Linear cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method

Abstract

The invention belongs to the technical field of traffic safety, and particularly relates to a linear cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method, which comprises the following steps of: constructing a vehicle-mounted ad hoc network with a clustering structure, and setting network parameters and signal transmission rules by taking a roadside unit (RSU) as a cluster head; when a traffic accident occurs in the cluster, the accident vehicle sends an original early warning signal to other vehicle nodes in the cluster and the cluster head; each vehicle node in the cluster receives the early warning signal, carries out local detection and hypothesis testing based on signal intensity, generates a local detection result, and sends the local detection result to the cluster head; the cluster head performs weighted linear fusion based on a preset weight vector related to a node accident distance to form a global detection statistic; judging the authenticity of the early warning signal in combination with a preset global judgment threshold; and deciding and executing broadcast forwarding or discarding operation of the early warning signal. Therefore, the problems of broadcast storm, resource shortage, high signal loss rate, low detection reliability and the like in the prior art are solved.

Description

Technical Field

[0001] This invention belongs to the field of traffic safety technology, specifically relating to a linearly cooperative method for detecting traffic accident early warning signals using an in-vehicle self-organizing network. Background Technology

[0002] Intelligent Transportation Systems (ITS) integrate technologies such as communication networks, computer information processing, multi-sensor detection, and automatic control to achieve real-time perception and comprehensive status monitoring of the traffic environment. As the core information transmission carrier of this system, vehicular ad hoc networks (V2VNs) support bidirectional communication between vehicles and between vehicles and roadside infrastructure. With their advantages of stable node power supply and dynamic networking based on road topology, they have become a key research focus in the field of intelligent transportation. Currently, such networks have been initially applied in scenarios such as traffic flow alerts and non-stop toll collection on highways.

[0003] However, existing vehicle-mounted ad hoc network technologies face significant challenges in the critical safety application of real-time traffic accident early warning. First, when a traffic accident occurs, safety information such as collision warnings and road anomalies is generated in a concentrated manner, forming a sudden data stream. Simultaneously, the large-scale convergence of vehicle nodes in the accident area can easily trigger a network broadcast storm, rapidly exhausting local network bandwidth and computing resources, leading to a surge in warning signal propagation delay and a significant increase in packet loss rate. Second, the high-speed movement of vehicles introduces channel interference such as Doppler shift, and the often insufficient deployment density of roadside units further deteriorates wireless channel conditions, resulting in a persistently high warning signal loss rate in real-world scenarios. The simple retransmission strategy used to compensate for this deficiency further exacerbates the channel load, creating a performance degradation cycle. Finally, existing warning signal detection methods fail to fully consider the dynamic impact of the high mobility of vehicle nodes on the received signal strength, making it difficult to achieve an effective balance between low false alarm rate and high detection rate. False alarms trigger unnecessary vehicle deceleration, easily causing subsequent traffic congestion; while missed detections may prevent following vehicles from avoiding danger in time, increasing the risk of secondary accidents. Therefore, there is an urgent need for a new solution for detecting traffic accident warning signals that can adapt to the dynamic topology and harsh channel characteristics of vehicle-mounted ad hoc networks, while also ensuring real-time performance and reliability. Summary of the Invention

[0004] This application provides a linear collaborative method for detecting traffic accident warning signals in vehicle-mounted ad hoc networks, in order to solve problems such as broadcast storms and resource scarcity, high signal loss rate, and low detection reliability in existing technologies.

[0005] The first aspect of this application provides a linearly cooperative method for detecting traffic accident warning signals in a vehicle-mounted ad hoc network, comprising the following steps: S1: Constructing a clustered vehicle-mounted ad hoc network, with a Roadside Unit (RSU) as the cluster head, and setting network parameters and signal transmission rules; S2: When a traffic accident occurs within a cluster, the vehicle involved in the accident sends an original warning signal to other vehicle nodes within the cluster and the cluster head; S3: Each vehicle node within the cluster receives the warning signal, performs local detection and hypothesis testing based on signal strength, generates a local detection result containing signal strength statistics, and sends the local detection result to the cluster head; S4: After receiving the local detection results from each vehicle node, the cluster head performs weighted linear fusion of all detection results based on a preset weight vector related to the accident distance of the node, forming a global detection statistic; S5: The cluster head performs a global cooperative decision based on the global detection statistic and a preset global decision threshold to determine the authenticity of the warning signal; S6: The cluster head decides and executes the broadcast forwarding or discarding operation of the warning signal based on the global decision result.

[0006] Preferably, in step S1, the construction of the cluster structure is specifically as follows: vehicle nodes form clusters based on the principle of "nearest affiliation" with roadside units (RSUs), and the RSUs serve as cluster heads, responsible for information coordination within and between clusters.

[0007] Preferably, in step S1, the signal transmission rule is as follows: only the cluster head has the authority to broadcast warning signals to all member nodes within the cluster; the accident vehicle is only allowed to send warning signals to the cluster head and other member nodes within the cluster.

[0008] Preferably, in step S3, the formula for calculating the signal strength statistics is: ; ; in, For the first Local detection statistics for each vehicle node; The number of time slots included in the detection period; For time slot numbering; For the first The vehicle node at the _ Signal strength received in each time slot; For local noise detection, For the first Channel gain of each vehicle node This represents the change in signal strength.

[0009] Preferably, in step S3, the decision is made based on the probability distribution characteristics of the signal strength statistics under different scenarios; specifically, when it is assumed that there is no actual traffic accident, the statistics follow a central chi-square distribution after normalization; when it is assumed that there is a traffic accident, the statistics follow a non-central chi-square distribution after normalization, and its non-central parameter is determined by the local signal-to-noise ratio at that node.

[0010] Preferably, in step S4, the weight vector allocation rule is as follows: weights are dynamically allocated based on the spatial distance between the vehicle node and the location of the traffic accident. The closer the node is to the accident point, the higher the proportion of its detection result in the subsequent global fusion.

[0011] Preferably, in step S4, the formula for calculating the global detection statistic is: ; in, The global detection statistic to be calculated; This represents the total number of vehicle nodes participating in the detection within the cluster. Let be the weight of the i-th vehicle node; This is the signal strength statistic received by the cluster head from the i-th vehicle node. This is the transpose of the weight vector w; To include all The global detection statistics vector.

