MESH networks and network control methods and systems based on joint scoring

By using a network control method based on joint scoring, low-power nodes are dynamically selected and activated as emergency routes, which solves the stability problem of MESH networks under disaster events and enables cross-regional propagation of critical alarm data and ensures battery life.

CN123123872APending Publication Date: 2026-11-17SHENZHEN HEIMAN TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611294209.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

After a disaster event causes the mains circuit to fail from the backbone, the existing MESH network cannot effectively restore the cross-regional and multi-hop propagation of critical alarm data, and the existing recovery scheme is difficult to adapt to the actual impact range of the disaster and the real-time status of the nodes, resulting in insufficient network stability.

Method used

A network control method based on joint scoring is adopted. By integrating factors such as local security event intensity, neighbor alarm consistency and network fragmentation, the confidence level of disaster events is calculated. Low-power nodes with sufficient energy and coverage are selected as emergency routes, and the route carrying capacity is dynamically activated to ensure the propagation of critical alarm data.

Benefits of technology

During disasters, it automatically and stably restores the cross-regional transmission of critical alarm data, avoiding node energy waste and channel congestion, ensuring network stability and battery life, and adapting to rapid response in different disaster scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123872A_ABST
    Figure CN123123872A_ABST
Patent Text Reader

Abstract

This invention discloses a MESH network and a network control method and system based on joint scoring. Low-power nodes operate as low-power terminals under normal conditions. During a disaster, the confidence level of the disaster event is calculated by integrating at least one of the following: local security event intensity, neighbor alarm consistency, network failure rate, and power supply anomaly rate. The node then enters an emergency route candidate state. Candidate nodes selected through a hard threshold including a security task energy retention threshold are evaluated using a joint score containing energy margin and topology coverage gain indicators. Higher scores result in shorter competition waiting times. The node whose score expires first sends a route bearer declaration. Other candidate nodes decide whether to continue competing or cancel their competition based on the increased coverage gain. The winning node activates the route bearer and performs temporary multi-hop forwarding. Once the confidence level falls below the exit threshold, the node reverts to low-power operation. After the mains backbone fails, critical alarm data propagates rapidly via multi-hop propagation, ensuring the energy for node security tasks and avoiding energy consumption and congestion on all normally open routes across the network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of disaster event alarm technology, specifically relating to a MESH network and a network control method and system based on joint scoring. Background Technology

[0002] In smart home and building fire safety systems, battery-powered safety devices such as smoke detectors, heat detectors, and carbon monoxide detectors are generally networked using wireless mesh networks that support multi-hop forwarding, such as Zigbee and Thread. The communication backbone of these networks typically consists of AC-powered lighting fixtures, sockets, gateways, or dedicated routing nodes, which are continuously online to handle neighbor maintenance and data forwarding. Battery-powered alarm devices, in order to meet the requirement of several years of battery life, operate as low-power terminals, sending their own data only when periodically woken up or when an alarm is triggered, and do not relay data for other devices. However, in real fires, flames, high temperatures, or electrical faults often damage the mains power lines or distribution equipment first, causing gateways and a large number of conventional routing nodes to go offline simultaneously. Scenarios such as large-scale power outages, power distribution failures, and gateway disconnections also lead to similar consequences. At this time, although the battery alarm devices that are still alive can complete local detection and audible and visual alarms by relying on their own batteries, they cannot transmit alarm messages across rooms and floors or reach external communication outlets because they have lost their parent node and multi-hop forwarding backbone. The cascading alarm and remote notification capabilities that the networked alarm system should provide fail precisely at the moment when they are most needed.

[0003] One intuitive approach is to configure all battery alarm devices as always-on routing nodes. However, continuous reception, neighbor maintenance, route discovery, and data forwarding significantly increase average power consumption, making it impossible for alarm devices to meet battery life requirements. Furthermore, a large number of routing nodes operating simultaneously exacerbate channel contention, neighbor table bloat, and redundant forwarding. Therefore, the crux of the problem is not whether battery nodes possess routing capabilities, but rather under what conditions they should be activated, which nodes should be selected, how many nodes should be activated, for how long they should operate, and how to safely deactivate them after energy depletion or the disaster has subsided. Most existing power outage network recovery schemes rely on pre-designating routing nodes with backup power supplies, pre-storing primary and backup routing topologies, or re-electing root nodes after coordinator failure. Such static or single-criteria schemes are difficult to adapt to the actual impact range of disasters and the real-time status of nodes: the pre-designated backup nodes may happen to be located in the disaster area or have also lost power, and the pre-storage topology may not match the actual failure situation; if a simple strategy of using the strongest signal or the highest power as the route is adopted, the selected nodes often have overlapping coverage areas, are too concentrated in one location, or are far away from critical failure areas. Some are unable to take on relay duties because they are performing high-intensity alarms on their own, and some do not have enough energy to support forwarding while maintaining their own detection and alarm tasks. Taking on routing duties may impair their own security functions or cause them to suddenly fail due to energy depletion, triggering network fragmentation again; in addition, when the mains power fluctuates, the smoke concentration fluctuates near the alarm threshold, or the gateway repeatedly comes online, the nodes in such schemes are prone to frequently switching between taking on and not taking on routing duties, causing repeated network instability.

[0004] Therefore, after the main circuit fails due to a disaster, how to enable the surviving battery-powered nodes to automatically, in a controlled manner, and in an appropriate number to temporarily assume routing without compromising their own safety detection and alarm capabilities, so as to quickly restore the cross-regional and multi-hop propagation of critical alarm data, and to safely and stably withdraw from routing after the disaster is over or the energy decreases, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the shortcomings of the existing MESH network in terms of network stability after the main circuit fails due to a disaster event, thereby providing a MESH network and a network control method and system based on joint scoring.

[0006] A network control method based on joint scoring, applied to MESH networks, the method comprising: Low-power nodes integrate local security event intensity and at least one of neighbor alarm consistency, network failure rate, and power supply anomaly rate to calculate disaster event confidence. When the confidence level of the disaster event reaches the entry threshold, or the intensity of the local security event exceeds the emergency pass-through threshold, the low-power node enters the emergency route candidate state. Nodes entering the emergency route candidate state undergo hard threshold screening, which includes at least a security task reserve energy threshold screening: the node's remaining energy, after deducting the local security task reserve energy, still meets the energy requirements for undertaking the emergency route, where the local security tasks include local sensing and alarms; nodes that fail the hard threshold screening exit the emergency route candidate state. A joint score is calculated for candidate nodes selected by the hard threshold. The joint score includes at least an energy margin index and a topology coverage gain index. The energy margin index is determined based on the remaining energy of the candidate node after deducting the energy reserved by the local security task. The topology coverage gain index is determined based on the coverage added by the candidate node relative to the already activated emergency route set. Candidate nodes determine their own contention waiting time based on the joint score. The higher the joint score, the shorter the contention waiting time. The candidate node whose contention waiting time expires first sends a route bearer declaration. After receiving the route bearer declaration, other candidate nodes recalculate their own new coverage gain relative to the set of declared nodes. Candidate nodes whose new coverage gain is lower than the suppression threshold cancel the competition, while candidate nodes that can still provide new coverage continue to participate in the competition. The winning candidate node activates its routing capacity and enters emergency carrying state to provide multi-hop data forwarding for other nodes; When the confidence level of the disaster event is lower than the exit threshold, the node in the emergency bearing state exits the routing bearing and resumes the low-power terminal working state. The exit threshold is lower than the entry threshold.

[0007] Furthermore, the determination of neighbor alarm consistency includes: Nodes in alarm state send event summary messages, which include at least device identifier, alarm type, risk level, area identifier, serial number, and authentication tag; The node receiving the event summary message authenticates and deduplicates the message, and only counts the alarm reports with different device identifiers, successful authentication and unique serial numbers as valid neighbor reports; Neighbor alarm consistency is determined based on at least one of the following: the number of valid neighbor reports, the highest risk level among them, the dispersion of the areas involved, and the concentration of the reporting time.

[0008] Furthermore, when the intensity of the local security event does not exceed the emergency pass-through threshold, the low-power node enters the emergency route candidate state through the multi-source consistency path, that is, when the confidence of the disaster event reaches the entry threshold for multiple consecutive time windows, or when multiple independent neighbor nodes report the same disaster status, it enters the emergency route candidate state. The confidence level of the disaster event is also calculated based on trusted system trigger information, which includes at least one of fire linkage inputs verified by identity verification and replay checks, emergency commands signed by the gateway, and manual on-site triggers.

