Self-organizing earthquake early warning method based on emergency information publishing system

Abstract

This invention discloses a self-organizing earthquake early warning method based on an emergency information dissemination system, specifically relating to the field of earthquake early warning technology. It acquires the initial feature vector of an earthquake event and broadcasts it to edge nodes. Each node constructs a local propagation weight and relay queue based on the received information, realizing the construction of a multi-hop self-organizing propagation path. During propagation, it dynamically monitors local reception density and relay delay, identifies signal attenuation areas, and triggers collaborative relay requests. The central node schedules available equipment to form temporary relay groups and automatically deploys hotspot communication protocols to improve information coverage and transmission reliability. Finally, it constructs a complete propagation topology and calculates the response timestamps of each node to evaluate system performance. This invention possesses high fault tolerance, rapid deployment capability, and good adaptive scalability, making it suitable for constructing emergency early warning systems in complex disaster scenarios.

Description

Technical Field

[0001] This invention relates to the field of earthquake early warning technology, and specifically to a self-organizing earthquake early warning method based on an emergency information dissemination system. Background Technology

[0002] Earthquake early warning systems play a crucial role in earthquake disaster prevention, especially in the epicenter and surrounding areas, providing valuable time for emergency responses such as evacuation, power outages, and traffic control. However, most current earthquake early warning systems rely heavily on fixed monitoring nodes and centralized information dissemination architectures, exhibiting the following significant shortcomings:

[0003] First, when ground infrastructure near the epicenter, such as communication towers, substations, and data centers, is damaged in the first wave of earthquake impact, the original earthquake early warning information links may be completely interrupted, making it impossible to transmit early warning information in a timely manner, which seriously affects the protection of life safety.

[0004] Secondly, in special environments such as traffic tunnels, underground parking garages, and valleys, traditional earthquake early warning systems have poor signal coverage, resulting in blind spots. Especially in environments with terrain obstruction and enclosed structures, early warning signals cannot spread rapidly, severely limiting the system's effectiveness.

[0005] Third, existing systems cannot achieve self-organized collaboration among terminal nodes within a region in the early stages of a disaster. The lack of a mechanism for rapid reconstruction, information dissemination, and consensus generation under conditions of localized network communication paralysis results in insufficient system response capabilities in the initial stages of an earthquake.

[0006] Therefore, a self-organizing earthquake early warning method based on an emergency information release system is proposed. This method can achieve multi-path, multi-hop information self-propagation and self-organizing response mechanism by relying on mobile devices and edge nodes in the region, under the condition that the traditional early warning system fails or is partially damaged, thereby ensuring the effective transmission of early warning information and coordinated response in the region. Summary of the Invention

[0007] The purpose of this invention is to provide a self-organizing earthquake early warning method based on an emergency information release system to address the shortcomings of the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a self-organizing earthquake early warning method based on an emergency information dissemination system, comprising:

[0009] After acquiring earthquake trigger signals and identifying them as earthquake events according to set rules, an initial feature vector of the earthquake event is constructed. ;

[0010] Will Broadcast to the edge node cluster within the event triggering area, and receive Subsequently, based on its geographical location, Li and the epicenter Calculate propagation delay And construct the local received feature vector Ei;

[0011] Each edge node generates an information propagation weight Wi based on the received Ei, and forms a local relay queue Qi according to the order of Wi. It then broadcasts E0 to neighboring nodes in order of priority, thus constructing the first-layer self-organizing information transmission path P1.

[0012] After path P1 is formed, each node monitors the information backhaul redundancy Rj between itself and its neighboring nodes. If Rj is lower than the threshold Θ, a redundant path P2 is generated, and P2 is used to supplement the broadcast warning information in parallel.

[0013] During information propagation, each node dynamically records the local reception density Di and relay delay δi. When Di is less than the set lower limit Dmin or δi exceeds the set upper limit δmax, the node will mark it as a signal attenuation area, initiate a local cooperative relay request, and report the request relay signal Si to the communicable central node C.

[0014] After receiving multiple Si, the central node C schedules available devices to form a temporary relay group G and automatically deploys the hotspot communication protocol.

[0015] After all warning signal paths have stabilized, a complete propagation topology map is generated, and the warning response timestamps for each node are calculated.

[0016] Preferably, the initial feature vector Including magnitude M, epicenter location Initial trigger time and preliminary wave velocity estimation .

[0017] Preferably, the feature vector Ei includes: magnitude, epicenter location, initial trigger time, local location, theoretical propagation delay, local reception timestamp, and difference error.

[0018] Preferably, the step of forming a local relay queue Qi based on Wi sorting includes:

[0019] Edge node Ni calculates the information propagation weights according to the parameters in its local received feature vector Ei, following a weighted model. The expression is: ;in, For the propagation delay of node Ni, For node signal strength indicators, This represents the proportion of remaining computing resources for a node. For signal error, ~ Preset weighting coefficients;

[0020] Node Ni scans the set of neighboring nodes Nj within its surrounding communication range and calculates the relative weight difference ΔWij for each neighboring node;

[0021] When ΔWij is greater than or equal to the threshold β, the corresponding node is added to the front of the priority relay queue Qi; otherwise, it is processed later.

