is deemed a failure, F cnt ++, shut down GNSS power and return to sleep mode to wait for the next cycle.

[0041] 2.21 PPS Edge Hardware Capture and Phase Deviation ΔN Calculation Configure Timer A to run in free mode (counting clock = high-frequency clock from crystal oscillator via PLL, assuming a nominal 1 second corresponds to a count value F). nom =48000000).

[0042] A 1PPS pulse is input to GPIO_EXTI (external interrupt), and the edge-triggered input capture or interrupt service routine reads the Timer A counter value CNT. cap，k .

[0043] Define the theoretical expected count value of the k-th PPS: CNT expected，k = (CNT) ref +k×F nom mod 2 32 CNT ref A snapshot of the reference counter taken during the last successful taming (in actual implementation, this can be replaced by "the actual count ΔCNT between two PPS intervals"). k =CNT cap，k -CNT cap，k-1 (To eliminate the absolute benchmark, more directly): ΔCNT k =CNT cap，k -CNT cap，k-1 (Processed using 32-bit modular arithmetic); Deviation indication: If ΔCNT k >F nom →Local speed is too slow (crystal oscillator frequency is too low); if ΔCNT k <F nom →The local speed is too fast (the crystal oscillator frequency is too high); In this example, K = 8 consecutive effective PPS are used to calculate the average period count: ΔCNT avg = (CNT) cap，K-CNT cap，1 ) / (K-1); Instantaneous frequency deviation ppm = ((ΔCNT) avg -F nom ) / F nom )×10 6 ; The sign and magnitude of this ppm value directly reflect the direction of frequency deviation of the local crystal oscillator under the current temperature / voltage conditions.

[0044] 2.3 Write the compensation amount to the frequency compensation register In this embodiment, the SoC provides a Clock Trim register (common in BLE / Zigbee industrial SoCs), which can be configured for fine-tuning steps of ±1 oscillator every N oscillation cycles. The mapping relationship is as follows: trim_step (register write value) = f (ppm), where f (·) is determined by the ppm-per-LSB conversion factor given in the chip datasheet (approximately 0.8ppm / LSB in the example, the direction of which is determined by the ppm sign).

[0045] The crystal oscillator discipline correction module writes the calculated trim_code into the Clock Trim register. This is equivalent to: fractional tick injection can be performed at the software layer, maintaining an accumulator acc, and handling each Timer A overflow cycle (or every N...) base (Count) Check acc += trim_err, and when acc ≥ threshold, perform a hidden +1 or -1 count correction and then acc -= threshold.

[0046] 2.4 Taming Quality Gating Calculate the variance var(ΔCNT) of K consecutive ΔCNTs, combined with the effective satellite number S. v : If var(ΔCNT) < σ 2 thresh (Example σ) 2 thresh Corresponding to approximately ±2 ppm RMS) and S v ≥4: Mark clk_state = DISCIPLINED and write the current trim_code to a non-volatile memory backup. otherwise: If clk_state = FREE_RUNNING_DRIFT, use the most recent backup trim_code (if any) and do not update trim.

[0047] The purpose of this gating is to reject erroneous pulls caused by excessive PPS jitter due to multipath / interference. Ultimately, the GNSS module is powered off, and Timer A enters the next long sleep cycle at the trimmed frequency.

[0048] Table 2 shows the technical effect data (measured range in the examples). 3. Implementation details of dual-routing table architecture 3.1 Backbone Routing Table (BRT) The coordinator is the root node (address 0x0001). The coordinator periodically or when a topology change event triggers a RouteProbe frame (TTL = maximum hop count). Upon receiving the frame, each node sends back its own {NodeID, LastHopID, RSSI, GPS_dist_from_root} via the "reverse path" (forwarding in descending order of hop count). Based on this, the coordinator builds a forwarding tree at the cost of RSSI-weighted hop count. path_cost(i) = Σ hop w1·hop_count+w2·f(-RSSI)+w3·GPS_distance_proj The coordinator generates a BRT entry {dest_id, next_hop_id, path_cost, gen_seq} for each node and sends it to each node as a BRT_Update unicast frame. Upon receiving the entry, each node uses gen_seq to determine whether to update it, thus preventing old frames from overwriting new tables.