[0012] Preferably, in step S5, the preset global decision threshold is determined based on the acceptable global false alarm rate of the system; the global collaborative decision is achieved through dual condition judgment: first, the calculated global detection statistics are compared with the decision threshold; second, the calculated global detection probability is evaluated to see if it reaches the preset reliability threshold; only when the statistics are not less than the threshold value and the detection probability reaches the threshold is the warning signal finally determined to be real, otherwise it is determined to be false.

[0013] Preferably, in step S6, the decision-making and execution operation specifically involves: if the global decision result is that the signal is real, the cluster head broadcasts the warning signal to all vehicle nodes within the cluster and forwards it to the cluster heads of adjacent clusters; if the global decision result is that the signal is false, the cluster head directly discards the warning signal.

[0014] A second aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement a linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method as described in the above embodiments.

[0015] Therefore, this application has the following beneficial effects:

[0016] This application's embodiments construct a clustered vehicular ad hoc network with Roadside Units (RSUs) as cluster heads and set differentiated transmission rules that allow only cluster heads to broadcast and accident vehicles to send signals only in a directional manner. This suppresses disorderly flooding of nodes from the network architecture and communication source, effectively mitigating broadcast storms caused by the concentration of safety information flow and dense convergence of nodes during sudden accidents, and reducing the ineffective consumption of network resources and competition conflicts. By utilizing local detection statistics generated based on the sum of squares of multi-slot signal strength and their probability models that follow central and non-central chi-square distributions under the assumption of no accident, signal detection can adapt to the Doppler frequency shift and fast signal fading caused by high-speed vehicle movement, improving robustness to adverse channel conditions. At the same time, by using an ideal orthogonal control channel dedicated to uploading local detection results, competition and interference between data and control channels are avoided, significantly reducing signal loss and end-to-end delay during transmission. A two-tier detection mechanism of "node local hypothesis testing - cluster head global linear collaboration" is adopted. During the global fusion phase, fusion weights are dynamically allocated based on the spatial distance between vehicle nodes and accident points. This ensures that nodes closer to the accident and with more reliable observations have a higher weight in the decision-making process. This achieves optimized integration and complementarity of multi-node observation information at the system level, maximizing the detection probability of true warning signals while ensuring a controllable preset false alarm rate, effectively balancing the risks of false alarms and missed detections. Therefore, it solves the problems of broadcast storms and resource scarcity, high signal loss rate, and low detection reliability in existing technologies.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0019] Figure 1 This is a flowchart of a linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to an embodiment of this application;

[0020] Figure 2 This is a schematic diagram of a traffic accident warning signal broadcasting model according to an embodiment of this application;

[0021] Figure 3 This is a schematic diagram of a linear cooperative detection system architecture provided according to an embodiment of this application. Detailed Implementation

[0022] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0023] The following describes a linearly cooperative method for detecting traffic accident warning signals in a vehicle-mounted ad hoc network, based on embodiments of the present application, with reference to the accompanying drawings. Addressing the high signal loss rate mentioned in the background section, this application provides a linearly cooperative method for detecting traffic accident warning signals in a vehicle-mounted ad hoc network. This method constructs a clustered vehicle-mounted ad hoc network with Roadside Units (RSUs) as cluster heads and sets differentiated transmission rules where only cluster heads have broadcast permissions and accident vehicles can only send signals in a directed manner. This suppresses disorderly flooding of nodes from the network architecture and communication source, effectively mitigating broadcast storms caused by the concentration of safety information flow and dense aggregation of nodes during sudden accidents, and reducing ineffective consumption and competition conflicts of network resources. By leveraging local detection statistics generated from the sum of squares of multi-slot signal strength and their probabilistic models following central and non-central chi-square distributions under the assumption of no accident, signal detection can adapt to Doppler shift and rapid signal fading caused by high-speed vehicle movement, improving robustness to adverse channel conditions. Simultaneously, by dedicating an ideal orthogonal control channel to uploading local detection results, competition and interference between data and control channels are avoided, significantly reducing signal loss and end-to-end latency during transmission. A two-stage detection mechanism of "node local hypothesis testing - cluster head global linear collaboration" is adopted. During the global fusion stage, fusion weights are dynamically allocated based on the spatial distance between vehicle nodes and accident points, giving higher weight to nodes closer to the accident and with more reliable observations. This achieves optimized integration and complementarity of multi-node observation information at the system level, maximizing the detection probability of true warning signals while ensuring a controllable preset false alarm rate, effectively balancing the risks of false alarms and missed detections. Therefore, this solves the problems of broadcast storms and resource scarcity, high signal loss rate, and low detection reliability in existing technologies.

[0024] Specifically, Figure 1 This is a flowchart illustrating the linear cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method provided in an embodiment of this application.

[0025] like Figure 1 As shown, the linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method includes the following steps:

[0026] In step S1, a clustered vehicle self-organizing network is constructed, with the roadside unit (RSU) as the cluster head, and network parameters and signal transmission rules are set.

[0027] Among them, the Roadside Unit (RSU) refers to a fixed infrastructure node deployed along the road that has wireless communication and information processing capabilities. As the cluster head, it is responsible for communication coordination, signal reception, and global decision-making within the cluster.

[0028] It should be noted that network parameters refer to a set of key variables predefined or configured for building and running a clustered detection system. These specifically include: the total number of vehicle nodes participating in detection within the cluster; the number of time slots included in the detection period divided by the system; the set of measures of noise impact experienced by each vehicle node during local signal detection (i.e., local noise variance); the set of measures of noise impact experienced by each node during the process of reporting detection results to the cluster head (i.e., global noise variance); and the channel gain parameter used to characterize the strength of the influence of the channel environment on the received signal of each node. Furthermore, relevant parameters may further include the local signal-to-noise ratio of each node, determined by the channel gain and noise variance.

[0029] It is understood that, by setting the Roadside Unit (RSU) as the fixed cluster head of the clustered network, this embodiment provides a stable control center for the orderly management and communication of vehicle nodes. Its coordination and decision-making functions structurally prevent disorderly flooding of nodes. Simultaneously, a series of pre-set network parameters provide clear quantitative benchmarks and calculation bases for each stage, including local detection, result transmission, and global fusion, enabling the entire collaborative detection process to be executed in a standardized and predictable manner. These parameters together form the foundation for the stable and efficient operation of the linear collaborative mechanism.