[0022] Preferably, the construction of the first-layer self-organizing information transmission path P1 includes:

[0023] By collecting E0 acknowledgment packets successfully transmitted between nodes, a first-level path matrix P1 = {pij | E0 successfully transmitted from Ni to Nj} is established based on the timing chain between nodes.

[0024] The path stability coefficient Ψij = (ρij / ΔTij) is calculated based on the path delay ΔTij and the acknowledgment rate ρij. When Ψij is lower than the threshold Γ, the corresponding link is automatically removed from P1.

[0025] Preferably, the recording of local reception density Di and relay delay δi includes:

[0026] Node Ni counts the number of neighboring nodes Nrec that successfully receive and forward early warning messages from different sources within its communication radius Rcom using a sliding observation window Tw, and calculates the local reception density Di based on this, where Di is equal to Nrec divided by the node's communication coverage area.

[0027] Node Ni records the set of single-hop delays for each relay within the observation window Tw, and calculates the average relay delay δi using the exponentially weighted moving average method.

[0028] Preferably, the scheduling available equipment forms a temporary relay group G, including:

[0029] The central node C calculates the scheduling priority Score of each requesting point based on the attenuation level Ldecay, the location of the requesting node Li, and the expected relay radius Rreq carried in the request relay signal Si.

[0030] The request points are sorted in descending order of Score, with high-priority areas processed first, and candidate relay devices within the reach radius of each target location are searched.

[0031] The central node C constructs a scheduling table, matches each high-priority request point with one or more relay devices, and sorts them according to the estimated value to select the optimal combination and initially form a temporary relay group.

[0032] Preferably, generating a complete propagation topology includes:

[0033] The propagation topology graph is a weighted directed acyclic graph, where the node set V is all valid nodes that have received E0 information, the edge set E is the valid forwarding behavior between point pairs, and the weight of the edge is the propagation delay.

[0034] The propagation topology graph is used for regional generation and merging. It is used to divide the overall network into multiple logical sub-regions according to physical location or administrative division. The edge nodes in each region construct a local topology graph. After the central node collects each subgraph, it constructs a boundary edge set based on the propagation behavior between the boundary nodes. The central node performs a directed graph merging operation on each subgraph and edge set to form the final propagation topology graph of the entire network, retaining the global shortest propagation path.

[0035] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0036] 1. This invention introduces a directed acyclic propagation topology graph based on node propagation logs, enabling modeling and global visualization management of the actual propagation path of earthquake early warning information in regional networks. It can accurately reconstruct the entire process of information diffusion from the source node to each edge node. Compared to the traditional centralized broadcast method, which cannot track propagation paths, this invention possesses strong traceability and structural transparency, providing emergency management departments with graphical and data-driven decision support.

[0037] 2. This invention calculates the early warning response timestamps for all participating nodes using a shortest path algorithm, and generates response time heatmaps and statistical matrices accordingly. This effectively quantifies the response speed, coverage, and end-point timeliness bottlenecks of the early warning system. This mechanism not only significantly improves the accuracy of system performance evaluation but also has good scalability, adapting to various heterogeneous network structures and sudden disaster scenarios, demonstrating significant engineering application value. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0039] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] For examples, please refer to Figure 1 As shown in this embodiment, the self-organizing earthquake early warning method based on an emergency information dissemination system includes:

[0042] After acquiring earthquake trigger signals and identifying them as earthquake events according to set rules, an initial feature vector of the earthquake event is constructed. ;

[0043] Will Broadcast to the edge node cluster within the event triggering area, and receive Subsequently, based on its geographical location, Li and the epicenter Calculate propagation delay And construct the local received feature vector Ei;

[0044] Each edge node generates an information propagation weight Wi based on the received Ei, and forms a local relay queue Qi according to the order of Wi. It then broadcasts E0 to neighboring nodes in order of priority, thus constructing the first-layer self-organizing information transmission path P1.

[0045] After path P1 is formed, each node monitors the information backhaul redundancy Rj between itself and its neighboring nodes. If Rj is lower than the threshold Θ, a redundant path P2 is generated, and P2 is used to supplement the broadcast warning information in parallel.

[0046] During information propagation, each node dynamically records the local reception density Di and relay delay δi. When Di is less than the set lower limit Dmin or δi exceeds the set upper limit δmax, the node will mark it as a signal attenuation area, initiate a local cooperative relay request, and report the request relay signal Si to the communicable central node C.

[0047] After receiving multiple Si, the central node C schedules available devices to form a temporary relay group G and automatically deploys the hotspot communication protocol.

[0048] After all warning signal paths have stabilized, a complete propagation topology map is generated, and the warning response timestamps for each node are calculated.

