A management method for edge computing in internet of vehicles

Abstract

The application discloses a kind of about edge computing of Internet of Vehicles management method, focus on solving computing and power resource allocation problem, consider packet loss probability, queuing model and total transmission delay, based on the high reliability and low delay characteristics of mobile edge computing (MEC) model of Internet of Vehicles, make total achievable rate maximization, and total achievable rate refers to the maximum transmission capacity that network can realize in unit time.Established based on the MEC model of Internet of Vehicles, and utilize full duplex communication to transmit data, while using non-convex optimization algorithm to solve optimization problem;Including the following contents:First, a full duplex communication MEC model based on Internet of Vehicles is proposed, considering the influence of packet loss and retransmission delay, using full duplex communication to achieve nearly twice the spectral efficiency;Second, due to the NP difficulty of the optimization problem, a continuous convex similarity algorithm using first-order Taylor series expansion is used to eliminate the difference of non-convex structure;Third, a low-complexity processing method suitable for MEC server computing is proposed;Fourth, the channel gain is obtained by sensing the position and speed of the vehicle.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) communication technology, and more specifically, to a management method for edge computing in V2X. Background Technology

[0002] With the increasing number of vehicles, traffic accidents have become a global public safety issue. When drivers are in dense fog or beyond their line of sight, low visibility often poses serious dangers. If real-time information about nearby vehicles, including speed, direction, and location, were available, drivers could take immediate action to avoid accidents. The application of vehicle-to-everything (V2X) technology has effectively facilitated connectivity between vehicles, between vehicles and infrastructure, and between people and vehicles. This ensures that drivers can access real-time road information and conditions, and also enable intelligent control of their vehicles, thereby improving overall vehicle performance.

[0003] With the rapid development of wireless communication technology, vehicle-to-vehicle (V2V) communication has attracted significant attention from governments, businesses, and academic organizations. Generally, in-vehicle information interaction—between vehicles, between vehicles and pedestrians, and between vehicles and infrastructure—relies on cellular wireless networks or dedicated short-range communications that can cover the entire cell and provide high convenience and mobility. Furthermore, the development of wireless access technologies will significantly improve communication performance, such as timeliness, reliability, and network throughput, thereby promoting the development of in-vehicle communication. Simultaneously, vehicle communication technology provides a comfortable commuting experience for pedestrians and drivers through efficient traffic management and safety. The increasing number of computationally intensive vehicle computing services brings challenges such as limited vehicle power, channel congestion, and spectrum management. Therefore, vehicle computing services require specialized computing units to process large amounts of dense safety and environmental information.

[0004] Mobile edge computing (MEC) is a feasible platform for timely processing of computational problems, reducing channel congestion and handling large amounts of environmental information. MEC servers can be installed near base stations and roadside units using wired connections, and need to be located near transmitters to minimize backhaul latency. Vehicles or other mobile communication devices can transmit field application information and computational problems to base stations or roadside units in cellular wireless networks, and MEC servers can handle computational problems requiring low latency and high computing power. The current development of vehicle-to-everything (V2X) networks has also accelerated the development of MEC. Although V2X-based MEC can solve latency-sensitive and computationally intensive computational problems, there is still a lack of an efficient management scheme for utilizing computing, storage, and spectrum resources to ensure different quality of service for vehicle applications. Therefore, this invention focuses on computing and power resources, designing a management method for edge computing in V2X networks to optimize resource utilization.

[0005] The perception layer is the front-end layer in the operation of the vehicle-to-everything (V2X) network. Its main function is information collection, facilitating overall system control. Currently, the perception layer of the V2X network is mainly composed of two parts: sensing devices and sensor networks. Firstly, sensing devices are the core components, primarily using RFID wireless radio frequency devices, infrared sensors, and video monitoring equipment. Video monitoring equipment is one of the advanced detection systems used in current V2X systems; for example, it collects rearview camera data from a car's reversing mirror system through vehicle video detection. Secondly, the vehicle perception layer mainly utilizes sensor network systems to implement overall vehicle network technology applications, ensuring the high efficiency of the vehicle system.