[0030] For example, in a typical intelligent transportation application scenario on a highway, the system deployment begins with roadside units (RSUs) set up at fixed intervals along the road. When a vehicle enters the wireless coverage area of ​​a particular RSU, it automatically registers and joins a communication cluster managed by that RSU. On a busy road segment, one RSU and approximately 20 vehicles within its coverage area constitute a collaborative detection cluster. Based on this physical structure, specific operating parameters need to be set for this cluster, including determining the total number of participating vehicles, dividing the detection window into several consecutive time slots, predefining noise parameters for each vehicle that reflect its receiving environment and transmission channel conditions, and acquiring its signal attenuation characteristics. Simultaneously, the system enforces a core communication rule: only the RSU has the authority to broadcast information to all vehicles within the cluster; when a traffic accident occurs within the cluster, the involved vehicles are only allowed to send targeted alarms to the RSU and a few geographically closest vehicles, effectively avoiding network congestion that could be caused by all nodes blindly forwarding information. This establishes the necessary network order and data interaction foundation for subsequent efficient and reliable collaborative signal detection.

[0031] In this embodiment of the application, the construction of the cluster structure is specifically as follows: vehicle nodes form clusters based on the principle of "nearest affiliation" with roadside units (RSUs), and the RSU is used as the cluster head, which is responsible for information coordination within and between clusters.

[0032] The "nearest cluster" principle refers to the automatic selection and joining of a vehicle node to the communication cluster managed by the RSU with the strongest signal or the closest location, based on the real-time wireless signal quality or geographical distance between the vehicle and different Roadside Units (RSUs). This dynamic clustering mechanism enables vehicles to maintain a stable connection with the optimal cluster head while moving at high speeds, thus adapting to continuous changes in vehicle location and network topology.

[0033] It should be noted that the responsibilities of the RSU coordinator as the cluster head include: managing the joining and leaving of vehicle nodes within the cluster, maintaining the communication timing and resource allocation within the cluster, receiving and integrating the local detection results reported by each node to execute global decisions, and, after confirming the real early warning signal, forwarding information to the cluster heads of neighboring clusters to complete cross-cluster collaborative early warning.

[0034] Understandably, the dynamic clustering mechanism based on "nearest affiliation" can not only effectively reduce transmission loss between vehicle nodes and cluster heads and improve the real-time performance of signaling interaction, but also avoid direct coordination overhead between vehicle nodes through centralized coordination of RSUs, thereby enhancing network controllability and overall stability and providing an efficient and orderly communication foundation for subsequent collaborative detection.

[0035] In this embodiment of the application, the signal transmission rules are as follows: only the cluster head has the authority to broadcast warning signals to all member nodes within the cluster; the accident vehicle is only allowed to send warning signals to the cluster head and other member nodes within the cluster.

[0036] The authority to broadcast warning signals refers to the fact that, after confirming that the accident warning is genuine and effective, the cluster head (RSU) is the sole operator responsible for sending the final warning information to all vehicle nodes within the cluster simultaneously.

[0037] It should be noted that "targeted transmission" means that when a vehicle is involved in an accident, it is only authorized to send initial alarm information to the cluster head and some nearby member nodes within the cluster, and cannot perform flooding broadcasts across the entire cluster or network. This rule imposes a mandatory constraint on the scope of information dissemination in terms of communication permissions.

[0038] Understandably, this hierarchical authorization signal transmission rule fundamentally reshapes the flow path of early warning information. It limits the initial release of incident information to the minimum necessary scope, effectively curbing broadcast storms that may be caused by a large number of nodes simultaneously forwarding similar information. At the same time, it establishes the cluster head as the sole center for information aggregation and authoritative dissemination, enabling network bandwidth and computing resources to be concentrated and efficiently used for global decision-making and critical information distribution, thereby significantly reducing the ineffective consumption and competitive conflicts of network resources.

[0039] In step S2, when a traffic accident occurs within the cluster, the vehicle involved in the accident sends an original warning signal to other vehicle nodes within the cluster and to the cluster head.

[0040] Among them, the accident vehicle refers to the mobile node in the vehicle-mounted ad hoc network that actually experienced a traffic accident.

[0041] It should be noted that the original warning signal refers to the initial alarm data packet generated and sent by the accident vehicle after the accident, without any intermediate node verification or forwarding. Its signal structure must contain several "key fields" describing the basic attributes of the accident. These key fields include: the real-time geographical coordinates of the accident location (e.g., latitude and longitude), the accident type classification identifier (e.g., rear-end collision, rollover, collision with a fixed object, etc.), and the precise timestamp of the accident. These fields provide the most crucial and original basis for the local detection of vehicle nodes and the global decision-making of the cluster head in subsequent steps.

[0042] It is understood that the embodiments of this application clearly define the vehicle at the first accident scene as the legally designated initial issuer of the warning information, ensuring the directness and authenticity of the information source. The propagation range of the initial signal is strictly limited: the accident vehicle can only send the signal to the cluster head and other nodes within the cluster, and cannot broadcast it outside the cluster or across the entire network. This strictly controls the initial spread of data packets within a local area, thereby informing the most likely affected neighboring vehicles (cluster members) and the network control center (cluster head) of the danger as soon as possible. Simultaneously, it fundamentally prevents network flooding that could be immediately caused by the accident vehicle's omnidirectional broadcast, creating conditions for subsequent orderly collaborative detection.

[0043] For example, in a highway clustering scenario, a car within a cluster rear-ends another vehicle due to insufficient emergency braking. The vehicle's onboard event detection system is immediately triggered, generating a raw warning data packet. This packet encapsulates real-time "accident location: 118.75°E, 31.98°N" obtained from the GPS module, "accident type: rear-end collision" determined by a sensor fusion algorithm, and the precise time of occurrence recorded by the system. Subsequently, the vehicle's communication unit strictly adheres to signal transmission rules, using a dedicated short-range communication channel to send this raw warning signal to the cluster head (i.e., the RSU responsible for that section) and three other vehicles within a 100-meter radius of its cluster. Through this step, the accident information completes a crucial leap from occurrence to initial limited distribution, promptly alerting high-risk vehicles and providing a clear and standardized signal source for subsequent steps in the most resource-efficient way.

[0044] In step S3, each vehicle node in the cluster receives the warning signal, performs local detection and hypothesis testing based on the signal strength, generates a local detection result containing signal strength statistics, and sends the local detection result to the cluster head.

[0045] Local detection refers to an independent decision-making process completed locally at the node. Each vehicle node acts as an independent sensing and processing unit, performing quantitative analysis and preliminary judgment based on the physical signals it receives, and its conclusions do not depend on information from other nodes.