[0049] BRT is used for: node to coordinator uplink (monitoring, alarms, heartbeats) and coordinator to node downlink (configuration changes, OTA triggers, network-wide broadcasts are copied and forwarded hop by hop via BRT).

[0050] BRT update triggering conditions include only: child node heartbeat timeout and disconnection, RSSI falling below the lower limit causing parent switch, new node joining the network, and non-periodic flooding.

[0051] 3.2 Proximity-based NFT forwarding Node in Ch work Upon receiving any valid Mesh frame, extract the sender ID and RSSI, and update the local NFT: NFT entry {neighbor_id, RSSI, last_seen, optional_GPS_lat, optional_GPS_lon}; Aging: If now-last_seen > T nbr_timeout (Example 120s~300s), entry deleted.

[0052] Independent NFT sending queue: Near-field control frames (such as "light up depth 3 hops") only check the NFT to determine unicast / limited broadcast, do not check BRT, and do not trigger BRT maintenance messages.

[0053] 4. Direction determination and directional advancement in chain-like scenarios Chain-like scenes additionally utilize GNSS coordinates: Each node records its own value (φ) upon successful taming. self , λ self ), and is carried in the optional extension header of the Mesh frame; The coordinator / engineering configuration provides the approximate azimuth θ of the main line. axis (Or the axis vector can be calculated from the coordinates of the two reference endpoints); Project the neighbor coordinates onto the axis: proj = (φ nbr -φ ref )·cosθ+(λ nbr -λ ref )·sinθ; proj > proj self →downstream; proj < proj self →upstream; When a "segment-by-segment lighting / linkage control" frame is received, it is forwarded only to downstream neighbors, with a maximum forwarding depth D. max ∈{1, 2, 3} is configurable for distribution. This mechanism reduces the potential O(N²) flooding to O(N) × D. max .

[0054] When GPS is temporarily unavailable (obstructed), it degenerates into "learning direction": the highest frequency of the k most recent source neighbor IDs in a frame is defaulted to upstream.

[0055] 5. Detailed process of silent monitoring of network access protocol Definition: Network access channel Ch in Fixed (Channel 16 within the selected frequency band in the embodiment), operating channel Ch work The coordinator (ED) will select the best candidate.

[0056] Router node power-on initialization: GNSS off → Timer started → RF tuning in →RX on, TX off (pure reception) →Start T listen Timer (e.g., 8s).

[0057] Coordinator in Ch in Periodically broadcast Invite frames (period such as 2 seconds, duration such as a 15-second window): Invite payload: {CoordinatorID, NetID, Ch work Nonce inv ,WinDuration,AuthTag}; Auth Tag=AES-CMAC(NetKey,CoordinatorID||NetID||Nonce inv ); Node at T listen Received Invite → Verify Auth Tag → Temporarily Store Ch work With Nonce inv →Construct JoinReq: Payload: {DevUID, HASH (DevUID||NetKey), Nonce inv Caps} Encryption verification field = AES-CMAC(NetKey, Payload) TX will randomly back off once (only once) within the WinDuration specified by Invite, and will still fire RX for the rest of the time. The coordinator receives the JoinReq → checks the whitelist (or only verifies the Auth Tag + DevUID hash) → if successful: Allocate a short address ShortAddr, generate JoinResp{ShortAddr, Ch work SyncBaseTime, T wake_intvl ,AuthTag'} The node receives JoinResp → writes ShortAddr → radio frequency jump Ch work →Entering normal operating state If T listen No Invite received after timeout: Node off RF → DeepSleep (T retry ), T retry By adopting exponential backoff (e.g., ×1.5), the maximum delay buffer can reach 30 days, enabling a construction mode where lights / sensors are installed first, and then the network is powered on and activated uniformly.

