A hybrid traffic scenario synaesthesia resource allocation method, system, device and medium
By establishing a V2V communication model from vehicle to vehicle in hybrid traffic scenarios and optimizing resource allocation in deep reinforcement learning, the problem of low vehicle resource allocation efficiency in hybrid traffic scenarios is solved, and more efficient resource utilization and communication quality improvement is achieved.
Patent Information
- Application Number
- CN202310729810.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-20
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2043-06-20
AI Technical Summary
The prior art is difficult to effectively allocate computing resources between vehicles in hybrid traffic scenarios, especially when traditional vehicles exist, and fail to fully consider the influence of third-party participants, resulting in serious contradictions in resource supply and demand, and impact on transmission rates and energy consumption. .
By establishing a V2V communication model of vehicle-to-vehicle in hybrid traffic scenarios, the energy consumption and delay of intelligent connected vehicles and traditional vehicles are calculated, the optimal unloading action is generated using deep reinforcement learning, resource allocation is optimized, and the penetration rate is considered to the transmission delay and The impact of energy consumption realizes resource allocation from vehicles to vehicles.
It improves the efficiency of vehicle task unloading in hybrid traffic scenarios, alleviates the contradiction between resource supply and demand, reduces the total energy consumption and total delay of the vehicle, and significantly improves the communication quality.
Smart Images

Figure CN116744370B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of wireless communications, and in particular relates to a method, system, device and medium for allocating synaesthesia resources in a mixed traffic scenario. Background Art
[0002] In recent years, as more and more intelligent networked vehicles (CAVs) have entered the market, we will usher in a mixed traffic era where traditional vehicles HDVs and intelligent networked vehicles (CAVs) coexist. Vehicles with computing-intensive task offloading needs to offload tasks to available autonomous driving vehicles around them through vehicle-to-vehicle (V2V) communication. However, there are limited autonomous driving vehicles with idle computing resources in a certain area, and the presence of traditional vehicles HDVs results in a distance between vehicles that is different from that in the fully intelligent networked vehicle (CAV) scenario, which affects the transmission rate.
[0003] In the existing technology, there is little research on resource allocation strategies for collaborative computing between vehicles in mixed traffic scenarios, while there are abundant algorithms available for reference in similar scenarios. Among many studies, mobile edge computing is considered to be a promising technology and is widely used in the process of task offloading of devices.
[0004] The paper "Distributed Deep Learning-based Offloading for Mobile Edge Computing Networks, Mobile Netw Appl 27, 1123–1130 (2022)" proposes a deep learning-based distributed offloading algorithm for mobile edge computing MEC networks, which uses multiple parallel deep neural networks DNN to generate offloading decisions. Its application scenario involves multiple wireless devices and an edge server. Regarding the question of whether the wireless device offloads the task to the edge server, the author generates an offloading decision by minimizing the weighted sum of the energy consumption of all wireless devices and the task completion delay, so as to save energy and ensure the service quality of the wireless device.
[0005] The paper "Deep Reinforcement Learning Based Computing Offloading for MEC-assisted Heterogeneous Vehicular Networks, 2020 IEEE 20th International Conference on Communication Technology (ICCT), Nanning, China, 2020, pp. 927-932." proposed a vehicle edge computing offloading solution based on deep reinforcement learning, designed two computing offloading schemes, serial and parallel offloading schemes, and formulated the optimization problem as a Markov decision process problem, which helped to improve the total computing rate at the cost of high computing latency.
[0006] The above-mentioned existing technologies have improved the efficiency of task offloading in their application scenarios, but it is difficult to directly apply them to mixed traffic scenarios. The impact of HDVs needs to be considered in mixed traffic scenarios, but it is obvious that the above-mentioned methods do not consider the impact of third-party participants. Summary of the invention
[0007] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide a method, system, device and medium for allocating synaesthesia resources in mixed traffic scenarios. By modeling the latency and energy consumption in edge computing and local computing, the system utility of the weighted sum of energy consumption and latency is obtained; with the purpose of minimizing the system utility, the optimal unloading action is generated through deep reinforcement learning. The present invention can improve the efficiency of vehicle-to-vehicle task unloading in mixed traffic scenarios, alleviate the contradiction between resource supply and demand, and has the characteristic of being able to quickly obtain the optimal unloading action.