[0046] It should be noted that the core of local detection is to distinguish between true and false signals through hypothesis testing. Hypothesis testing relies on clear decision rules, that is, by comparing the signal strength statistics with a preset threshold, a preliminary decision result is output. The specific formula is as follows: (i=1,2,…,M); in, This is a hypothetical scenario where there is no actual traffic accident and the received signal is an erroneous signal; This is a hypothetical scenario where an actual traffic accident occurs and the received signal is a real signal; Let be the local detection statistics for the i-th vehicle node; Let be the local decision threshold for the i-th vehicle node; This represents the total number of vehicle nodes within the cluster.

[0047] After the judgment rule is implemented, the probability of "erroneously determining the authenticity of the signal" needs to be calculated to ensure that the error risk of local detection is controllable. The specific formula is as follows: ; in, Let be the local false alarm probability of the i-th vehicle node; The complementary cumulative distribution function; It is the cumulative distribution function; Let be the local decision threshold for the i-th vehicle node; This is a hypothetical scenario where no actual traffic accident occurs. This is a hypothetical scenario where an actual traffic accident occurs.

[0048] In addition to the risk of errors, it is also necessary to quantify the probability of "correctly detecting the real signal" (local detection probability) to ensure that local detection can effectively identify the real warning signal. The specific formula is as follows: ; in, Let be the local detection probability of the i-th vehicle node; The complementary cumulative distribution function; Let be the local decision threshold for the i-th vehicle node; This is a hypothetical scenario where an actual traffic accident occurs.

[0049] After a vehicle node generates a local detection result, it needs to transmit it to the cluster head via a channel. However, global detection noise will be superimposed during the transmission process. Therefore, a formula is needed to accurately describe the relationship between the signal received by the cluster head and the original detection result. The specific formula is as follows: , , ; in, This is the global detection statistics vector received by the cluster head; This is the signal strength statistic received by the cluster head from the first vehicle node; This is the signal strength statistic received by the cluster head from the second vehicle node; The first received by the cluster head Signal strength statistics for each vehicle node; This is the vector of local detection statistics for the vehicle node; This represents the local detection statistics for the first vehicle node; This refers to the local detection statistics for the second vehicle node; For the first Local detection statistics for each vehicle node; This is the global detection noise vector; This refers to the global detection noise superimposed during the transmission of the first vehicle node; This refers to the global detection noise superimposed during the transmission process of the second vehicle node; For the first Global detection noise superimposed during the transmission process of each vehicle node; The total number of vehicle nodes within the cluster; This represents the transpose operation of a vector.

[0050] It is understood that the embodiments of this application generate stable local detection statistics from a defined calculation method, define the judgment logic of signal authenticity through clear decision rules, quantify the reliability of local detection by combining false alarm probability and detection probability, and finally characterize the noise impact in the transmission process of detection results through a special modeling method. This ensures the accuracy of local detection and provides real and complete input data for cluster head global collaborative detection.

[0051] For example, in a clustered scenario on an urban main road during the morning rush hour, vehicle V2 is traveling at the 60km / h speed limit. Vehicle V1, 50 meters ahead, brakes suddenly due to a rear-end collision. V2 quickly receives the initial warning signal from V1. At this time, the urban main road experiences heavy traffic and complex environmental interference such as traffic light changes and pedestrian crossings. However, V2 initiates local detection according to a preset procedure: first, it integrates the received signal strength from five time slots within the detection period, generating stable local detection statistics through a predefined calculation method; then, it invokes preset local limitations, combines the distribution characteristics of the signal strength statistics, and executes a decision rule to initially determine that the warning signal is genuine; to ensure the reliability of the detection results, V2 further verifies the results using the formulas for false alarm probability and detection probability, confirming that the risk of incorrect judgment is within a controllable range. After completing the local detection, V2 transmits the local detection results, including the signal strength statistics, decision conclusion, and probability verification results, to the cluster head RSU-B via an ideal orthogonal control channel. During transmission, although the data was affected by the complex electromagnetic environment of the main urban road, and the data was superimposed with global detection noise with zero mean and no spatial correlation, the orthogonal control channel effectively avoided the competition interference between the data and control channels, ensuring that the cluster head could completely receive the local detection results of V2, and providing high-quality input data for subsequent global collaborative decision-making.

[0052] In this embodiment of the application, the formula for calculating the signal strength statistics is as follows: ; ; in, For the first Local detection statistics for each vehicle node; The number of time slots included in the detection period; For time slot numbering; For the first The vehicle node at the _ Signal strength received in each time slot; For local noise detection, For the first Channel gain of each vehicle node This represents the change in signal strength.

[0053] It is understood that the embodiments of this application constitute the algorithmic basis for local detection. The squaring operation performed on the received signal strength aims to obtain its instantaneous power, while the summation of power values ​​across multiple consecutive time slots integrates the signal energy, thereby outputting a more stable statistic less affected by random noise fluctuations. This model clearly distinguishes between the signal components originating from the accident vehicle and the noise components introduced during transmission detection, laying a clear mathematical framework for subsequent probabilistic analysis and decision-making based on statistical theory.

[0054] In this embodiment of the application, the decision is made based on the probability distribution characteristics of the signal strength statistics under different scenarios. Specifically, when it is assumed that there is no actual traffic accident, the statistics follow a central chi-square distribution after normalization. When it is assumed that there is a traffic accident, the statistics follow a non-central chi-square distribution after normalization, and its non-central parameter is determined by the local signal-to-noise ratio at the node.

[0055] The central chi-square distribution and the non-central chi-square distribution are probability distribution models that describe the statistical characteristics of specific random variables. In this embodiment, after normalization by dividing the statistic by its noise variance, the probability distribution of the statistic will change based on the two fundamental assumptions: no accidents or accidents.

[0056] It should be noted that normalization eliminates the influence of noise power dimensions, making the statistics comparable across different nodes. The local signal-to-noise ratio (SNR) is a key parameter, directly determining the magnitude of the non-central parameter of the non-central chi-square distribution followed by the statistics under the assumption of an accident. A higher SNR means a stronger received signal, and the probability of the statistics taking a larger value increases significantly.

[0057] Understandably, incorporating the detection problem into the rigorous framework of statistical hypothesis testing is the theoretical cornerstone for achieving highly reliable decisions. By precisely defining the probability distribution of statistics under different assumptions, the quantitative relationship between the detection probability and the false alarm probability as a function of the decision threshold can be rigorously derived mathematically. This transforms system performance from empirical design to quantifiable prediction and precise optimization, providing a scientific and reliable design basis for rationally setting local decision thresholds to achieve the optimal trade-off between high detection rate and low false alarm rate.

