Time-sensitive network time delay control method and system for automatic driving

Abstract

The invention provides an automatic driving-oriented time-sensitive network time delay control method and system, and relates to the technical field of network communication, and the method comprises the steps: obtaining data packet transmission time delay information and network topology information, and generating a candidate transmission path set based on node connectivity layering and state space search; historical time delay data are collected, and prediction calculation and weight distribution are carried out; and iteratively updating the time delay weight based on the hierarchical time scale, and selecting a path with the highest comprehensive score to forward the data packet. According to the invention, the network transmission time delay of the automatic driving system can be effectively reduced, and the stability and reliability of data transmission are improved.

Description

Technical Field

[0001] The present invention relates to the field of network communication technologies, and in particular to a time-sensitive network delay control method and system for autonomous driving. Background Art

[0002] With the rapid development of autonomous driving technology, vehicle network communication systems have placed higher demands on the timeliness of data transmission. Time-sensitive networking, as a deterministic network technology, can provide low-latency, high-reliability data transmission for autonomous driving, which is crucial to the safety and performance of autonomous driving.

[0003] In autonomous driving scenarios, vehicles need to receive and process large amounts of data from their surroundings in real time, including sensor data, traffic information, and control commands. The transmission latency of this data directly impacts the decision-making efficiency and safety performance of autonomous driving systems. Currently, common time-sensitive networking technologies primarily reduce network transmission latency through optimized network scheduling algorithms, traffic shaping, and preemption mechanisms. However, these existing technologies still have many shortcomings in the complex and ever-changing autonomous driving environment.

[0004] Existing autonomous driving network delay control technologies still have some defects. In terms of routing selection, existing technologies lack the ability to adapt to dynamic changes in network topology and cannot flexibly adjust transmission paths according to real-time network status, resulting in the inability to guarantee timely and reliable transmission of data packets in the event of network congestion or node failure. Existing technologies lack accuracy in delay prediction and often use simple historical data averages as the basis for prediction. They cannot effectively deal with delay fluctuations caused by sudden changes in network traffic, affecting the real-time performance of autonomous driving systems. Existing technologies lack comprehensive consideration of the correlation between network node load status and delay, and cannot formulate differentiated transmission strategies based on the characteristics of different nodes, resulting in uneven distribution of network resources and low overall transmission efficiency. Summary of the Invention

[0005] The embodiments of the present invention provide a time-sensitive network delay control method and system for autonomous driving, which can solve the problems in the prior art.

[0006] A first aspect of an embodiment of the present invention provides a time-sensitive network delay control method for autonomous driving, comprising:

[0007] Obtain the data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information;

[0008] Generate a network node connection relationship graph based on network topology information, stratify nodes based on their connectivity values ​​in the network node connection relationship graph, perform state space search operations to generate a node state transition sequence, and obtain a set of candidate transmission paths;

[0009] Collect historical delay data of each candidate transmission path, calculate the average value based on the preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation;

[0010] Based on the delay difference comparison results between candidate transmission paths, the delay weight coefficient of each candidate transmission path is determined by subset partitioning, and a delay weight distribution sequence is generated;

[0011] It iteratively updates the delay weight distribution sequence based on the hierarchical time scale, scores the candidate transmission paths according to the node load rate data and node delay evaluation parameters, and selects the candidate transmission path with the highest comprehensive score as the optimal transmission path;

[0012] Data packets are forwarded through the optimal transmission path. When the real-time delay change value is detected to exceed the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

[0013] In an optional embodiment, nodes are hierarchized based on their connectivity values ​​in a network node connection relationship graph, and a state space search operation is performed to generate a node state transition sequence. The obtained candidate transmission path set includes:

[0014] Counting the number of connection edges of each node in the network node connection relationship graph to obtain a node connectivity value, and dividing the nodes in the network node connection relationship graph into multiple connection layers according to a preset first connectivity threshold and a preset second connectivity threshold;

[0015] Obtain the geographic coordinate information of each node, perform regional clustering operations on the nodes in each connection layer, calculate the Euclidean distance between nodes, and generate a node regional distribution sequence;

[0016] Calculate the Euclidean distance between each node, obtain the connectivity ratio of each node, record the historical number of successful transmissions and the total number of transmissions between each node, and calculate the weighted transmission state transition value;

[0017] determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value;

[0018] The state space is searched multiple times using a random sampling method to select a state transition path whose replacement value meets a preset cost threshold, and the state transition path is mapped to the actual node access sequence to generate a set of candidate transmission paths.

[0019] In an optional embodiment, determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value includes:

[0020] Determine the inter-node transfer probability according to the transmission state transfer value;

[0021] Taking the starting node as the search starting point, performing a state space search operation, setting the starting node as the current search node, and initializing a dynamic search depth value;

[0022] Selecting a next-hop node of the current search node based on the inter-node transition probability, and recording the next-hop node in the visited node set; counting the historical access frequency of the next-hop node, and when the historical access frequency is greater than a preset access threshold, multiplying the inter-node transition probability of the next-hop node by a preset attenuation coefficient;

[0023] When the next-hop node is the target node, the current search path is recorded as a node state transition sequence; when the next-hop node is not the target node and is not in the visited node set, the next-hop node is set as the current search node, the dynamic search depth value is updated based on the preset depth adjustment coefficient, and the state space search operation is continued; when the next-hop node is in the visited node set, return to the previous level node to reselect the next-hop node; when the search depth reaches the dynamic search depth value, the path cost is calculated based on the cost value of each node in the visited node set and the distance from the target node, and when the path cost is greater than the preset path cost threshold, return to the previous level node to reselect the next-hop node;

[0024] Calculating the cost of each transition in the node state transition sequence;

[0025] The node state transition sequence is optimized based on the cost value, and a node state transition sequence whose cost value is less than a preset cost threshold and whose node hop count is less than a preset hop count threshold is selected as the final node state transition sequence.

[0026] In an optional embodiment, collecting historical delay data of each candidate transmission path, performing average calculation based on a preset time window, and obtaining a predicted delay value of each candidate transmission path through weighted calculation includes:

[0027] Collecting historical delay data of each candidate transmission path, regularly sampling the historical delay data according to a preset sampling time interval, and generating a delay sampling sequence;

[0028] Based on a preset first time window and a preset second time window, the time-delay sampling sequence is divided into data, wherein the second time window is larger than the first time window;

[0029] Calculate the mean of the sampled data in the first time window and the sampled data in the second time window to determine the short-term delay mean and the long-term delay mean respectively;

[0030] Calculating a deviation rate between the short-term delay average and the long-term delay average, and determining a delay prediction weight coefficient according to the deviation rate;

[0031] Performing weighted calculation on the delay prediction weight coefficient and the delay sampling value at the current moment to obtain the predicted delay value of each candidate transmission path;

[0032] The prediction accuracy of the predicted delay value is counted, and when the prediction accuracy is lower than a preset accuracy threshold, the delay prediction weight coefficient is updated based on a preset adjustment step size, and the predicted delay value is recalculated.