[0009] Furthermore, the local security task retention energy includes local alarm retention and local self-test retention; The hard threshold screening also includes at least one of routing capability threshold, local risk threshold, temperature and battery health threshold, protocol and security threshold and resource threshold, and each threshold in the hard threshold screening is a condition that cannot be compensated by the joint score. A node that does not meet any threshold will be removed from the emergency route candidate state. The routing capability threshold requires nodes to have the ability to continuously receive data, maintain neighbors, forward routes, and manage child nodes; the local risk threshold requires that the processor, signal acquisition, and radio frequency resources occupied by local sensing and alarm tasks do not exceed a preset upper limit; the temperature and battery health threshold requires that the node's temperature status, battery undervoltage status, battery internal resistance, and fault flags be within the range that allows it to continuously receive data; the protocol and security threshold requires nodes to have valid network identity, network keys, or emergency network authorization materials; and the resource threshold requires that at least one of the node's memory, forwarding queue, neighbor table, and child node capacity be higher than the corresponding minimum value.

[0010] Furthermore, the joint score also includes at least one of the following indicators: Link quality metrics are determined based on at least one of the following: bidirectional acknowledgment success rate, packet loss rate, and signal strength stability between candidate nodes and neighboring nodes within a time window; The spatial redundancy index is determined based on the degree of overlap between candidate nodes and activated emergency routes in at least one dimension: coverage, uplink path, physical area, and common fault domain. The higher the degree of overlap, the lower the spatial redundancy index. The resource availability metric is the minimum value among the candidate node's processor availability, memory availability, forwarding queue availability, and the available child node capacity availability. Risk penalty items are determined based on at least one of the following factors for candidate nodes: battery aging, temperature rise, local alarm task occupancy, frequency of historically undertaken routes, and expected relay load, and are used to deduct from the joint score.

[0011] Furthermore, the method also includes controlling the number of activated emergency routes: The number of target routes is determined based on the number of currently uncovered targets, the effective carrying capacity of a single emergency route, and the redundancy requirements of key areas, and the number of target routes is limited to between a preset minimum number of routes and a preset maximum number of routes; wherein, the uncovered targets are determined based on alarm devices that have lost their parent nodes, key areas that need to restore coverage, and configuration requirements, and the effective carrying capacity is determined based on at least one of the following: the available child node capacity of the candidate node, the queue processing capacity, and the channel occupancy; When the number of activated emergency routes reaches the target number of routes, or when coverage and redundancy requirements are met, the remaining candidate nodes cancel their competition. Furthermore, minimum coverage constraints are set for the regions divided by the network. When selecting activation nodes, the key regions and different network partitions are first guaranteed to have coverage, and then the remaining candidate nodes are activated based on the new coverage and energy consumption.

[0012] Furthermore, the method also includes dynamic replacement migration during emergency support: The node in emergency bearing state calculates the expected bearing time based on its remaining energy budget after deducting the energy reserved for local safety tasks, as well as the expected receiving load and forwarding load; When the estimated carrying time is lower than the minimum carrying time, or the energy margin index is lower than the backup threshold, or the intensity of the local security event increases and more alarm energy needs to be retained, the node currently in emergency carrying state broadcasts a backup request. The backup request carries the node's coverage area, child node summary, routing load and estimated exit time. Surrounding candidate nodes are re-evaluated and compete based on the coverage gain generated by taking over the coverage area. The winning substitute node is activated to carry the routing capacity first. The original bearer node and the substitute node enter a dual-active handover period: the original bearer node continues to forward messages and stops accepting new child nodes, the substitute node sends a bearer availability announcement to guide the child nodes to migrate in batches, the critical alarm source is migrated first and after it is confirmed that it has been successfully transmitted through the new path, the original bearer node releases routing resources and restores the low-power terminal working state.

[0013] Furthermore, for a node in emergency bearing state to exit routing bearing, it must simultaneously meet the following conditions: the confidence level of the disaster event is lower than the exit threshold for multiple consecutive time windows, the regular routing nodes and gateways have recovered and stabilized, the critical alarm data has been propagated, and the duration of emergency bearing has reached the minimum maintenance time. The exit process includes: freezing new child node access, notifying existing child nodes to migrate to regular routing nodes or other emergency routing nodes; after confirming that there are alternative paths for critical nodes, reducing the forwarding queue length and receive duty cycle, and entering observation mode; During the observation period, if the confidence level of the disaster event reaches the entry threshold again, the route bearer is directly restored without re-engaging in contention; after the observation period ends, the temporary route state and neighbor state are cleared, and normal low-power operation is restored.

[0014] A network control system based on joint scoring, applied to a MESH network, wherein the system is deployed in low-power nodes of the MESH network, comprising: The confidence calculation module is used to calculate the confidence level of a disaster event by integrating the intensity of local security events and at least one of neighbor alarm consistency, network failure degree, and power supply anomaly degree. The hard threshold filtering module is used to perform hard threshold filtering on the node when the confidence level of the disaster event reaches the entry threshold or the intensity of the local security event exceeds the emergency pass threshold. The hard threshold filtering includes at least the security task reserve energy threshold filtering: the node's remaining energy, after deducting the local security task reserve energy, still meets the energy requirements for undertaking emergency routing. The local security tasks include local sensing and alarm. The joint scoring module is used to calculate a joint score for the node that passes the hard threshold screening. The joint score includes at least an energy margin index and a topology coverage gain index. The energy margin index is determined based on the node's remaining energy after deducting the energy reserved by the local security task. The topology coverage gain index is determined based on the coverage added by the node relative to the activated emergency route set. The contention activation module is used to determine the contention waiting time of this node based on the joint score. The higher the joint score, the shorter the contention waiting time. When the contention waiting time expires, a route bearer declaration is sent. After receiving the route bearer declarations of other candidate nodes, the module decides whether to continue the contention or cancel the contention based on the new coverage gain of this node relative to the set of declared nodes. The bearer management module is used to activate the routing bearer capability after winning the competition, enter the emergency bearer state to provide multi-hop data forwarding for other nodes, calculate the expected bearer duration based on the energy budget after deducting the energy reserved by the local security task from the remaining energy and the expected load during the bearer period, broadcast a substitute request when the expected bearer duration is lower than the minimum bearer duration to trigger the substitute node to activate first, and the local node to exit after dual-active handover, and exit the routing bearer and restore the low-power terminal working state when the disaster event confidence is lower than the exit threshold of the entry threshold.

[0015] A MESH network includes a battery-powered low-power node and at least one of a conventional routing node and a gateway; the low-power node includes at least one of a smoke alarm, a temperature alarm, a carbon monoxide alarm, a water immersion sensor, and a door / window sensor, and is configured to perform the method described above. In this process, the low-power nodes operate as low-power terminals during normal operation and do not forward data to other nodes. After a disaster event triggers the emergency routing process, low-power nodes that have been activated for routing capacity through hard threshold screening, joint scoring, and competitive activation enter the emergency carrying state and temporarily undertake multi-hop data forwarding. After the disaster event is resolved, the activated low-power nodes exit the emergency carrying state and resume low-power terminal operation.

[0016] Beneficial effects: This invention enables the MESH network to automatically select nodes from the surviving low-power nodes to temporarily assume routing responsibility after the main circuit fails due to a disaster event, thus restoring the cross-regional and multi-hop propagation of critical alarm data; while in normal operation, all low-power nodes maintain low-power terminal mode, and routing responsibility is only temporarily assumed by a few nodes during disasters, so the overall power consumption level and battery life indicators of the network are not affected.

[0017] By integrating the intensity of local security events with at least one of the following factors—neighbor alarm consistency, network instability, and power supply anomaly—the confidence level of disaster events is calculated, avoiding false triggering of route activation by a single network failure or a single sensor spike. When the intensity of a local security event exceeds the emergency pass-through threshold, it can enter the emergency route candidate state without waiting for the accumulation of multi-source information. In general scenarios, it is triggered when the confidence level of the disaster event reaches the entry threshold, balancing the response speed of high-risk scenarios with the triggering stability of general scenarios. A security task energy retention threshold excludes nodes with insufficient remaining energy to simultaneously support their own sensing and alarm tasks and route carrying during the candidate screening phase. This mechanism ensures that nodes undertaking routing will not lose their local security detection and alarm capabilities due to energy consumption during forwarding.

[0018] The joint score, composed of energy margin and topology coverage gain indicators, ensures that selected nodes not only have the energy to undertake routing but also provide additional coverage relative to the already activated emergency route set. This overcomes the problems of coverage duplication and location concentration caused by selecting nodes solely based on signal strength or battery level. After mapping the joint score to a contention waiting time, nodes with higher scores send route-bearing declarations first. Each candidate node autonomously competes based on its own conditions, without the need for centralized decision-making by a central node. This allows for route selection even in disaster scenarios where central nodes such as gateways also fail. Upon receiving a route-bearing declaration, the candidate node recalculates its additional coverage gain relative to the already declared node set and decides whether to continue competing or cancel the competition. This causes the number of activated nodes to automatically converge as the additional coverage is satisfied, forming a minimum emergency route backbone that meets propagation requirements. This avoids energy waste and channel congestion caused by all battery nodes simultaneously having routes open.