[0006] In the 1G to 5G era, communication and sensing existed independently. For example, 4G communication systems only included communication, while radar systems were only responsible for speed measurement and sensing imaging. This separate design resulted in a waste of wireless spectrum and hardware resources, and the independent functions also led to high information processing latency. In the future, the communication spectrum will overlap with the traditional sensing spectrum, enabling joint scheduling of communication and sensing resources. The goal of future wireless communication is to provide services based on sensing and communication functions. To achieve this goal, Integrated Communication and Sensing (ISAC) has emerged and is attracting increasing attention from academia and industry.

[0007] Communication-sensing integration refers to a novel information processing technology that simultaneously achieves wireless sensing and wireless communication functions based on software and hardware resource sharing or information sharing. It can effectively improve system spectrum efficiency, hardware efficiency, and information processing efficiency. Domestic and international research institutions have conducted extensive research on the application scenarios and key technologies of communication-sensing integration. Among these, application scenarios mainly focus on assisted autonomous driving in vehicle-to-everything (V2X) scenarios and assisted drone path planning and drone monitoring in drone scenarios. In intelligent transportation scenarios, communication-sensing integration is mainly used for road monitoring, assisted autonomous driving, and high-definition map acquisition. In low-altitude drone scenarios, communication-sensing integration is mainly used for drone monitoring, path planning, and obstacle avoidance. Key technologies for communication-sensing integration mainly focus on integrated architecture, waveform design, interference cancellation, and network fusion technologies. Communication-sensing integration includes basic functions such as sensing data collection and processing, and sensing control. After collecting sensing-related data, the base station needs to report the collected data to the sensing control functional unit. Based on the processing and analysis of the sensing-related data, the sensing control functional unit generates corresponding strategies or control information and delivers it to relevant nodes or functions for execution through interface messages. The goal of integrated communication and sensing is to simultaneously support communication and sensing functions on the same spectrum and on the same equipment. This improves spectrum utilization, reduces equipment costs, and enables efficient collaboration and mutual benefit between the two functions. Through joint design of the air interface and protocols, and sharing of hardware and software equipment, integrated communication and sensing achieves communication and sensing functions on the same spectrum resources. This allows wireless networks to obtain information about target objects or the environment by analyzing direct, reflected, and scattered signals of wireless communication signals while conducting data communication, thereby improving spectrum utilization and equipment reuse.

[0008] In terms of communication, integrated communication and sensing requires the exchange of sensing information to assist in intelligent sensing and distributed computing. Based on intelligent sensing information, it enables optimized deployment and scheduling of communication resources to accurately meet the QoS requirements of services; it also enables intelligent deployment and operation and maintenance of the network, reducing industrial network costs. In terms of sensing, it requires adjustments to the field environment, service characteristics, wireless network environment, and network capabilities to help adaptive network management and maintenance, and to determine the optimal scheduling of network communication and computing resources.

[0009] The application of integrated communication and perception in the Internet of Vehicles (IoV) leverages the high-precision, refined perception, high-bandwidth, low-latency, high-quality communication, and intelligent cloud-edge-device collaborative computing capabilities based on integrated communication, sensing, and computing technologies. This fully supports the 3D high-definition display, precise tracking and positioning, and accurate recognition technologies required for immersive extended reality and holographic communication. It is capable of providing deeply immersive interactive scenarios such as extended reality and holographic communication, thereby building a bridge for intelligent interconnection between humans, machines, and things, and efficient communication between intelligent agents. This achieves seamless integration between the virtual and the real, helping human society move towards a new era of deep integration between the virtual and the real. Summary of the Invention

[0010] The purpose of this invention is to solve the above-mentioned technical problems by proposing a management method for edge computing in the Internet of Vehicles (IoV) environment.