[0033] In an optional embodiment, based on the delay difference comparison result between the candidate transmission paths, determining the delay weight coefficient of each candidate transmission path by subset division, and generating the delay weight distribution sequence includes:

[0034] Calculate the delay difference between each candidate transmission path and all other candidate transmission paths, and establish a delay difference set for each transmission path;

[0035] Comparing the positive and negative attributes of each delay difference in the delay difference value set of each candidate transmission path, classifying the delay difference greater than zero into the positive value subset, and classifying the delay difference less than zero into the negative value subset;

[0036] The delay differences in the positive subset of each candidate transmission path are accumulated to obtain the delay disadvantage value, and the delay differences in the negative subset are accumulated to obtain the delay advantage value;

[0037] Based on the ratio between the delay disadvantage value and the delay advantage value, the weight coefficient of the corresponding candidate transmission path is determined through exponential operation; the weight coefficients of all candidate transmission paths are normalized and the delay weight distribution sequence is determined.

[0038] In an optional embodiment, iteratively updating the delay weight distribution sequence based on the hierarchical time scale, scoring the candidate transmission paths according to the node load rate data and the node delay evaluation parameter, and selecting the candidate transmission path with the highest comprehensive score as the optimal transmission path includes:

[0039] Obtaining a delay weight distribution sequence of a candidate transmission path, and combining the delay weight distribution sequence with a current path delay state to form a state parameter;

[0040] Perform weight learning on each candidate transmission path, calculate the current weight value based on the delay weight distribution sequence of the candidate transmission path, use the ratio of the candidate transmission path's historical transmission success rate to the weight volatility as the weight prediction coefficient, and modify the current weight value based on the weight prediction coefficient to determine the expected weight value;

[0041] A hierarchical time-scale weight update mechanism is constructed. In the first time scale, the current weight value is updated based on the change of the path state. In the second time scale, the calculation parameters of the expected weight value are updated based on the path stability. The weighted sum of the weight mean of the first time scale and the weight prediction value of the second time scale is used to obtain the iterative weight value.

[0042] Collect the load rate data of each node on the candidate transmission path, and calculate the node delay evaluation parameters based on the data transmission rate and data packet round-trip time between nodes;

[0043] Calculate the load balancing score based on the load rate data, calculate the delay score based on the node delay evaluation parameters, and use the weighted sum of the load balancing score and the delay score as the path base score;

[0044] The path basic score is multiplied by the iterative weight value to obtain the path comprehensive score, and the candidate transmission path with the highest path comprehensive score and greater than the preset score threshold is selected as the optimal transmission path.

[0045] A second aspect of an embodiment of the present invention provides a time-sensitive network delay control system for autonomous driving, including:

[0046] The first unit is used to obtain data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information;

[0047] The second unit is used to generate a network node connection relationship graph based on the network topology information, hierarchize the nodes based on the connectivity values ​​of the nodes in the network node connection relationship graph, perform a state space search operation to generate a node state transition sequence, and obtain a set of candidate transmission paths;

[0048] The third unit is used to collect historical delay data of each candidate transmission path, perform average calculation based on a preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation;

[0049] The fourth unit is configured to determine a delay weight coefficient of each candidate transmission path by subset partitioning based on the delay difference comparison result between the candidate transmission paths, and generate a delay weight distribution sequence;

[0050] The fifth unit is used to iteratively update the delay weight distribution sequence based on the hierarchical time scale, score the candidate transmission paths according to the node load rate data and the node delay evaluation parameters, and select the candidate transmission path with the highest comprehensive score as the optimal transmission path;

[0051] The sixth unit is used to forward data packets through the optimal transmission path. When it is detected that the real-time delay change value exceeds the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

[0052] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0053] processor;

[0054] a memory for storing processor-executable instructions;

[0055] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0056] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0057] In an embodiment of the present invention, a set of candidate transmission paths is constructed by combining node hierarchical layering and state-space search. Historical delay data is used for prediction and dynamic allocation of delay weights. This allows accurate identification of the optimal transmission path in the network, effectively reducing data transmission delay in the autonomous driving system and improving data packet transmission efficiency. An iterative update mechanism based on a hierarchical time scale is adopted, combining node load rate and delay evaluation parameters for comprehensive scoring, making path selection more adaptive and enabling timely adjustment of transmission strategies based on changes in network status, effectively addressing network congestion and fluctuations, and enhancing the reliability and stability of the autonomous driving network. Real-time monitoring and rapid response to network delay are achieved, and the optimal path can be immediately recalculated when delay changes outside a preset range are detected. This dynamic delay control mechanism is particularly suitable for application scenarios with high real-time requirements, such as autonomous driving, and significantly improves the system's adaptability to network environment changes and the transmission quality of time-sensitive data. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a flow chart of a time-sensitive network delay control method for autonomous driving according to an embodiment of the present invention;

[0059] Figure 2 This is a flowchart for dynamic probabilistic state transition path search and optimization. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0061] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0062] Figure 1 FIG. 1 is a flow chart of a time-sensitive network delay control method for autonomous driving according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0063] Obtain the data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information;

[0064] Generate a network node connection relationship graph based on network topology information, stratify nodes based on their connectivity values ​​in the network node connection relationship graph, perform state space search operations to generate a node state transition sequence, and obtain a set of candidate transmission paths;

[0065] Collect historical delay data of each candidate transmission path, calculate the average value based on the preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation;

[0066] Based on the delay difference comparison results between candidate transmission paths, the delay weight coefficient of each candidate transmission path is determined by subset partitioning, and a delay weight distribution sequence is generated;

[0067] It iteratively updates the delay weight distribution sequence based on the hierarchical time scale, scores the candidate transmission paths according to the node load rate data and node delay evaluation parameters, and selects the candidate transmission path with the highest comprehensive score as the optimal transmission path;

[0068] Data packets are forwarded through the optimal transmission path. When the real-time delay change value is detected to exceed the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

[0069] In a specific embodiment, in a time-sensitive network environment for autonomous driving, the reliability and real-time performance of data packet transmission directly impact vehicle safety control and decision-making. Obtaining data packet transmission latency and network topology information for autonomous vehicles is fundamental to path selection. The onboard communication module continuously monitors communication quality with roadside units (RSUs), cloud servers, and surrounding vehicles, collecting end-to-end data packet transmission latency data. Network discovery protocols are used to obtain currently available network node information, including node identification, connection relationships, and transmission capabilities. For example, in a typical urban road scenario, an autonomous vehicle may be connected to multiple RSUs, mobile base stations, and edge computing nodes, forming a network topology consisting of 15 nodes. By parsing the timestamp information in network layer protocol packets, the transmission latency of data packets from sending to receiving is measured, and the processing latency of each node is recorded. This latency data is statistically analyzed to calculate parameters such as the average processing latency, latency fluctuation range, and packet loss rate of each node, which are then comprehensively analyzed to form node latency evaluation parameters. Taking an edge computing node as an example, its average processing delay is 3.5 milliseconds, the delay fluctuation range is 1.2 milliseconds, and the packet loss rate is 0.03%. Based on this calculation, the node delay evaluation parameter value is 4.2.