[0019] Setting the exit threshold to be lower than the entry threshold results in a hysteresis exit. Nodes in emergency bearing state will not frequently start or stop routing bearing due to fluctuations in disaster event confidence around the threshold. The network maintains stable role under disaster boundary conditions. After exiting, the node resumes low-power terminal operation. The additional energy consumption brought by routing bearing is limited to the duration of the disaster. This allows the network to simultaneously obtain reliable alarm propagation during the disaster and ensure battery life under long-term operation. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the overall architecture of the normal low-power network and the disaster emergency routing bearer network of the present invention; Figure 2 This is a schematic diagram of the dual-mode protocol stack and hardware functions of the low-power node of the present invention. Figure 3 This is a schematic diagram of the disaster event confidence calculation and emergency mode triggering process of the present invention; Figure 4 This is a schematic diagram of the hard threshold and energy-topology joint scoring algorithm for emergency route candidate nodes according to the present invention; Figure 5 This is a schematic diagram illustrating the timing of distributed contention, suppression, and route bearer activation of candidate nodes in this invention. Figure 6 This is a schematic diagram of the topology changes of the main circuit before and after failure and the replacement by emergency nodes in the present invention; Figure 7 This is a schematic diagram of the emergency routing energy budget, dynamic replacement, and hysteresis exit state machine of the present invention; Figure 8 This is a schematic diagram illustrating the implementation path for activating routing capacity under different MESH protocols according to the present invention. Detailed Implementation

[0022] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0023] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0024] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "joining," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0025] Example 1: This embodiment provides a network control method based on joint scoring, applied to a MESH network. The MESH network can be Thread, Matter over Thread, Zigbee, Sub-G private MESH, a custom network based on IEEE 802.15.4, or other wireless networks supporting multi-hop forwarding. Low-power nodes in the MESH network can be smoke detectors, temperature detectors, carbon monoxide detectors, water immersion sensors, door and window sensors, or other battery-powered safety devices. During normal operation, these low-power nodes function as low-power terminals, not forwarding data to other nodes, and only handling local sensing and alarm functions. Neighbor maintenance and data forwarding are handled by a communication backbone composed of AC-powered lights, sockets, gateways, or dedicated routing nodes. In this state, the overall network power consumption is low, and battery-powered devices can meet the lifespan requirement of several years.

[0026] like Figure 1 As shown, after a disaster, the network does not activate routing for all battery nodes simultaneously; instead, it only activates nodes that can increase coverage and redundancy. The main terms and state definitions used in this embodiment are as follows:

[0027] In step 1, the low-power node integrates the intensity of local security events with at least one of the following: neighbor alarm consistency, network breakage, and power supply anomaly, to calculate the confidence level of the disaster event. The node collects multiple types of inputs using a sliding time window to avoid triggering network reconstruction based solely on a single packet loss or a single sensor spike; each input is preferably calculated separately within a timestamped sliding window, with alarm windows ranging from 5 to 30 seconds and network and power supply status windows ranging from 10 to 60 seconds, and can be adjusted according to the product sampling period.

[0028] Local security incident intensity This reflects the disaster risk indicated by the node's own sensors. The node first calculates five sub-scores for smoke concentration, temperature level, temperature rise rate, CO concentration, and abnormal duration. A two-point calibration method is used for continuous quantities: the sub-score is 0 when the measured value does not exceed the warning threshold, the sub-score is 1 when it reaches the formal alarm threshold, and the value is taken according to the proportion when it is between the two thresholds. If it exceeds the alarm threshold, it is still treated as 1.

[0029] The temperature rise rate is obtained by dividing the difference between the current temperature and the temperature 30-60 seconds ago by that time interval; the duration sub-score is obtained by dividing the abnormal duration by the confirmation time and is limited to 0-1. The basic implementation can be weighted by 35% smoke, 15% temperature level, 20% temperature rise rate, 20% CO, and 10% duration. If a certain sensor is missing, the weight of that sensor item will be proportionally allocated to the remaining valid items; when the device has entered a formal local alarm state, The threshold is directly increased to no less than 0.9, enabling nodes in alarm status to obtain a high local risk assessment immediately, laying the foundation for subsequent high-risk direct triggering.

[0030] Neighbor alarm consistency This reflects the extent to which a disaster is confirmed across multiple points in space. Neighbor node information can be obtained from two paths: normal network maintenance messages and authentication event messages during a disaster. During normal operation, nodes obtain neighbor identities, link quality, and recent online time from the neighbor table, parent-child relationship table, link-layer source address, periodic heartbeats, or check-in / polling responses from low-power devices. When devices are installed or added to the network, the gateway stores room, floor, or area binding relationships and issues compressed area identifiers to authorized nodes. During a disaster, nodes in alarm status send event summary messages, which include at least the device identifier, alarm type, risk level, area identifier, serial number, and authentication tag. Nodes receiving event summary messages authenticate and deduplicate them, counting only alarm reports with different device identifiers, successful authentication, and unique serial numbers as valid neighbor reports. Retransmitted packets from the same node are counted only once. Based on this, four sub-scores are calculated: the effective alarm neighbor count score is the smaller of "effective alarm neighbor count / confirmed neighbor count" and 1; the risk score is the highest value among all effective neighbor risk levels and mapped to early warning 0.4, alarm 0.7, and emergency 1.0; the regional dispersion score is the smaller of "number of different rooms or floors involved / number of confirmed areas" and 1; the time concentration score is 1 when the interval between the first and last reports does not exceed the synchronization window, and gradually decreases to 0 according to the excess ratio after exceeding the synchronization window. The above four items are combined at 40%, 25%, 20%, and 15% respectively. The number of confirmed neighbors can be 2-3, and the number of confirmed areas can be 2. After authentication and deduplication, repeated broadcasts from a single device cannot increase this input, thus suppressing false triggers caused by spoofing or repeated alarms.

[0031] Network Fragmentation This reflects the failure level of the backbone network. Each node calculates its parent node disconnection score, regular route offline ratio, network heartbeat loss score, and reachable path decline score: the parent node disconnection score is the smaller of "number of consecutive probe failures / number of acknowledgment failures" and 1; the regular route offline ratio is the number of regular routes determined to be offline within the window divided by the baseline number of regular routes before the disaster; the network heartbeat loss score is the smaller of "number of consecutive lost heartbeats / number of lost acknowledgments" and 1; the path decline score is "1 The value is "Currently accessible independent paths / Pre-disaster accessible independent paths" and is limited to 0-1. It can be synthesized at 30%, 25%, 25%, and 20%; if only one parent node switch occurs and a replacement parent node is found within one reconnection cycle, the upper limit of the parent node disconnection score is limited to 0.2, so that a single parent node switch is only considered a minor anomaly, and it is only raised to a higher level when multiple consecutive detection failures occur and the number of alternative paths decreases synchronously.

[0032] Power supply anomaly This reflects evidence of mains power system failure. Node statistics show the power outage or offline status of trusted devices with mains power attributes within the window: the mains power node power outage ratio is the number of confirmed power outage mains power nodes divided by the number of online mains power nodes before the disaster; the gateway or bridge power outage score is 1 when a certified power outage reason, undervoltage reset reason, or backup power switching flag is received, and 0.4 when only ordinary offline; the synchronous power outage score is determined based on whether the power outage time of multiple different areas is concentrated, and the higher value is taken when multiple nodes lose power simultaneously within 5-15 seconds. It can be synthesized by the following proportions: 50% power loss of mains nodes, 30% power loss of gateways / bridges, and 20% power loss of synchronization. When there is no evidence of mains power status, ordinary communication disconnection is not directly equated with power failure, so as to avoid misjudging communication disturbances as power supply disasters.

[0033] Trusted system trigger For discrete inputs: Fire alarm linkage inputs, gateway signature emergency commands, or on-site manual triggers are set to 1 only when the device identity is valid, the message authentication is successful, the serial number or frame counter has not rolled back, the timestamp is within the valid window, and the service type belongs to the allowed emergency type; inputs that are authenticated but have insufficient time freshness and require secondary confirmation can be set to 0.5; inputs that fail any security check are set to 0, and the reason for rejection is recorded. Improving disaster event assessment is not enough; it cannot bypass hard thresholds such as battery overheating, hardware failure, or key expiration.