[0058] 6. Security and Encryption All Mesh service frames are protected by hardware AES-128 (ECB / CCM optional), with the key being a 128-bit NetKey issued by the coordinator; The device's unique UID is hashed and used as the network access identity credential; the plaintext UID is not transmitted over the air. Distinguish between the operation and maintenance key (allowing reading status / triggering reports) and the administrator key (allowing modification of topology / removal of devices / updating NetKey). The NetKey is rolled over monthly, and the old key → new key overlap window prevents disconnection.

[0059] Example 2: Two-dimensional campus / factory area mesh + cluster hybrid deployment This invention discloses a low-power mesh networking system and method for intermittent satellite timing discipline crystal oscillators, such as... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown, the specific implementation method of this embodiment is as follows: when the nodes are distributed in a two-dimensional plane (industrial park equipment monitoring, warehouse area sensor network), the direction determination no longer uses a single axis projection: In addition to RSSI, NFT neighbors record GPS coordinates, and nodes establish polar coordinates with themselves as the origin: neighbors are divided into sectors according to azimuth angle (e.g., one sector every 60°). Targeted forwarding degenerates into sector-restricted forwarding: control frames carry TTL≤2, and each node only unicasts to 1~2 of its best RSSI neighbors in the "forward sector" (depending on the direction of control intent), instead of broadcasting to the entire network; The BRT path_cost is supplemented with a GPS distance term: cost = α·hop + β·(–RSSI) + γ·great_circle_dist_to_coordinator, so that the parent selection is not only based on the number of hops or a single RSSI threshold, which can adapt to the non-uniform radio frequency attenuation caused by steel structure obstruction in the factory area.

[0060] The remaining GNSS discipline closed loop, silent network entry, and dual-table isolation are consistent with Example 1.

[0061] Example 3: Pure Crystal Oscillator Time-Based Degradation Mode under Long-Term GNSS Blockage This invention discloses a low-power mesh networking system and method for intermittent satellite timing discipline crystal oscillators, such as... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown, the specific implementation method of this embodiment is as follows: it is applicable to situations where nodes inside tunnels, nodes in deep ravines, or where continuous rain causes GNSS to be unable to locate for several days. When consecutive N fail_thresh(Example = 3~5 times) When all wake-ups fail, the node enters the FREE_RUNNING_DRIFT state; The last DISCIPLINED backup trim_code is still used (at least locking the crystal oscillator to a better frequency point after the last taming, which is far better than the original crystal oscillator that was not trimmed at the factory). The NFT keep-alive protection band is automatically doubled (e.g., widened from ±15ms to ±30ms), and the BRT hello cycle is appropriately lengthened to reduce the chain breakage false alarm rate; Once GNSS is restored (the node is moved to an open location or briefly taken out by maintenance personnel), it will re-enter DISCIPLINED upon successful wake-up, and the protection zone will retract.

[0062] This variation demonstrates that even without GNSS for an extended period, a node will not degenerate into completely unconstrained drift because the initial successful taming during the initial installation has established a frequency correction baseline, which is superior to the completely untamed scheme.

[0063] Example 4: Multi-ISM Band Anti-Interference Channel Migration This invention discloses a low-power mesh networking system and method for intermittent satellite timing discipline crystal oscillators, such as... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown, the specific implementation method of this embodiment is as follows: the system is configured with at least two of SUB-1G (433 / 868 / 915MHz family) and 2.4GHz, and the frequency band switching does not change the fixed number of the access channel Chin (e.g., regardless of the frequency band, Chin is always the same). in =16 The actual center frequency mapped to varies with the frequency band, but the protocol layer number remains unchanged.

[0064] Coordinator monitoring Ch work ED energy: If energy > Threshold duration > T jam Coordinator: Select new working channel Ch′ work ; In the current Ch in Broadcast Migration Frame {NetID, NewBand, Ch′ work ,Nonce,AuthTag}; After the already joined node receives the message, it will redirect to Ch′. work ; Unregistered nodes are always in Ch inSilently awaiting Invite / Migration, it will not "get lost" due to frequency band migration.