[0070] Based on the acquired network topology information, a network node connectivity graph is generated, which is a prerequisite for generating candidate transmission paths. Each network node is represented as a vertex in the graph, and the communication links between nodes are represented as edges, constructing an undirected connected graph. For each node in the graph, its connectivity is calculated, representing the number of directly connected nodes. Nodes are then stratified based on connectivity, with nodes with higher connectivity assigned to higher tiers and nodes with lower connectivity assigned to lower tiers. In the aforementioned 15-node network, analysis shows that nodes 2, 7, and 11 have connectivity of 6, 5, and 4, respectively, and are classified as the first tier. Nodes 1, 4, 8, and 13 have connectivity of 3, and are classified as the second tier. The remaining nodes have connectivity of 1 or 2, and are classified as the third tier. Based on the stratification results, a search algorithm is executed to search the state space, starting with the node where the onboard communication module is located and ending at the target service node. During the search, nodes at higher tiers are prioritized for state expansion, while inter-node distances and estimated transmission delays are considered as heuristic information. Through state space search, a series of node state transition sequences from the starting point to the end point are generated. Each sequence corresponds to a possible transmission path. After screening and optimization, eight candidate transmission paths are finally obtained, forming the candidate transmission path set.

[0071] Collecting historical latency data for each candidate transmission path and performing latency predictions are key steps in evaluating path performance. For each candidate transmission path, end-to-end transmission delays are recorded by periodically sending probe packets, establishing a historical latency database. The average latency for each path is calculated based on a preset time window, such as the latency data within the last 10 minutes. During the calculation, a weighted average is applied to the data within the time window, with delay data closer to the current moment given greater weight to reflect recent changes in network status. For example, for candidate path 3, assuming 60 sets of latency data were collected within the last 10 minutes, with the average latency for the last minute being 22 milliseconds, the average latency for the previous 1-5 minutes being 25 milliseconds, and the average latency for the previous 6-10 minutes being 27 milliseconds, then a weighted average calculation (with a weight ratio of 5:3:2) yields a predicted latency value of 23.8 milliseconds for this path. Similarly, the predicted delay values ​​of the other seven candidate paths are calculated to be 18.5 ms, 21.2 ms, 25.6 ms, 19.7 ms, 24.3 ms, 20.1 ms, and 22.4 ms, respectively.

[0072] Determining the weight of each path based on the delay difference comparison between candidate transmission paths is a crucial factor in resource allocation. For each pair of candidate transmission paths i and j, the delay difference Dij is calculated: the predicted delay of path i minus the predicted delay of path j. Based on this calculation, a delay difference set is established for each path. For path 1 (predicted delay 18.5 milliseconds), the delay difference set is {-2.7, -7.1, -1.2, -5.8, -1.6, -3.9}, all negative values, indicating that this path has superior delay performance to all other paths. Each path's delay difference set is then subsetted, with positive differences assigned to the positive subset and negative differences assigned to the negative subset. The sum of the positive subset (delay disadvantage) and the sum of the negative subset (delay advantage) are calculated for each path. For the second path (predicted delay of 21.2 milliseconds), the delay difference set is {2.7, -4.4, 1.5, -3.1, 1.1, -1.2}. The sum of the positive subset is 5.3, and the sum of the negative subset is -8.7. The weight coefficient is determined through exponential calculation based on the ratio of the delay disadvantage to the absolute value of the delay advantage. When the delay advantage of the path is zero, the weight coefficient is set to the minimum value of 0.01. Otherwise, the weight coefficient is equal to the result of the power operation of the natural logarithm e to the negative of the ratio of the delay disadvantage to the absolute value of the delay advantage. The calculated weight coefficients for the eight paths are 1.0, 0.543, 0.257, 0.073, 0.721, 0.124, 0.682, and 0.387, respectively. These weight coefficients are normalized to obtain the delay weight distribution sequence {0.264, 0.143, 0.068, 0.019, 0.190, 0.033, 0.180, 0.102}.

[0073] Iteratively updating the delay weight distribution sequence based on a hierarchical time scale is a key mechanism for adapting to dynamic network changes. Three time scales are set: short-term (1 minute) and long-term (15 minutes). In the short-term scale, real-time delay data for each path is updated every 10 seconds; in the long-term scale, the complete path search and evaluation process is re-executed every 5 minutes. This hierarchical time strategy ensures rapid response to network state changes while avoiding the system overhead caused by frequent path switching. Furthermore, a comprehensive score is assigned to each candidate transmission path based on node load data and node delay evaluation parameters. The score calculation considers three factors: predicted path delay, average node load on the path, and node delay evaluation parameters. For example, path 1 has a predicted delay of 18.5 milliseconds, an average node load of 65%, and an average node delay evaluation parameter of 3.8. The weighted score yields a comprehensive score of 87.3. Similarly, comprehensive scores are calculated for the remaining paths, and the first path with the highest score is ultimately selected as the optimal transmission path.

[0074] Data packets are forwarded through the optimal transmission path while continuously monitoring changes in the actual transmission delay. When it is detected that the real-time delay change value exceeds the preset delay change range, the path reselection mechanism is triggered. For example, if the preset delay change range is set to ±20%, if the actual transmission delay of the first path suddenly increases from 18.5 milliseconds to 23.5 milliseconds, exceeding the preset range, the delay weight coefficient and comprehensive score of each path will be recalculated immediately using the latest collected historical delay data and the updated node delay evaluation parameters. During a reselection process, due to a sharp increase in the load rate of a key node on the first path to 95%, the comprehensive score of the path dropped to 75.8, while the comprehensive score of the fifth path rose to 83.5, and the fifth path was automatically switched to for data transmission to ensure continuous and efficient communication.

[0075] Through the above method, the time-sensitive network for autonomous driving can continuously optimize the transmission path of data packets according to the dynamic changes in network status, ensuring that critical information can be reliably transmitted under strict latency constraints, providing stable network support for autonomous driving systems, and effectively improving the real-time and reliability of vehicle control, environmental perception, and decision-making.

[0076] In an optional embodiment, nodes are hierarchized based on their connectivity values ​​in a network node connection relationship graph, and a state space search operation is performed to generate a node state transition sequence. The obtained candidate transmission path set includes:

[0077] Counting the number of connection edges of each node in the network node connection relationship graph to obtain a node connectivity value, and dividing the nodes in the network node connection relationship graph into multiple connection layers according to a preset first connectivity threshold and a preset second connectivity threshold;

[0078] Obtain the geographic coordinate information of each node, perform regional clustering operations on the nodes in each connection layer, calculate the Euclidean distance between nodes, and generate a node regional distribution sequence;

[0079] Calculate the Euclidean distance between each node, obtain the connectivity ratio of each node, record the historical number of successful transmissions and the total number of transmissions between each node, and calculate the weighted transmission state transition value;

[0080] determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value;

[0081] The state space is searched multiple times using a random sampling method to select a state transition path whose replacement value meets a preset cost threshold, and the state transition path is mapped to the actual node access sequence to generate a set of candidate transmission paths.

[0082] In one specific embodiment, the number of edges connecting each node in the network node connection relationship diagram is counted to obtain the node's connectivity value. For example, in a network consisting of 50 nodes, node A is connected to 8 edges, so its connectivity is 8; node B is connected to 3 edges, and its connectivity is 3. Based on a pre-set first connectivity threshold of 5 and a second connectivity threshold of 10, the nodes are divided into three connection layers: nodes with a connectivity less than 5 are classified into the first connection layer (such as node B), nodes with a connectivity between 5 and 10 are classified into the second connection layer (such as node A), and nodes with a connectivity greater than 10 are classified into the third connection layer. This stratification helps prioritize nodes with better connectivity when selecting subsequent paths.