[0034] Accidental Trigger Penalty This is derived from evidence that contradicts actual disasters. The system sets penalty values ​​for common interferences and adds them up to a cap: a rapid increase in humidity within a short period while temperature, CO, and neighbor alarms are all normal is judged as steam interference, increasing by 0.25; a single parent node switch followed by a rapid recovery of the alternative path increases by 0.15; all neighbor reports originating from the same device are retransmitted, increasing by 0.20; sensor self-test anomalies or significant data distortion increase by 0.20; and construction, testing, or maintenance modes have been authorized and enabled, increasing by 0.30. Take the smaller value between the cumulative result and 0.6; when the machine has entered the strong sound and light alarm or fire linkage command authentication is valid, the penalty item shall not reduce the final confidence level below the high-risk direct access threshold to prevent interference with evidence from offsetting the judgment of real disaster.

[0035] After all inputs are calculated, they are combined into a disaster event confidence score: ; In the formula For the confidence level of disaster events, These are, respectively, the intensity of local security events, neighbor consistency, network fragility, power supply anomaly, and trusted system triggering. For the corresponding weights, For accidentally triggering the penalty item, This indicates that the calculation results will be limited to between 0 and 1. Before fusion, each input is normalized to 0-1 according to the product calibration range. Normalization is preferably implemented using a segmented mapping or lookup table method of "normal lower limit - confirmation upper limit" to avoid relying on complex floating-point operations: the original quantity is recorded as 0 when it is below the normal lower limit, as 1 when it is above the confirmation upper limit, and increases step by step according to the interval position when it is in between; discrete states are mapped to several fixed levels according to normal, mild abnormality, severe abnormality, and confirmed abnormality. In the engineering implementation, the original value, mapping level, and triggering reason of each input can be saved for easy retesting and adjustment.

[0036] Weights are not required to be solved continuously during operation, but can be dynamically switched by a pre-defined event configuration table. The system should at least set the following weight configurations: "Local Fire Confirmation," "Multi-point Alarm," "Network Backbone Failure," "Large-area Power Outage," and "Trusted Linkage Mandatory." When the local smoke, temperature rise, or CO has reached the formal alarm conditions, the local safety event weight should be increased. When the local risk is low but multiple neighbors in different areas alarm in unison, the neighbor consistency weight should be increased. When the gateway and multiple regular paths fail simultaneously, the network failure weight should be increased. When multiple mains power nodes lose power simultaneously, the power supply anomaly weight should be increased. When a certified fire linkage command is received, the trusted system trigger weight should be increased, but hardware capability, energy, and temperature hard thresholds should still be enforced.

[0037] As an example, this machine confirms that the fire configuration can be set. The configuration values ​​for multi-point neighbor alarms can be set to 0.25, 0.40, 0.15, 0.10, and 0.10; the configuration values ​​for network backbone failures can be set to 0.15, 0.20, 0.45, 0.10, and 0.10; the configuration values ​​for large-scale power outages can be set to 0.15, 0.15, 0.20, 0.40, and 0.10; and the configuration values ​​for trusted fire alarm linkage can be set to 0.10, 0.10, 0.10, 0.10, and 0.60, with the sum of all positive weights being 1. These values ​​are examples and can be adjusted using test data.

[0038] Weight switching in each window first determines the primary event type: Select "Local Fire Confirmation" when the local machine has reached the formal alarm threshold; select "Multi-point Neighbor Alarm" when at least two independent nodes and two areas report the same type of disaster within the synchronization window; select "Network Backbone Failure" when the network key trips and two or more regular paths fail simultaneously; select "Large-area Power Outage" when the offline ratio of mains nodes exceeds the set value and the gateway shows evidence of undervoltage or power failure; select "Trusted Fire Linkage" when the fire linkage input is valid. When multiple conditions are met simultaneously, the priority order is trusted fire linkage, local fire confirmation, multi-point alarm, large-area power outage, and network failure. New configurations require two consecutive windows to be valid before switching, and at least three windows must be maintained after switching to avoid frequent weight jumps under critical conditions. Each configuration and its applicable conditions can be calibrated in the laboratory using simulated scenarios such as steam, single-point false alarms, partial network outages, entire floor power outages, and real alarms.

[0039] In step 2, as Figure 3 As shown, when the confidence level of the disaster event reaches the entry threshold, or the intensity of the local security event exceeds the emergency pass-through threshold, the low-power node enters the emergency route candidate state. This step uses dual-path triggering to balance response speed and triggering stability. In the high-risk pass-through path, when the local fire risk exceeds the emergency pass-through threshold, or the local device has entered a strong audible and visual alarm state, it can enter the candidate selection without waiting for neighbor votes, ensuring rapid response in high-risk scenarios. In the multi-source consistency path, when the intensity of the local security event does not exceed the emergency pass-through threshold, the disaster event confidence level is required to be higher than the entry threshold for K consecutive windows. The system enters the candidate screening process only when at least M independent neighbors report the same disaster status, ensuring trigger stability under normal network fluctuations. Furthermore, the disaster event confidence level is calculated based on trusted system trigger information, which includes at least one of the following: fire-fighting linkage input verified through identity verification and replay checks, emergency commands signed by the gateway, and manual on-site triggers. The exit mechanism requires the disaster event confidence level to be below a lower exit threshold for multiple consecutive windows. and meet the minimum emergency hold time. And the network recovery time reached This, in conjunction with step 8, forms a hysteresis.

[0040] In step 3, nodes entering the emergency route candidate state undergo hard threshold screening. This hard threshold screening includes at least a security task reserve energy threshold screening: the node's remaining energy, after deducting the energy reserved for its own security tasks, still meets the energy requirements for undertaking emergency route support. The local security tasks include local sensing and alarm functions. Nodes failing the hard threshold screening exit the emergency route candidate state. The hard threshold represents a security condition that cannot be compensated for by high link quality or high coverage gain; if either condition is not met, the node cannot undertake emergency route support. Figure 4 As shown, hard thresholds are determined independently item by item; failure in any one threshold results in immediate removal from the candidate pool, and cannot be compensated by a higher link or coverage score. Threshold data must be marked with freshness; failures due to timeout or inability to read are treated as non-compliance, and the reason for removal is recorded. The determination methods for each hard threshold item and the handling of non-compliance are as follows:

[0041] Routing capability thresholds are verified through the device capability bitmap, protocol stack compilation and configuration, and startup self-test to confirm continuous reception, neighbor table, routing table, forwarding queue, and authentication functions are all available. Protocol and security thresholds check network identity, key validity period, frame counter, and emergency authorization materials.

[0042] The local safety reserve energy includes the local alarm reserve energy and the local self-test reserve energy. The calculation process for the safety reserve energy is as follows: first estimate the remaining available energy. Devices with coulomb counters directly read the remaining power. Devices without coulomb counters obtain a conservative value by looking up a table based on the battery open-circuit voltage, load voltage drop, internal resistance, and temperature. Then, the energy that the device must retain is calculated. The alarm retention amount equals the average power consumption during the alarm period multiplied by the minimum alarm duration required by regulations or product specifications. The self-test retention amount equals the energy consumption of a single complete self-test multiplied by the planned number of retention cycles. The safety margin can be taken as 10% to 30% of the sum of the first two items. Only... If, after deducting the aforementioned reserve amount, the energy consumption is still greater than the estimated energy consumption for one emergency carrying cycle, and the battery voltage is not lower than the routing undervoltage threshold, then the node passes the energy hard threshold. Since this threshold excludes nodes with insufficient energy during the candidate screening stage, the node undertaking the routing will not lose its local detection and alarm capabilities due to forwarding energy consumption.

[0043] The judgment method for local risk threshold, temperature and battery health threshold and resource threshold is as follows: The node counts CPU utilization, RF duty cycle, sensor sampling task timeout and alarm audio task usage in the last 1 to 5 seconds. If the CPU utilization continuously exceeds 80% to 90%, critical sensor tasks time out, the internal temperature of the casing or battery exceeds the allowable continuous receiving temperature, or the battery internal resistance increases by more than a set percentage compared to the factory baseline, it is judged as unsuitable to undertake routing. The specific threshold is determined by hardware temperature rise and battery type test. The free memory must be greater than the sum of the memory required for the new routing function of the protocol stack and the safety margin. The forwarding queue must be able to accommodate the packets expected to arrive within a peak alarm cycle. The neighbor table, routing table and child node table must each reserve at least one or a configured number of available slots. Any resource below the minimum value, statistical data exceeding the freshness time limit or reading failure will be treated as a hard threshold failure.

[0044] In step 4, a joint score is calculated for the candidate nodes selected through the hard threshold. The joint score includes at least an energy margin index and a topology coverage gain index. Each index is preferably obtained using existing protocol stack statistics and short-term probe results. All indexes retain their statistical window and update time; expired information is not included in the current score.