[0011] To achieve the above objectives, the present invention adopts the following technical solution:

[0012] A management method for edge computing in the Internet of Vehicles (IoV), the steps of which are as follows:

[0013] S1. Establish a radar sensing-assisted topology link. Roadside units and base stations obtain the area of ​​the current sensing area and the number, location, and speed of vehicles within the detectable range through radar assistance, construct a capacity calculation model, and establish an optimization scheme.

[0014] S2. With the help of the communication sensing system, the superimposed signals are transmitted by the dual-function base station, providing services to both communication users and sensing targets at the same time. The beammap gain of the radar target is derived and used as a sensing indicator.

[0015] S3. After sensing vehicle information, the SCALE algorithm is used to optimize MEC task allocation and power. With the help of the lower bound expression, the converted achievable rate is obtained. Then, the interruption probability expression is expanded using the first-order Taylor series to convert it into a convex problem before solving it.

[0016] S4. Optimize the power allocation for vehicles using the Lagrange duality method to obtain the maximum achievable speed for all vehicles.

[0017] In a preferred embodiment of the present invention, the maximization of the achievable rate for all vehicles in step S4 specifically refers to:

[0018] Information about the target is determined by varying transmission delays, including its location and speed. Once the target is detected, it is tracked. The first [unclear phrase - likely referring to a specific event or process] in the uplink channel... achievable speed between a vehicle and a base station It can be represented as

[0019] (1)

[0020] (1) In the formula, It is the first The vehicle's transmission power, It is the base station's transmission power; Indicates the first Channel gain for each vehicle, and This represents additive white Gaussian noise and residual self-interference coefficients. The downlink rate of the base station refers to the rate at which the base station sends data to the user equipment. The uplink rate of the roadside unit refers to the data transmission rate when the mobile terminal sends information to the roadside unit. Other achievable rates for the base station and roadside unit during upload and download processes, i.e. , and As shown below

[0021]

[0022]

[0023]

[0024] in Let represent the residual self-interference coefficient, and satisfy . .

[0025] In a preferred embodiment of the present invention, the process of solving for the achievable rate is assumed to involve an infinite number of vehicles exchanging data packets with the MEC server at any given time, resulting in packet loss. Furthermore, due to vehicle power limitations and interference from other base stations and roadside units, the first... During the process of downloading information from a base station or roadside unit, the first vehicle... The probability of each data packet being lost is respectively used as and express

[0026]

[0027]

[0028] The inequalities in equations (5) and (6) represent the first... The probability that the signal-to-noise ratio of a vehicle is less than the minimum signal-to-noise ratio required to successfully download a data packet.

[0029] Assuming each MEC server operates for an unlimited number of users, a fast sequence of new data packets is bound to a queue, satisfying the M / G / 1 queuing theory; M, G, and 1 represent that the queuing time follows a Poisson distribution, while the service time follows a general distribution. A general distribution refers to a probability distribution in probability theory and statistics that does not belong to a specific probability distribution (such as the normal distribution, Poisson distribution, exponential distribution, etc.). Each queue has only one MEC server, and each data packet is retransmitted once when transmission fails.

[0030] Assuming a limited number of vehicles and that the number of data packets transmitted by each vehicle follows a negative exponential distribution, the average queuing time for each vehicle's data packets at the base station and roadside unit is... and Represented as:

[0031]

[0032]

[0033] In the formula, and From the first The number of data packets arriving from the vehicle to the base station and roadside unit per unit time; and It is the MEC server from the first The number of data packets processed per unit time from vehicle to base station and roadside unit;

[0034] Define the average transmission delay using , Represented as

[0035]

[0036]

[0037] Where E(·) represents the mathematical expectation. and This indicates the data packet transmission time from the vehicle to the base station and roadside unit.

[0038] In a preferred embodiment of the present invention, the specific step in step S4 of the edge computing management method of the present invention for obtaining the optimized achievable rate is as follows: The maximum sum of the achievable rates for all vehicle users is obtained by considering the packet transmission rate, packet loss probability, and average transmission delay.