[0065] Example 5: Taming the Clock and Coordinating Lossless OTA Upgrades This invention discloses a low-power mesh networking system and method for intermittent satellite timing discipline crystal oscillators, such as... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown, the specific implementation method of this embodiment is as follows: the firmware image fragment number is 0...M-1, and the coordinator issues "OTA_START{ImgHash, M, Ch" through BRT. work "micro_slot_map" leverages the characteristic that all nodes in the network are aligned to UTC within ± milliseconds after taming: One second is divided into M micro-slots (each node is assigned one or more slot indices). The node only opens RX to receive fragments within its own slot window, and DeepSleep the rest of the time. The fragmented packets {seq, payload, CRC} are forwarded hop-by-hop along the BRT by the coordinator, and the receiving nodes write the CRC to the OTA buffer after verifying the CRC. All fragments are collected → ImgHash is verified → If CRC is OK, the Swap partition is restarted; otherwise, a NACK request is made to retransmit the list. During the upgrade, NFT near-field linkage is paused (business isolation) to prevent irrelevant frames from interspersing the upgrade air interface.

[0066] In the actual test, the 1MB firmware fragment arrival rate was ≥99.9%, and the total time was about 50~60s. During this time, the GNSS sleep-wake cycle was not affected (the taming can be done after waking up).

[0067] II. Comparative Example Comparative Example 1: Mesh Nodes Based on GNSS Constant Power Supply Technical solution: The node always keeps the GNSS module powered on, outputting 1PPS per second or reading UTC time to cover the local counter.

[0068] Drawbacks: Although the clock accuracy is high, the GNSS module continuously consumes 20-45mA of current, resulting in extremely short battery life (typically <6 months), which cannot meet the needs of long-term field monitoring. Furthermore, no frequency compensation is performed on the local crystal oscillator; once the GNSS connection is lost, the clock immediately drifts back to the original crystal oscillator.

[0069] The difference from Comparative Example 1: This invention wakes up the GNSS only for a very short time and maintains accuracy by taming the crystal oscillator rather than continuously reading time, thus resolving the contradiction between endurance and accuracy.

[0070] Comparative Example 2: Mesh Network Based on Multi-Hop Relative Synchronization Technical solution: The root node connects to GNSS, and the child nodes pass clock offset corrections (such as adjusting the local tick) level by level through the parent-child node tracing chain.

[0071] Drawback: Synchronization error accumulates with each hop. Assuming each hop introduces an error of ±5ms, the error between the 10th hop node and the root node can reach ±50ms. Furthermore, if the root node itself does not have a high-precision crystal oscillator, the entire network will still drift.

[0072] The difference from Comparative Example 2: In this invention, each node independently has a disciplined closed loop, does not rely on deviation correction of multi-hop transmission to maintain local accuracy, each node's clock is independent and highly accurate, and the network scale expansion is not limited by the accumulation of synchronization errors.

[0073] Comparative Example 3: Network Access Mechanism Based on Active Scanning Flooding Technical solution: After power-on, the node actively scans multiple channels, sends Probe Requests or beacon frames to probe the network, and selects a parent node based on the RSSI dual threshold.

[0074] Defects: In the early stages of network access, the air interface is filled with a large number of probe packets, resulting in serious collisions and retransmissions; the peak power consumption is extremely high; and the features highly overlap with existing patents, posing a risk of infringement.

[0075] The difference from Comparative Example 3: This invention adopts silent listening + coordinator invitation, and the node "only listens and does not speak" after power-on, which completely eliminates the air interface pollution caused by active scanning.

[0076] Comparative Example 4: Mesh Network with Single Routing Table Technical solution: All nodes maintain a unified routing table. Whether it is long-distance data reporting or near-field control linkage, the same table is queried and the same forwarding logic is used (usually flooding or RREQ).