[0045] The energy margin index is determined based on the remaining energy of the candidate node after deducting the energy reserved for the local safety task: ; In the formula Candidate nodes Energy margin, For the remaining usable energy, Reserved quantity for local alarm. This is a reserve quantity for the machine's self-test. The routing budget is calculated by multiplying the continuous receive power consumption by the planned carrying duration, adding the expected number of forwarded packets multiplied by the energy consumption per packet sent, adding the expected number of received packets multiplied by the energy consumption per packet received, and adding the energy consumption estimate calculated by the protocol stack. A value less than 1 indicates that the node cannot complete a full planning cycle, and the system can reject the node or shorten its expected carrying time.

[0046] The topology coverage gain metric is determined based on the additional coverage gained by a candidate node relative to the set of activated emergency routes: ; In the formula Candidate nodes Topology coverage gain, This refers to the set of nodes or key areas already covered by existing emergency routes. Candidate nodes Coverable neighbor set, Let be the target set whose coverage needs to be restored, ∪ denotes the set union operation, and |·| denotes the number of elements in the set. Ensure that the denominator is not less than 1. Only statistics China belongs to And it has not yet appeared The goal is to proactively suppress candidate nodes that highly overlap with existing route coverage. The acquisition method is as follows: candidate nodes send authentication probes with random time slots, and the identity of the node receiving the response and the region identifier are formed. The activated emergency route carries an coverage summary in the bearer declaration, which is then merged to form... ; It consists of the currently disconnected alarm nodes, key stairwells or exits, and configuration requirements; nodes whose responses exceed the freshness time, authentication fail, or link confirmation rate is below the minimum value will not be allowed entry. .

[0047] The joint scoring also includes link quality metrics, spatial redundancy metrics, resource margin metrics, and risk penalty items. Link quality metrics are determined based on the bidirectional acknowledgment success rate, packet loss rate, and signal strength stability between the candidate node and its neighbors within a time window. ; In the formula Candidate nodes Link quality The score is for bidirectional accessibility. The packet loss rate of this candidate node. For RSSI stability score, The weights are used to calculate the total number of messages sent, the number of acknowledgments received, the total number of messages received, and the number of successful replies in the most recent N bidirectional probes or service messages. The bidirectional reachability score is the smaller value between the success rate of sending acknowledgments and the success rate of receiving replies. The RSSI stability score is equal to the number of failed messages divided by the total number of messages. It is also equal to 1 minus the ratio of the standard deviation of the RSSI within the window to the maximum allowable fluctuation, and is limited to 0-1. Up to 50% can be achieved in both directions, 1 PER accounts for 30%, and RSSI stability accounts for 20% in the synthesis; N can be 8 to 32 times. When there are insufficient messages, the data reliability is reduced. Single RSSI is not used as a substitute. Only when the single RSSI is high and the reverse confirmation is unstable can a high link score be obtained.

[0048] The spatial redundancy index is determined based on the degree of overlap between candidate nodes and activated emergency routes in at least one dimension: coverage, uplink path, physical region, and common fault domain. The higher the degree of overlap, the lower the spatial redundancy index. ; In the formula Candidate nodes Spatial redundancy, For the activated emergency route set, Candidate nodes With activated routes Overlap ratio. Coverage overlap can be calculated by dividing the number of shared coverage targets by the union of the coverage targets of both; when using the same parent node or the same unique uplink, path overlap is assigned a value of 1, otherwise it is assigned a value of 0 or a lower value; 1 is assigned for locations in the same room, 0.5 for adjacent rooms, and 0 for locations across floors; those powered by the same distribution circuit or in the same vulnerable location can be considered as a common fault domain. The four items can be combined to form the overlap ratio at 40%, 30%, 20%, and 10%. Take 1 minus the maximum overlap between the candidate node and any activated route. This is reduced when the candidate node is in the same room as an existing route and uses the same uplink path, and increased when it can provide different floors or different parent paths.

[0049] The resource availability metric is the minimum value among the candidate node's processor availability, memory availability, forwarding queue availability, and the available child node capacity availability. ; In the formula Candidate nodes The remaining resources. These represent the available capacity reserves for CPU, memory, forwarding queue, and child node capacity, respectively. CPU reserve is equal to the current idle rate minus the minimum allowed idle rate, divided by the difference between the good idle rate and the minimum allowed idle rate. Memory, queue, and child node reserves are all mapped and limited to 0-1 by subtracting the minimum safe amount from the current available amount, divided by the difference between the good amount and the minimum safe amount. For example, if CPU reserve is 0.8, memory is 0.7, queue is 0.5, and child node slots are 0.9, then... We set it to 0.5 to prevent the queue from becoming a bottleneck in the system.

[0050] Risk penalty items are determined based on at least one of the following factors for candidate nodes: battery aging, temperature rise, local alarm task usage, frequency of historical route assignments, and expected relay load. These factors are used to deduct from the joint score. Penalty sub-items are generated for poor battery health, temperature approaching limits, high local alarm task usage, frequent route assignments in the past, high overlap in coverage with activated nodes, and expected relay load exceeding the rated load. Each item is graded as no risk 0, mild risk 0.5, and severe risk 1, then multiplied by the configured weight and summed, with a limit of 0 to 1. In the example, the weights for battery aging, temperature rise, local tasks, historical route assignments, overlapping coverage, and excessive load can be set to 0.20, 0.15, 0.20, 0.10, 0.15, and 0.20, respectively.

[0051] Calculate the joint score based on the above indicators: ; In the formula Candidate nodes The combined score These are the metrics for link quality, topology coverage gain, energy margin, spatial redundancy, and resource margin. As a risk penalty item, Each weight is assigned. These weights do not need to be fixed; they can be increased in the early stages of a fire or during severe network disruptions. and Improved during prolonged power outages In multi-story residential buildings, improve Prioritize the formation of independent paths across floors. As a numerical example, a node measured... The current configuration weights are 0.25, 0.25, 0.20, 0.15, and 0.15 respectively, with a penalty weight of 0.20. Therefore, the node score is:

[0052] The score is only used for ranking candidates after passing the hard threshold and cannot compensate for failure to pass the hard threshold.

[0053] In step 5, candidate nodes determine their own contention waiting time based on the joint score. The higher the joint score, the shorter the contention waiting time. The candidate node whose contention waiting time expires first sends a route bearer declaration. Figure 5 As shown, the mapping method from rating to competition time is as follows: ; In the formula Candidate nodes The waiting time for competition, To minimize waiting time, The joint score for this candidate node. For the duration of the competition window, A small random jitter generated by the device's unique identifier is used to resolve collisions caused by similar or simultaneous transmissions of scores. The route bearer declaration must include at least the node identifier, score level, estimated bearer duration, capacity of child nodes that can be accommodated, coverage summary, and declaration sequence number. This step also includes two confirmation methods: centralized and decentralized. If the gateway, coordinator, or Thread Leader is still reachable, it approves the route bearer request based on the global topology. If the central node has failed, candidate nodes use distributed timeout confirmation. After sending a declaration, if a node does not receive a rejection or coverage declaration with a higher score covering the same area within the confirmation window, it automatically enters emergency bearer state and can request confirmation from at least a few neighbors to avoid accidental activation of isolated nodes. Therefore, even in disaster scenarios where the central node also fails, each candidate node can still autonomously complete route selection.

[0054] In step 6, after receiving the route bearer declaration, other candidate nodes recalculate their new coverage gain relative to the set of declared nodes. Candidate nodes with a new coverage gain lower than the suppression threshold cancel their competition, while candidate nodes that can still provide new coverage continue to participate in the competition. If the target number of routes is already met, candidate nodes also cancel their competition and return to the suppression-maintaining state. This step also includes controlling the number of activated emergency routes. Selecting only the highest-scoring node may still result in insufficient coverage or an excessive number of routes; therefore, target route quantity and coverage constraints are set. ; In the formula For the target number of routes, This represents the number of critical nodes that are currently not covered. This refers to the number of effective child nodes that a single emergency route can stably support under the current interference and load conditions. For the redundancy path requirement, To minimize the number of routes, This represents the maximum number of routes. Indicates rounding up. This indicates that the results will be restricted to... to between. Ensure that at least a basic core group exists. Limit the number of battery nodes that are always on at the same time. The method of determination is: to Each target in the list is compared with the current reachability table. If a target has no activated emergency route to provide two-way confirmation in the last two detection cycles, or if there is a route but the confirmation success rate is lower than the minimum value, it is recorded as not covered. Multiple ordinary nodes in the same room can be aggregated into a coverage unit according to the configuration, but smoke detectors, stairwells, exits and cross-floor relay points should be counted separately. Instead of directly using the maximum number of child nodes in the protocol stack theory, the slot capacity, queue capacity, and channel capacity are calculated separately: slot capacity is the number of slots currently available for child nodes; queue capacity is the number of packets that can be processed in a reporting cycle divided by the number of packets expected per child node in that cycle; and channel capacity is the allowable emergency channel occupancy time divided by the average occupancy time of a single child node. The minimum of these three values ​​is then multiplied by the nearest link confirmation success rate and rounded down. To allow for a margin for burst traffic, a safety factor of 0.7 to 0.9 can be multiplied, and this can be calibrated in type testing. Set according to critical areas: 0 for ordinary areas with no backup requirements, at least 1 for areas with a stairwell, a single exit, or a critical link across floors, and 1 for each of two independent critical areas that require backup. It must be at least 1, and no less than the number of currently detected independent network partitions; The minimum value among the following three factors is selected: the number of nodes that can be activated simultaneously according to the product configuration, the number of candidate nodes that pass the energy hard threshold, and the number of nodes that the channel capacity allows. In actual selection, the system first ensures coverage of critical areas and different network partitions, and then adds candidates in the order of maximum coverage and low energy consumption. The process stops once the coverage and redundancy requirements are met. If the maximum number of nodes still cannot cover all targets, the system prioritizes fire alarms, evacuation and fault messages, and reports insufficient coverage, rather than activating nodes indefinitely.

