A Distributed Closed-Loop Control Method for Vehicle Network Channel Congestion

Abstract

This invention belongs to the field of vehicle-to-everything (V2X) control technology and discloses a distributed closed-loop control method for V2X channel congestion. This method proposes a closed-loop feedback channel resource congestion control strategy. The closed-loop feedback control framework adopted by this strategy consists of a reference model and an adaptive feedback control model. In the reference model, a packet reception rate model and a network effect maximization model are constructed, and the optimal packet reception probability is solved as the ideal state of the V2X. In the adaptive closed-loop feedback control model, a PI controller adjusts the error between the ideal state and the actual network state, allowing the actual network state to infinitely approximate the ideal network state. This invention dynamically adjusts network parameters based on model reference adaptive control, ensuring the reliability of communication between vehicles in dynamic network environments and maintaining the system's expected stability and robustness in dynamically changing environments.

Description

Technical Field

[0001] This invention belongs to the field of vehicle network control technology and relates to a distributed closed-loop control method for vehicle network channel congestion. Background Technology

[0002] Vehicle-to-everything (V2X) communication, as a crucial application of the Internet of Things (IoT) in intelligent transportation systems, primarily employs three communication modes: vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-pedestrian (V2P). V2V communication enables direct data interaction between adjacent vehicles. Among various wireless communication technologies, the Dedicated Short Range Communication (DSRC) protocol, utilizing the IEEE 802.11p physical layer / MAC layer protocol, achieves reliable low-latency information transmission in dynamically changing network environments. Cooperative Vehicle Safety Systems (CVSSs) are one of the most challenging applications in V2X. As shown in Figure 1, in a CVSS system, each vehicle periodically broadcasts its own status information (such as position and speed) and safety event information (such as emergency braking) to adjacent vehicles via the IEEE 802.11p protocol. These messages, known as beacons, contain motion status information such as vehicle position, speed, and direction of travel. Vehicles utilize the received beacons to perceive the status of surrounding vehicles and detect potential collision risks in real time. In various traffic scenarios (such as highways and intersections), periodic beacon transmission can achieve a collision risk prediction accuracy of up to 90%.

[0003] Wireless channel resources in vehicle-to-everything (V2X) networks are limited. In high-density scenarios, numerous vehicles compete for and share channels, periodically broadcasting beacon messages, severely increasing channel load and causing congestion. When congestion occurs, data packets collide extensively, preventing vehicles from tracking the status of neighboring vehicles in real time and reducing the collision warning accuracy of cooperative vehicle safety systems. Therefore, a channel congestion control strategy needs to be designed to reduce channel load in dynamic network environments while ensuring reliable vehicle tracking performance.

[0004] Currently, significant progress has been made in research on channel congestion control in vehicular networks (V2V), which can be categorized into three main types from a control theory perspective: transmission power adjustment, beacon transmission rate control, and hybrid power and rate control. These methods adaptively adjust physical layer / MAC layer parameters (such as transmission power and beacon rate) using models such as cooperative game theory and network utility maximization. However, the parameters in these models (such as weighting coefficients and fairness coefficients) remain fixed during optimization, and these strategies primarily rely on predefined models to calculate channel congestion control actions in an open-loop manner. In V2V networks, channel states (such as channel congestion rate) and traffic conditions (such as vehicle density) change dynamically. Existing channel congestion control methods lack feedback mechanisms and cannot correct and optimize previous channel congestion control decisions based on environmental changes, thus making it difficult to maintain reliable V2V communication in dynamically changing V2V environments. Summary of the Invention

[0005] The purpose of this invention is to propose a distributed closed-loop control method for vehicle network channel congestion. This method dynamically adjusts network parameters based on model reference adaptive control to ensure the reliability of communication between vehicles in a dynamic network environment, thereby maintaining the expected stability and robustness of the system in a dynamically changing environment.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A distributed closed-loop control method for channel congestion in vehicle-to-everything (V2X) networks includes the following steps:

[0008] Step 1. Construct a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, construct a data packet reception rate model, and design a network utility maximization model, i.e., a reference model, using a rolling time domain approach.

[0009] The optimal vehicle beacon transmission rate is obtained by solving the dual decomposition and gradient descent methods to obtain the ideal network state.

[0010] Step 2. Construct an adaptive feedback control model. Based on the ideal network state obtained from the reference model and the actual network state obtained, introduce a PI controller and use the error between the ideal network state and the actual network state as the input of the PI controller.

[0011] The PI controller uses MIT parameter tuning, which dynamically optimizes the proportional gain and integral gain of the PI controller based on the error signal between the ideal network state and the actual network state, thereby generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

[0012] Furthermore, based on the aforementioned distributed closed-loop control method for vehicle network channel congestion, this invention also proposes a corresponding distributed closed-loop control system for vehicle network channel congestion. Both are based on the same inventive concept and employ the following technical solutions:

[0013] A distributed closed-loop control system for vehicle network channel congestion includes the following modules:

[0014] The reference model building module is used to build a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, build a data packet reception rate model, and design a network utility maximization model using a rolling time domain approach.

[0015] The optimal vehicle beacon transmission rate is obtained by solving the dual decomposition and gradient descent methods to obtain the ideal network state.

[0016] An adaptive feedback control model building module is used to introduce a PI controller based on the ideal network state obtained from the reference model and the actual network state obtained, and to use the error between the ideal network state and the actual network state as the input of the PI controller.

[0017] The PI controller uses MIT parameter tuning, which dynamically optimizes the proportional gain and integral gain of the PI controller based on the error signal between the ideal network state and the actual network state, thereby generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

[0018] Furthermore, based on the aforementioned distributed closed-loop control method for vehicular network channel congestion, this invention also proposes a computer device, which includes a memory and one or more processors. Executable code is stored in the memory. When the processor executes the executable code, it implements the steps of the aforementioned distributed closed-loop control method for vehicular network channel congestion.

[0019] Furthermore, based on the aforementioned distributed closed-loop control method for vehicle network channel congestion, this invention also proposes a computer-readable storage medium storing a program that, when executed by a processor, is used to implement the steps of the aforementioned distributed closed-loop control method for vehicle network channel congestion.

[0020] The present invention has the following advantages:

[0021] As described above, this invention discloses a distributed closed-loop control method for channel congestion in vehicular networks. This method proposes a closed-loop feedback channel resource congestion control strategy, implemented based on a closed-loop feedback control framework. This framework consists of two parts: a reference model and an adaptive feedback control model. The reference model construction details a packet reception rate model and a network effect maximization model to obtain the optimal beacon rate, thereby achieving the optimal packet successful reception probability (representing the ideal state of the vehicular network). In the adaptive closed-loop feedback control model, a PI controller adjusts the error between the ideal and actual network states, allowing the actual network state to approximate the ideal state infinitely. This invention dynamically adjusts network parameters based on model reference adaptive control to ensure the reliability of vehicle-to-vehicle communication in dynamic network environments. Model reference adaptive control is a closed-loop control method suitable for handling uncertainties and external interference in dynamic systems, maintaining the desired stability and robustness of the system in dynamically changing environments. In this invention, the wireless channel is treated as a dynamically controlled object, and embedded sensors continuously monitor the state of the controlled object to capture channel conditions. This method follows a "observation-evaluation-control" cyclical mechanism, achieving superior vehicle tracking performance in a dynamically uncertain network environment. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a vehicle-to-everything (V2X) communication scenario in an embodiment of the present invention;

[0023] Figure 2 This is a diagram showing the direct collision area and the hidden collision area in an embodiment of the present invention;

[0024] Figure 3 This is a Markov model diagram of the retreat process in an embodiment of the present invention;

[0025] Figure 4 This is a diagram of the distributed closed-loop control framework in an embodiment of the present invention;

[0026] Figure 5 This is a diagram of the MRAC model for beacon rate control in an embodiment of the present invention;

[0027] Figure 6 This is a diagram of the Model Reference Adaptive Control (MRAC) architecture for beacon rate adaptation in an embodiment of the present invention.