[0049] In this invention, the first step is to acquire the earthquake triggering signal and determine it as an earthquake event according to the set rules, and then construct the initial feature vector E0 of the earthquake event.

[0050] Specifically, the system collects surface vibration signals in real time through accelerometers, seismographs, or other earthquake sensing devices deployed at earthquake monitoring points. When continuous high-intensity vibrations are detected exceeding a set time window threshold, and their waveform characteristics meet the set earthquake identification criteria (such as P-wave abrupt changes, spectral concentration, and vibration energy distribution characteristics), the system will determine the current event as an earthquake-triggered event.

[0051] Immediately after the event is confirmed, the initial feature vector of the earthquake event is constructed. This vector is used to describe the basic parameters of the current earthquake event and serves as the basic data structure for subsequent information dissemination and collaborative judgment.

[0052] Initial feature vector It includes the following four elements:

[0053] Magnitude M: The earthquake magnitude is quickly estimated based on the initial amplitude and frequency analysis model and is used to measure the intensity of an earthquake.

[0054] Epicenter The initial location of this earthquake was calculated using a time-difference inversion algorithm based on multi-point trigger source signals, and is represented by latitude and longitude coordinates.

[0055] Initial trigger time The absolute timestamp of the first confirmed earthquake event, accurate to milliseconds, is used to synchronize the information propagation process within the region.

[0056] Preliminary wave velocity estimation Based on historical earthquake databases and current geological structure presets, combined with the difference between trigger time and node reception, the estimated earthquake P-wave propagation velocity is used for subsequent propagation delay calculation and early warning path construction.

[0057] The initial feature vector As a structured data format, it is packaged in real time and pushed to all emergency information nodes, mobile terminals and edge gateways within the event-affected area, serving as the starting point for information dissemination in the entire self-organizing early warning network.

[0058] Initial eigenvectors of earthquake events Once the system is built, it immediately triggers the first broadcast of early warning information within the area. The broadcast is executed in parallel through multiple available communication channels, including but not limited to cellular mobile communication, low-power wide-area network (LPWAN), Wi-Fi Mesh network, satellite link, and short-range wireless transmission protocols (such as Bluetooth Mesh), ensuring that information distribution with high fault tolerance can still be achieved even under conditions where some communication is impaired or restricted.

[0059] The broadcast range is limited to the location of the epicenter. A cluster N of edge nodes is formed within a radius R (e.g., 20-100 kilometers) centered on the edge node cluster. The edge nodes include pre-registered and accessed terminal nodes with communication and computing capabilities, such as mobile terminals, vehicle-mounted devices, wearable devices, and portable emergency base stations.

[0060] Each edge node Ni receives the initial feature vector Subsequently, based on its own geographic location information Li (provided by GPS, BeiDou, or other positioning systems) and the location of the epicenter... Given the spatial distance between them, and considering the preset seismic wave propagation velocity V0, calculate its theoretical propagation delay ΔTi: ;in, Let node Ni be the epicenter. The Euclidean distance between them, in kilometers. The value is a preliminary estimate of the wave velocity in kilometers per second, and ΔTi is the theoretical time delay of the seismic wave propagating to node Ni in seconds.

[0061] Node Ni based on received

[0062] Based on the calculated propagation delay ΔTi, a local reception feature vector Ei is constructed to describe the current node's local awareness of the earthquake event. The feature vector Ei includes the following fields:

[0063] Magnitude M (inherited from) );

[0064] Epicenter (inherited from) );

[0065] Initial trigger time (inherited from) );

[0066] Local location Li (the node's own location);

[0067] Theoretical propagation delay ΔTi;

[0068] Locally received timestamp (i.e., the actual system time when E0 was received).

[0069] Difference error It is used to assess the deviation between local reception and theoretical propagation, and to help determine signal quality or node clock offset.

[0070] The locally received feature vector Ei will serve as the basic parameter input for nodes to participate in a series of actions such as self-organization and collaboration, path construction, load assessment and reconstruction requests. It will also be used to build a multi-node response model within the region, enhancing the spatial distribution perception capability and dynamic adaptability of the early warning system.

[0071] After constructing the local received feature vector Ei, each edge node Ni calculates the corresponding information propagation weights based on multiple parameters in Ei. This weight is used to measure the quality of node Ni's ability to perform the role of information relay in its current state. The higher the value, the more suitable the node is to act as an information forwarding node. ;in, For the propagation delay of node Ni, For node signal strength indicators, This represents the proportion of remaining computing resources for a node. For signal error, ~ Preset weighting coefficients;

[0072] The specific definitions are as follows:

[0073] Propagation delay ΔTi: represents the time required for the seismic wave to theoretically propagate to node Ni, derived from the calculation in step S200;

[0074] Signal strength Si: The signal received strength measured by Ni when receiving E0, in dBm, typically ranging from −120 to −30.