[0083] After obtaining the geographic coordinate information of each node, perform regional clustering operations on the nodes in each connection layer. Taking the second connection layer as an example, assume that this layer contains 20 nodes, each with its own geographic coordinates (x, y). Calculate the Euclidean distance between any two nodes. For example, the Euclidean distance between node A (10, 15) and node C (13, 17) is 5. Set the clustering radius to 10 and group nodes with a distance less than this value into a regional cluster. Through clustering operations, this connection layer may be divided into 3 regional clusters, containing 7, 8, and 5 nodes respectively. Perform similar operations on all connection layers to finally generate a node regional distribution sequence.

[0084] Calculate the Euclidean distance between each node in the network. For example, the distance between node D (20, 30) and node E (25, 35) is 7.07. Obtain the connectivity ratio of each node. For example, if the connectivity of node F is 7 and the connectivity of node G is 14, then the connectivity ratio of F to G is 0.5. Record the historical transmission status between each node. For example, the transmission from node H to node I was successful 85 times in the past 100 times, and its historical transmission success rate is 0.85. Based on these three indicators, calculate the transmission state transition value by weight. Assuming the distance weight is 0.3, the connectivity ratio weight is 0.3, and the historical transmission success rate weight is 0.4, the transmission state transition value between nodes can be calculated. For example, the transmission state transition value of the node pair (H, I) may be 0.75.

[0085] The probability of inter-node transitions is determined based on the transmission state transition value. A higher transition probability indicates a more superior link. A dynamic search depth is set, initially to 3 and dynamically adjusted to 5 or 7 based on network size. A state space search operation is performed to generate a node state transition sequence. For example, starting from source node S, the state transition sequence [S→A→C→D→T] may be generated, where T is the target node. The total cost of this sequence is calculated, assuming it is 28.5. The search is performed multiple times to generate multiple state transition sequences, such as [S→B→E→T] with a cost of 32.1 and [S→F→G→H→T] with a cost of 26.8. Node state transition sequences are optimized based on the cost, and the sequence with the lowest cost is selected as the preferred path.

[0086] A random sampling method is used to perform multiple traversal searches of the state space. In each search, the next hop node is randomly selected with a certain probability to avoid falling into a local optimal solution. For example, performing 100 random sampling searches yields 100 possible paths. A preset cost threshold is set to 30, and paths with a replacement cost less than 30 are selected, such as [S→A→C→D→T] and [S→F→G→H→T]. These state transition paths are mapped to the actual node access sequence to generate a set of candidate transmission paths.

[0087] In this embodiment, by stratifying the connectivity of network nodes and performing regional clustering in combination with the node geographic coordinate information, the organizational structure of network nodes is made clearer and the complexity of path search is reduced; in the path selection process, the node connectivity ratio, Euclidean distance and historical transmission success rate are simultaneously considered, which can comprehensively reflect the actual transmission capacity between nodes, thereby improving the accuracy of path evaluation; state space search is performed based on the transmission state transition value and dynamic search depth, and optimization is performed in combination with the cost value, which helps to achieve dynamic optimization of the transmission path; the state space is traversed multiple times through random sampling methods to avoid falling into local optimal solutions, thereby improving the probability of selecting the global optimal path.

[0088] In an optional embodiment, determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value includes:

[0089] Determine the inter-node transfer probability according to the transmission state transfer value;

[0090] Taking the starting node as the search starting point, performing a state space search operation, setting the starting node as the current search node, and initializing a dynamic search depth value;

[0091] Selecting a next-hop node of the current search node based on the inter-node transition probability, and recording the next-hop node in the visited node set; counting the historical access frequency of the next-hop node, and when the historical access frequency is greater than a preset access threshold, multiplying the inter-node transition probability of the next-hop node by a preset attenuation coefficient;

[0092] When the next-hop node is the target node, the current search path is recorded as a node state transition sequence; when the next-hop node is not the target node and is not in the visited node set, the next-hop node is set as the current search node, the dynamic search depth value is updated based on the preset depth adjustment coefficient, and the state space search operation is continued; when the next-hop node is in the visited node set, return to the previous level node to reselect the next-hop node; when the search depth reaches the dynamic search depth value, the path cost is calculated based on the cost value of each node in the visited node set and the distance from the target node, and when the path cost is greater than the preset path cost threshold, return to the previous level node to reselect the next-hop node;

[0093] Calculating the cost of each transition in the node state transition sequence;

[0094] The node state transition sequence is optimized based on the cost value, and a node state transition sequence whose cost value is less than a preset cost threshold and whose node hop count is less than a preset hop count threshold is selected as the final node state transition sequence.

[0095] In a specific embodiment, during network transmission, the state transition between nodes generally depends on various network conditions, such as bandwidth, latency, and packet loss rate. By quantifying these conditions, a transmission state transition value can be calculated. Specifically, assume that there are nodes A, B, C, D, and E in the network, where A is the starting node and E is the destination node. The transmission state transition value between nodes can be obtained through network detection. For example, the transmission state transition value from node A to B is 0.8, from A to C is 0.6, from B to D is 0.7, from C to D is 0.5, and from D to E is 0.9.

[0096] The determination of inter-node transition probabilities is directly based on the transmission state transition value, which can be obtained by normalizing the transition value. For node A, the transition probabilities of its neighboring nodes B and C are 0.8 / (0.8+0.6)=0.57 and 0.6 / (0.8+0.6)=0.43, respectively.

[0097] The state space search operation begins at the starting node A. The initial dynamic search depth is 3, indicating the maximum depth allowed for the initial search. The preset depth adjustment coefficient is 0.8, the preset access threshold is 2, the preset decay coefficient is 0.5, the preset path cost threshold is 5.0, the preset cost threshold is 10.0, and the preset hop count threshold is 4.

[0098] During the search, the current search node is initially set to the starting node A. Based on the inter-node transition probabilities, the next hop node is selected from A's neighboring nodes B and C. Assume that the value produced by the random number generator is 0.3, which is less than B's transition probability of 0.57. Therefore, B is selected as the next hop node. B is recorded in the visited node set {B}. The historical visit frequency of B is also counted as 1, which does not exceed the preset access threshold of 2. Therefore, there is no need to adjust the transition probability.

[0099] Since B is not the target node E and is not in the visited node set (the visited node set is now {B}), B is set as the current search node. The dynamic search depth value is updated to 3 × 0.8 = 2.4, rounded up to 2. The state space search continues, selecting the next hop node D from B's neighboring nodes. D is added to the visited node set {B, D}, and the historical visit frequency of D is counted as 1.

[0100] D is not the target node and is not in the visited node set. Set D as the current search node. Update the dynamic search depth to 2 × 0.8 = 1.6, rounded to 2. Continue the search, selecting D's neighboring node E as the next hop. Since E is the target node, record the current search path A→B→D→E as a node state transition sequence.

[0101] At the same time, other possible paths are explored. For example, starting from A, select C as the next hop node, then select D, and finally reach E, forming the path A→C→D→E.