[0077] Drawbacks: Near-field control frames (such as streetlights lighting up one by one) will flood the entire network, consuming a large amount of bandwidth and causing congestion and packet loss in long-distance services; and the routing table is updated frequently, increasing unnecessary wake-up and computational power consumption.

[0078] Difference from Comparative Example 4: This invention uses a backbone routing table (BRT) + neighbor forwarding table (NFT), physically isolates long and short services, and limits near-field services to local forwarding within 5 hops, which greatly reduces network load and latency.

[0079] III. Experimental Examples The experimental environment is set uniformly: Hardware platform: It adopts an industrial-grade RF SoC (48MHz main frequency), with an external ordinary ±20ppm passive crystal oscillator (undisciplined state), and the GNSS module is a typical low-power model (such as UBLOX MAX series or equivalent).

[0080] Power supply environment: Standard configuration includes a 3.7V 5000mAh lithium-ion battery (typical energy of approximately 18.5Wh).

[0081] Network topology: chain topology, with a node spacing of 30 meters and a total of 50 routing nodes.

[0082] Business model: Each node reports a monitoring data packet (approximately 50 bytes) containing GPS coordinates every 5 minutes.

[0083] Experiment Example 1: Comparison of Crystal Oscillator Discipline Accuracy and Time Drift Objective: To verify the effect of the GNSS intermittent wake-up discipline mechanism on the correction of local crystal oscillator frequency deviation.

[0084] method: Ten routing nodes were selected and operated in an outdoor environment at normal temperature (25℃±5℃). The GNSS wake-up cycle was set to 30 minutes. The phase deviation ΔN after each successful wake-up was recorded, the actual ppm value after taming was calculated, and compared with the original crystal oscillator deviation before taming.

[0085] Table 3 shows the test data for Experiment Example 1. Conclusion: By capturing phase deviation and writing to the frequency compensation register, this invention successfully improves the accuracy of ordinary inexpensive crystal oscillators to a level close to that of TCXOs (temperature compensated crystal oscillators), eliminates the cumulative drift under long-term operation, and ensures the stability of Mesh network timing (such as sleep window alignment).

[0086] Experiment Example 2: Outdoor Battery Endurance Test Objective: To verify the effect of GNSS intermittent hibernation and taming mechanisms on improving the overall equipment endurance.

[0087] method: Two groups of five nodes each were configured, employing the "GNSS constant power supply" (Comparative Example 1) and "intermittent wake-up and discipline" modes of this invention, respectively. The wake-up cycle was set to 30 minutes, the radio frequency transmission power to 10 dBm, and the daily service packet volume to 288 times (once every 5 minutes).

[0088] Table 4 shows the test data for Experiment Example 2. Conclusion: This invention enables GNSS to operate without constant power supply by employing a disciplined closed-loop system, requiring only operation during wake-up. This reduces average power consumption to the microampere level, resulting in a battery-powered node's endurance increased by more than 6 times, fully meeting the deployment requirements for unattended outdoor scenarios.

[0089] Experiment Example 3: Network Latency and Air Interface Load under Dual Routing Table Architecture Objective: To verify the technical effectiveness of isolating the backbone routing table (BRT) and the neighbor forwarding table (NFT).

[0090] method: Construct a 50-node chain network. Scenario simulation: The coordinator issues a "segment-by-segment lighting" control command (near-field service), while nodes upload monitoring data (long-distance service). Compare the performance of the "single routing table" and the "dual routing table" of this invention.

[0091] Table 5 shows the test data for Experiment Example 3. Conclusion: The dual-routing-table architecture effectively isolates long and short services, avoids ineffective flooding of control frames across the network, significantly reduces network latency and air interface conflicts, and improves network robustness under complex services.

[0092] Experiment Example 4: Interference Resistance and Power Consumption Performance of Silent Network Entry Mechanism Objective: To verify the superiority of silent listening combined with invitation-only network access over active scanning.