[0075] Remaining computing resource ratio Ci: refers to the percentage of CPU or memory resources currently available to Ni, collected using the system API interface;

[0076] Reception error εi: Represents the deviation between the received timestamp and the theoretical propagation time (i.e., the measured error), used to determine the accuracy of the local clock and the stability of the signal;

[0077] The above four parameters are normalized and mapped to a uniform scoring range (e.g., 0 to 1), and then multiplied by a preset weighting coefficient. The weighting coefficients can be obtained through training on historical data or set manually. Recommended initial values ​​are: =0.35, =0.25, =0.25, =0.15. This setting ensures both information transmission speed and node stability and computational sustainability.

[0078] Node Ni obtains its own propagation weight Then, it will broadcast detection to all neighboring nodes Nj within its communication range (e.g., via Wi-Fi scanning, Bluetooth discovery, or multicast handshake protocol) and collect the weight values ​​Wj of the neighboring nodes.

[0079] Node Ni calculates the relative weight difference accordingly. This is used to determine the priority difference between nodes. If ΔWij is greater than the set threshold β, then Ni is considered superior to Nj and has a higher forwarding priority, and Nj should be placed at the back of the relay queue.

[0080] The threshold β is a dynamically adjustable value that can be adaptively adjusted according to the node density ρ. The recommended calculation method is as follows:

[0081] Where ρ is the number of nodes within a unit communication radius, log e It is a natural logarithm function. This method can raise the competition threshold in densely populated areas and avoid multiple nodes broadcasting repeatedly.

[0082] Finally, Ni arranges all detected neighboring nodes in descending order of ΔWij to form a local relay queue Qi, which serves as the basis for decision-making in the next round of information forwarding.

[0083] Node Ni performs phased broadcasting based on the relay queue Qi. Each time, it only sends a message to the first k nodes in the queue (e.g., the first 3). The information packet is sent to other nodes, and no further transmission is made until a confirmation is received. This strategy reduces broadcast collisions and improves channel utilization.

[0084] In addition, the system sets an adaptive broadcast interval τi to control the time delay between each broadcast. The formula for calculating τi is: ;in, The minimum broadcast interval is δ (50 milliseconds is recommended), δ is the local network congestion coefficient (measured by packet loss rate), and ρ is the neighbor node density. This formula demonstrates that in cases of network congestion or dense node density, the broadcast interval should be appropriately extended to reduce the probability of channel collisions.

[0085] If a node finds that the signal acknowledgment rate (ACK packet reception rate) is lower than the threshold η (e.g., 0.5) after multiple broadcast attempts, it will automatically trigger the redundant broadcast mechanism to broadcast in parallel with low power to nodes ranked lower in the queue, ensuring that the information propagation chain is not interrupted.

[0086] During the secondary broadcast, nodes synchronously exchange E0 data and acknowledgment responses, and record all link pairs (Ni → Nj) that successfully broadcast and respond. Each link forms a valid single-hop propagation path.

[0087] Based on the collected link pairs, construct the first-level path matrix P1, which is a directed graph structure with nodes Ni and edges representing successfully propagated links pij.

[0088] To ensure path stability, a path stability coefficient Ψij is calculated for each link, defined as: Ψij = acknowledgment rate ρij / path propagation delay ΔTij; where ρij is the success rate of node Nj's acknowledgment to Ni (i.e., the proportion of acknowledgment packet reception to the total number of transmissions), and ΔTij is the average time taken for Ni to propagate E0 to Nj. If Ψij is lower than a set threshold Γ (e.g., 0.1), it is considered an unstable link and is not included in the P1 path graph.

[0089] Through the above path filtering mechanism, a first-layer information propagation topology P1 with optimal propagation efficiency and the lowest risk of information loss is obtained.

[0090] After constructing a local relay queue Qi based on the information propagation weight Wi and forming the first-layer information propagation path P1, the earthquake early warning information begins to propagate along P1 in a relay manner across multiple nodes. However, due to unstable factors often present in the post-disaster environment, such as damage to communication base stations, signal obstruction, and node drift, information transmission between some nodes may be interrupted, repeated, or omitted, leading to partial failure of the path.

[0091] To improve the completeness of early warning information coverage and the fault tolerance of the system, this invention designs a dynamic detection mechanism based on information backhaul redundancy Rj to determine whether there are weak connections or unstable nodes in the current path, and automatically generate a redundant path P2 to supplement the potentially risky parts of the P1 path, thereby realizing multi-path parallel transmission of information.

[0092] After successfully forwarding the warning information E0 to its neighboring node Nj, node Ni requires Nj to return an acknowledgment packet (ack) within a set time window ΔTack (e.g., within 100 milliseconds) to indicate successful information reception. The system sets an observation period Tobs (e.g., 2 seconds). Within this period, node Ni counts the number of ack packets received from Nj (Nj_ack) and the number of broadcasts it initiated to Nj (Nj_send).