[0039]

[0040] The restrictions are as follows:

[0041]

[0042] in To the maximum acceptable packet loss probability, The set representing natural numbers, The set representing positive integers. This is the maximum downlink transmission delay; where These represent the maximum transmission power of the base station, roadside unit, and vehicle, respectively. , Indicates the first The vehicle uploads data packets using a base station. , This indicates that data packets will not be uploaded using roadside units. and It is the minimum required rate that can be achieved by the base station and roadside unit.

[0043] In a preferred embodiment of the present invention, the specific steps of transforming the edge computing management method of the present invention into a convex problem and then performing convex optimization in step S3 are as follows:

[0044] The SCALE algorithm is used to solve non-convex problems as a relaxation method to avoid convexity, employing the following lower bound inequality:

[0045]

[0046] If z= The approximate constant is chosen as

[0047]

[0048]

[0049] Approximate vector and The optimal value can be obtained by iteration, defined

[0050]

[0051]

[0052]

[0053]

[0054] As a vehicle The lower limit of achievable transmission rate, pass This is achieved using a non-convex relaxation transformation, resulting in the rate expression being transformed as follows:

[0055]

[0056] The same method is used to transform the constraints.

[0057]

[0058]

[0059]

[0060] The interruption probability is expanded using a first-order Taylor series, and is expressed as follows:

[0061]

[0062]

[0063] in, These are the corresponding average power and Similarly, it can be and Transform into a convex form;

[0064] The following are relaxed results for the maximum realizable rate problem:

[0065]

[0066] In a preferred embodiment of the present invention, step S4 of the edge computing management method of the present invention specifically includes the following steps:

[0067] S401: Set the power variable to a constant, and the only optimization variable is the connection variable. The optimization problem is transformed into:

[0068]

[0069] in The specific optimization strategy employed is the Lagrange optimization method.

[0070] S402: The connection variable is set to the calculated result. The optimization variable and constraint in equation (26) include power. The power is optimized using the SCALE algorithm, which transforms it into a convex problem. The constraint becomes convex, and the objective function is also convex.

[0071]

[0072] st

[0073]

[0074] The power problem can be transformed into a dual problem. ,in It is a Lagrange multiplier, and its dual function can be expressed as:

[0075]

[0076]

[0077] in yes The Lagrange transformation is equivalent to the original optimization problem; based on Lagrange duality, the original problem is transformed into a dual problem to be solved, and the transformation result is as follows:

[0078]

[0079]

[0080] Due to dual equations Differentiable everywhere, we use gradient descent to calculate the Lagrange multipliers, and simultaneously use the Lagrange multipliers to solve for... In .

[0081] Inequality constraints Where g represents the constraint, the optimization problem is transformed into maximizing using the Lagrange multiplier method. Seeking a solution and Lagrange multipliers :

[0082]

[0083] right Find the partial derivative:

[0084]

[0085] make Seeking That's all.

[0086] Similarly, The same method can be used to solve it.

[0087] Find the Lagrange multipliers:

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096]

[0097] in is the learning rate for gradient descent.

[0098] Compared with existing technologies, this invention provides an optimized power control mechanism based on a software-defined network architecture, which has the following advantages:

[0099] 1. A full-duplex MEC model based on vehicle-to-everything (V2X) is constructed. Full-duplex communication is adopted, which can improve the spectrum resource utilization rate by more than double compared with half-duplex communication. At the same time, packet loss probability, queuing theory and total transmission delay are considered to maximize the overall achievability of high reliability and low latency of the V2X-based MEC model.

[0100] 2. A scaling algorithm based on the first-order Taylor series approximation is used to eliminate the differences in convex structures, thereby reducing the complexity of spectrum resource management.

[0101] 3. A simple and practical method is demonstrated for selecting the most suitable MEC server to handle computational problems.