[0055] As an example of route count calculation, there are currently 9 key targets that are not covered. Type testing and real-time resource statistics demonstrate the stable carrying capacity of a single emergency route. The value is 4, so one redundant path needs to be added to the stairwell. System settings First, we calculate the value by rounding up to 9 / 4, which gives 3. Then, we add a redundancy of 1, resulting in a target quantity of 4. This value falls within the range of 2 to 5, so a maximum of 4 emergency routes can be selected in this round. If the third route already covers all targets and the stairwell redundancy is also satisfied, the process can be stopped early. The system can also divide the residence into floors, rooms, stairwells, or key exit areas, with a minimum coverage constraint set for each area. Even if a high-scoring node has already covered a large number of ordinary nodes, if it still cannot cover the stairwell or the far bedroom, the system still allows the second-highest-scoring node covering the key area to be activated. Node selection is thus transformed from a simple ranking problem into a minimum coverage set problem with energy constraints.

[0056] In step 7, the winning candidate node activates its routing capacity, enters emergency bearer state, and provides multi-hop data forwarding for other nodes, specifically accepting terminals that have lost their parent nodes to reattach or select new uplink paths. To avoid channel congestion caused by suddenly activating all protocol stack functions, activation proceeds in the following order: first, a continuous or high duty cycle receive window is opened to collect neighbor and broken topology information; then, neighbor tables, routing tables, broadcast deduplication tables, and finite-length forwarding queues are established or restored; next, a bearer availability announcement is sent, allowing terminals that have lost their parent nodes to reattach; then, queues are configured according to alarm service priorities, with fire alarms, cascading, and fault messages taking precedence over ordinary telemetry; finally, bearer health information is periodically sent, including remaining energy, queue occupancy, number of child nodes, and estimated duration. When a terminal reselects a parent node, not only link quality is considered, but also the remaining bearer budget of the emergency route, the current number of child nodes, and path independence are used as selection factors to avoid all terminals simultaneously flocking to a single high-power node. Figure 6 The diagram shows the topology changes of the main circuit before and after failure, as well as the replacement by emergency nodes.

[0057] During the carrying period, nodes in emergency carrying state calculate their estimated carrying time based on their remaining energy budget (after deducting the energy reserved for local security tasks) and the projected receiving and forwarding loads. The energy budget allowed for routing is: ; In the formula Candidate nodes Allowed energy budget for routing This is for temperature compensation margin.

[0058] The estimated duration of operation is: ; In the formula Based on the estimated duration of operation, To continuously receive power consumption, and These represent the estimated number of packets forwarded and received per unit time, respectively. and For single-packet transmission and reception of energy, Calculate power consumption for the protocol stack. This represents the additional power consumption of sensors during disaster conditions. The expected message rate can be updated exponentially based on the actual load in the most recent window. When the expected carrying time is less than the minimum carrying time, or the energy margin index is lower than the backup threshold, or the intensity of the local security event increases and more alarm energy needs to be retained, the node initiates a backup migration in advance, instead of waiting for the battery to run out of voltage and suddenly disconnect.

[0059] The dynamic replacement migration process is as follows: The node currently in emergency bearer state broadcasts a replacement request, which carries the node's coverage area, child node summary, routing load, and estimated exit time; surrounding candidate nodes are re-evaluated based on the takeover coverage gain generated by taking over the coverage area, and a shortened competition is conducted; the winning replacement node first activates its routing bearer capability; subsequently, the original bearer node and the replacement node enter a dual-active handover period. The original bearer node continues to forward existing alarm messages but stops accepting new ordinary child nodes, while the replacement node activates its routing bearer capability and broadcasts an availability announcement to guide the child nodes to migrate in batches. Critical alarm sources are prioritized for migration, and it is confirmed that they have been successfully sent at least a certain number of times via the new path. The original bearer node securely synchronizes the necessary deduplication sequence number, alarm source summary, or routing status to the replacement node. After the migration is completed or the maximum handover time is reached, the original bearer node releases routing resources and returns to a low-power state, retaining its local sensing and audible / visual alarm energy. By activating the replacement node first and then exiting the old node, the second network split caused by a sudden power outage of the routing node is reduced. Figure 7 The diagram shows the emergency routing energy budget, dynamic replacement, and hysteresis exit state machine.

[0060] Regarding protocol stack resource scheduling during the bearer operation, the emergency bearer state should avoid ordinary sensor telemetry occupying the alarm link, and multi-level queues can be set up:

[0061] The routing node can also dynamically adjust the capacity of child nodes based on queue occupancy. When the P0 / P1 queue remains at a high level, it stops accepting new ordinary child nodes and triggers neighboring candidate nodes to share the coverage.

[0062] This embodiment is adaptable to different protocol standards. For Thread or Matter over Thread, the Router Eligible End Device in Thread has the standard capability to be promoted to Router. During the normal phase, the node can be attached as a REED. After winning the score, it sends a Router upgrade request to the Leader, obtains the Router ID, establishes links to other Routers, and undertakes forwarding. If the Leader is unreachable but the Thread network still exists, it is executed according to the Leader recovery and Router selection mechanism of the Thread protocol itself, combined with the candidate activation conditions of this scheme. When exiting, this scheme controls the minimum hold time, energy replacement, and degradation timing. For Zigbee, since Sleepy End Devices typically cannot dynamically transform into Routers while maintaining their original online status, this solution does not consider changing the Zigbee node type in situ as the only implementation method. Possible technical paths include: pre-configuring node hardware and firmware with full Router capabilities, allowing controlled departure from the terminal identity during emergencies, and rapid re-entry into the original network as a Router using pre-authorized security materials; retaining the original Zigbee service identity while simultaneously launching a second network instance or an independent emergency overlay network, mapping the alarm source identity at the application layer, and having surviving bridging nodes communicate with the original network; for chips supporting multiple networks or protocols, the normal network maintains low-power terminals, while the emergency network instance undertakes continuous reception and relay; if the product is designed to allow regular Router identities, this solution can control its service carrying level, queue capacity, and emergency forwarding priority without violating the standard requirement that Routers cannot sleep. For Sub-G or private MESH, route bearer flags, neighbor tables, and forwarding queues can be directly defined. During normal operation, continuous reception is disabled or only time-slot listening is enabled. After an emergency node wins, high duty cycle reception, route forwarding, broadcast deduplication, and child node acceptance are enabled. Due to the controllable protocol, scoring backoff, coverage declaration, replacement migration, and controlled flooding can be fully implemented. For example... Figure 8 The diagram shows the implementation paths for activating routing capacity under different MESH protocols.

[0063] In terms of security, emergency route activation cannot be triggered solely by unauthenticated broadcasts; otherwise, attackers could potentially drain the battery by forging disaster messages. The system can employ the following measures: emergency trigger messages are authenticated using the original network key, device link key, Matter session security, or a separate emergency group key; high-risk passthroughs are only allowed to be triggered by local sensor status, and remote single messages cannot bypass the hard threshold; neighbor consistency only counts messages with different device identities verified through a replay counter; route bearer claims carry monotonic sequence numbers, event identifiers, and short-term valid authentication tags to prevent replay of old claims; when anomalies trigger frequently but lack local risk and power supply anomalies, the trustworthiness of the corresponding neighbors is reduced; exit and replacement migration also require authentication to prevent attackers from forcibly shutting down critical routes.