[0093] Then, the information backhaul redundancy Rj corresponding to node Nj is defined as: Rj = Nj_ack / Nj_send.

[0094] The value of Rj ranges from 0 to 1. A higher value indicates a more stable communication link between nodes and a higher degree of redundancy. Conversely, a lower value of Rj indicates that the link is unreliable due to packet loss, high latency, or failed acknowledgment.

[0095] A preset redundancy threshold Θ is set, with a recommended value between 0.5 and 0.7. This value can be adjusted flexibly based on node density and application scenario. When Rj is lower than Θ, it indicates that the current link does not have a continuous and stable information relay capability, and node Ni will mark the connection as a "potential failure path".

[0096] If a node detects that the path ratio (Rj) between itself and one or more neighboring nodes is lower than the threshold Θ, it automatically initiates redundant path generation logic. The steps are as follows:

[0097] Node Ni broadcasts a status update packet to all neighboring nodes within its communication range. The packet includes: warning information ID, current forwarding path (its own link in P1), target node Nj, and redundancy value Rj. Upon receiving the packet, neighboring nodes update their own status (including Wi value, connectivity with Ni, and whether they have received the update). (Information, etc.) Reply to Ni.

[0098] Ni filters out nodes that have not yet appeared in the P1 path and have high propagation weight based on feedback from neighboring nodes. Nodes whose values ​​are above the set average are selected as candidate redundant relay nodes G = {Nk1, Nk2, ..., Nkn}. Each candidate node must satisfy the following condition:

[0099] The distance Ni is within the communication radius Rcom;

[0100] This segment of information is not currently being forwarded from its original path.

[0101] Higher than the weighted average of regional nodes;

[0102] The signal strength is high and computing resources are available.

[0103] Ni selects the first k nodes (e.g., k=2) as redundant relay nodes according to the Wi sorting of nodes in set G, and establishes a backup path link from Ni to Nj, i.e., redundant path P2. P2 can send the same E0 data packets in the form of multicast or parallel unicast, and uses a different channel or time window for time-division transmission with the original P1 path to reduce collisions.

[0104] Once the redundant path P2 is established, node Ni will maintain two forwarding channels simultaneously: the original path P1 and the redundant path P2. The broadcast of warning information E0 employs a parallel distribution mechanism: the information content is consistent, but different transmission identifiers and path labels are set to facilitate subsequent nodes in identifying the information source and avoid redundant relaying.

[0105] This invention introduces local reception density Di as a spatial index to describe the activity of information propagation within the communication radius of a node, and is used to measure the number of effective neighboring nodes that successfully receive the same earthquake early warning event information within a unit area.

[0106] Each node Ni successfully receives seismic events within its communication radius Rcom within a sliding time window Tw (e.g., set to 5 seconds). And the number of neighboring nodes that exchange ACK responses with it, Nrec. The communication radius Rcom is the maximum effective signal transmission distance of the device, with a recommended value of 200 meters.

[0107] Node Ni is calculated using the following formula: Where Acom represents the area of ​​a circular region centered at node Ni and with a communication radius Rcom, its calculation formula is: That is, approximately Square meters. Nrec represents the number of neighboring nodes that have formed a valid channel acknowledgment with Ni within the Tw time window.

[0108] Di is measured in "nodes per square meter", which can be converted to nodes per square kilometer (multiplied by 10). 6 Di is used for comparison with the lower limit threshold Dmin. The higher Di is, the higher the activity level of information reception in the area; conversely, it may indicate that the area is at the edge of propagation or in a blind zone.

[0109] Relay delay δi is used to describe the completion of node Ni The average response time between receiving and forwarding information is an important time indicator for measuring the forwarding efficiency of a node.

[0110] Node Ni, within the time window Tw, for each... Forwarding behavior record start time Ts (received) The forwarding delay for this round is calculated by combining the time of arrival (T) and the completion time Te (the time when the downstream node returns the ACK acknowledgment packet). The average relay delay δi is calculated by summing up the delays from multiple forwarding attempts and then smoothing the data using the Exponentially Weighted Moving Average (EWMA) method. Where α is the smoothing factor (recommended value is 0.3). This represents the weighted delay value after the nth observation. This is the result calculated at the previous time step. This is the current observation value.

[0111] This method can effectively filter out occasional bursts of latency and reflect the true forwarding performance under stable node operation. When δi continues to rise, it means that the node may be experiencing a processing bottleneck due to congestion, high hardware load, or link instability.

[0112] This invention designs a joint judgment mechanism. When Di is lower than a set lower limit Dmin, or δi exceeds a set upper limit δmax, the system considers the current node Ni to be in the signal attenuation region. The specific judgment logic is as follows:

[0113] Dmin: The minimum expected local receiver density. Based on the system deployment strategy, the recommended value is at least 3 effective receiver nodes per square kilometer. (Unit: number per square meter);

[0114] δmax: The maximum relay delay tolerated by the node. It is set according to the seismic wave propagation speed and the real-time requirement of information processing, and the recommended value is 2 seconds.