[0058] In step S4, after receiving the local detection results from each vehicle node, the cluster head performs weighted linear fusion of all detection results based on a preset weight vector related to the accident distance of the node to form a global detection statistic.

[0059] The weight vector is a vector composed of the weights corresponding to each vehicle node. The weight allocation is negatively correlated with the spatial distance from the node to the location of the traffic accident.

[0060] It should be noted that weighted linear fusion refers to the process by which the cluster head integrates information from multiple nodes through a linear combination of weights and the local detection results of each node. The core is to highlight the detection value of highly reliable nodes.

[0061] The global detection statistics are quantitative indicators generated at the cluster head level by integrating data from multiple nodes, and they serve as the core basis for subsequent global decisions.

[0062] Understandably, in this embodiment, the weight vector is allocated through a negative correlation between distance and weight, allowing the detection results of nearby nodes, which are less affected by channel interference and have more accurate observations, to occupy a higher proportion, effectively strengthening the decision-making influence of effective information. The weighted linear fusion not only avoids the interference of misjudgment by a single node on the global decision, but also realizes the complementary integration of multi-node detection information, making up for the one-sidedness of single-node observations. The global detection statistics, as the core indicator after fusion, accurately integrate the effective information of all nodes in the cluster, significantly improving the signal-to-noise ratio and decision relevance of the data, providing high-quality and highly reliable input data for subsequent global collaborative decision-making of the cluster head, and ensuring the accuracy of the global decision.

[0063] For example, in a highway clustering scenario, the cluster head's RSU-A coverage area includes five vehicle nodes (V1 is the accident vehicle, and V2-V5 are 30 meters, 80 meters, 150 meters, and 200 meters away from the accident point, respectively). After the cluster head receives the local detection results sent by V2-V5, it calls a preset weight vector (V2 weight 0.35, V3 weight 0.25, V4 weight 0.2, V5 weight 0.1, and V1 as the accident vehicle weight 0.1). At this point, V2, being the closest, has the least Doppler shift interference and the highest reliability, thus having the highest weight; V5, being the farthest, has the most significant signal attenuation and the lowest weight. The cluster head combines the detection results of each node with their corresponding weights according to a preset weighted linear fusion logic to calculate the final global detection statistics. The entire process fully respects the highly reliable detection results of nearby nodes while integrating supplementary information from other nodes, effectively avoiding misjudgments caused by extreme channel conditions of a single node, and enabling the generated global detection statistics to more objectively and accurately reflect the true characteristics of the warning signal.

[0064] In this embodiment of the application, the weight vector allocation rule is as follows: the weights are dynamically allocated according to the spatial distance between the vehicle node and the location of the traffic accident. The closer the node is to the accident point, the higher the proportion of its detection result in the subsequent global fusion.

[0065] It is understandable that the embodiments of this application are closely aligned with actual scenarios. Nodes closer to the accident point are less affected by channel interference and signal attenuation, and the detection results are more reliable. By weighting, effective information is strengthened and invalid information is weakened, which further improves the signal-to-noise ratio and decision relevance of the global detection statistics.

[0066] In this embodiment of the application, the formula for calculating the global detection statistic is: ; in, The global detection statistic to be calculated; This represents the total number of vehicle nodes participating in the detection within the cluster. Let be the weight of the i-th vehicle node; This is the signal strength statistic received by the cluster head from the i-th vehicle node. This is the transpose of the weight vector w; To include all The global detection statistics vector.

[0067] It is understood that the formula in this application has the advantages of high computational efficiency and strong adaptability. It can quickly complete multi-node data fusion without complex hardware support, perfectly adapting to the real-time requirements of intelligent transportation scenarios. At the same time, the dynamic adjustment characteristic of the weight vector allows the formula to flexibly adapt to the differences in node distribution in different road scenarios such as highways and urban arterial roads, ensuring that accurate and reliable global detection statistics can be generated in various scenarios.

[0068] In step S5, the cluster head performs a global collaborative decision based on the global detection statistics and the preset global decision threshold to determine the authenticity of the warning signal.

[0069] Among them, the global decision threshold refers to the judgment threshold determined by the global false alarm rate preset by the system, which is used to distinguish between real signals and false signals; the global collaborative decision is the process by which the cluster head integrates the weighted fusion results of multiple nodes and probability characteristics to make the final judgment on the authenticity of the warning signal, which is the core link of the "local detection + global fusion" two-level detection mechanism.

[0070] It should be noted that the core of global collaborative decision-making is to output results through "comparison of global detection statistics and decision thresholds". This comparison logic must be clearly defined to ensure the decision-making process is standardized and repeatable. The specific formula is as follows: ; in, This serves as the global decision threshold. This is a global detection statistic; This is a hypothetical scenario where there is no actual traffic accident and the warning signal is incorrect. This is a hypothetical scenario where an actual traffic accident occurs and the warning signal is real.

[0071] To ensure the accuracy of the decision threshold and probability calculation, it is necessary to first characterize the distribution characteristics of the global detection statistics under different hypothetical scenarios. Variance is one of the core indicators. First, calculate the variance under scenario D0 (no accidents). The specific formula is as follows: ; ; in, Global detection statistics in scenario D0 The variance; The weight vector; weight vector transpose; This is a diagonal matrix in scenario D0; The number of time slots included in the detection period; It is a diagonal matrix with the corresponding vectors as diagonal elements; This is the local detection noise variance vector; Indicates to The diagonal matrix formed by squaring each element in the matrix; This is the global detection noise variance vector.

[0072] Similarly, the variance of scenario D1 (with an accident) needs to be calculated and compared with that of scenario D0 to provide a basis for subsequent calculations of false alarm probability and detection probability. The specific formula is as follows: ; ; in, Global detection statistics in scenario D1 The variance; The weight vector; weight vector transpose; This is a diagonal matrix in the D1 scene; The number of time slots included in the detection period; To The diagonal matrix formed by squaring each element in the matrix; To detect the signal energy transmitted by the system within the detection period; Let h be a diagonal matrix formed by the elements of h as the diagonal. It is the squared-modulus matrix of the local channel gain; This is the local detection noise variance vector; This is the global detection noise variance vector.

[0073] Variance calculation in scenario D1 depends on signal energy. This key parameter needs to be calculated first using a formula to ensure the completeness and accuracy of the variance calculation. The specific formula is as follows: ; in, For signal energy; For time slot numbering; The number of time slots included in the detection period; This represents the change in signal strength. for The square of the modulus.