[0064] The following is a typical application scenario. In a three-story residential Zigbee smoke detector emergency coverage network scenario, the original gateway, smart sockets, and lights in the house act as Zigbee routers, and 12 battery-powered smoke detectors act as Sleepy End Devices. After a power distribution failure on the second floor, the gateway and multiple mains-powered routers go offline, and the smoke detectors on the second floor detect smoke and issue local alarms. Each smoke detector calculates based on its own alarm, neighbor alarms, and the parent node's disconnection. Among them, the second-floor corridor smoke detector S5, the stairwell smoke detector S6, and the third-floor corridor smoke detector S9 passed the hard threshold; S5 has high link quality but its coverage overlaps with S6, S6 covers the upper and lower floors and has a high energy margin, and S9 can cover the third-floor bedroom; after scoring, S6 declares its carrying capacity first, S5 exits due to low added coverage gain, and S9 still covers the unconnected area and is subsequently activated; the two nodes form a minimum emergency backbone, realizing the cascading propagation of the second-floor alarm to the third and first floors. In the Matterover Thread smoke detector node REED upgrade scenario, multiple combined smoke detectors with full-function Thread capabilities operate as REEDs during the normal phase. When the border router loses power and the number of neighboring Routers decreases rapidly, the nodes calculate the disaster confidence. The stairwell node with the highest score first applies for the Router role from the Thread Leader, obtains the Router ID, and establishes a routing link. Another remote node also obtains the Router role because it still has coverage gain; as the regular Router recovers, the nodes degrade sequentially after meeting the stability time and minimum hold time to avoid simultaneous exit. In a dynamic backup scenario during a prolonged power outage, emergency routing node A has been working continuously for 40 minutes. Its energy model predicts that it can only carry for 15 minutes, which is lower than the set minimum of 20 minutes. A broadcasts a backup request. Candidate nodes B and C are re-evaluated according to the takeover coverage gain. B wins and is activated first. A and B enter a dual-active window. The critical smoke detector sub-node migrates to B first and completes multiple alarm confirmations. Then, the ordinary nodes migrate. After confirming that the coverage has been restored, A exits the routing and returns to a low-power state.

[0065] The technical effectiveness of this embodiment can be verified through the following indicators:

[0066] The weights, thresholds, time windows, number of routes, and energy model parameters in this embodiment can all be calibrated according to the product, battery type, residential area, network protocol, and certification requirements, and should not be construed as limitations on the scope of protection. The above formulas can be replaced by lookup tables, piecewise functions, fuzzy rules, machine learning models, or other equivalent calculation methods, as long as the technical effect of controlled dynamic activation of route carrying capacity based on disaster events, energy, and topology information is achieved.

[0067] Example 2: This embodiment provides a network control system based on joint scoring, applied to a MESH network, wherein the system is deployed in the low-power nodes of the MESH network. For example... Figure 2 As shown, the emergency routing bearer control layer inside the low-power node is located between the security service layer and the protocol stack. It can obtain alarm status, power, link and resource information, and issue parameters such as role application, re-association, continuous reception, routing table activation, queue capacity and power control to different protocol stacks. The node's dual-mode protocol stack runs as a low-power terminal in the normal phase and as a routing configuration in the emergency bearer state.

[0068] The system includes a confidence calculation module, which is used to calculate the confidence level of a disaster event by integrating the intensity of the local security event and at least one of the following: neighbor alarm consistency, network failure degree, and power supply anomaly degree. The hard threshold filtering module is used to perform hard threshold filtering on the node when the confidence level of the disaster event reaches the entry threshold or the intensity of the local security event exceeds the emergency pass threshold. The hard threshold filtering includes at least the security task reserve energy threshold filtering: the node's remaining energy, after deducting the local security task reserve energy, still meets the energy requirements for undertaking emergency routing. The local security tasks include local sensing and alarm. The joint scoring module is used to calculate a joint score for the node that passes the hard threshold screening. The joint score includes at least an energy margin index and a topology coverage gain index. The energy margin index is determined based on the node's remaining energy after deducting the energy reserved by the local security task. The topology coverage gain index is determined based on the coverage added by the node relative to the activated emergency route set. The contention activation module is used to determine the contention waiting time of this node based on the joint score. The higher the joint score, the shorter the contention waiting time. When the contention waiting time expires, a route bearer declaration is sent. After receiving the route bearer declarations of other candidate nodes, the module decides whether to continue the contention or cancel the contention based on the new coverage gain of this node relative to the set of declared nodes. The bearer management module is used to activate the routing bearer capability after winning the competition, enter the emergency bearer state to provide multi-hop data forwarding for other nodes, calculate the expected bearer duration based on the energy budget after deducting the energy reserved by the local security task from the remaining energy and the expected load during the bearer period, broadcast a substitute request when the expected bearer duration is lower than the minimum bearer duration to trigger the substitute node to activate first, and the local node to exit after dual-active handover, and exit the routing bearer and restore the low-power terminal working state when the disaster event confidence is lower than the exit threshold of the entry threshold.

[0069] The specific working process of each module corresponds one-to-one with the steps in Implementation Example 1: The confidence calculation module performs step 1, which involves the acquisition of each input, fusion calculation, and dual-path trigger judgment; the hard threshold screening module performs step 3, which involves the judgment of each threshold and the recording of exit reasons; the joint scoring module performs step 4, which involves the calculation of each index and the synthesis of the joint score; the contention activation module performs step 5, which involves the mapping of contention waiting time and the sending of declarations, and step 6, which involves the judgment of suppression; the bearer management module performs step 7, which involves phased activation, multi-level queue scheduling, energy budget calculation, and substitute migration, and step 8, which involves exit control. The above modules can be implemented by the processor in the node executing the program instructions stored in the memory, or they can be implemented using dedicated logic circuits. Each module can be set up independently or in combination.

[0070] Example 3: This embodiment provides a MESH network, including a battery-powered low-power node and at least one of a conventional routing node and a gateway; the low-power node includes at least one of a smoke alarm, a temperature alarm, a carbon monoxide alarm, a water immersion sensor, and a door / window sensor, and is configured to perform the method described in Embodiment 1.

[0071] In this context, the low-power nodes operate as low-power terminals during normal operation and do not forward data for other nodes. The communication backbone is provided by conventional routing nodes and gateways powered by mains electricity. After a disaster event triggers the emergency routing process, low-power nodes that have been activated for routing capacity through hard threshold screening, joint scoring, and competitive activation enter the emergency carrying state and temporarily undertake multi-hop data forwarding. After the disaster event is resolved, the activated low-power nodes exit the emergency carrying state and resume low-power terminal operation.

[0072] like Figure 1As shown, after a disaster, the network does not allow all battery nodes to start routing simultaneously. Instead, it performs event judgment, candidate screening, joint scoring, distributed competition, and quantity suppression, activating only nodes that can increase coverage and redundancy. In this embodiment, the network maintains the periodic sleep of terminal nodes and their multi-year battery life during normal operation. During disasters, it automatically forms a minimum emergency backbone covering critical areas, improving alarm message delivery rate and path redundancy. It also ensures that nodes undertaking routing retain sufficient energy to perform local sensing, audible and visual alarms, self-testing, and minimum safe operation. Furthermore, it is compatible with Thread, Matter over Thread, Zigbee, Sub-G private MESH, and custom networks based on IEEE 802.15.4 through a protocol adaptation layer.

[0073] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0074] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A network control method based on joint scoring, applied to MESH networks, characterized in that, The method includes: Low-power nodes integrate local security event intensity with at least one of neighbor alarm consistency, network failure rate, and power supply anomaly rate to calculate disaster event confidence. When the confidence level of the disaster event reaches the entry threshold, or the intensity of the local security event exceeds the emergency pass-through threshold, the low-power node enters the emergency route candidate state. Nodes entering the emergency route candidate state undergo hard threshold screening, which includes at least a security task reserve energy threshold screening: the node's remaining energy, after deducting the local security task reserve energy, still meets the energy requirements for undertaking the emergency route, where the local security tasks include local sensing and alarms; nodes that fail the hard threshold screening exit the emergency route candidate state. A joint score is calculated for candidate nodes selected by the hard threshold. The joint score includes at least an energy margin index and a topology coverage gain index. The energy margin index is determined based on the remaining energy of the candidate node after deducting the energy reserved by the local security task. The topology coverage gain index is determined based on the coverage added by the candidate node relative to the already activated emergency route set. Candidate nodes determine their own contention waiting time based on the joint score. The higher the joint score, the shorter the contention waiting time. The candidate node whose contention waiting time expires first sends a route bearer declaration. After receiving the route bearer declaration, other candidate nodes recalculate their own new coverage gain relative to the set of declared nodes. Candidate nodes whose new coverage gain is lower than the suppression threshold cancel the competition, while candidate nodes that can still provide new coverage continue to participate in the competition. The winning candidate node activates its routing capacity and enters emergency carrying state to provide multi-hop data forwarding for other nodes; When the confidence level of the disaster event is lower than the exit threshold, the node in the emergency bearing state exits the routing bearing and resumes the low-power terminal working state. The exit threshold is lower than the entry threshold.