[0115] If Di < Dmin or δi > δmax, the node Ni marks itself as a signal attenuation node and triggers the following processing flow:

[0116] Once the node Ni is determined to be in the signal attenuation area, it immediately enters the relay request preparation process and generates a request relay signal Si. This signal is a structured data packet and contains the following fields:

[0117] Event identification ID (corresponding to the unique number of the seismic event);

[0118] Sending node ID (such as MAC address or node unique number);

[0119] Node's current position information Li (GPS coordinates or relative to the reference point);

[0120] Measured Di and δi values in real time;

[0121] Attenuation level Ldecay: It is divided into levels 0 - 3 according to the deviation degree of Di and δi. Specifically: ; Among them, and are weighting coefficients, and the recommended values are both 0.5. After rounding to the nearest integer, the discrete level is obtained (0: normal, 1: mild attenuation, 2: moderate, 3: severe);

[0122] Request relay coverage radius Rreq, the calculation method is: , k is an empirical factor, recommended k = 2; available resource report (remaining battery, CPU occupancy rate, channel quality index, etc.).

[0123] The node Ni reports Si to the central node C through the priority channel. The priority channel can include the cellular network control channel (if available), the emergency LoRa channel, or the high-priority forwarding path of the multi-hop Mesh network.

[0124] The central node C listens and receives the request relay signals Si uploaded by each edge node in the network in real time. This signal is generated and reported by the aforementioned node Ni in the signal attenuation area, and the included fields are: event number, node unique identifier, node position Li, local reception density Di, relay delay δi, signal attenuation level Ldecay, relay coverage requirement radius Rreq, available resource indicators (battery power, channel status), etc.

[0125] To rationally allocate limited emergency relay resources, the central node C needs to prioritize all received requests Si. To this end, the system constructs a decay scheduling scoring model, Score, to quantify the urgency of each request. The scoring model is as follows: Where Ldecay is the signal attenuation level (integer 0~3); Di is the reception density (number of nodes per unit area); δi is the forwarding delay of this node; δmax is the maximum tolerable relay delay (recommended value is 2 seconds); α, β, and γ are weighting factors used to adjust the importance of each parameter, and the recommended values ​​are s1=0.4, s2=0.3, and s3=0.3, respectively.

[0126] The central node sorts all Si in descending order of Score value, and high-priority requests will be processed first to ensure that resources are allocated to the areas with the highest network risk.

[0127] After completing the scheduling and sorting, central node C searches for schedulable relay devices within its service radius for each high-priority request point, including the following categories:

[0128] Vehicle-mounted emergency terminal equipped with a wireless communication module;

[0129] A mobile multi-rotor unmanned aerial vehicle platform;

[0130] Portable emergency micro base stations deployed temporarily;

[0131] Volunteer or law enforcement terminals equipped with hotspot functionality.

[0132] After sorting and filtering, the central node selects several devices to form a temporary relay group G, and logically numbers them as follows: Each relay group is responsible for serving one or more requesting areas. Relay group G members may include primary relay nodes and secondary relay nodes. The primary node is responsible for hotspot creation, and the secondary nodes are responsible for link relaying, load balancing, or backup forwarding.

[0133] After relay group G is established, central node C deploys communication protocol instructions for its downstream hotspots to quickly establish a local self-organizing wireless network, supporting multi-hop forwarding and terminal broadcasting of earthquake early warning information. The deployment process includes:

[0134] The central node selects a suitable communication protocol based on the environmental requirements of the target area, the capabilities of the terminal equipment, and the available frequency bands, such as:

[0135] It uses the 2.4 GHz Wi-Fi protocol to adapt to most smart terminals;

[0136] The LoRa protocol is suitable for low-power, long-distance communication.

[0137] ZigBee or BLE Mesh is suitable for short-distance, dense node networking.

[0138] If the relay node supports cellular links, then LTE-M or NB-IoT is enabled as the uplink channel.

[0139] A configuration file is distributed to each master relay node, with parameters including:

[0140] SSID (Hotspot Name) and encryption key;

[0141] Operating frequency band (e.g., 433 MHz, 2.4 GHz, 5 GHz);

[0142] Maximum concurrent connections (set according to device capabilities, such as 10~50);

[0143] Channel bandwidth (e.g., 20 MHz) and channel occupancy limit (recommended not to exceed 80%).

[0144] Packet forwarding rules (broadcast, directed, or multicast);

[0145] Communication protocol support level (e.g., 802.11b / g / n compatibility).

[0146] Upon receiving the configuration, each relay group member device automatically completes the protocol loading and service broadcasting process. The master node first initiates hotspot broadcasting, while auxiliary nodes monitor the master node's signal strength and status, and automatically join the Mesh network or enable slave hotspots based on load conditions to build redundant links.