[0028] Figure 7 This is a convergence graph of the distributed beacon control algorithm implemented using the rolling time-domain method in this embodiment of the invention;

[0029] Figure 8 This is a beacon rate diagram of the distributed beacon control algorithm under different traffic conditions in a specific embodiment of the present invention;

[0030] Figure 9 The cumulative distribution function (CDF) of the optimal beacon rate under different traffic conditions is shown in a specific example of the present invention.

[0031] Figure 10 This is a convergence graph of the controller in a 400-node scenario in a specific embodiment of the present invention;

[0032] Figure 11 This is a convergence graph of the controller in an 800-node scenario in a specific embodiment of the present invention;

[0033] Figure 12 This is the optimal beacon rate diagram for a 400-node scenario in a specific example of the present invention;

[0034] Figure 13 This is the optimal beacon rate diagram for an 800-node scenario in a specific example of the present invention;

[0035] Figure 14 This is a channel busy rate diagram of four methods in a 400-node scenario in a specific example of the present invention;

[0036] Figure 15 This is a channel busy rate diagram of four methods in an 800-node scenario in a specific example of the present invention;

[0037] Figure 16 This is a cumulative distribution function graph of channel busy rate for four methods in a 400-node scenario in a specific example of the present invention;

[0038] Figure 17 This is a cumulative distribution function graph of channel busy rate for four methods in an 800-node scenario in a specific example of the present invention;

[0039] Figure 18 This is a tracking error diagram of four methods in a 400-node scenario in a specific example of the present invention;

[0040] Figure 19 This is a tracking error diagram of four methods in an 800-node scenario in a specific example of the present invention. Detailed Implementation

[0041] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0042] Example 1

[0043] In vehicle-to-everything (V2X) networks, vehicles periodically broadcast beacons via wireless channels to achieve collaborative perception and collision warning functions. However, limited channel capacity is easily saturated by excessive beacon traffic, leading to channel congestion and the loss of a large number of data packets due to collisions. Current channel congestion control methods mainly rely on open-loop mechanisms, which are difficult to adapt to dynamic network changes (such as dynamic changes in vehicle density and vehicle movement status), ultimately resulting in a decline in data packet reception performance. To address these issues, this embodiment proposes a distributed closed-loop control method for V2X channel congestion. This method first constructs a data packet reception model to describe the data packet success rate between vehicles; secondly, it designs a distributed closed-loop feedback control framework based on a model reference adaptive control method. This feedback control framework consists of two parts: a reference model and an adaptive feedback control model. The reference model uses a rolling time-domain optimization method to determine the ideal network state that meets the preset target; the adaptive feedback control model generates a real-time channel congestion control strategy, i.e., a beacon rate control strategy, by minimizing the deviation between the ideal network state and the actual network state.

[0044] The distributed closed-loop control method for vehicle network channel congestion in this embodiment includes the following steps:

[0045] Step 1. Construct a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, construct a data packet reception rate model, and design a network utility maximization model (i.e., a reference model) using a rolling time domain approach; solve for the optimal vehicle beacon transmission rate through dual decomposition and gradient descent to obtain the ideal network state.

[0046] This invention addresses vehicle-to-everything (V2X) communication in a highway scenario. Vehicles rely on Dedicated Short Range Communication (DSRC) technology to send and receive messages. Each vehicle is equipped with a dedicated DSRC radio, a Global Positioning System (GPS), and onboard sensors. It periodically broadcasts its motion status information (position, speed, acceleration, etc.) to nearby vehicles via beacons to improve vehicle tracking accuracy and avoid collisions. This invention treats the highway scenario as a one-dimensional linear structure with vehicles randomly distributed on the road. Furthermore, all vehicles communicate using a Carrier Sense Multiple Access Collision Avoidance (CSMA / CA) mechanism and have the same transmit power and communication range. This invention uses Packet Receive Rate (PRR) to quantify the reliability of data transmission, defined as the amount of information successfully received by the receiver at a distance x from the source. Constructing the PRR model requires determining the actual beacon transmission rate at the sender and the probability of successful transmission at the receiver for each attempt. Because hidden nodes significantly impact inter-vehicle communication performance, they are incorporated into the linear topology during modeling.

[0047] The highway scenario can be modeled as a one-dimensional linear structure, assuming that the nodes (vehicles) in this topology are randomly distributed along a one-dimensional straight line, as shown in Figure 2. For the receiver to successfully receive messages from the source, it must simultaneously avoid the effects of both direct and hidden collisions.

[0048] A direct collision refers to a situation where nodes within each other's transmission range send messages in the same time slot.

[0049] Assuming the model is a one-dimensional linear topology and each node has the same transmit power, the communication range of each node is fixed (denoted as ). Assume the source node is located at point 0 on the road (i.e., in Figure 2). (Node), whose communication range is from arrive . The node is located at the target distance Place, in Within the node's communication range, its communication range is from arrive The receiving end is at The direct collision region of the points is denoted as It is defined as the distance from the sending end The location is the set of nodes (excluding the source node) within the overlapping area of ​​the transmission ranges of the source node and the receiver. Therefore, The expression is as follows:

[0050] (1)

[0051] in, This represents the set of nodes within the transmission range of the source node. Indicates the target receiving node (i.e. Figure 2 In The set of nodes within the transmission range of a node. All nodes are randomly distributed along a one-dimensional path. Let... This represents the set of all nodes on the road.

[0052] In vehicle-to-everything (V2X) communication, vehicles periodically broadcast beacon messages. Let's assume... Let be the beacon arrival rate of node i. When a node has a beacon message to send, it first senses the channel state. If the channel is detected to be idle during the Distributed Coordination Function Inter-Frame Interval (DIFS), the node broadcasts the message; otherwise, the vehicle will... Randomly select a competition window within the range to begin retreating. This represents the minimum contention window in the 802.11p protocol. The random backoff process reduces the likelihood of collisions with beacon messages sent by other nodes. The Markov chain of the backoff process is as follows: Figure 3 As shown. Let This indicates the backoff counter state at time t. When a node detects that the channel is idle within a time slot, the backoff counter will... The probability decreases by 1. This indicates the probability of beacon busy. When a node detects that the channel is busy within a time slot, the backoff counter will increment by a probability of 0. Maintain the current state. Beacon message transmission begins when the backoff counter reaches 0.

[0053] To obtain The total beacon transmission process of all nodes is approximated as a rate of The homogeneous Poisson process.

[0054] If in a time slot If at least one node within node i's communication range is transmitting a beacon message, then the channel will be detected as busy within that time slot. Therefore, the channel busy probability can be defined as:

[0055] (2)

[0056] in, Let represent the set of nodes within the transmission range of node i. This represents the probability of being in state w during the retreat process.