[0074] After making a global decision, its error risk (global false alarm probability) needs to be quantified to ensure that the system's false alarm rate is controlled within an acceptable range. The specific formula is as follows: ; in, This represents the global false alarm probability. The complementary cumulative distribution function; It is the cumulative distribution function; This serves as the global decision threshold. This represents the number of times the vehicle node samples the signal. The standard deviation vector of local detection noise transpose; This is the column vector of weights for the vehicle nodes within the cluster; For signal energy; This is the local channel gain vector; This is the transpose of the weight vector w; This is a diagonal matrix in scenario D0; This is a diagonal matrix in scene D1.

[0075] While quantifying the risk of error, it is also necessary to quantify the probability of "correctly detecting the real signal" (global detection probability) to ensure that the global decision can effectively identify the real warning signal. The specific formula is as follows: ; in, This represents the global detection probability. The complementary cumulative distribution function; This serves as the global decision threshold. This represents the number of times the vehicle node samples the signal. This is the column vector of weights for the vehicle nodes within the cluster; For signal energy; This is the local channel gain vector; This is the transpose of the weight vector w; This is a diagonal matrix in scene D1.

[0076] It is understood that this application's embodiments take the decision rules as the core starting point, accurately describe the signal distribution characteristics under different assumptions through scenario-specific variance, supplement it with the key supporting parameter of signal energy, and finally quantify the reliability of the decision by calculating the global false alarm probability and detection probability, forming a complete global decision closed loop. This logic not only ensures the standardization and repeatability of the decision process, but also provides a reliability guarantee for the decision results through dual probability verification, effectively achieving a dynamic balance between "low false alarm rate" and "high detection rate".

[0077] In this embodiment, the preset global decision threshold is determined based on the acceptable global false alarm rate of the system; the global collaborative decision is achieved through dual condition judgment: first, the calculated global detection statistics are compared with the decision threshold; second, the calculated global detection probability is evaluated to see if it reaches the preset reliability threshold; only when the statistics are not less than the threshold value and the detection probability reaches the threshold is the warning signal finally determined to be real, otherwise it is determined to be false.

[0078] Understandably, the embodiments of this application construct a protection mechanism that combines quantitative indicators with probabilistic safeguards through a dual-condition decision rule. This mechanism relies on both statistical comparisons with thresholds to quickly filter signals and on detection probability thresholds to solidify the reliability of the decision results. This mechanism effectively avoids single-shot misjudgments caused by noise fluctuations, extreme channel conditions, and other factors, making the global decision more stable and adaptable in the complex and ever-changing vehicular ad hoc network environment. Simultaneously, it compensates for the limitations of a single decision standard, significantly improving the robustness of the overall decision and providing more reliable decision support for traffic safety early warning, thus helping to achieve a dynamic balance between low false alarm rate and high detection rate.

[0079] In step S6, the cluster head decides and executes the broadcast forwarding or discarding operation of the warning signal based on the global decision result.

[0080] Broadcast forwarding refers to the operation of the cluster head synchronously broadcasting the warning signal to all vehicle nodes within its cluster and forwarding the signal to the cluster heads of neighboring clusters. The core purpose is to achieve full coverage of the warning range.

[0081] It should be noted that the discard operation refers to the cluster head directly discarding the warning signal that has been determined to be false, without forwarding or broadcasting it. The core purpose is to avoid invalid information occupying network resources.

[0082] It is understood that the embodiments of this application form a control logic of "accurate propagation of real signals and timely interception of invalid signals" through the propagation path of cluster head centralized decision signals: the broadcasting and forwarding of real signals ensures that surrounding vehicles can quickly receive warnings, buying time for hazard avoidance operations such as slowing down and detouring; the direct discarding of false signals avoids the waste of network bandwidth and computing resources, prevents traffic congestion caused by false alarms, and balances traffic safety and network efficiency.

[0083] For example, after the cluster head RSU-B confirms the authenticity of the warning signal through global collaborative decision-making, it immediately initiates an efficient signal propagation process: on the one hand, it synchronously pushes complete warning information to all vehicle nodes within its cluster via a dedicated broadcast channel. This information includes detailed information such as the precise geographical location of the accident, the type of accident, the risk level, and recommended evasive actions (e.g., deceleration level, maintaining a safe following distance), ensuring that vehicles within the cluster quickly obtain crucial decision-making information. On the other hand, it forwards the warning signal in real-time to the cluster heads of adjacent clusters via cross-cluster communication interfaces, extending the warning range across clusters and covering potentially affected vehicles in a larger area surrounding the accident. Upon receiving the warning information, surrounding vehicles respond quickly and execute evasive actions: immediately reducing speed, activating hazard warning lights to alert oncoming vehicles, and intelligently planning detour routes or choosing safe areas to park and wait based on their own driving trajectory and the accident location. The entire response process is efficient and orderly, effectively reserving sufficient time and operational space to avoid secondary accidents. If the global judgment result is a false signal, the cluster head RSU-B will directly execute the signal discard operation without broadcasting or forwarding in any form. This avoids invalid information occupying network bandwidth, computing resources and vehicle communication channels, and prevents unnecessary deceleration or detours caused by false alarms. This eliminates traffic congestion or disorder caused by false alarms at the source.

[0084] In this embodiment of the application, the decision-making and execution operation is specifically as follows: if the global decision result is that the signal is true, the cluster head broadcasts the warning signal to all vehicle nodes in the cluster and forwards it to the cluster heads of adjacent clusters; if the global decision result is that the signal is false, the cluster head directly discards the warning signal.

[0085] It is understood that the embodiments of this application, through concise, clear, and efficient cluster head processing logic, precisely adapt to the real-time requirements of intelligent transportation scenarios: intra-cluster broadcasting ensures that all vehicles within the cluster can quickly obtain early warning information, achieving comprehensive coverage; cross-cluster forwarding extends the early warning range to adjacent areas, expanding the risk avoidance coverage radius; and directly discarding false signals quickly releases network bandwidth and computing resources, avoiding ineffective occupation. The organic combination of these three elements forms a complete closed loop for signal propagation and control, ensuring both the efficient dissemination of genuine early warnings and preventing resource waste and traffic interference caused by false signals, providing an efficient and feasible execution solution for traffic safety early warnings in actual road scenarios.