[0102] For each node state transition sequence, calculate the cost of each transition. The cost can be based on various factors, such as transmission delay and energy consumption. Assuming the cost of A→B is 2.0, the cost of B→D is 3.0, and the cost of D→E is 1.5, the total cost of the path A→B→D→E is 6.5. If the cost of A→C is 1.5, the cost of C→D is 2.5, and the cost of D→E is 1.5, the total cost of the path A→C→D→E is 5.5.

[0103] During the search process, if a node's historical visit frequency exceeds a preset access threshold, its transition probability needs to be adjusted. For example, if node D is visited again and its historical visit frequency increases to 3, exceeding the preset access threshold of 2, D's transition probability is multiplied by a preset attenuation factor of 0.5 to reduce the likelihood of repeated visits.

[0104] When the search depth reaches the dynamic search depth, the cost of the current path is evaluated. For example, if starting from A, passing through B and C, and arriving at D, the search depth is 3, which is equal to the dynamic search depth. The calculated path cost is 2.0 for A→B, 2.2 for B→C, and 2.5 for C→D, for a total of 6.7. This exceeds the preset path cost threshold of 5.0, so the process returns to the previous node and reselects the next hop.

[0105] All node state transition sequences are optimized based on cost. The total cost of path A→C→D→E is 5.5, which is less than the preset cost threshold of 10.0. Its node hop count is 3, which is less than the preset hop count threshold of 4. The total cost of path A→B→D→E is 6.5, also meeting the requirements. Comparing the two paths, A→C→D→E has the lower cost, so A→C→D→E is selected as the final node state transition sequence.

[0106] like Figure 2 As shown, a dynamic probability state transition path search and optimization flowchart is shown.

[0107] In this embodiment, by determining the transition probability between nodes based on the transmission state transition value and performing state space search in combination with the dynamic search depth, the search complexity can be reduced while ensuring the adequacy of the search; by utilizing the statistical results of the visited node set and the historical access frequency, and introducing a probability decay mechanism when the preset access threshold is exceeded, repeated access and invalid search are effectively avoided; during the search process, the search depth is dynamically updated according to the access situation and the preset depth adjustment coefficient, so that the search process has adaptability and flexibility; the path cost is evaluated in combination with the node cost value and the distance to the target node, and the node state transition sequence is optimized through the dual constraints of the cost threshold and the hop count threshold, thereby ensuring the high quality and rationality of the selected path.

[0108] In an optional embodiment, collecting historical delay data of each candidate transmission path, performing average calculation based on a preset time window, and obtaining a predicted delay value of each candidate transmission path through weighted calculation includes:

[0109] Collecting historical delay data of each candidate transmission path, regularly sampling the historical delay data according to a preset sampling time interval, and generating a delay sampling sequence;

[0110] Based on a preset first time window and a preset second time window, the time-delay sampling sequence is divided into data, wherein the second time window is larger than the first time window;

[0111] Calculate the mean of the sampled data in the first time window and the sampled data in the second time window to determine the short-term delay mean and the long-term delay mean respectively;

[0112] Calculating a deviation rate between the short-term delay average and the long-term delay average, and determining a delay prediction weight coefficient according to the deviation rate;

[0113] Performing weighted calculation on the delay prediction weight coefficient and the delay sampling value at the current moment to obtain the predicted delay value of each candidate transmission path;

[0114] The prediction accuracy of the predicted delay value is counted, and when the prediction accuracy is lower than a preset accuracy threshold, the delay prediction weight coefficient is updated based on a preset adjustment step size, and the predicted delay value is recalculated.

[0115] In one specific embodiment, to collect historical delay data for each candidate transmission path, multiple probe nodes can be deployed in the network. These nodes periodically send probe packets to the target server and record the round-trip time. The sampling interval can be set to 500 milliseconds, that is, delay data is collected every 500 milliseconds. For each candidate path, a delay data record table is maintained, containing a timestamp and the corresponding delay value, forming a delay sampling sequence. For example, some sampled data for a path may be: [2023-10-01 08:00:00, 25ms], [2023-10-01 08:00:00.5, 27ms], [2023-10-01 08:00:01, 26ms], etc.

[0116] After the delay sampling sequence is generated, the data is divided into preset time windows. The first time window (short-term window) can be set to 10 seconds, and the second time window (long-term window) can be set to 60 seconds. The short-term window reflects immediate changes in network status, while the long-term window reflects overall network stability. In specific implementations, a sliding window mechanism is maintained to continuously update the data within the window over time. For example, if the current time is 08:01:00, the short-term window includes data from 08:00:50 to 08:01:00 (a total of 20 sampling points), and the long-term window includes data from 08:00:00 to 08:01:00 (a total of 120 sampling points).

[0117] When calculating the mean of the divided data, we calculate the arithmetic mean of all sampling points within the short-term window and the long-term window. Assuming the sum of the latency values ​​of the 20 sampling points within the short-term window is 520ms, the short-term mean latency is 26ms. If the sum of the latency values ​​of the 120 sampling points within the long-term window is 2880ms, the long-term mean latency is 24ms.

[0118] Calculating the deviation rate between the short-term and long-term average delays is a key step in latency prediction. The deviation rate is calculated by dividing the difference between the short-term and long-term averages by the long-term average, and taking the absolute value. Continuing with the previous example, the deviation rate is |26 - 24| / 24 = 0.0833, or approximately 8.33%. The deviation rate reflects the changing trend of network status; a larger deviation rate indicates greater network fluctuations.

[0119] An adaptive weighting mechanism is used to determine the latency prediction weight coefficient based on the deviation rate. When the deviation rate is small (e.g., less than 5%), the current sample value is given a higher weight (e.g., 0.7) and the historical average is given a lower weight (e.g., 0.3). When the deviation rate is large (e.g., greater than 20%), the historical average is given a higher weight (e.g., 0.8) and the current sample value is given a lower weight (e.g., 0.2). When the deviation rate is in the intermediate range, the weight coefficient changes linearly. In the above example, the deviation rate is 8.33%, which is in the intermediate range. The system might assign a weight of 0.6 to the current sample value and a weight of 0.4 to the historical average.

[0120] When performing weighted calculations, the current sampled latency value is combined with the historical average according to the weights. Assuming the current sampled latency value is 28 ms with a weight of 0.6, and the historical average (which can be the long-term latency average) is 24 ms with a weight of 0.4, the predicted latency is 28 × 0.6 + 24 × 0.4 = 26.4 ms. The predicted latency is calculated for each candidate path and stored in the predicted latency table.

[0121] Evaluating forecast accuracy is crucial for ensuring forecast quality. The difference between each predicted value and the actual observed value is recorded, and the forecast error rate is calculated. For example, if the predicted latency is 26.4ms and the actual observed latency at the next moment is 25ms, the forecast error rate is |26.4 - 25| / 25 = 0.056, or approximately 5.6%. Maintain the average error rate over a sliding window (e.g., the most recent 100 forecasts) as a metric for evaluating forecast accuracy.

[0122] When the prediction accuracy falls below a preset threshold (e.g., the average error rate exceeds 10%), a weight adjustment mechanism is initiated. The adjustment step size can be set to 0.05, and the weight coefficient is adjusted based on the direction of the prediction error. If the predicted value is generally higher than the actual value, the weight of the current sample value is reduced and the weight of the historical average is increased; vice versa. After the adjustment, the system recalculates the predicted latency using the new weight coefficient and continues to monitor the prediction accuracy, forming a closed-loop optimization mechanism.