[0057] The transition probabilities of the Markov chain in Figure 3 can be derived as follows:

[0058] (3)

[0059] (4)

[0060] (5)

[0061] Equation (3) indicates that when the channel is detected to be idle, the backoff counter is decremented by 1; Equation (4) indicates that when the channel is detected to be busy, the backoff counter remains in its current state; Equation (5) indicates that when the backoff counter is decremented to 0, a beacon message is transmitted.

[0062] in This represents the value of the backoff counter. Let... Let w represent the stationary distribution of the Markov chain. Based on the steady-state condition of the Markov chain, the state transition probability of the backoff process in state w can be obtained:

[0063] (6)

[0064] in To back off data packets to a probability of sending zero data packets, according to the normalization condition of the Markov chain, we obtain:

[0065] (7)

[0066] Using the state transition probability (6) and the normalization condition (7), we can obtain:

[0067] (8)

[0068] set up This represents the probability that a node has messages to send in any given time slot. To obtain... Two conditions must be met: first, the node must have a ready beacon message to send; second, the beacon message will only be transmitted when the backoff counter is reduced to 0.

[0069] The probability that a beacon message is present in the vehicle queue at any given time is:

[0070] (9)

[0071] Here, E[S] represents the average service time, which is the time elapsed from when a beacon message enters the queue until it is successfully transmitted or a collision occurs. E[S] includes the transmission time. and retreat time Two parts. The formula for deriving the transmission time is as follows:

[0072] (10)

[0073] in, This indicates the time it takes for the beacon to travel through the wireless channel to the receiving end. This represents an inter-frame interval. The average backoff time can be derived from the Markov chain in Figure 3, as follows:

[0074] (11)

[0075] Therefore, the average service time for the data packet is:

[0076] (12)

[0077] The probability that a node will attempt to transmit at any given time can be obtained as follows:

[0078] (13)

[0079] If the area If no node within the region initiates transmission in the same time slot as the source node, then a direct collision will not occur at point x. The probability of a direct collision at point x at the receiving end refers to the probability of a collision within the region. The probability that at least one node in the network is transmitting concurrently with the source node in the same time slot.

[0080] set up Let x represent the probability that the receiver at a distance x from the source node i has no direct collision. Then we can obtain:

[0081] (14)

[0082] in ,in This indicates the number of nodes within the direct collision area.

[0083] Unlike direct collisions, hidden collisions occur when nodes outside each other's communication range (i.e., hidden nodes) simultaneously transmit messages to the same receiver, resulting in a conflict at the receiver. Let the hidden collision region at receiver x be denoted as... This region includes all nodes that are within the communication range of the receiver x but outside the communication range of the sender. The definition is as follows:

[0084] (15)

[0085] in, This represents the set of nodes outside the transmission range of the source node.

[0086] To avoid a hidden collision between the source node's transmission and the transmission destined for the receiver at a distance x, the following two conditions must be met: first, when the source node starts transmitting the beacon, no node in region H is transmitting; second, during the entire transmission period of the source node, no node in region H initiates transmission.

[0087] The total beacon transmission process of all nodes within the hidden node region is approximated as a transmission rate of... The homogeneous Poisson process. This represents the number of hidden terminals in the source node. The probability of no hidden collisions occurring. It can be represented as:

[0088] (16)

[0089] in, ,in This indicates the number of nodes within the hidden collision region.

[0090] Therefore, the probability that the receiver at a distance x from the source node successfully receives the beacon can be expressed as:

[0091] (17)

[0092] in .

[0093] Therefore, the success rate of the receiver at a distance x from the source point can be obtained as follows:

[0094] (18)

[0095] Formula (17) is the data packet reception probability model, while formula (18) is the constructed data packet reception rate model.

[0096] Existing channel congestion control methods mostly employ open-loop control mechanisms to adjust network parameters. These methods rely on predefined system models, and the model parameters remain fixed throughout the optimization process. When the network environment changes dynamically, these methods lack the ability to provide feedback correction and optimization of previous congestion control strategies, making it difficult to adapt to dynamic changes in the network environment and consequently failing to guarantee the reliability of inter-vehicle communication. To address these issues, this invention proposes a closed-loop feedback control framework based on model reference adaptive control. By designing a feedback controller to monitor changes in the network environment and dynamically adjust network parameters, the actual network state under dynamic network conditions approaches the ideal network state. Figure 4 This paper describes a distributed closed-loop feedback channel congestion control framework employing the Model Reference Adaptive Control (MRAC) method. In this framework, each node (vehicle) operates an independent controller. The controller generates beacon rate control actions based on the deviation between the ideal network state and the actual network state, guiding the actual network state towards the ideal network state. This is a distributed framework; each vehicle possesses an MRAC controller, which consists of two parts: a reference model and an adaptive feedback control model. Because it is distributed, each vehicle generates its own error. (Right now )and . Figure 5 A model reference adaptive control framework for channel congestion control in vehicular networks is described. This framework consists of two core components: a reference model and an adaptive feedback control model. Figure 6 A model-reference adaptive control architecture for beacon rate adaptation is described. The reference model first outputs the expected probability of successful packet reception based on a preset maximum packet reception rate (PRR) target. This was compared with the actual data packet reception success probability collected from a wireless channel based on the IEEE 802.11p protocol. By comparison, the error signal is obtained. The error signal is input to the PI controller, which, in conjunction with dynamically adjusted proportional and integral gains, generates a beacon rate control quantity according to the control law. (i.e., the beacon rate of the node) and acts on the controlled object, the wireless channel, through the transfer function. The model outputs a new actual network state. Simultaneously, the model is adjusted based on the MIT rule, and the parameters of the PI controller are dynamically optimized using the sensitivity derivative of the error signal, so that the actual output of the wireless channel continuously approaches the expected output of the reference model.

[0097] The optimization objective of the reference model is to improve the tracking accuracy of each node while avoiding channel congestion. Tracking accuracy primarily depends on the packet success rate (PRR) metric of each node. Therefore, this framework takes maximizing the PRR as the system objective. Based on this objective, the reference model generates the desired packet reception probability according to the current vehicle density. The reference model first evaluates the current traffic conditions and then generates a channel control strategy to achieve the desired packet reception probability under the current vehicle density. Since traffic conditions in vehicular networks are constantly changing dynamically, only short-term vehicle density assessments are suitable. Therefore, the reference model uses a short-term rolling optimization approach to evaluate traffic conditions. The adaptive feedback control model contains two loops: a conventional feedback loop and an adaptive feedback loop. The conventional feedback loop consists of a controller and the controlled object (i.e., a wireless channel based on IEEE 802.11p). The controller generates a real-time channel congestion control strategy (i.e., a beacon rate control strategy) based on the error between the ideal network state output by the reference model (i.e., the ideal packet reception probability) and the current real network state (i.e., the actual packet reception probability) collected from the actual network environment. Because traffic conditions (such as vehicle density) in vehicular networks change rapidly, controller parameters should be dynamically adjusted rather than fixed to enhance the robustness of channel congestion control strategies in dynamic traffic scenarios. Adaptive feedback loops adjust controller parameters to ensure the actual network state matches the ideal network state output by the reference model. Therefore, the goal of the feedback control model is to minimize the error between the ideal and actual network states, guiding the actual network state towards the ideal state.

[0098] The reference model is based on the ideal state of the target output network of the preset system. This invention will construct a reference model to generate the ideal network state and the corresponding channel control strategy to achieve this state. In vehicular networks, each node relies on periodically broadcast beacons to perceive the motion state (such as vehicle position, speed, etc.) of neighboring nodes. In high-density scenarios, a large number of vehicles compete for channel resources to broadcast state information, which will generate high channel load, leading to channel congestion and a large number of data packet collisions.