[0086] The linear cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method proposed in this application constructs a clustered vehicle-mounted ad hoc network with Roadside Units (RSUs) as cluster heads. It sets differentiated transmission rules, allowing only cluster heads to broadcast and accident vehicles to send signals only in a directional manner. This suppresses disorderly flooding of nodes from the network architecture and communication source, effectively mitigating broadcast storms caused by the concentration of safety information flow and dense node aggregation during sudden accidents, and reducing ineffective consumption and competition conflicts of network resources. By utilizing local detection statistics generated based on the sum of squares of multi-slot signal strength and their probability models following central and non-central chi-square distributions under the assumption of no accident, signal detection can adapt to Doppler shift and rapid signal fading caused by high-speed vehicle movement, improving robustness to adverse channel conditions. Simultaneously, by using an ideal orthogonal control channel dedicated to uploading local detection results, competition interference between data and control channels is avoided, significantly reducing signal loss and end-to-end delay during transmission. A two-tier detection mechanism of "node local hypothesis testing - cluster head global linear collaboration" is adopted. During the global fusion phase, fusion weights are dynamically allocated based on the spatial distance between vehicle nodes and accident points. This ensures that nodes closer to the accident and with more reliable observations have a higher weight in the decision-making process. This achieves optimized integration and complementarity of multi-node observation information at the system level, maximizing the detection probability of true warning signals while ensuring a controllable preset false alarm rate, effectively balancing the risks of false alarms and missed detections. Therefore, it solves the problems of broadcast storms and resource scarcity, high signal loss rate, and low detection reliability in existing technologies.

[0087] The following will illustrate a linear cooperative method for detecting traffic accident warning signals in vehicular ad hoc networks through a specific embodiment, including:

[0088] Traffic flow is heavy on the city's main roads during the evening rush hour, with vehicles traveling slowly at 40-60 km / h on six-lane roads in both directions, resulting in minor congestion in some sections. To ensure the efficiency of regional traffic safety early warning, relevant departments have deployed roadside units (RSUs) every 500 meters along the roads to construct a clustered vehicle-mounted self-organizing network. The RSU-B wireless coverage area is a 250-meter radius. If five vehicles (V1, V2, V3, V4, and V5) travel within this area, they automatically join the communication cluster headed by RSU-B (i.e., cluster B) based on the "nearest affiliation" principle, forming a collaborative detection network. The system pre-configures network operating parameters: the total number of vehicle nodes participating in detection within the cluster is 5, the detection period includes 5 time slots, and parameters such as local noise variance, global noise variance, and channel gain of each vehicle node are configured according to the characteristics of the road segment channel environment. Simultaneously, signal transmission rules are clearly defined—only the RSU-B has the authority to broadcast warning signals to all member nodes within the cluster; accident vehicles can only send warning signals to the RSU-B and other member nodes within the cluster, and flooding broadcasts to the entire cluster or network are strictly prohibited to avoid wasting network resources and broadcast storms. Figure 2 As shown, the clustered architecture, through the centralized coordination of RSU-B, achieves orderly management and communication scheduling of nodes within the cluster, laying a stable network foundation for subsequent early warning signal detection. Adjacent areas also deploy RSU-A (corresponding to cluster A) and RSU-C (corresponding to cluster C), forming an adjacent cluster linkage architecture with cluster B, preparing for cross-cluster early warning forwarding.

[0089] While approaching an intersection, V1, in an attempt to avoid a pedestrian crossing the road, braked suddenly and rear-ended V6, which was traveling normally behind it. The traffic accident occurred instantly. According to pre-set rules, V1, as the vehicle involved in the accident, immediately activated its onboard warning signal generation module, encapsulating a raw warning signal containing key information: the accident location was 116.3974°E, 39.9088°N (northeast lane of the intersection), the accident type was labeled "rear-end collision," and the precise time of the accident was recorded as 18:15:32. Subsequently, V1's communication unit transmitted the raw warning signal directionally to the cluster leader RSU-B and the other four vehicles (V2, V3, V4, and V5) via a dedicated short-range communication channel. The signal transmission strictly adhered to the "directed transmission" rule, preventing propagation to irrelevant nodes outside the cluster. This ensured that vehicles in high-risk areas quickly received the warning information while avoiding invalid signals consuming network bandwidth.

[0090] After V2, V3, V4, and V5 within the cluster successively receive the original warning signal sent by V1, they immediately initiate a local detection process. Taking V2 as an example, its onboard detection module first parses the received signal, extracting key fields such as the accident location and type. Subsequently, it conducts multi-timeslot acquisition and analysis based on signal strength. Within five consecutive time slots, it continuously receives and integrates signal strength data, and determines the authenticity of the signal through a preset hypothesis testing logic. V2 is only 30 meters away from the accident point, and its signal is less affected by channel interference and has stable strength. The local detection determines that the warning signal is genuine. Although V3, V4, and V5 are at different distances from the accident point (60 meters, 100 meters, and 150 meters respectively), and their signal strengths differ, they all complete signal verification through independent local detection processes, generating local detection results that include signal strength statistics and preliminary judgment conclusions. Subsequently, the four vehicles transmitted their local detection results to the cluster head RSU-B via the ideal orthogonal control channel. Although the transmission was affected by the complex electromagnetic environment of urban roads and a small amount of global noise, the orthogonal channel effectively avoided the competition and interference between the data and control channels, ensuring that the RSU-B received all the detection results completely.

[0091] After receiving the local detection results from V2 to V5, RSU-B initiates the global fusion process. The system's preset weight vector is dynamically allocated based on the distance between each vehicle and the accident point: V2 is the closest and has the highest weight (0.35); V3 is next (0.25); V4 has a weight of 0.2; and V5 is the farthest and has a weight of 0.1. Simultaneously, accident vehicle V1 is assigned a weight of 0.1, comprehensively reflecting its detection value as the source of the accident. Following a weighted linear fusion logic, RSU-B integrates the local detection results of each node with their corresponding weights, fully emphasizing the decision-making weight of highly reliable detection results from nearby nodes while also considering supplementary information from other nodes. This effectively avoids the impact of misjudgments from a single node on the global decision, ultimately forming a global detection statistic that objectively reflects the true situation of the warning signal, providing high-quality input data for subsequent global judgment.