[0147] After deployment, the temporary relay group G reports its state vector to the central node C every communication cycle Tcycle (Tcycle = 10 seconds is recommended), which includes the following fields:

[0148] Current number of online connections;

[0149] Uplink and downlink channel occupancy (percentage);

[0150] Average latency and packet loss rate;

[0151] Battery level and remaining service time (TTL).

[0152] Central node C determines whether the relay node is operating stably based on the state vector. If any of the following conditions are found:

[0153] The average communication delay exceeds 3 seconds;

[0154] Packet loss rate consistently exceeds 10%;

[0155] The node's battery level is less than 20%.

[0156] TTL is about to expire (less than 60 seconds);

[0157] This will trigger relay optimization operations, including automatically assigning alternative relays, adjusting the node structure within the group, extending service time, or issuing evacuation instructions.

[0158] Once the earthquake early warning event cycle ends or the original communication network is restored, the central node C issues a relay group release command, and all G group member devices will:

[0159] Turn off hotspot broadcast service;

[0160] Release channel and system resources;

[0161] Write to the event log and report it;

[0162] Return to the original task state or standby state, and enter low-power mode.

[0163] Earthquake early warning information During the propagation process, each edge node participating in the forwarding records its own propagation log, called the information propagation log, and its recording format includes:

[0164] Node ID (unique identifier);

[0165] Received The upstream node ID of the information (if any);

[0166] This node receives timestamp of information ;

[0167] This node forwards Timestamp (TF);

[0168] The number of responses from downstream nodes and the confirmation time (used to verify the effectiveness of forwarding).

[0169] The ID of the earthquake event to which it belongs (used to distinguish between multiple concurrent events);

[0170] Current node location Li (in WGS84 latitude and longitude format).

[0171] The propagation log The logs are periodically packaged and reported to the central node C through a node caching mechanism. The central node C has a log collection module that uses event ID as the aggregation basis to clean, filter, and structure the logs uploaded by all edge nodes.

[0172] After collecting sufficient propagation logs, central node C performs a propagation topology graph construction operation. This graph is a weighted directed acyclic graph (DAG), defined as follows:

[0173] The vertex set V represents all successfully received vertices. Information nodes;

[0174] The edge set E represents the valid forwarding relationships between nodes;

[0175] Each edge e(i, j) ∈ E represents the action of successfully forwarding from node i to node j. The weight of the edge is the propagation time, i.e., the time it takes for node j to receive the data. Subtract the forwarding time of node i from the total time.

[0176] The composition steps include:

[0177] The central node scans all propagation logs. If node j's log records that its upstream node is node i, and the reception time is... If , indicating that node i has a valid forwarding relationship with node j, then a directed edge e(i, j) is established, and the edge weight is set to . .

[0178] Perform loop detection and topology sorting on all existing edges, and remove paths with time anomalies or illogical behavior (e.g., edges with negative propagation time, no acknowledgment response, or abnormal time exceeding the set upper limit Δtmax, which is recommended to be 10 seconds).

[0179] After the propagation topology graph Graph_total is constructed, the central node C needs to calculate the early warning response timestamp Ti′ for each node in the graph, which reflects the first successful reception by that node during this earthquake event. The timing of the information.

[0180] For any node Vi ∈ V, its response timestamp Ti′ is defined as: Ti′ = shortest path time + ;in, The initial trigger time of the earthquake event is given by the shortest path time from the epicenter node. (Can be set as the earliest propagating node or the central node) The total time taken for the shortest propagation path to node Vi.

[0181] The shortest path is calculated using either Breadth-First Search (BFS) or Dijkstra's algorithm, finding the path with the minimum sum of edge weights within the DAG structure. The calculation method is as follows:

[0182] Initialize source point Ti′ = ;

[0183] For all outgoing edges e(i, j), set candidate Ti′j = Ti′i + edge weight e(i, j);

[0184] If Ti′j has not been recorded or the candidate value is smaller, then update Ti′j;

[0185] Repeat the above process until all nodes Ti′ have been traversed and determined.

[0186] If a node has multiple propagation paths, the path with the shortest propagation time is selected as the effective path by default, and its upstream link chain is recorded for visualization and reverse tracing analysis.

[0187] After the central node C completes all Ti′ calculations, it constructs a response time matrix T = {Ti′ | i ∈ V}. This matrix records the early warning response time of each node. Combined with the node location Li, a response time heatmap can be drawn to visualize the spread effect of the early warning signal in the entire area.

[0188] Based on this, the following statistical indicators are further generated:

[0189] Average response time Tavg: the arithmetic mean of all Ti′, used to evaluate the overall early warning response speed;

[0190] Maximum response time Tmax: The maximum value among all Ti′, identifying the system's end-point response time limit;

[0191] Preset several time intervals (such as 0–2 seconds, 2–5 seconds, and more than 5 seconds), count the number and proportion of nodes at each level, and form a level histogram;

[0192] If Tavg > Tquake (estimated average arrival time of seismic waves), the central node will generate a performance alert, indicating that the propagation efficiency of the early warning system may be insufficient.