[0093] method: Simulate powering on 100 nodes in batches (20 nodes every minute). Compare the peak power consumption and collision rate of "active scanning + flooding beacon" and the "silent listening" method of this invention during the network access phase.

[0094] Table 6 shows the test data for Experiment Example 4. Conclusion: This invention eliminates the blind active scanning when nodes are powered on, and significantly reduces air interface congestion and device power consumption in the early stages of network access through the orderly invitation mechanism of the coordinator, making it particularly suitable for large-scale batch deployment.

[0095] Experiment Example 5: Taming Stability under Temperature Variation Objective: To verify the ability of the discipline mechanism to suppress crystal oscillator temperature drift over a wide temperature range.

[0096] method: The node was placed in a high and low temperature test chamber with a temperature range of -20°C to +70°C. The natural ppm drift of the undisciplined crystal oscillator and the actual operating ppm under the discipline mechanism of this invention were recorded.

[0097] Table 7 shows the test data for Experiment Example 5. Conclusion: Even in extremely cold or high temperature environments, although ordinary passive crystal oscillators drift drastically (more than ±30ppm), the local clock can still be locked within a high-precision range of ±2.5ppm through the intermittent discipline closed loop of this invention, verifying the applicability of this architecture in harsh outdoor environments.

[0098] The above descriptions are merely preferred embodiments, experimental examples, and comparative examples of the present invention, and are not intended to limit the scope of protection of the present invention. The present invention uses the standard 1PPS pulse obtained from intermittent GNSS wake-ups for closed-loop correction of local crystal oscillator frequency deviations (rather than simple time overwriting), enabling each battery-powered node to maintain a docile clock sufficient to support Mesh timing consistency with an average power consumption in the microampere range. Based on this, dual routing tables isolate long- and short-range services, and a silent listening invitation system replaces active flooding for network entry, forming a complete, low-power, and timing-robust self-organizing network solution. Those skilled in the art can make equivalent substitutions and optimizations to various parameters (wake-up period, ppm threshold, hop count limit, sector resolution, encryption mode) without departing from the structural framework of the present invention, all of which fall within the scope of protection of the present invention.

Claims

1. A low-power mesh networking system with an intermittent satellite timing discipline crystal oscillator, characterized in that, It includes a coordinator and a plurality of routing nodes. The coordinator and each of the routing nodes are respectively equipped with a radio frequency transceiver unit, a local crystal oscillator timing unit and a GNSS positioning and timing unit. The radio frequency transceiver unit adopts industrial unlicensed low-power radio frequency, supports at least one unlicensed ISM frequency band, and has anti-interference radio frequency transceiver capability. The GNSS positioning and timing unit is configured to work intermittently in a sleep-timed wake-up mode: during non-wake-up periods, the power supply to the radio frequency front-end is turned off while only the wake-up timer is kept running; during wake-up periods, the unit is powered on to capture satellite signals and output at least one standard 1PPS pulse and the corresponding UTC timestamp, and then re-enters sleep mode. The local crystal timing unit includes a free-running counter that uses a local crystal as its clock source; The routing node is also equipped with a crystal oscillator discipline correction module. The crystal oscillator discipline correction module is configured to: capture the latch count value of the free-running counter at the rising edge of the 1PPS pulse each time the GNSS positioning and timing unit successfully wakes up and outputs a valid 1PPS pulse, calculate the phase deviation ΔN between the latch count value and the theoretical expected count value corresponding to the 1PPS pulse, estimate the instantaneous frequency deviation of the local crystal oscillator from the phase deviation ΔN of at least two consecutive 1PPS cycles, and write the compensation amount corresponding to the instantaneous frequency deviation into the frequency compensation register of the local crystal oscillator timing unit so that the free-running counter runs at the compensated frequency during the subsequent GNSS sleep period; The system operates on a dual-routing-table architecture, including a backbone routing table and a neighbor forwarding table: the backbone routing table describes end-to-end forwarding paths spanning multiple hops and is used for long-distance service frames; the neighbor forwarding table is autonomously constructed locally by the routing nodes based on the quality of the detected neighbor links, and the number of hops in each entry does not exceed a preset short-distance limit, and is used for near-field control frames and local linkage frames; the aging policy and sending queue of the backbone routing table entries are isolated from those of the neighbor forwarding table entries. During the network entry phase, after the routing node is powered on, it adjusts the radio frequency transceiver unit to a fixed network entry channel and enters a pure receive state for silent listening. It does not send active scanning and probe frames on the network entry channel. It only sends a network entry request when it receives an invitation frame containing valid access credentials broadcast by the coordinator on the network entry channel. The coordinator then performs whitelist verification or key authentication to decide whether to allow the routing node to join the network. The routing node and the coordinator are each configured with a security engine, which performs integrity verification and encryption protection on network access credentials and service frames. Let the nominal 1 second of the free running counter correspond to a count number F nom , the latched count value when the kth 1PPS pulse arrives is CNT cap，k , the theoretically expected count value is CNT expected，k , then the phase deviation ΔN k = CNT cap，k - CNT expected，k ; The crystal oscillator taming correction module is composed of ΔN1 to ΔN K According to Δppm=[(ΔN K -ΔN1) / (K-1)] / F nom ×10 6 The instantaneous frequency deviation is calculated, wherein K is the sample number of continuous effective 1PPS pulses and K≥4. After the compensation amount is written into the frequency compensation register, it is applied to the free-running counter in at least one of the following ways: per N trim Insert or subtract a sub-count every counting cycle; Alternatively, adjust the digital control word of the programmable load capacitor array connected in parallel with the local crystal oscillator; Alternatively, adjust the fractional division parameter of the clock divider.