[0008] A method for allocating synaesthesia resources in a mixed traffic scenario comprises the following steps:
[0009] Step 1, establish a mixed traffic scenario vehicle-to-vehicle (V2V) communication model;
[0010] Step 2, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the energy consumption and delay of the intelligent connected vehicle (CAV) serving as the edge server in the mixed traffic scenario, i.e., the service vehicle (SV);
[0011] Step 3, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the local energy consumption and delay;
[0012] Step 4: According to the energy consumption and delay formulas of vehicle-to-vehicle V2V communication in mixed traffic scenarios in steps 2 and 3, a resource optimization problem R related to delay and energy consumption is established;
[0013] Step 5: By minimizing the objective function Q(d,x) of the resource optimization problem R, the optimal offloading action {x m}, complete the allocation of vehicle-to-vehicle V2V resources in the intelligent connected vehicle (CAVs) scenario.
[0014] The specific method of step 1 is:
[0015] The hybrid traffic scenario includes CAVs and HDVs. CAVs communicate with each other via a vehicle-to-vehicle (V2V) link. CAVs are divided into client vehicles (CVs) and service vehicles (SVs). When executing an offloading decision, client vehicles (CVs) offload tasks to service vehicles (SVs) or process them locally.
[0016] Use the Internet of Vehicles simulation software to randomly generate N vehicles, and randomly select N of them according to the penetration rate p. CAV The number of CAVs involved in the calculation is 1, and the rest are traditional HDVs. CAV It can be obtained by the following formula:
[0017] N CAV =pN
[0018] Among them, N CAV represents the number of CAVs, p represents the penetration rate, and N represents all vehicles.
[0019] The specific method of step 2 is:
[0020] 2.1, calculate the total energy consumption of the client vehicle CV to offload its task m to the service vehicle SV for processing:
[0021]
[0022] Among them, d m represents the workload of the mth task, represents the energy consumed by the client vehicle CV to upload its workload m to the service vehicle SV, where α is the weight of the server energy consumption. When α = 0, only the energy consumption of the client vehicle CV is considered. This includes the energy consumption of sending the task and the energy consumption of the server utility that executes this task;
[0023] 2.2, calculate the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, which is given by:
[0024]
[0025] Among them, d mrepresents the workload of the mth task, c represents the bandwidth of communication between the customer vehicle CV and the service vehicle SV;
[0026] 2.3, calculate the service vehicle SV processing delay, which is given by the following formula:
[0027]
[0028] Among them, d m represents the workload of the mth task, f c Indicates the server processing rate;
[0029] 2.4, according to the uninstall decision {x m}, calculate the total delay of the client vehicle CV offloading the task to the service vehicle SV for processing, which is given by the following formula:
[0030]
[0031] in, represents the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, represents the service vehicle SV processing delay, x m represents the offloading decision of task m.
[0032] The specific method of step 3 is:
[0033] 3.1, calculate the energy consumption of the client vehicle CV to perform its task m locally, which is given by:
[0034]
[0035] Among them, d m represents the workload of the mth task, e l Represents the local energy consumption of each data bit of the client vehicle CV;
[0036] 3.2, calculate the total processing time of the client vehicle CV to perform its task m, which is given by:
[0037]
[0038] Among them, d m represents the workload of the mth task, t l The local processing time for each data bit representing the client vehicle CV;
[0039] 3.3, according to the unloading decision of the customer vehicle CV {x m}, calculate the total delay for the client vehicle CV to complete its task locally, which is given by:
[0040]
[0041] in, represents the total processing time of the client vehicle CV to perform its task m, x m represents the offloading decision of task m.
[0042] The specific method of step 4 is:
[0043] 4.1 Assuming that the vehicle-to-vehicle V2V communication speed and energy loss per kilometer remain unchanged, the relationship between the transmission delay and transmission energy consumption of vehicle-to-vehicle V2V communication in mixed traffic scenarios and the penetration rate p, that is, the CAV occupancy rate is:
[0044] The permeability p is defined as follows:
[0045]
[0046] Where p represents permeability, N CAV Represents the number of intelligent connected vehicles, N HDV represents the number of conventional vehicles;
[0047] Define a one-way road of length S0. On this road, the communication vehicles include CAVs and HDVs. Assume that CAVs are evenly distributed on this road. The traffic density on this road is constant. The distance between any two adjacent vehicles is fixed to D0. The number of vehicles that this road can accommodate is fixed to N0. Assume that the distance between two adjacent CAVs is D CAV , the following relationship exists:
[0048] (N0-1)D0=S0
[0049] N CAV +N HDV =N0
[0050] (N CAV +N HDV -1) D0 = S0
[0051] When S0>>D CAV , the following formula holds:
[0052] (N CAV -1)D CAV =S0
[0053] Right now:
[0054] (N CAV -1)D CAV =(N CAV +N HDV -1)D0
[0055] When N CAV >>1, N0>>1, the following equation is true:
[0056]
[0057] When the communication speed remains unchanged, the communication delay is proportional to the communication distance. Proportional to the distance, communication delay is inversely proportional to the penetration rate. Assuming that the energy consumption is lost by the same amount per kilometer during the transmission process, the transmission energy consumption is also proportional to the distance and inversely proportional to the penetration rate.