[0099] This invention dynamically allocates beacon rates based on changes in traffic conditions to control channel load and avoid congestion. Due to the highly dynamic and unpredictable nature of vehicular networks (V2X), network parameters (such as transmission power and beacon rates) need to be dynamically adjusted. Therefore, a rolling time-domain optimization method is used to dynamically calculate the optimal beacon rate for each node. Furthermore, fairness is a key objective in V2X beacon rate allocation. Unfairness arises when some nodes receive a large amount of bandwidth while others suffer from resource scarcity. To avoid unfair behavior in network resource allocation, the beacon rate control problem is modeled as a network utility maximization (NUM) problem. Each node is associated with a utility function, and the goal of beacon rate allocation is to maximize the total utility of the nodes.

[0100] Fairness can be defined in various ways, and commonly used fairness models include proportional fairness models:

[0101] (19)

[0102] in This represents a weighting coefficient. The value of can be used to prioritize specific communication needs to achieve weighted fairness. If nodes in the network have the same needs for all (i.e., ...), ... ), then it can be based on The value of generates different fairness models. Specifically, 0-proportional fairness ( The aim is to maximize total network throughput. Proportional fairness ( This is used to balance channel resource utilization efficiency and allocation fairness. It maximizes total network throughput while avoiding extreme unfairness. Maximize-minimum fairness (MPF) To ensure absolute fairness in resource allocation, the goal is to maximize the minimum rate allocation among all nodes. Since 0-proportional fairness may lead to arbitrary unfair channel resource allocation, and maximum-minimum fairness often sacrifices resource utilization efficiency in certain scenarios, this framework adopts a proportional fairness model for channel resource allocation.

[0103] To ensure fair channel resource allocation under different network environments and dynamic traffic conditions, this invention employs a rolling domain approach to construct a reference model. This model generates the desired network state and seeks the corresponding optimal control strategy. The model determines the optimal beacon rate for each node by maximizing the network performance objective (PRR) within a certain time frame.

[0104] At each time step The reference model is constructed as an optimization problem of the following form:

[0105] (20)

[0106] in, This represents the proportional equity effect function. This represents the beacon reception rate function at time step k. Represents a specific sequence at time step k. The acceptance rate function under the following conditions This represents a specific sequence at time step k for node i. The beacon rate below; This represents the beacon rate of node i within time step k; This represents the beacon rate of other nodes besides node i within the direct and hidden collision regions at time step k. Represents a specific sequence at time step k. The beacon rates of other nodes except node i within the direct and hidden collision regions.

[0107] in This represents the prediction time domain; to avoid channel congestion, a price factor is introduced using the Lagrange dual decomposition method. This allows for the relaxation of constraints to limit the total channel load of a given node to a maximum load threshold. the following.

[0108] The total load of a given node is calculated as the sum of the load generated by its neighboring nodes and its own load.

[0109] ; ;

[0110] in The physical meaning is the sum of the channel loads of all time series at time step k. Represents a specific sequence at time step k. Channel load under [the current situation]. Let i represent the set of neighboring nodes of node i, including node i itself. Each node is a vehicle. Each node sends beacons at a different rate and constant transmission power. These beacons are received by neighboring nodes within its communication range.

[0111] Simultaneously, the beacon rate of the node Limited to the interval [ , Within [the box], the formula is as follows:

[0112] ;

[0113] in Indicates the minimum beacon rate. This indicates the maximum beacon rate.

[0114] The beacon rate optimization process is described as follows:

[0115] set up This represents the sampling interval. In each control step k (corresponding to control time k) Each node collects channel state information from its neighbors and traffic state information from the road environment. Then, the optimization problem of equation (20) is solved to determine the time interval. beacon control sequence within .

[0116] In each control step, only the first control action of the control sequence (i.e., This is applied to each node. Finally, the prediction time domain is shifted forward by one sampling interval, and the above process is repeated at the next time step (k+1).

[0117] Since the packet reception rate model in formula (18) depends only on the network parameters at the current moment, it means that the PRR model is a memoryless model. Therefore, one-step model predictive control is adopted. To determine the optimal beacon rate in optimization problem (20).

[0118] This invention employs the dual decomposition method to solve the optimization problem in formula (20). First, the Lagrangian function and relaxation constraints of the optimization problem are constructed; substituting formulas (17) and (18) into the optimization problem (20), we obtain:

[0119] (twenty one)

[0120] in, This represents the packet reception rate at time step k. This represents the beacon rate of node i at time step k; The Lagrange multiplier represents the price a node pays for using channel resources to transmit beacon data at a certain rate.

[0121] In formula (21) The function is the Lagrange function, which is obtained by introducing a Lagrange multiplier, i.e., the price. Integrating constraints into the objective function essentially transforms constrained optimization into unconstrained optimization.

[0122] As channel congestion increases, the corresponding cost also increases; given a set of non-negative prices The optimal rate allocation can be obtained by solving the Lagrange dual problem shown below:

[0123] (twenty two)

[0124] in Let represent the Lagrange dual function, where the Lagrange dual is the maximum value of the Lagrange function at the beacon rate. That is, given a set of non-negative prices, find the optimal rate allocation in the Lagrange dual.

[0125] Each node i must know the prices of its neighboring nodes in order to calculate the optimal beacon rate;

[0126] There exists a set of optimal prices This ensures that the prices derived from the Lagrange dual problem are consistent with the optimal solution of the optimization problem. The process of determining these optimal prices constitutes the dual problem, which is formally defined as follows:

[0127] (twenty three)

[0128] in, Let represent the dual function, and this formula represents the dual problem associated with the optimization problem; Let $\mathbf{a}$ represent a set of optimal prices. Since the objective function in the optimization problem is strictly concave and the constraints are linear, the dual problem is strictly convex and differentiable. Therefore, the dual problem has a unique set of optimal prices. .

[0129] Furthermore, for any given set of prices dual function The gradient is expressed as:

[0130] (twenty four)

[0131] in This represents the optimal beacon rate of node i at time step k.

[0132] To solve the rate allocation problem in a distributed manner, each node i needs to exchange price parameters, i.e., Lagrange multipliers, with its neighboring nodes. These prices are then used as inputs to the local optimization problem defined by the Lagrange dual problem of formula (22). Subsequently, a set of beacon rates at time step k is given. The gradient descent algorithm is performed to maximize the Lagrangian function in the Lagrangian dual problem of formula (22), where The gradient is:

[0133] (25)

[0134] This invention designs a distributed beacon control algorithm to obtain the optimal beacon rate for each node. The algorithm is implemented using a rolling time-domain approach. In the algorithm, and These represent the initial beacon rate and initial price for all nodes, respectively. and ). and These are the maximum and minimum beacon rates; parameters. and This is a weighting factor that affects the convergence speed of the algorithm. At each time step, the node first measures the traffic conditions of the current road segment and obtains price information from neighboring nodes, then uses gradient descent to solve (20) to obtain the optimal solution. Steps 1.4 to 1.6 describe the process within the time interval... The process involves obtaining the optimal beacon rate. In each step, all nodes update their own prices and broadcast them to their neighbors, adjusting their own beacon rates only based on information from their neighbors. Finally, the prediction time domain is shifted forward one step, and the optimization process is repeated in the new time domain by measuring the latest traffic conditions.

[0135] Specifically, the process of finding the optimal vehicle beacon transmission rate is as follows:

[0136] Step 1.1. Initialize the beacon rate of all nodes. price .