[0102] 4. Improve the utilization rate of spectrum resources by using integrated sensing technology. Attached Figure Description

[0103] Figure 1 This is a flowchart of the present invention;

[0104] Figure 2 This is a system model diagram of the present invention;

[0105] Figure 3 This is a structural diagram of the MAQL algorithm of this invention. Detailed Implementation

[0106] To illustrate in detail the technical solutions adopted by the present invention to achieve the intended technical objectives, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Furthermore, the technical means or technical features in the embodiments of the present invention can be replaced without creative effort. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0107] First, let's explain the technical terms used in this application:

[0108] Vehicle-to-everything (V2X) networks support various vehicle-related applications, such as information security and entertainment, as well as vehicle sensing and data processing services. A typical 5G network solution, V2X networks require ultra-high reliability, low-latency data communication transmission, and high-throughput or massive connectivity. They necessitate advanced mobile communication systems to enable timely data transmission, sharing, and computation between vehicles and between vehicles and other devices, including V2N (Vehicle-to-Network), V2V (Vehicle-to-Vehicle), V2P (Vehicle-to-Pedestrian), and V2I (Vehicle-to-Infrastructure).

[0109] This invention relates to a management method for edge computing in the Internet of Vehicles (IoV), such as... Figure 1 As shown, the steps of the edge computing management method are as follows:

[0110] S1. Establish a radar sensing-assisted topology link. Roadside units and base stations obtain the area of ​​the current sensing area and the number, location, and speed of vehicles within the detectable range through radar assistance, construct a capacity calculation model, and establish an optimization scheme.

[0111] S2. With the help of the communication sensing system, the superimposed signals are transmitted by the dual-function base station, providing services to both communication users and sensing targets at the same time. The beammap gain of the radar target is derived and used as a sensing indicator.

[0112] S3. After sensing vehicle information, the SCALE algorithm is used to optimize MEC task allocation and power. With the help of the lower bound expression, the converted achievable rate is obtained. Then, the interruption probability expression is expanded using the first-order Taylor series to convert it into a convex problem before solving it.

[0113] S4. Optimize the power allocation for vehicles using the Lagrange duality method to obtain the maximum achievable speed for all vehicles.

[0114] like Figure 2 As shown in this embodiment, a full-duplex communication vehicle network model based on edge computing is presented. Full-duplex transmission supports simultaneous transmission and reception within the same frequency band, potentially doubling the throughput compared to half-duplex transmission. The main problem in full-duplex transmission stems from interference caused by its self-transmission, which can be suppressed using various self-interference suppression techniques. It is assumed that the self-interference in full-duplex transmission is eliminated within an acceptable range. The roadside unit is a DFRC system employing a millimeter-wave high-mass uniform linear array, consisting of an Nt transmitting antenna and an Nr receiving antenna. By utilizing full-duplex radio technology on the transmitting and receiving antennas, the roadside unit can receive potential signal echoes for sensing while maintaining uninterrupted downlink communication. Vehicles can upload environmental information and computational problems to the base station or roadside unit according to different scheduling schemes; based on... Figure 2 The example diagram assumes there are 800 vehicles, 20 roadside units, an area of ​​72 square kilometers, and 3 base stations.

[0115] In this embodiment, the model is simplified, considering only one MEC server connected to the base station and multiple MEC servers connected to roadside units. The model consists of 8,000 mobile vehicles. By sensing the position, rate, and channel gain of the vehicles, the channel gain is obtained, with the goal of maximizing the achievable rate of all mobile vehicles.

[0116] Information about the target is determined by varying transmission delays, including its location and speed. Once the target is detected, it is tracked. The first [unclear phrase - likely referring to a specific event or process] in the uplink channel... achievable speed between a vehicle and a base station It can be represented as

[0117]

[0118] In equation (42), It is the first The vehicle's transmission power, It is the base station's transmission power; Indicates the first Channel gain for each vehicle, and This represents additive white Gaussian noise and residual self-interference coefficients. The downlink rate of the base station refers to the rate at which the base station sends data to the user equipment. The uplink rate of the roadside unit refers to the data transmission rate when the mobile terminal sends information to the roadside unit. Other achievable rates for the base station and roadside unit during upload and download processes, i.e. , and As shown below

[0119]

[0120]

[0121]

[0122] in Let represent the residual self-interference coefficient, and satisfy . .