2. The method according to claim 1, characterized in that, The determination of the consistency of neighbor alarms includes: Nodes in alarm state send event summary messages, which include at least device identifier, alarm type, risk level, area identifier, serial number, and authentication tag; The node receiving the event summary message authenticates and deduplicates the message, and only counts the alarm reports with different device identifiers, successful authentication and unique serial numbers as valid neighbor reports; Neighbor alarm consistency is determined based on at least one of the following: the number of valid neighbor reports, the highest risk level among them, the dispersion of the areas involved, and the concentration of the reporting time.

3. The method according to claim 1, characterized in that, When the intensity of the local security event does not exceed the emergency pass-through threshold, the low-power node enters the emergency route candidate state through the multi-source consistency path. That is, when the confidence of the disaster event reaches the entry threshold for multiple consecutive time windows, or when multiple independent neighbor nodes report the same disaster state, the node enters the emergency route candidate state. The confidence level of the disaster event is also calculated based on trusted system trigger information, which includes at least one of fire linkage inputs verified by identity verification and replay checks, emergency commands signed by the gateway, and manual on-site triggers.

4. The method according to claim 1, characterized in that, The local security task reserve energy includes the local alarm reserve amount and the local self-test reserve amount; The hard threshold screening also includes at least one of routing capability threshold, local risk threshold, temperature and battery health threshold, protocol and security threshold and resource threshold, and each threshold in the hard threshold screening is a condition that cannot be compensated by the joint score. A node that does not meet any threshold will be removed from the emergency route candidate state. The routing capability threshold requires nodes to have the ability to continuously receive data, maintain neighbors, forward routes, and manage child nodes; the local risk threshold requires that the processor, signal acquisition, and radio frequency resources occupied by local sensing and alarm tasks do not exceed a preset upper limit; the temperature and battery health threshold requires that the node's temperature status, battery undervoltage status, battery internal resistance, and fault flags be within the range that allows it to continuously receive data; the protocol and security threshold requires nodes to have valid network identity, network keys, or emergency network authorization materials; and the resource threshold requires that at least one of the node's memory, forwarding queue, neighbor table, and child node capacity be higher than the corresponding minimum value.

5. The method according to claim 1, characterized in that, The joint score also includes at least one of the following indicators: Link quality metrics are determined based on at least one of the following: bidirectional acknowledgment success rate, packet loss rate, and signal strength stability between candidate nodes and neighboring nodes within a time window; The spatial redundancy index is determined based on the degree of overlap between candidate nodes and activated emergency routes in at least one dimension: coverage, uplink path, physical area, and common fault domain. The higher the degree of overlap, the lower the spatial redundancy index. The resource availability metric is the minimum value among the candidate node's processor availability, memory availability, forwarding queue availability, and the available child node capacity availability. Risk penalty items are determined based on at least one of the following factors for candidate nodes: battery aging, temperature rise, local alarm task occupancy, frequency of historically undertaken routes, and expected relay load, and are used to deduct from the joint score.

6. The method according to claim 1, characterized in that, The method also includes control over the number of activated emergency routes: The number of target routes is determined based on the number of currently uncovered targets, the effective carrying capacity of a single emergency route, and the redundancy requirements of key areas, and the number of target routes is limited to between a preset minimum number of routes and a preset maximum number of routes; wherein, the uncovered targets are determined based on alarm devices that have lost their parent nodes, key areas that need to restore coverage, and configuration requirements, and the effective carrying capacity is determined based on at least one of the following: the available child node capacity of the candidate node, the queue processing capacity, and the channel occupancy; When the number of activated emergency routes reaches the target number of routes, or when the coverage and redundancy requirements are met, the remaining candidate nodes cancel their competition. Furthermore, minimum coverage constraints are set for the regions divided by the network. When selecting activation nodes, the key regions and different network partitions are first guaranteed to have coverage, and then the remaining candidate nodes are activated based on the new coverage and energy consumption.

7. The method according to claim 1, characterized in that, The method also includes dynamic replacement migration during emergency support: The node in emergency bearing state calculates the expected bearing time based on its remaining energy budget after deducting the energy reserved for local safety tasks, as well as the expected receiving load and forwarding load; When the estimated carrying time is lower than the minimum carrying time, or the energy margin index is lower than the backup threshold, or the intensity of the local security event increases and more alarm energy needs to be retained, the node currently in emergency carrying state broadcasts a backup request. The backup request carries the node's coverage area, child node summary, routing load and estimated exit time. Surrounding candidate nodes are re-evaluated and compete based on the coverage gain generated by taking over the coverage area. The winning substitute node is activated first to carry the routing capacity. The original bearer node and the substitute node enter a dual-active handover period: the original bearer node continues to forward packets and stops accepting new child nodes, the substitute node sends a bearer availability announcement to guide the child nodes to migrate in batches, the critical alarm source is migrated first and after it is confirmed that it has been successfully transmitted through the new path, the original bearer node releases routing resources and restores the low-power terminal working state.

8. The method according to claim 1, characterized in that, A node in emergency bearing state must simultaneously meet the following conditions to exit routing bearing: the confidence level of the disaster event is lower than the exit threshold for multiple consecutive time windows, the regular routing nodes and gateways have recovered and stabilized, the critical alarm data has been propagated, and the duration of emergency bearing has reached the minimum maintenance time. The exit process includes: freezing new child node access, notifying existing child nodes to migrate to regular routing nodes or other emergency routing nodes; after confirming that there are alternative paths for critical nodes, reducing the forwarding queue length and receive duty cycle, and entering observation mode; During the observation period, if the confidence level of the disaster event reaches the entry threshold again, the route bearer is directly restored without re-engaging in contention; after the observation period ends, the temporary route state and neighbor state are cleared, and normal low-power operation is restored.

9. A network control system based on joint scoring, applied to a MESH network, characterized in that, The system is deployed in the low-power nodes of the MESH network and includes: The confidence calculation module is used to calculate the confidence level of a disaster event by integrating the intensity of local security events and at least one of neighbor alarm consistency, network failure degree, and power supply anomaly degree. The hard threshold filtering module is used to perform hard threshold filtering on the node when the confidence level of the disaster event reaches the entry threshold or the intensity of the local security event exceeds the emergency pass threshold. The hard threshold filtering includes at least the security task reserve energy threshold filtering: the node's remaining energy, after deducting the local security task reserve energy, still meets the energy requirements for undertaking emergency routing. The local security tasks include local sensing and alarm. The joint scoring module is used to calculate a joint score for the node that passes the hard threshold screening. The joint score includes at least an energy margin index and a topology coverage gain index. The energy margin index is determined based on the node's remaining energy after deducting the energy reserved by the local security task. The topology coverage gain index is determined based on the coverage added by the node relative to the activated emergency route set. The contention activation module is used to determine the contention waiting time of this node based on the joint score. The higher the joint score, the shorter the contention waiting time. When the contention waiting time expires, a route bearer declaration is sent. After receiving the route bearer declarations of other candidate nodes, the module decides whether to continue the contention or cancel the contention based on the new coverage gain of this node relative to the set of declared nodes. The bearer management module is used to activate the routing bearer capability after winning the competition, enter the emergency bearer state to provide multi-hop data forwarding for other nodes, calculate the expected bearer duration based on the energy budget after deducting the energy reserved by the local security task from the remaining energy and the expected load during the bearer period, broadcast a substitute request when the expected bearer duration is lower than the minimum bearer duration to trigger the substitute node to activate first, and the local node to exit after dual-active handover, and exit the routing bearer and restore the low-power terminal working state when the disaster event confidence is lower than the exit threshold of the entry threshold.

10. A MESH network, characterized in that, The device includes a battery-powered low-power node, and at least one of a conventional routing node and a gateway; the low-power node includes at least one of a smoke alarm, a temperature alarm, a carbon monoxide alarm, a water immersion sensor, and a door / window sensor, and is configured to perform the method as described in any one of claims 1 to 8. In this process, the low-power nodes operate as low-power terminals during normal operation and do not forward data to other nodes. After a disaster event triggers the emergency routing process, low-power nodes that have been activated for routing capacity through hard threshold screening, joint scoring, and competitive activation enter the emergency carrying state and temporarily undertake multi-hop data forwarding. After the disaster event is resolved, the activated low-power nodes exit the emergency carrying state and resume low-power terminal operation.