2. The low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, The crystal oscillator discipline correction module also maintains a discipline quality index: when the variance of the phase deviation ΔN of K consecutive valid 1PPS pulses is lower than a preset variance threshold and the number of valid GNSS satellites is not less than 4, the clock state of the routing node is marked as disciplined, and the disciplined local clock drives the Mesh network timing reference. When GNSS wake-up fails to acquire a valid 1PPS, the routing node continues to operate using the compensation amount most recently written to the frequency compensation register and marks the clock state as a free-running drift state.

3. A low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, Each neighbor entry in the neighbor forwarding table includes the neighbor node identifier, received signal strength, and last update time. If a neighbor entry is not refreshed within a preset neighbor timeout period, it is eliminated. In chain-like or one-dimensional deployment scenarios, the routing node calculates the trunk axis projection position relationship based on its own GPS coordinates obtained from GNSS and the GPS coordinates of neighbor entries, distinguishing neighbors into upstream and downstream directions. Near-field control frames are only forwarded to neighbors determined to be downstream, and the default maximum forwarding depth can be configured to 1 to 5 hops.

4. A low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, The backbone routing table is centrally constructed by the coordinator: the coordinator collects the hop count, parent node and link quality indicators of each routing node layer by layer through control frames, generates a reverse forwarding tree with the coordinator as the root, and unicasts the backbone routing table entries of each routing node to the corresponding routing node through sending frames; the update trigger condition of the backbone routing table is a partial or full refresh triggered by a topology change event, rather than a periodic full network flood.

5. A low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, The channel number of the network access channel remains fixed during system operation; the coordinator periodically broadcasts an invitation frame containing the network identifier, the current working channel number, and a temporary random number on the network access channel; when constructing a network access request frame during silent listening, the routing node performs a hash operation on the device's unique identifier and encapsulates it with an encrypted verification field protected by a preset network key, and the network access request frame is sent within the sending window indicated by the invitation frame; if the routing node does not receive any invitation frame after the network access listening period expires, the routing node shuts down the RF front-end, enters deep sleep, and delays retrying with a wake-up timer, supporting a network access delay cache of up to 30 days.