[0137] Step 1.2. At each time step k, each node measures the current traffic conditions through the received beacon information (the beacon contains vehicle location information), analyzes the location information of the source node and the receiving node, and determines whether a node is a node in the direct collision area or a node in the hidden collision area by using the location and the distance criteria between the direct collision area and the hidden collision area. In this way, the nodes in the direct collision area and the hidden collision area are counted to evaluate the number of neighboring nodes competing with them for the shared channel.

[0138] Step 1.3. Solve the optimization problem to determine the time interval of node i. The optimal beacon rate within the range.

[0139] Step 1.4. In each iteration step q, perform the gradient descent process, as shown in the following formula:

[0140] ;

[0141] in This represents the beacon rate of node i at time step k during the (q+1)th iteration. Let represent the beacon rate of node i at time step k during the q-th iteration. This represents the learning rate in gradient descent. This represents the beacon rate of all nodes at time step k. The Lagrangian function represents the time step k. This represents the Lagrange multiplier associated with the relaxation constraint at time step k.

[0142] Step 1.5. Each node i receives information from its neighboring nodes. price .

[0143] Step 1.6. Each node i updates its price according to the following steps. ; This represents the cost required for node i to transmit the beacon using channel resources at time step k.

[0144] .

[0145] in This represents the price of node i at time step k in the (q+1)th iteration. This represents the price of node i at time step k in the q-th iteration. Indicates the learning rate. Let represent the beacon rate of node i at time step k during the q-th iteration.

[0146] Step 1.7. After completing the current cycle, move the prediction time domain forward one step; by measuring the new traffic conditions at time step k+1, repeat the optimization process in the new time domain, i.e., repeat steps 1.2 and 1.6 above.

[0147] Step 2. Construct an adaptive feedback control model. Based on the ideal network state obtained from the reference model and the actual network state, introduce a PI controller. Use the error between the ideal network state and the actual network state as the input of the PI controller. The PI controller adopts MIT parameter adjustment. Based on the error signal between the ideal network state and the actual network state, dynamically optimize the proportional gain and integral gain of the PI controller to generate a real-time channel congestion control strategy, namely the beacon rate control strategy.

[0148] By solving the rolling time-domain optimization problem of formula (20), the target data packet reception performance of each node under the current traffic conditions can be determined. The rolling time-domain optimization method mainly relies on the network utility model defined by formula (20) and the current traffic conditions to derive the optimal network parameters. This method belongs to the open-loop control method. Unlike other wireless networks, traffic conditions in vehicle networks change rapidly, and any environmental uncertainty may cause the actual data packet reception performance to deviate from the target state. In order to ensure stable network performance in a dynamic and uncertain network environment, this invention proposes a closed-loop feedback control framework for real-time adjustment of beacon rate. This method continuously monitors performance deviation, compensates for suboptimal beacon control actions, and guides the actual network state to approach the ideal network state. To build this framework, a measurable reference index needs to be set, which should be able to directly reflect the ideal channel performance in vehicle networks based on the IEEE 802.11p protocol. In actual vehicle networks, the probability of successful data packet reception (i.e., the probability of successful data packet reception) can be efficiently measured through existing monitoring mechanisms such as channel state feedback. ).

[0149] Therefore, this invention selects the optimal data packet reception success probability at time step k (i.e., (k) serves as the reference point for the closed-loop feedback control framework. The distributed beacon control algorithm enables each node to calculate its optimal beacon rate at time step k. Given a set of optimal beacon rates... The probability of successfully receiving the target data packet can be derived from formula (17), as follows:

[0150] (26)

[0151] As shown in equation (17), the probability of successful packet reception depends only on the current beacon rate and is independent of historical rates, which means that the network state has no memory. Furthermore, equation (17) also reveals... and The nonlinear dependency between them. To simplify controller design, in the current network scenario, a linear function is used at the stable equilibrium point. The nonlinear relationship is approximated by the surrounding area. The linear perturbation near the stationary point of the resulting linear time-invariant (LTI) model is defined as follows:

[0152] (27)

[0153] in express The infinitesimal change in the quantity. The impact of the disturbance on the probability performance of successful beacon reception can be modeled as:

[0154] (28)

[0155] in express an infinitesimal change quantity, express The infinitesimal change quantity. This is a matrix showing the impact of changes in the node's beacon rate on the probability of data packet reception, and its formula is expressed as follows:

[0156] .

[0157] in:

[0158] (29)

[0159] (30)

[0160] (31)

[0161] Combining (29) and (30), the common terms can be extracted: .

[0162] Therefore, we can conclude that:

[0163] (32)

[0164] Around the stable point The nonlinear model is linearized as follows:

[0165] (33)

[0166] in , .

[0167] To improve the accuracy of vehicle tracking performance under dynamic traffic conditions, this invention proposes an adaptive feedback control model for channel congestion control. As shown in Figure 6, this model consists of a conventional feedback loop and an adaptive feedback loop. The former consists of a controller and a controlled object (a wireless channel based on the IEEE 802.11p protocol), while the latter dynamically adjusts the controller parameters. This adaptive feedback control model includes three basic control-related parameters, namely the measured values... Control quantity and reference value Among them, the measured values Represents the probability of successful data packet reception at time step k for node i; control variable. The beacon rate of node i at time step k; reference value. The probability of successful reception of the target data packet at node i at time step k represents the ideal state of the network. Due to the highly dynamic nature of traffic conditions, the controller parameters need to be adaptively adjusted to guide the network state towards the desired state. This adjustment model utilizes feedback error (i.e., the difference between the reference value and the measured value) to optimize the controller parameters. This controller is based on a proportional-integral (PI) controller design; the proportional element can quickly reduce the deviation of the system from the initial response to the process of stabilization, but cannot completely eliminate it; the integral element eliminates the steady-state error by continuously integrating the error.

[0168] The output signal of the PI controller at time t Represented as:

[0169] (34)

[0170] in The proportional gain of node i at time step k is used to reduce error oscillations; Let be the integral gain of node i at time step k, which is used to eliminate steady-state errors in the vehicle network by accumulating historical errors.

[0171] The error at node i at time step k is represented as follows:

[0172] (35)

[0173] in The optimal data packet reception probability at the current moment. This represents the actual probability of receiving data packets at the current moment.

[0174] In real-world vehicle-to-everything (V2X) scenarios, the network environment is highly dynamic, meaning that the parameters in formula (34) (i.e. and The parameters of the PI controller must be dynamically adjusted over time, rather than remaining fixed. To address these dynamic changes and mitigate the adverse effects of external disturbances on network performance, this invention employs the MIT rule to dynamically adjust the PI controller parameters, thereby achieving adaptation to dynamic environmental conditions. The MIT rule is a parameter adjustment rule based on gradient descent. By optimizing the error gradient through steepest descent, it can dynamically adjust the controller parameters to minimize the error between the actual network state and the desired network state.

[0175] The loss function is defined as follows:

[0176] (36)

[0177] in This represents the error at node i at time step k. ;in and These are the control parameters, which correspond to the proportional gain and integral gain of the PI controller, respectively.

[0178] To minimize the loss function The controller parameters are adjusted using the MIT rule, and the adjustment formula is as follows:

[0179] (37)

[0180] in, The vectors representing integral gain and proportional gain. Represents the loss function. For adaptive gain; and These are two positive scalars used to determine the convergence speed of parameter adaptation; The term is called the sensitivity derivative, used to quantize the adjustable parameter vector. For error The impact; applying the MIT rule to the PI controller, we get:

[0181] (38)

[0182] (39)

[0183] in .