[0123] This method accurately predicts the latency of each candidate transmission path, providing a reliable basis for network transmission path selection and effectively improving network transmission efficiency and user experience. This method is highly adaptable and can handle latency variations in various complex network environments, suggesting broad application prospects.

[0124] In an optional embodiment, based on the delay difference comparison result between the candidate transmission paths, determining the delay weight coefficient of each candidate transmission path by subset division, and generating the delay weight distribution sequence includes:

[0125] Calculate the delay difference between each candidate transmission path and all other candidate transmission paths, and establish a delay difference set for each transmission path;

[0126] Comparing the positive and negative attributes of each delay difference in the delay difference value set of each candidate transmission path, classifying the delay difference greater than zero into the positive value subset, and classifying the delay difference less than zero into the negative value subset;

[0127] The delay differences in the positive subset of each candidate transmission path are accumulated to obtain the delay disadvantage value, and the delay differences in the negative subset are accumulated to obtain the delay advantage value;

[0128] Based on the ratio between the delay disadvantage value and the delay advantage value, the weight coefficient of the corresponding candidate transmission path is determined through exponential operation; the weight coefficients of all candidate transmission paths are normalized and the delay weight distribution sequence is determined.

[0129] In a specific embodiment, calculating the delay difference between candidate transmission paths is a basic step of this method. For the n candidate transmission paths in the autonomous driving network, the end-to-end delay value of each path is measured respectively, which is recorded as T1, T2, ..., Tn. In order to evaluate the delay performance of each path relative to other paths, it is necessary to calculate the delay difference Dij between path i and all other paths j. Dij is calculated as Ti minus Tj, which represents the delay difference of path i relative to path j. For example, in the case of 5 candidate paths, the delay of path 1 is 25 milliseconds and the delay of path 2 is 30 milliseconds, then D12 is -5 milliseconds, indicating that path 1 is 5 milliseconds faster than path 2. In this way, a delay difference set containing n-1 elements is established for each candidate transmission path.

[0130] Classifying delay difference sets is a crucial step in evaluating path performance. For each candidate transmission path i, its delay difference set {Di1, Di2, ..., Din} (excluding Dii) is classified according to its positive and negative attributes. Delay differences greater than zero are classified into the positive subset Pi, while delay differences less than zero are classified into the negative subset Ni. A positive delay difference indicates that the path has a longer delay than the comparison path; a negative delay difference indicates a shorter delay. In practice, for example, if the delay difference set for path 3 is {5, -3, -2, 4}, its positive subset P3 is {5, 4}, and its negative subset N3 is {-3, -2}.

[0131] The calculation of delay advantage and disadvantage directly impacts the path evaluation results. For each candidate transmission path i, all delay differences in the positive subset Pi are accumulated to obtain the delay disadvantage value PSumi, which represents the cumulative time that the path is slower than other paths. All delay differences in the negative subset Ni are accumulated to obtain the delay advantage value NSumi, which represents the cumulative time that the path is faster than other paths. For path 3, for example, its delay disadvantage value PSum3 is 9 milliseconds (5 + 4), and its delay advantage value NSum3 is -5 milliseconds (-3 + (-2)). The smaller the delay disadvantage value and the larger the absolute value of the delay advantage value, the better the path performance.

[0132] The weight coefficient is determined using an exponential calculation method, reflecting the nonlinear relationship between path performance and weight distribution. For candidate transmission path i, the weight coefficient Wi is calculated based on its delay disadvantage value PSumi and delay advantage value NSumi. When the delay advantage value NSumi is 0, Wi = 1; when NSumi is not 0, Wi is equal to the power operation result with e as the base and the negative of the ratio of PSumi to |NSumi| as the exponent. For example, for path 3, its weight coefficient W3 = e (-(9 / 5)) ≈0.166. This exponential calculation method ensures that paths with better latency performance receive higher weight coefficients.

[0133] Weight normalization ensures the rationality of resource allocation. The sum of the weight coefficients of all candidate transmission paths, Sum_W, is calculated, and then the weight coefficient of each path is divided by Sum_W to obtain the normalized weight value NWi. In this way, the sum of the normalized weights of all paths is equal to 1, forming the delay weight distribution sequence {NW1, NW2, ..., NWn}. In a network with five paths, assuming the calculated weight coefficients are W1 = 0.368, W2 = 0.223, W3 = 0.166, W4 = 0.135, and W5 = 0.108, then Sum_W = 1.0, and the normalized weight distribution sequence is {0.368, 0.223, 0.166, 0.135, 0.108}.

[0134] In practical applications, using a self-driving vehicle communication network as an example, the end-to-end delays of the four available transmission paths were measured to be 18 milliseconds, 22 milliseconds, 25 milliseconds, and 20 milliseconds, respectively. By calculating the delay differentials, the delay differential set for path 1 is {-4, -7, -2}, all of which are negative, indicating that path 1 has better delay performance than all other paths. The delay differential set for path 2 is {4, -3, 2}, with the positive subset being {4, 2} and the negative subset being {-3}. The delay differential set for path 3 is {7, 3, 5}, all of which are positive, indicating that path 3 has the worst delay performance. The delay differential set for path 4 is {2, -2, -5}, with the positive subset being {2} and the negative subset being {-2, -5}.

[0135] Calculate the delay disadvantage and advantage of each path: PSum1 = 0, NSum1 = -13 for path 1; PSum2 = 6, NSum2 = -3 for path 2; PSum3 = 15, NSum3 = 0 for path 3; PSum4 = 2, NSum4 = -7 for path 4. Based on this, calculate the weight coefficient: W1 = e (-(0 / 13)) =1,W2=e (-(6 / 3)) ≈0.135, W3=1 (because NSum3=0), W4=e (-(2 / 7)) The sum of the weight coefficients Sum_W = 2.886. After normalization, the delay weight distribution sequence is {0.347, 0.047, 0.347, 0.260}.

[0136] This delay weight distribution sequence is applied to network resource allocation, such as bandwidth allocation or packet scheduling priority setting. For example, with a total bandwidth of 100 Mbps, path 1 is allocated 34.7 Mbps, path 2 is allocated 4.7 Mbps, path 3 is allocated 34.7 Mbps, and path 4 is allocated 26.0 Mbps. This allocation method fully considers the latency characteristics of each path, prioritizing paths with better latency performance to receive more resources.

[0137] In this embodiment, the time-sensitive network can dynamically adjust the resource allocation strategy based on the path delay performance, improve the real-time communication efficiency and reliability of the autonomous driving system, meet the strict delay requirements of functions such as vehicle control and environmental perception, and provide a solid network foundation for safe driving.

[0138] In an optional embodiment, iteratively updating the delay weight distribution sequence based on the hierarchical time scale, scoring the candidate transmission paths according to the node load rate data and the node delay evaluation parameter, and selecting the candidate transmission path with the highest comprehensive score as the optimal transmission path includes:

[0139] Obtaining a delay weight distribution sequence of a candidate transmission path, and combining the delay weight distribution sequence with a current path delay state to form a state parameter;

[0140] Perform weight learning on each candidate transmission path, calculate the current weight value based on the delay weight distribution sequence of the candidate transmission path, use the ratio of the candidate transmission path's historical transmission success rate to the weight volatility as the weight prediction coefficient, and modify the current weight value based on the weight prediction coefficient to determine the expected weight value;

[0141] A hierarchical time-scale weight update mechanism is constructed. In the first time scale, the current weight value is updated based on the change of the path state. In the second time scale, the calculation parameters of the expected weight value are updated based on the path stability. The weighted sum of the weight mean of the first time scale and the weight prediction value of the second time scale is used to obtain the iterative weight value.