[0092] Based on the generated global detection statistics, RSU-B initiates a global collaborative decision-making process. The system's pre-set global false alarm rate is 0.05, which is used to determine the corresponding global decision threshold. RSU-B first compares the global detection statistics with this threshold, while simultaneously calculating the global detection probability to assess the reliability of the decision result. If the global detection statistics are not less than the preset threshold and the global detection probability reaches 0.93, meeting the system's preset reliability threshold, and both conditions are met, the warning signal is ultimately confirmed as genuine and valid. The entire decision-making process strictly adheres to a dual protection logic of "quantitative indicators + probability assurance," effectively avoiding single-shot misjudgments that may be caused by noise fluctuations and extreme channel conditions, ensuring the accuracy and reliability of the decision result.

[0093] Upon confirming the authenticity of the warning signal, RSU-B immediately performs a signal broadcast forwarding operation. On one hand, it synchronously pushes complete warning information to all vehicle nodes (V1-V5) within cluster B via a dedicated intra-cluster broadcast channel, clearly indicating the accident location, type, and evasive action recommendations (reduce speed to below 30 km / h, maintain a safe distance, and carefully avoid the accident lane). On the other hand, it forwards the warning signal in real time to the adjacent cluster heads RSU-A (corresponding to cluster A) and RSU-C (corresponding to cluster C) through the cross-cluster communication interface. Vehicles within cluster B respond quickly upon receiving the warning: V2-V5 immediately reduce speed, activate hazard lights, and adjust their driving trajectory using the vehicle navigation system to avoid the accident lane; vehicles in clusters A and C, upon receiving the warning signals forwarded by their respective cluster heads, also reduce speed in advance and prepare for evasive action, effectively preventing secondary accidents. Figure 3 As shown, the signal propagation process achieves an organic combination of full coverage within the cluster and extension across clusters, forming a complete closed loop for early warning propagation; if the judgment result is a false signal, RSU-B will directly discard the signal without any forwarding, so as to avoid invalid information from interfering with traffic order.

[0094] In summary, the embodiments of this application fully present the practical application process of the linear cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method. Through the entire chain of operations, including cluster network construction, directional signal transmission, local detection, weighted fusion, global decision-making, and broadcast forwarding, it effectively suppresses broadcast storms, reduces signal loss rate, and achieves a balance between low false alarm rate and high detection rate. It is adaptable to complex traffic scenarios on urban arterial roads, requires no additional dedicated hardware, and operates efficiently and reliably, providing a practical solution for traffic safety early warning and significantly improving the safety and stability of road traffic.

[0095] Next, referring to the accompanying drawings, a linear cooperative vehicle-mounted self-organizing network traffic accident early warning signal detection system according to an embodiment of this application is described.

[0096] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described linearly cooperative method for detecting traffic accident warning signals in a vehicle-mounted ad hoc network.

[0097] In the description of this specification, the references to "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0098] Furthermore, 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 technical features indicated. Thus, 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.

[0099] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0100] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0101] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium. When executed, the program includes one or a combination of the steps of the method embodiments.

[0102] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A linearly cooperative method for detecting traffic accident early warning signals in vehicular ad hoc networks, characterized in that, include: S1: Construct a clustered vehicle self-organizing network with the Roadside Unit (RSU) as the cluster head, and set network parameters and signal transmission rules; S2: When a traffic accident occurs within the cluster, the vehicle involved in the accident sends an original warning signal to other vehicle nodes within the cluster and the cluster head; S3: Each vehicle node within the cluster receives the warning signal, performs local detection and hypothesis testing based on the signal strength, generates a local detection result containing signal strength statistics, and sends the local detection result to the cluster head; S4: After receiving the local detection results from each vehicle node, the cluster head performs weighted linear fusion of all detection results based on a preset weight vector related to the accident distance of the node to form a global detection statistic. S5: The cluster head performs a global collaborative decision based on the global detection statistics and the preset global decision threshold to determine the authenticity of the warning signal; S6: The cluster head decides and executes the broadcast forwarding or discarding operation of the warning signal based on the global decision result.

2. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S1, the construction of the cluster structure is specifically as follows: vehicle nodes form clusters based on the principle of "nearest affiliation" with roadside units (RSUs), and the RSUs serve as the cluster heads, responsible for information coordination within and between clusters.

3. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S1, the signal transmission rules are as follows: only the cluster head has the authority to broadcast warning signals to all member nodes within the cluster; the accident vehicle is only allowed to send warning signals to the cluster head and other member nodes within the cluster.

4. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S3, the formula for calculating the signal strength statistics is as follows: ； ； in, For the first Local detection statistics for each vehicle node; The number of time slots included in the detection period; For time slot numbering; For the first The vehicle node at the _ Signal strength received in each time slot; For local noise detection, For the first Channel gain of each vehicle node This represents the change in signal strength.

5. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S3, a decision is made based on the probability distribution characteristics of the signal strength statistics under different scenarios. Specifically, when it is assumed that there is no actual traffic accident, the statistics follow a central chi-square distribution after normalization. When it is assumed that there is a traffic accident, the statistics follow a non-central chi-square distribution after normalization, and its non-central parameter is determined by the local signal-to-noise ratio at that node.

6. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S4, the weight vector allocation rule is as follows: weights are dynamically allocated based on the spatial distance between the vehicle node and the location of the traffic accident. The closer the node is to the accident point, the higher the proportion of its detection result in the subsequent global fusion.

7. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S4, the formula for calculating the global detection statistic is: ； in, The global detection statistic to be calculated; This represents the total number of vehicle nodes participating in the detection within the cluster. Let be the weight of the i-th vehicle node; This is the signal strength statistic received by the cluster head from the i-th vehicle node. This is the transpose of the weight vector w; To include all The global detection statistics vector.

8. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S5, the preset global decision threshold is determined based on the acceptable global false alarm rate of the system; the global collaborative decision is achieved through dual condition judgment: first, the calculated global detection statistics are compared with the decision threshold; Secondly, assess whether the calculated global detection probability reaches a preset reliability threshold; only when the statistic is not less than the threshold and the detection probability reaches the threshold is the warning signal finally determined to be genuine, otherwise it is determined to be false.

9. The linearly cooperative vehicle-mounted ad hoc network traffic accident early warning signal detection method according to claim 1, characterized in that, In step S6, the decision-making and execution operation specifically involves: if the global decision result is that the signal is real, the cluster head broadcasts the warning signal to all vehicle nodes within the cluster and forwards it to the cluster heads of adjacent clusters; if the global decision result is that the signal is false, the cluster head directly discards the warning signal.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When a computer program or instruction is executed, it implements the linearly cooperative vehicle-mounted self-organizing network traffic accident early warning signal detection method as described in any one of claims 1 to 9.

Images