[0184] Represents the loss function. This represents the actual data packet reception probability at time step k. This represents the beacon rate of node i at time step k. Represents the Laplace transform factor. This represents the ideal data packet reception probability at time step k.

[0185] By solving the above equation, the control parameters are obtained. and The proportional gain and integral gain of the PI controller are dynamically optimized and combined with the PI controller (i.e., after dynamically optimizing the two parameters of the PI controller, the PI controller uses the optimized parameters to generate a congestion control strategy), thus generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

[0186] The performance of the proposed Distributed Closed-Loop Control (DCFC) method was evaluated using the NS-3 simulation platform. This platform supports simulations of various network protocols, including wired and wireless networks. Its built-in library provides a realistic two-ray ground propagation loss model to simulate signal propagation between nodes. In the simulation, a one-kilometer-long four-lane highway model was constructed, employing the two-ray ground propagation loss model. The transmit power of each node was set to 20 dBm, meaning that the signal coverage range of each node is approximately 300 meters. The experimental parameter settings are summarized in Table 1.

[0187] To verify the effectiveness of the proposed method, DCFC was compared with three other benchmark methods: ABC, SAE, and LIMERIC. ABC dynamically adjusts the beacon rate based on the risk of rear-end collisions; SAE optimizes the beacon rate based on vehicle density; and LIMERIC uses linear control theory to optimize the beacon transmission rate based on the total rate error observed by neighboring nodes.

[0188] Table 1 Simulation Parameters

[0189]

[0190] First, the convergence of the beacon rate adjustment process described in this invention is verified. The number of nodes is set to 400, randomly distributed along the highway. For example... Figure 7 As shown, four nodes were selected on the highway at distances of 20 meters, 50 meters, 80 meters, and 110 meters from the receiver, respectively. Figure 7 The beacon adjustment process for these four nodes is illustrated. It can be seen that all nodes successfully converged to their respective optimal beacon rates. Specifically, the farther a node is from the receiver, the lower its optimal beacon rate. This is because the number of nodes located in the direct collision or hidden collision regions increases with distance, exacerbating channel contention. To alleviate congestion, the proposed algorithm adaptively reduces the beacon rates of these nodes, effectively controlling channel load while maintaining tracking accuracy.

[0191] Next, the scalability of the adaptive beacon rate of the distributed beacon control algorithm of this invention is evaluated. Experiments were conducted in scenarios with 400, 600, and 800 nodes deployed on highways, respectively. Figure 8 As shown, when the number of nodes is 400, the optimal beacon rate is mainly distributed between 4-6 beacons / second; when the number of nodes is 600, the optimal beacon rate range narrows to 3-4 beacons / second; and when the number of nodes is 800, the rate is further controlled to 2-3 beacons / second. This indicates that the distributed beacon control algorithm can dynamically adjust the beacon rate of each node according to the vehicle density. Figure 9 The cumulative distribution function (CDF) of the beacon rate is shown. It can be seen that the distributed beacon control algorithm can effectively regulate the beacon rate as vehicle density increases. Specifically, when the number of nodes increases from 400 to 800, the median optimal beacon rate decreases from 5.8 beacons / second to 2.5 beacons / second.

[0192] Figure 10 and Figure 11 The error correction (i.e., the deviation between the reference model output and the actual network state) achieved by the DCFC method is shown for 400 and 800 nodes, respectively. Nodes 171 and 286 were randomly selected from traffic scenarios with 400 and 800 nodes for observation. It can be observed that as the beacon rate control law is implemented, the error between the ideal state and the actual network state gradually decreases. All errors converge to zero within 15 iterations.

[0193] The results show that the controller can promote the actual network state to an ideal state under different network conditions. Next, to further verify the effectiveness of this method, it is compared and analyzed with three benchmark strategies: ABC, SAE, and LIMERIC.

[0194] Figure 12 and Figure 13This paper demonstrates the beacon rate adjustments for four methods—DCFC, LIMERIC, ABC, and SAE—under different traffic scenarios. When the number of nodes is 400, the beacon rate for all nodes in the LIMERIC method is adjusted to 6-10 beacons / second. Such a high beacon rate generates excessive channel load, leading to channel congestion. In the SAE method, the beacon rate for all nodes is controlled within the range of 4-8 beacons / second. The ABC method exhibits a polarized rate allocation among nodes because it adjusts the beacon rate based on the assessed hazard level. When a node faces a high collision risk, it broadcasts at the maximum beacon rate; conversely, when the collision risk is low, the beacon rate is reduced to the minimum. In the DCFC method proposed in this invention, because more nodes participate in resource competition in the middle area of ​​the highway, its beacon rate is maintained in the range of 4.2-6 beacons / second, lower than the beacon rates at the ends of the road.

[0195] When the number of nodes increases to 800, the beacon rate for each method decreases with the increase in the number of nodes. This is mainly because in high-density scenarios, more data packets compete for the wireless channel, leading to excessive channel load. Therefore, all methods reduce the beacon rate of individual nodes to alleviate channel load and reduce channel congestion. Specifically, the LIMERIC method maintains a beacon rate of around 5 beacons / second for most nodes; the SAE method controls the beacon rate of all nodes between 2 and 4 beacons / second; and the DCFC method controls the beacon rate below 3.5 beacons / second in the middle area of ​​the highway.

[0196] Next, we will evaluate the channel resource utilization of these four different channel congestion control methods.

[0197] Channel busy rate (CBR) describes the channel utilization of a wireless channel. To ensure reliable vehicle tracking performance, the channel busy rate should be kept between 0.6 and 0.65. Once the channel busy rate exceeds this threshold, channel congestion occurs.

[0198] Figure 14 The diagram shows the channel congestion rate for each method when the number of nodes is 400. It can be seen that the channel congestion rate for all methods does not exceed 0.7, with the DCFC method controlling the channel congestion rate at around 0.6-0.65. Therefore, this method can keep the channel status within an ideal range. Figure 16The cumulative distribution function (CDF) of channel busy rate for the four methods is presented when the number of nodes is 400. The results show that in the ABC and SAE methods, almost all nodes have a channel busy rate below 0.55, indicating that these two methods cannot fully utilize channel resources. In the LIMERIC method, over 60% of the nodes have a channel busy rate in the range of 0.65-0.7.

[0199] Figure 15 shows the Channel Busy Rate (CBR) for the four methods when there are 800 nodes. The proposed DCFC method can control the CBR within the range of 0.6-0.65. This method is based on Model Reference Adaptive Control (MRAC), which allows each node to autonomously adjust the beacon rate by minimizing the error between the ideal network state and the actual network state. Therefore, even in a dynamic and uncertain environment, DCFC can still stabilize the CBR at an ideal level. In contrast, from Figure 17 It can be seen that the LIMERIC method results in over 80% of nodes having a channel busy rate exceeding 0.7, meaning that this method can lead to excessively high channel load. For example... Figure 17 As shown, the channel busy rates of the ABC and SAE methods are lower than 0.55 and 0.51, respectively, indicating that these two methods cannot fully utilize channel resources. In summary, under different traffic scenarios, the DCFC method proposed in this invention achieves better network performance and resource utilization compared to the other three benchmark methods.