[0123] In the process of solving for the achievable rate, it is assumed that an infinite number of vehicles exchange data packets with the MEC server at any given time, resulting in packet loss. Furthermore, due to vehicle power limitations and interference from other base stations and roadside units, the first... During the process of downloading information from a base station or roadside unit, the first vehicle... The probability of each data packet being lost is respectively used as and express

[0124]

[0125]

[0126] The inequalities in equations (46) and (47) represent the first... The probability that the signal-to-noise ratio of a vehicle is less than the minimum signal-to-noise ratio required to successfully download a data packet.

[0127] Assuming each MEC server operates for an unlimited number of users, a fast sequence of new data packets is bound to a queue, satisfying the M / G / 1 queuing theory; M, G, and 1 represent that the queuing time follows a Poisson distribution, while the service time follows a general distribution. A general distribution refers to a probability distribution in probability theory and statistics that does not belong to a specific probability distribution (such as the normal distribution, Poisson distribution, exponential distribution, etc.). Each queue has only one MEC server, and each data packet is retransmitted once when transmission fails.

[0128] Assuming a limited number of vehicles and that the number of data packets transmitted by each vehicle follows a negative exponential distribution, the average queuing time for each vehicle's data packets at the base station and roadside unit is... and Represented as:

[0129]

[0130]

[0131] In the formula, and From the first The number of data packets arriving from the vehicle to the base station and roadside unit per unit time; and It is the MEC server from the first The number of data packets processed per unit time from vehicle to base station and roadside unit;

[0132] Define the average transmission delay using , Represented as

[0133]

[0134]

[0135] Where E(·) represents the mathematical expectation. and This indicates the data packet transmission time from the vehicle to the base station and roadside unit.

[0136] The specific steps in step S4 of the edge computing management method to obtain the optimized achievable rate are as follows: The maximum sum of the achievable rates for all vehicle users is obtained by considering packet transmission rate, packet loss probability, and average transmission delay.

[0137]

[0138] The restrictions are as follows:

[0139]

[0140] in To the maximum acceptable packet loss probability, The set representing natural numbers, The set representing positive integers. This is the maximum downlink transmission delay; where These represent the maximum transmission power of the base station, roadside unit, and vehicle, respectively. , Indicates the first The vehicle uploads data packets using a base station. , This indicates that data packets will not be uploaded using roadside units. and It is the minimum required rate that can be achieved by the base station and roadside unit.

[0141] The specific steps for transforming the edge computing management method into a convex problem and then performing convex optimization in step S3 are as follows:

[0142] The SCALE algorithm is used to solve non-convex problems as a relaxation method to avoid convexity, employing the following lower bound inequality:

[0143]

[0144] If z= The approximate constant is chosen as

[0145]

[0146]

[0147] Approximate vector and The optimal value can be obtained by iteration, defined

[0148]

[0149]

[0150]

[0151]

[0152] As a vehicle The lower limit of achievable transmission rate, pass This is achieved using a non-convex relaxation transformation, resulting in the rate expression being transformed as follows:

[0153]

[0154] The same method is used to transform the constraints.

[0155]

[0156]

[0157]

[0158] The interruption probability is expanded using a first-order Taylor series, and is expressed as follows:

[0159]

[0160]

[0161] in, These are the corresponding average power and Similarly, it can be and Transform into a convex form;

[0162] The following are relaxed results for the maximum realizable rate problem:

[0163]

[0164] The specific steps of step S4 in the edge computing management method are as follows:

[0165] S401: Set the power variable to a constant, and the only optimization variable is the connection variable. The optimization problem is transformed into:

[0166]

[0167] in The specific optimization strategy employed is the Lagrange optimization method.