[0142] Collect the load rate data of each node on the candidate transmission path, and calculate the node delay evaluation parameters based on the data transmission rate and data packet round-trip time between nodes;

[0143] Calculate the load balancing score based on the load rate data, calculate the delay score based on the node delay evaluation parameters, and use the weighted sum of the load balancing score and the delay score as the path base score;

[0144] The path basic score is multiplied by the iterative weight value to obtain the path comprehensive score, and the candidate transmission path with the highest path comprehensive score and greater than the preset score threshold is selected as the optimal transmission path.

[0145] In one embodiment, a delay weight distribution sequence for a candidate transmission path is obtained. This sequence reflects the weighted distribution of the delays at each node along the path. The delay weight distribution sequence can be represented by a set of values. For example, for a path containing five nodes, the delay weight distribution sequence might be [0.15, 0.22, 0.18, 0.25, 0.20]. These weight values, together with the delay status of the current path, constitute state parameters. The delay status can include network quality indicators such as the average delay and packet loss rate along the path.

[0146] When performing weight learning for each candidate transmission path, the current weight is calculated based on the candidate transmission path's delay weight distribution sequence. For example, the weighted average of the weights of each node in the sequence is used as the current weight. If the weight distribution is [0.15, 0.22, 0.18, 0.25, 0.20] and the actual measured delays at each node are [5ms, 8ms, 6ms, 9ms, 7ms], the current weight is calculated as 0.15 × 5 + 0.22 × 8 + 0.18 × 6 + 0.25 × 9 + 0.20 × 7 = 7.17ms. The historical transmission success rate and weight volatility of the candidate transmission path are obtained. Assuming a historical transmission success rate of 0.95 and a weight volatility of 0.08, the weight prediction coefficient is 0.95 / 0.08 = 11.875. This prediction coefficient is used to adjust the current weight, for example, using a linear adjustment formula to obtain an expected weight of 6.85ms.

[0147] A hierarchical weight update mechanism is constructed, encompassing two timescales. Within the first timescale (e.g., 10 seconds), path status changes are monitored every second and the current weight value is updated. Suppose that within these 10 seconds, the measured weight value sequence is [7.17, 7.22, 7.15, 7.18, 7.25, 7.20, 7.16, 7.19, 7.21, 7.23], with an average of 7.196 ms. Within the second timescale (e.g., 1 minute), path stability is assessed every 10 seconds and the weight prediction parameters are updated. If the path appears stable within 1 minute, the system might adjust the weight prediction coefficient from 11.875 to 12.5, and calculate a weight prediction value of 6.80 ms based on the new coefficient. The weighted mean value of the first time scale and the weighted predicted value of the second time scale are weighted and summed. Assuming that the weight coefficients are 0.4 and 0.6, the weight value after iteration is 0.4×7.196+0.6×6.80=6.958ms.

[0148] Collect load data for each node on the candidate transmission path, including CPU utilization, memory usage, and network bandwidth utilization. For example, the CPU utilization of five nodes on the path is [45%, 60%, 38%, 52%, 41%], the memory utilization is [50%, 65%, 42%, 58%, 46%], and the network bandwidth utilization is [35%, 48%, 30%, 42%, 37%]. Node latency evaluation parameters are also calculated based on the inter-node data transmission rate and packet round-trip time. If the data transmission rate between two nodes is 100 Mbps, the packet size is 1 KB, and the round-trip time is 10 ms, the calculated latency evaluation parameter for the link is 0.8.

[0149] The load balancing score is calculated based on the load rate data. This can be calculated by negating the standard deviation of each node's load rate. A smaller standard deviation indicates a more balanced load and a higher score. For example, the standard deviation of the CPU utilization mentioned above is 8.44%. After negating and normalizing, the load balancing score is 0.75. To calculate the latency score based on the node latency evaluation parameters, the weighted average of the latency evaluation parameters of each link can be taken, for example, to obtain a latency score of 0.82. Assuming the weights of the load balancing score and latency score are 0.4 and 0.6, respectively, the path base score is 0.4 × 0.75 + 0.6 × 0.82 = 0.792.

[0150] Multiply the path base score by the iterative weight value to obtain the path composite score. Assuming the iterative weight value is 6.958ms, normalized to 0.85 for ease of calculation, the path composite score is 0.792 × 0.85 = 0.6732. If the preset score threshold is 0.65, and the path score of 0.6732 exceeds the threshold and is the highest among all candidate paths, then this path is selected as the optimal transmission path.

[0151] When a network status change is detected, such as a sudden increase of more than 20% in node load or a 30% increase in link latency, the system re-determines the delay weight distribution sequence and performs the path optimization process. This dynamic adjustment mechanism adapts to network changes, consistently selecting the optimal transmission path and improving data transmission efficiency and reliability.

[0152] The time-sensitive network delay control system for autonomous driving according to an embodiment of the present invention includes:

[0153] The first unit is used to obtain data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information;

[0154] The second unit is used to generate a network node connection relationship graph based on the network topology information, hierarchize the nodes based on the connectivity values ​​of the nodes in the network node connection relationship graph, perform a state space search operation to generate a node state transition sequence, and obtain a set of candidate transmission paths;

[0155] The third unit is used to collect historical delay data of each candidate transmission path, perform average calculation based on a preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation;

[0156] The fourth unit is configured to determine a delay weight coefficient of each candidate transmission path by subset partitioning based on the delay difference comparison result between the candidate transmission paths, and generate a delay weight distribution sequence;

[0157] The fifth unit is used to iteratively update the delay weight distribution sequence based on the hierarchical time scale, score the candidate transmission paths according to the node load rate data and the node delay evaluation parameters, and select the candidate transmission path with the highest comprehensive score as the optimal transmission path;

[0158] The sixth unit is used to forward data packets through the optimal transmission path. When it is detected that the real-time delay change value exceeds the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

[0159] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0160] processor;

[0161] a memory for storing processor-executable instructions;

[0162] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0163] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0164] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A time-sensitive network delay control method for autonomous driving, characterized by: include: Obtain the data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information; Generate a network node connection relationship graph based on network topology information, stratify nodes based on their connectivity values ​​in the network node connection relationship graph, perform state space search operations to generate a node state transition sequence, and obtain a set of candidate transmission paths; Collect historical delay data of each candidate transmission path, calculate the average value based on the preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation; Based on the delay difference comparison results between candidate transmission paths, the delay weight coefficient of each candidate transmission path is determined by subset partitioning, and a delay weight distribution sequence is generated; It iteratively updates the delay weight distribution sequence based on the hierarchical time scale, scores the candidate transmission paths according to the node load rate data and node delay evaluation parameters, and selects the candidate transmission path with the highest comprehensive score as the optimal transmission path; Data packets are forwarded through the optimal transmission path. When the real-time delay change value is detected to exceed the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

2. The method according to claim 1, characterized in that The nodes are layered based on their connectivity values ​​in the network node connection relationship graph, and a state space search operation is performed to generate a node state transition sequence. The candidate transmission path set obtained includes: Counting the number of connection edges of each node in the network node connection relationship graph to obtain a node connectivity value, and dividing the nodes in the network node connection relationship graph into multiple connection layers according to a preset first connectivity threshold and a preset second connectivity threshold; Obtain the geographic coordinate information of each node, perform regional clustering operations on the nodes in each connection layer, calculate the Euclidean distance between nodes, and generate a node regional distribution sequence; Calculate the Euclidean distance between each node, obtain the connectivity ratio of each node, record the historical number of successful transmissions and the total number of transmissions between each node, and calculate the weighted transmission state transition value; determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value; The state space is searched multiple times using a random sampling method to select a state transition path whose replacement value meets a preset cost threshold, and the state transition path is mapped to the actual node access sequence to generate a set of candidate transmission paths.