[0200] Next, the tracking performance of the four methods is evaluated by calculating the Euclidean tracking error of each node. In vehicular networks, each node tracks the position of its neighbors by receiving beacon signals. A node can calculate its tracking error using the received beacon information (i.e., the position and velocity of neighboring nodes). If a neighboring node fails to receive the beacon signal, it cannot perceive the latest state of that node, leading to an increase in tracking error.

[0201] Figure 18 and Figure 19The tracking error of the four methods under different numbers of nodes is shown. Compared with the other three methods, the DCFC method achieves lower tracking errors in scenarios with different numbers of nodes. When the number of nodes is 400, for a receiving node 20 meters away from the source, the tracking error of the DCFC method is 0.6 meters, 0.4 meters, and 1.6 meters lower than that of LIMERIC, ABC, and SAE, respectively. When the number of nodes is 800, the tracking error for the receiving node is 0.2 meters, 0.4 meters, and 4.1 meters lower than that of the above three methods, respectively. This shows that the DCFC method has stronger robustness and scalability under different network conditions. The main reasons are as follows: The ABC method prioritizes allocating the maximum beacon rate to nodes with high collision risk. Once the channel load threshold is reached, the remaining nodes will be allocated the minimum beacon rate, causing some nodes to send beacons at a frequency of 1 Hz; while the LIMERIC and SAE methods fail to consider the hidden node problem, which reduces the data packet reception success rate and thus increases the tracking error.

[0202] The experimental results above demonstrate that, compared with existing channel congestion control methods, the method proposed in this invention can effectively alleviate channel congestion and maintain a low tracking error under different vehicle density scenarios.

[0203] This invention proposes a distributed closed-loop feedback control method for vehicle-to-everything (V2X) channel congestion control. This method employs a model reference adaptive control mechanism, ensuring that the actual network state asymptotically converges to the ideal network state even under dynamically changing network conditions. This invention constructs the channel congestion control problem as a network utility maximization model. Based on a preset objective, this model obtains the ideal network state (i.e., the desired data packet reception probability) through short-term rolling time-domain optimization, and uses this as the reference benchmark (i.e., the desired objective) for the adaptive feedback control model. Simultaneously, this invention also designs an adaptive feedback control model to ensure the stability of vehicle channel congestion control under different traffic conditions. This model, combined with a proportional-integral (PI) controller, ensures that the wireless channel performance converges to the desired objective (i.e., the desired data packet reception probability). To mitigate the impact of highly dynamic traffic conditions on network performance, the controller parameters are dynamically adjusted to generate an adaptive channel congestion control strategy.

[0204] Example 2

[0205] This embodiment 2 describes a distributed closed-loop control system for vehicle network channel congestion, which is based on the same inventive concept as the distributed closed-loop control method for vehicle network channel congestion in embodiment 1 above.

[0206] The distributed closed-loop control system for vehicle network channel congestion in this embodiment includes the following modules:

[0207] The reference model building module is used to build a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, build a data packet reception rate model, and design a network utility maximization model using a rolling time domain approach.

[0208] The optimal vehicle beacon transmission rate is obtained by solving the dual decomposition and gradient descent methods to obtain the ideal network state.

[0209] An adaptive feedback control model building module is used to introduce a PI controller based on the ideal network state obtained from the reference model and the actual network state obtained, and to use the error between the ideal network state and the actual network state as the input of the PI controller.

[0210] The PI controller uses MIT parameter tuning, which dynamically optimizes the proportional gain and integral gain of the PI controller based on the error signal between the ideal network state and the actual network state, thereby generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

[0211] It should be noted that any content not mentioned in the above-mentioned functional modules of the vehicle network channel congestion distributed closed-loop control system described in this embodiment can be referred to the step description of the corresponding method in the above embodiment 1, and will not be repeated in detail here.

[0212] Example 3

[0213] This embodiment 3 describes a computer device including a memory and one or more processors. Executable code is stored in the memory. When the processor executes the executable code, it implements the steps of the distributed closed-loop control method for vehicle network channel congestion in embodiment 1 above.

[0214] In this embodiment, the computer device can be any device or apparatus with data processing capabilities, and will not be described in detail here.

[0215] Example 4

[0216] This embodiment 4 describes a computer-readable storage medium storing a program that, when executed by a processor, is used to implement the steps of the distributed closed-loop control method for vehicle network channel congestion in embodiment 1 above.

[0217] The computer-readable storage medium can be an internal storage unit of any device or apparatus with data processing capabilities, such as a hard disk or memory, or an external storage device of any device with data processing capabilities, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc.

[0218] Of course, the above description is only a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. It should be noted that any equivalent substitutions or obvious modifications made by those skilled in the art under the guidance of this specification fall within the scope of this specification and should be protected by the present invention.

Claims

1. A distributed closed-loop control method for vehicle network channel congestion, characterized in that, Includes the following steps: Step 1. Construct a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, construct a data packet reception rate model, and design a network utility maximization model, i.e., a reference model, using a rolling time domain approach. In step 1, at each time step k, the reference model is constructed as an optimization problem of the following form: ； in, This represents the proportional equity effect function. This represents the beacon reception rate function at time step k. Represents a sequence at time step k. The acceptance rate function under the following conditions This represents a sequence at time step k for node i. The beacon rate below; This represents the beacon rate of node i within time step k; This represents the beacon rate of other nodes besides node i within the direct and hidden collision regions at time step k. Represents a sequence at time step k. The beacon rates of all nodes except node i within the direct and hidden collision regions; where Indicates the prediction time domain, This represents the number of nodes within the direct collision region. This indicates the number of nodes within the hidden collision region; To avoid channel congestion, a price factor is introduced using the Lagrange dual decomposition method. Relaxing constraints to limit the total channel load of a given node to the maximum load threshold. The formula is as follows: ； ； in This represents the set of neighboring nodes of node i, including node i itself. Each node is a vehicle. Each node transmits beacons at a different rate and constant transmission power. These beacons are received by neighboring nodes within its communication range. Represents a sequence at time step k. Channel load under; Simultaneously, the beacon rate of the node Limited to the interval [ , Within [the box], the formula is as follows: ； in Indicates the minimum beacon rate. Indicates the maximum beacon rate; The optimal vehicle beacon transmission rate is obtained by solving the dual decomposition and gradient descent methods to obtain the ideal network state; Step 2. Construct an adaptive feedback control model. Based on the ideal network state obtained from the reference model and the actual network state obtained, introduce a PI controller and use the error between the ideal network state and the actual network state as the input of the PI controller. The PI controller uses MIT parameter tuning, which dynamically optimizes the proportional gain and integral gain of the PI controller based on the error signal between the ideal network state and the actual network state, thereby generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

2. The distributed closed-loop control method for vehicle network channel congestion according to claim 1, characterized in that, In step 1, the constructed data packet reception rate model is as follows: ； in This represents the data packet reception rate model. This represents a data packet reception probability model. This represents the set of all nodes on the road; Indicates the beacon rate of node i. This represents the beacon rate of nodes other than node i within the direct collision region and the hidden collision region. This represents the number of nodes within the direct collision region. This indicates the number of nodes within the hidden collision region; ； in This represents the probability that the receiver at a distance x from the source point will not have a direct collision. This represents the probability that no hidden collision has occurred at the receiver at a distance x from the source node; ； in This represents the beacon rate of all nodes except node i within the direct collision region; Indicates the average service time of data packets. This represents the beacon rate of any node within the direct collision region. This represents the probability that a data packet backs off to a zero-data packet and attempts to send it. Indicates the area of ​​direct collision. ; ； in This is the time it takes for the beacon to travel from the wireless channel to the receiving end. This represents the beacon rate of any node within the hidden collision region. This represents the number of hidden terminals at the source node. This represents the beacon rate of all nodes except node i within the hidden collision region.