[0168] S402: The connection variable is set to the calculated result. The optimization variable and constraint in equation (67) include power. The power is optimized using the SCALE algorithm, which transforms it into a convex problem. The constraint becomes convex, and the objective function is also convex.

[0169]

[0170] ,

[0171]

[0172] The power problem can be transformed into a dual problem. ,in It is a Lagrange multiplier, and its dual function can be expressed as:

[0173]

[0174]

[0175]

[0176] in yes The Lagrange transformation is equivalent to the original optimization problem; based on Lagrange duality, the original problem is transformed into a dual problem to be solved, and the transformation result is as follows:

[0177]

[0178]

[0179] Due to dual equations Differentiable everywhere, we use gradient descent to calculate the Lagrange multipliers, and simultaneously use the Lagrange multipliers to solve for... In .

[0180] Inequality constraints Where g represents the constraint, the optimization problem is transformed into maximizing using the Lagrange multiplier method. Seeking a solution and Lagrange multipliers :

[0181]

[0182] right Find the partial derivative:

[0183]

[0184] make Seeking That's all.

[0185] Similarly, The same method can be used to solve it.

[0186] Find the Lagrange multipliers:

[0187]

[0188]

[0189]

[0190]

[0191]

[0192]

[0193] (80)

[0194] (81)

[0195] (82)

[0196] in is the learning rate for gradient descent.