3. The method according to claim 2, characterized in that Determining an inter-node transition probability based on the transmission state transition value, performing a state space search operation based on a dynamic search depth value to generate a node state transition sequence, calculating a cost value for each transition, and optimizing the node state transition sequence based on the cost value includes: Determine the inter-node transfer probability according to the transmission state transfer value; Taking the starting node as the search starting point, performing a state space search operation, setting the starting node as the current search node, and initializing a dynamic search depth value; Selecting a next-hop node of the current search node based on the inter-node transition probability, and recording the next-hop node in the visited node set; counting the historical access frequency of the next-hop node, and when the historical access frequency is greater than a preset access threshold, multiplying the inter-node transition probability of the next-hop node by a preset attenuation coefficient; When the next-hop node is the target node, the current search path is recorded as a node state transition sequence; when the next-hop node is not the target node and is not in the visited node set, the next-hop node is set as the current search node, the dynamic search depth value is updated based on the preset depth adjustment coefficient, and the state space search operation is continued; when the next-hop node is in the visited node set, return to the previous level node to reselect the next-hop node; when the search depth reaches the dynamic search depth value, the path cost is calculated based on the cost value of each node in the visited node set and the distance from the target node, and when the path cost is greater than the preset path cost threshold, return to the previous level node to reselect the next-hop node; Calculating the cost of each transition in the node state transition sequence; The node state transition sequence is optimized based on the cost value, and a node state transition sequence whose cost value is less than a preset cost threshold and whose node hop count is less than a preset hop count threshold is selected as the final node state transition sequence.

4. The method according to claim 1, wherein Collect historical delay data for each candidate transmission path, calculate the average value based on the preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation, including: Collecting historical delay data of each candidate transmission path, regularly sampling the historical delay data according to a preset sampling time interval, and generating a delay sampling sequence; Based on a preset first time window and a preset second time window, the time-delay sampling sequence is divided into data, wherein the second time window is larger than the first time window; Calculate the mean of the sampled data in the first time window and the sampled data in the second time window to determine the short-term delay mean and the long-term delay mean respectively; Calculating a deviation rate between the short-term delay mean and the long-term delay mean, and determining a delay prediction weight coefficient according to the deviation rate; Performing weighted calculation on the delay prediction weight coefficient and the delay sampling value at the current moment to obtain the predicted delay value of each candidate transmission path; The prediction accuracy of the predicted delay value is counted, and when the prediction accuracy is lower than a preset accuracy threshold, the delay prediction weight coefficient is updated based on a preset adjustment step size, and the predicted delay value is recalculated.

5. The method according to claim 1, characterized in that Based on the delay difference comparison results between candidate transmission paths, the delay weight coefficient of each candidate transmission path is determined by subset division, and a delay weight distribution sequence is generated, including: Calculate the delay difference between each candidate transmission path and all other candidate transmission paths, and establish a delay difference set for each transmission path; Comparing the positive and negative attributes of each delay difference in the delay difference value set of each candidate transmission path, classifying the delay difference greater than zero into the positive value subset, and classifying the delay difference less than zero into the negative value subset; The delay differences in the positive subset of each candidate transmission path are accumulated to obtain the delay disadvantage value, and the delay differences in the negative subset are accumulated to obtain the delay advantage value; Based on the ratio between the delay disadvantage value and the delay advantage value, the weight coefficient of the corresponding candidate transmission path is determined through exponential operation; the weight coefficients of all candidate transmission paths are normalized and the delay weight distribution sequence is determined.

6. The method according to claim 1, wherein The delay weight distribution sequence is iteratively updated based on the hierarchical time scale. The candidate transmission paths are scored according to the node load rate data and node delay evaluation parameters. The candidate transmission path with the highest comprehensive score is selected as the optimal transmission path, including: Obtaining a delay weight distribution sequence of a candidate transmission path, and combining the delay weight distribution sequence with a current path delay state to form a state parameter; Perform weight learning on each candidate transmission path, calculate the current weight value based on the delay weight distribution sequence of the candidate transmission path, use the ratio of the candidate transmission path's historical transmission success rate to the weight volatility as the weight prediction coefficient, and modify the current weight value based on the weight prediction coefficient to determine the expected weight value; A hierarchical time-scale weight update mechanism is constructed. In the first time scale, the current weight value is updated based on the change of the path state. In the second time scale, the calculation parameters of the expected weight value are updated based on the path stability. The weighted sum of the weight mean of the first time scale and the weight prediction value of the second time scale is used to obtain the iterative weight value. Collect the load rate data of each node on the candidate transmission path, and calculate the node delay evaluation parameters based on the data transmission rate and data packet round-trip time between nodes; Calculate the load balancing score based on the load rate data, calculate the delay score based on the node delay evaluation parameters, and take the weighted sum of the load balancing score and the delay score as the path base score; The path basic score is multiplied by the iterated weight value to obtain the path comprehensive score, and the candidate transmission path with the highest path comprehensive score and greater than the preset score threshold is selected as the optimal transmission path.

7. A time-sensitive network delay control system for autonomous driving, configured to implement the method according to any one of claims 1 to 6, characterized in that: include: The first unit is used to obtain data packet transmission delay information and network topology information of the autonomous driving vehicle, and calculate the node delay evaluation parameter based on the transmission delay information; The second unit is used to generate a network node connection relationship graph based on the network topology information, hierarchize the nodes based on the connectivity values ​​of the nodes in the network node connection relationship graph, perform a state space search operation to generate a node state transition sequence, and obtain a set of candidate transmission paths; The third unit is used to collect historical delay data of each candidate transmission path, perform average calculation based on a preset time window, and obtain the predicted delay value of each candidate transmission path through weighted calculation; The fourth unit is configured to determine a delay weight coefficient of each candidate transmission path by subset division based on a delay difference comparison result between candidate transmission paths, and generate a delay weight distribution sequence; The fifth unit is used to iteratively update the delay weight distribution sequence based on the hierarchical time scale, score the candidate transmission paths according to the node load rate data and the node delay evaluation parameters, and select the candidate transmission path with the highest comprehensive score as the optimal transmission path; The sixth unit is used to forward data packets through the optimal transmission path. When it is detected that the real-time delay change value exceeds the preset delay change range, the delay weight coefficient of each path is recalculated using the latest collected historical delay data and node delay evaluation parameters.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images