3. The distributed closed-loop control method for vehicle network channel congestion according to claim 1, characterized in that, In step 1, the process of finding the optimal vehicle beacon transmission rate through dual decomposition and gradient descent is as follows: Step 1.

1. Initialize the beacon rate of all nodes. price ; Step 1.

2. At each time step k, each node measures the current traffic conditions to assess the number of neighboring nodes competing with it for the shared channel; Step 1.

3. Solve the optimization problem to determine the time interval of node i. Optimal beacon rate within the range; Step 1.

4. In each iteration step q, perform the gradient descent process, as shown in the following formula: ； in This represents the beacon rate of node i at time step k during the (q+1)th iteration; This represents the beacon rate of node i at time step k during the q-th iteration; This represents the learning rate in gradient descent. This represents the beacon rate of all nodes at time step k. The Lagrangian function represents the time step k. This represents the Lagrange multiplier associated with the relaxation constraint at time step k; Step 1.

5. Each node i receives information from its neighboring nodes. price ; Step 1.

6. Each node i updates its price according to the following steps. ; ； in This represents the price of node i at time step k in the (q+1)th iteration. This represents the price of node i at time step k in the q-th iteration. Indicates the learning rate. This represents the beacon rate of node i at time step k during the q-th iteration; Step 1.

7. After completing the current cycle, move the prediction time domain forward one step; by measuring the new traffic conditions at time step k+1, repeat the optimization process in the new time domain, i.e., steps 1.2 to 1.

6.

4. The distributed closed-loop control method for vehicle network channel congestion according to claim 3, characterized in that, In step 1.3, the dual decomposition method is used to solve the optimization problem. The specific process is as follows: First, construct the Lagrangian function for the optimization problem and relax the constraints; Substituting the constructed packet reception rate model into the optimization problem, we get: ； in, This represents the packet reception rate at time step k. This represents the beacon rate of node i at time step k; The Lagrange multiplier represents the price a node pays for using channel resources to transmit beacon data at a rate that is sufficient to achieve its intended channel resource usage. As channel congestion increases, the corresponding cost also increases; given a set of non-negative prices The optimal rate allocation can be obtained by solving the Lagrange dual problem shown below: ； in Let represent the Lagrange dual function. The Lagrange dual is the maximum value of the Lagrange function at the beacon rate, that is, given a set of non-negative prices, find the optimal rate allocation of the Lagrange dual. There exists a set of optimal prices This ensures that the prices derived from the Lagrange dual problem are consistent with the optimal solution of the optimization problem. The process of determining these optimal prices constitutes the dual problem, which is formally defined as follows: ； in, Let represent the dual function, and this formula represents the dual problem associated with the optimization problem; Since the objective function in the optimization problem is a strictly concave function and the constraints are linear, the dual problem is a strictly convex function and differentiable; therefore, the dual problem has a unique set of optimal prices. ; Furthermore, for any given set of prices dual function The gradient is expressed as: ； in This represents the optimal beacon rate of node i at time step k; To solve the rate allocation problem in a distributed manner, each node i needs to exchange price parameters, i.e., Lagrange multipliers, with its neighboring nodes. These prices are then used as input to the local optimization problem defined by the Lagrange dual problem; Subsequently, a set of beacon rates at time step k is given. Perform gradient descent to maximize the Lagrangian function in the Lagrangian dual problem, where The gradient is: 。 5. The distributed closed-loop control method for vehicle network channel congestion according to claim 1, characterized in that, In step 2, the adaptive feedback control model consists of a conventional feedback loop and an adaptive feedback loop. The conventional feedback loop consists of a controller and the controlled object, i.e., a wireless channel based on the IEEE 802.11p protocol. The adaptive feedback loop dynamically adjusts the controller parameters. The adaptive feedback control model includes three basic control-related parameters, namely, the measured values... Control quantity and reference value Among them, the measured values This represents the probability of successful data packet reception at time step k for node i, i.e., the actual network state; control variable. The beacon rate of node i at time step k; reference value. Let k be the probability of successfully receiving the target data packet at node i at time step k, i.e., the ideal network state. This controller is based on a proportional-integral (PI) controller design. The proportional element can quickly reduce the deviation of the system from the initial response to the process of stabilization, but it cannot completely eliminate it; the integral element eliminates the steady-state error by continuously integrating the error; the PI controller outputs a signal at time t. Represented as: ； in The proportional gain of node i at time step k is used to reduce error oscillations; The integral gain of node i at time step k is used to eliminate steady-state errors in the vehicle network by accumulating historical errors. The error at node i at time step k is represented as follows: ； in The optimal data packet reception probability at the current moment. This represents the actual probability of receiving data packets at the current moment.

6. The distributed closed-loop control method for vehicle network channel congestion according to claim 5, characterized in that, In step 2, the process of generating a real-time channel congestion control strategy using the MIT parameter adjustment method is as follows: The loss function is defined as follows: ； in This represents the error at node i at time step k. ;in and These are the control parameters, which correspond to the proportional gain and integral gain of the PI controller, respectively. To minimize the loss function, the MIT rule is used to adjust the controller parameters, and the adjustment formula is as follows: ； in, The vectors representing integral gain and proportional gain. Represents the loss function. For adaptive gain; and These are two positive scalars used to determine the convergence speed of parameter adaptation; The term is called the sensitivity derivative, used to quantize the adjustable parameter vector. For error The impact; Applying the MIT rule to the PI controller, we get: ； ； in ; Represents the loss function. This represents the actual data packet reception probability at time step k. This represents the beacon rate of node i at time step k. Represents the Laplace transform factor. This represents the ideal data packet reception probability at time step k; The control parameters are obtained by solving the above equation. and The proportional gain and integral gain of the PI controller are dynamically optimized and combined with the PI controller to generate a real-time channel congestion control strategy, namely the beacon rate control strategy.

7. A distributed closed-loop control system for vehicular network channel congestion for implementing the distributed closed-loop control method for vehicular network channel congestion as described in claim 1, characterized in that, The vehicle-to-everything (V2X) channel congestion distributed closed-loop control system includes the following modules: The reference model building module is used to build a vehicle-to-everything (V2X) communication scenario model, analyze the direct and hidden collision areas of vehicle communication, build a data packet reception rate model, and design a network utility maximization model using a rolling time domain approach. The optimal vehicle beacon transmission rate is obtained by solving the dual decomposition and gradient descent methods to obtain the ideal network state; An adaptive feedback control model building module is used to introduce a PI controller based on the ideal network state obtained from the reference model and the actual network state obtained, and to use the error between the ideal network state and the actual network state as the input of the PI controller. The PI controller uses MIT parameter tuning, which dynamically optimizes the proportional gain and integral gain of the PI controller based on the error signal between the ideal network state and the actual network state, thereby generating a real-time channel congestion control strategy, namely the beacon rate control strategy.

8. A computer device, comprising a memory and one or more processors; characterized in that, The memory stores executable code, which, when executed by the processor, is used to implement the steps of the distributed closed-loop control method for vehicle network channel congestion as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a program stored thereon; characterized in that, When executed by the processor, the program is used to implement the steps of the distributed closed-loop control method for vehicle network channel congestion as described in any one of claims 1 to 6.

Images