6. A low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, The radio frequency transceiver unit is compatible with at least one of the SUB-1G band and the 2.4GHz ISM band; the system divides the network access channel into a fixed channel and a dynamically selected working channel; the network access channel always falls on a predefined fixed numbered channel, and the working channel is selected by the coordinator from the set of available channels in the current frequency band through the energy detection ED; When the coordinator detects that the interference energy of the current working channel exceeds the limit, it triggers channel migration. The migration notice is broadcast on the new channel, and nodes that have not joined the network can still be discovered because the network entry channel number is fixed.

7. A low-power mesh networking system with intermittent satellite timing discipline crystal oscillator according to claim 1, characterized in that, The security engine is a symmetric encryption engine, specifically an AES-128 encryption engine. The credential field of the network access request frame and the payload of the service frame are protected by AES-128 operations. The coordinator is configured with two levels of permission keys: operation and maintenance and administrator. The keys are periodically updated and distributed to each routing node.

8. A low-power mesh networking method using an intermittent satellite timing discipline crystal oscillator, characterized in that, Includes the following steps: S1. After the routing node is powered on, it adjusts the radio frequency transceiver unit to the network access channel with a fixed number, enters a pure receive state and silently listens to the network access invitation frame broadcast by the coordinator, and does not send active scanning probe frames or beacon frames on the network access channel. S2. After receiving an invitation frame containing valid access credentials, the routing node sends a network access request within the sending window indicated by the invitation frame. The coordinator performs whitelist verification or key authentication. If successful, the routing node is allowed to join the network and the working channel parameters are notified. The S3 GNSS positioning and timing unit operates intermittently in a sleep-timed wake-up mode. During the wake-up period, it captures the standard 1PPS pulse and UTC timestamp of the satellite and then goes back to sleep. Each time it is successfully woken up, the crystal oscillator discipline correction module captures the latched count value of the local crystal oscillator free-running counter at the rising edge of the 1PPS pulse, calculates the phase deviation and estimates the instantaneous frequency deviation of the local crystal oscillator, and writes the corresponding compensation amount into the frequency compensation register so that the crystal oscillator runs at the compensated frequency during the sleep period. Let F be the nominal number of counts per second of the free-running counter. nom The latch count value at the arrival of the k-th 1PPS pulse is CNT. cap，k The theoretical expected count value is CNT. expected，k Then the phase deviation ΔN k =CNT cap，k -CNT expected，k ; The crystal oscillator discipline correction module consists of K consecutive effective 1PPS pulses from ΔN1 to ΔN. K According to Δppm=[(ΔN K -ΔN1) / (K-1)] / F nom ×10 6 The instantaneous frequency deviation is calculated, where K is the number of samples of consecutive effective 1PPS pulses and K≥4; After the compensation amount is written into the frequency compensation register, it is applied to the free-running counter in at least one of the following ways: per N trim Insert or subtract a sub-count every counting cycle; Alternatively, adjust the digital control word of the programmable load capacitor array connected in parallel with the local crystal oscillator; Alternatively, adjust the fractional division parameter of the clock divider; S4. The network operates a backbone routing table and a neighbor forwarding table. The backbone routing table is centrally constructed and distributed by the coordinator to describe end-to-end forwarding paths that span multiple hops. The neighbor forwarding table is constructed locally by the routing nodes based on the neighbors they hear, and the maximum number of hops is no more than 5. The aging policies and sending queues of the two types of routing tables are isolated from each other.

9. A low-power mesh networking method for intermittent satellite timing discipline crystal oscillators according to claim 8, characterized in that, In step S3, the compensation amount is applied to the local crystal oscillator timing unit in at least one of the following ways: per N trim Insert or subtract a sub-count in each counting cycle, adjust the digital control word of the programmable load capacitor array connected in parallel with the local crystal oscillator; when consecutive GNSS wake-ups fail to acquire a valid 1PPS, the routing node continues to operate with the most recent compensation amount and maintains near-field communication with a widened keep-alive protection band until the next wake-up is successful and it is re-tamed.

Images