[0193] Where Tquake is the distance from the epicenter to the average node divided by the propagation velocity of the seismic P-wave. For example, if the average distance from the epicenter to the target area is 60 kilometers, If the speed is 6 km / s, then Tquake takes 10 seconds.

[0194] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A self-organizing earthquake early warning method based on an emergency information dissemination system, characterized in that: include: After acquiring earthquake trigger signals and identifying them as earthquake events according to set rules, an initial feature vector of the earthquake event is constructed. ; Will Broadcast to the edge node cluster within the event triggering area, and receive Subsequently, based on its geographical location, Li and the epicenter Calculate propagation delay And construct the local received feature vector Ei; Each edge node generates an information propagation weight Wi based on the received Ei, and forms a local relay queue Qi according to the order of Wi. It then broadcasts E0 to neighboring nodes in order of priority, thus constructing the first-layer self-organizing information transmission path P1. The step of forming a local relay queue Qi based on Wi sorting includes: edge node Ni calculating information propagation weights according to the parameters in the local received feature vector Ei using a weighted model. The expression is: ;in, For the propagation delay of node Ni, For node signal strength indicators, This represents the proportion of remaining computing resources for a node. For signal error, ~ The preset weight coefficient is used; node Ni scans the set of neighboring nodes Nj within the surrounding communication range and calculates the relative weight difference ΔWij for each neighboring node; when ΔWij is greater than or equal to the threshold β, the corresponding node is added to the front of the priority relay queue Qi, otherwise it is processed later. The construction of the first-layer self-organizing information transmission path P1 includes: collecting E0 acknowledgment packets successfully transmitted between nodes, establishing a first-layer path matrix P1 = {pij | E0 successfully transmitted from Ni to Nj} based on the timing chain between nodes; calculating the path stability coefficient Ψij = (ρij / ΔTij) based on the path delay ΔTij and the acknowledgment packet return rate ρij; automatically removing the corresponding link from P1 when Ψij is lower than the threshold Γ; after the P1 path is formed, each node monitors the information backhaul redundancy Rj between itself and its neighboring nodes; if Rj is lower than the threshold Θ, a redundant path P2 is generated, and P2 is used to supplement the broadcast warning information in parallel; each node dynamically records the local reception density Di and relay delay δi during the information propagation process; when Di is less than the set lower limit Dmin or δi exceeds the set upper limit δmax, the node will mark it as a signal attenuation area, initiate a local cooperative relay request, and report the request relay signal Si to the communicable central node C; After receiving multiple Si, the central node C schedules available devices to form a temporary relay group G and automatically deploys the hotspot communication protocol. The recording of local reception density Di and relay delay δi includes: Node Ni counts the number of neighboring nodes Nrec that successfully receive and forward warning messages from different sources within its communication radius Rcom using a sliding observation window Tw, and calculates the local reception density Di based on this, where Di is equal to Nrec divided by the node's communication coverage area; Node Ni records the set of single-hop delays for each relay within the observation window Tw, and calculates the average relay delay δi using an exponentially weighted moving average method. The scheduling of available devices to form a temporary relay group G includes: the central node C calculating the scheduling priority Score of each requesting point based on the attenuation level Ldecay, the location Li of the requesting node, and the expected relay radius Rreq carried in the requesting relay signal Si; arranging the requesting points in descending order of Score, prioritizing high-priority areas, and searching for candidate relay devices within the reach radius of each target location; the central node C constructing a scheduling table, matching each high-priority requesting point with one or more relay devices, and pre-sorting and filtering out the optimal combination to initially form a temporary relay group; After all warning signal paths have stabilized, a complete propagation topology map is generated, and the warning response timestamps of each node are calculated. Generating the complete propagation topology map includes: The propagation topology graph is a weighted directed acyclic graph, where the node set V is all valid nodes that have received E0 information, the edge set E is the valid forwarding behavior between point pairs, and the weight of the edge is the propagation delay. The propagation topology graph is used for regional generation and merging. It is used to divide the overall network into multiple logical sub-regions according to physical location or administrative division. The edge nodes in each region construct a local topology graph. After the central node collects each subgraph, it constructs a boundary edge set based on the propagation behavior between the boundary nodes. The central node performs a directed graph merging operation on each subgraph and edge set to form the final propagation topology graph of the entire network, retaining the global shortest propagation path.

2. The self-organizing earthquake early warning method based on an emergency information dissemination system according to claim 1, characterized in that: The initial feature vector Including magnitude M, epicenter location Initial trigger time and preliminary wave velocity estimation .

3. The self-organizing earthquake early warning method based on an emergency information dissemination system according to claim 1, characterized in that: The feature vector Ei includes: magnitude, epicenter location, initial trigger time, local location, theoretical propagation delay, local reception timestamp, and difference error.

Images