[0197] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A management method for edge computing in the Internet of Vehicles, characterized in that, The steps of the edge computing management method are as follows: S1. Establish a radar sensing-assisted topology link. Roadside units and base stations obtain the area of ​​the current sensing area and the number, location, and speed of vehicles within the detectable range through radar assistance, construct a capacity calculation model, and establish an optimization scheme. S2. With the help of the communication sensing system, the superimposed signals are transmitted by the dual-function base station, providing services to both communication users and sensing targets at the same time. The beammap gain of the radar target is derived and used as a sensing indicator. S3. After sensing vehicle information, the SCALE algorithm is used to optimize MEC task allocation and power. With the help of the lower bound expression, the converted achievable rate is obtained. Then, the interruption probability expression is expanded using the first-order Taylor series to convert it into a convex problem before solving it. S4. Optimize the power allocation of vehicles using the Lagrange duality method to obtain the maximum achievable speed for all vehicles; The specific steps in S4 of the edge computing management method to obtain the maximum achievable rate for all vehicles are as follows: The maximum sum of the achievable rates for all vehicle users is obtained by considering packet transmission rate, packet loss probability, and average transmission delay. （1） The restrictions are as follows: ； in To the maximum acceptable packet loss probability, The set representing natural numbers, The set representing positive integers. This is the maximum downlink transmission delay; where These represent the maximum transmission power of the base station, roadside unit, and vehicle, respectively. , Indicates the first The vehicle uploads data packets using a base station. , This indicates that data packets are being uploaded using roadside units. and It is the minimum required rate that can be achieved by the base station and roadside unit; The maximization of the achievable speed for all vehicles in step S4 specifically refers to: Information about the target is determined by varying transmission delays, including its location and speed. Once the target is detected, it is tracked. The first [unclear phrase - likely referring to a specific event or process] in the uplink channel... achievable speed between a vehicle and a base station It can be represented as (2) (2) In the formula, It is the first The vehicle's transmission power, It is the base station's transmission power; Indicates the first Channel gain for each vehicle, and This represents additive white Gaussian noise and residual self-interference coefficients. The downlink rate of a base station refers to the rate at which the base station sends data to the user equipment. The uplink rate of a roadside unit refers to the data transmission rate when a mobile terminal sends information to a roadside unit. Other achievable rates for the base station and roadside units during upload and download processes are also included. , and As shown below (3) (4) (5) in Let represent the residual self-interference coefficient, and satisfy . ; In the process of solving for the converted achievable rate in step S3, it is assumed that there are an infinite number of vehicles exchanging data packets with the MEC server at any given time, resulting in packet loss. Due to vehicle power limitations and interference from other base stations and roadside units, the first... During the process of downloading information from a base station or roadside unit, the first vehicle... The probability of each data packet being lost is respectively used as and express (6) (7) The inequalities in equations (6) and (7) represent the first... The probability that the signal-to-noise ratio of a vehicle is less than the minimum signal-to-noise ratio required to successfully download a data packet; Assuming each MEC server operates with an unlimited number of users, a fast sequence of new data packets is bound to a queue, satisfying the M / G / 1 queuing theory; M, G, and 1 indicate that the queuing time follows a Poisson distribution, while the service time follows a general distribution. The general distribution refers to a probability distribution in probability theory and statistics that does not belong to a specific probability distribution. There is only one MEC server for each queue, and each data packet is retransmitted once when a transmission fails. Assuming a limited number of vehicles and that the number of data packets transmitted by each vehicle follows a negative exponential distribution, the average queuing time for each vehicle's data packets at the base station and roadside unit is... and Represented as: (8) (9) In the formula, and From the first The number of data packets arriving from the vehicle to the base station and roadside unit per unit time; and It is the MEC server from the first The number of data packets processed per unit time from vehicle to base station and roadside unit; Define the average transmission delay using , Represented as (10) (11) Where E(·) represents the mathematical expectation; and This indicates the data packet transmission time from the vehicle to the base station and roadside unit; In S3 of the edge computing management method, the SCALE algorithm is used to optimize MEC task allocation and power. Using a lower bound expression, the transformed achievable rate is obtained. Then, a first-order Taylor series expansion is used to expand the interruption probability expression, transforming it into a convex problem before solving it. The specific steps are as follows: The SCALE algorithm is used to solve non-convex problems as a relaxation method to avoid convexity, employing the following lower bound inequality: (12) If z= The approximate constant is chosen as (13) (14) definition (15) (16) (17) (18) As a vehicle The lower limit of achievable transmission rate, pass This is achieved using a non-convex relaxation transformation, resulting in the rate expression being transformed as follows: ； The same method is used to transform the constraints. ； ； ； The interruption probability is expanded using a first-order Taylor series, and is expressed as follows: ； ； Similarly, it can be and Transform into a convex form; The following are relaxed results for the maximum realizable rate problem: （25）。 2. The management method for edge computing in the Internet of Vehicles according to claim 1, characterized in that, The specific steps in S4 of the edge computing management method to optimize the power allocation of vehicles using the Lagrange dual method to obtain the maximum achievable rate for all vehicles are as follows: S401: Set the power variable to a constant, and the only optimization variable is the connection variable. The optimization problem is transformed into: （26） in The specific optimization strategy adopted is the Lagrange optimization method. S402: The connection variable is set to the calculated result. The optimization variable and constraint in equation (26) include power. The power is optimized using the SCALE algorithm, which transforms it into a convex problem. The constraint becomes convex, and the objective function is also convex. （27） ， ； The power problem can be transformed into a dual problem. ,in It is a Lagrange multiplier, and its dual function can be expressed as: （28） （29） in yes The Lagrange transformation is equivalent to the original optimization problem; based on Lagrange duality, the original problem is transformed into a dual problem to be solved, and the transformation result is as follows: （30） ； Due to dual equations Differentiable everywhere, we use gradient descent to calculate the Lagrange multipliers, and simultaneously use the Lagrange multipliers to solve for... In ; Inequality constraints Where g represents the constraint, the optimization problem is transformed into maximizing using the Lagrange multiplier method. Seeking a solution and Lagrange multipliers : （31） right Find the partial derivative: （32） make Seeking That's all; Similarly, The same method can be used to solve it. Find the Lagrange multipliers: （33） （34） （35） （36） （37） （38） （39） （40） （41） in is the learning rate for gradient descent.

Images