[0058] 4.2, according to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission delay in the fully intelligent connected vehicle (CAV) scenario, the transmission delay under the penetration rate is calculated, which is expressed as follows:
[0059]
[0060] in, Transmission delay in the fully intelligent connected vehicle (CAV) scenario;
[0061] According to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission energy consumption in the fully intelligent connected vehicle (CAV) scenario, the transmission energy consumption under this penetration rate is calculated, which is expressed as follows:
[0062]
[0063] in, It represents the transmission energy consumption in the scenario of fully intelligent connected vehicle (CAV);
[0064] 4.3, using the system utility Q(d,x), defined as the weighted sum of energy consumption and task completion delay:
[0065]
[0066] in, β represents the weight of energy consumption and task completion time;
[0067] 4.4, formulate the resource optimization problem R, for the task transfer of the client vehicle CV, by optimizing its unloading decision {x m}, in order to minimize the Q value, it can be expressed as follows:
[0068] R:
[0069] stx m ∈{0,1}
[0070] Among them, Q(d,x) represents the system utility, x mrepresents the offloading decision of task m.
[0071] The specific method of step 5 is:
[0072] 5.1, using multiple parallel deep neural networks DNN to generate binary offloading decisions, and selecting a set of offloading decisions x with the minimum system utility * As output, store (d,x * ) and used to train the DNN later, repeating the above process until the system utility Q(d,x) reaches the minimum value in this scenario;
[0073] 5.2, set the gain ratio to the ratio of the historical minimum system utility to the current system utility;
[0074] 5.3. According to the above settings, when the gain ratio converges, the unloading strategy corresponding to minimizing the system utility is obtained, and the resource allocation between the intelligent connected vehicles (CAVs) in the mixed traffic scenario is completed.
[0075] The specific method for the client vehicle CVs to offload tasks to the service vehicle SVs or process them locally is:
[0076] Assume that a customer vehicle CV has M independent tasks to be processed, expressed as: Use d m Denotes the workload of the mth task. The customer vehicle CV can decide whether to offload its task m to the service vehicle SV. This process is represented by a set of binary offloading decisions. The offloading decision is represented by the variable x m To express, x m represents the offloading decision of the mth task, where x m ∈{0,1},x m =1 means that the customer vehicle CV decides to offload task m to the service vehicle SV, x m =0 means that the client vehicle CV decides to execute task m locally.
[0077] The present invention also provides a hybrid traffic scenario synaesthesia resource allocation system, comprising:
[0078] Transmission delay calculation module: Using the penetration rate p, calculate the delay of the intelligent connected vehicle (CAV) as the edge server in the mixed traffic scenario, that is, the service vehicle (SV);
[0079] Module for calculating transmission energy consumption: Using the penetration rate p, calculate the energy consumption of the intelligent connected vehicle (CAV) as the edge server in the mixed traffic scenario, i.e., the service vehicle (SV);
[0080] Calculate local energy consumption module: calculate the energy consumption required for local processing tasks based on the size of the task;
[0081] Module for calculating local delay: Calculates the delay required for local processing tasks based on the size of the task;
[0082] Unloading decision generation module: establishes resource optimization problems related to energy consumption and latency and generates unloading decisions, and obtains binary unloading decisions by minimizing system utility to complete resource allocation.
[0083] The present invention also provides a hybrid traffic scenario synaesthesia computing resource allocation device, comprising:
[0084] Memory: a computer program storing the above-mentioned method for allocating synaesthesia resources in a mixed traffic scenario, which is a computer-readable device;
[0085] Processor: used to implement the method for allocating synaesthesia resources in a mixed traffic scenario when executing the computer program.
[0086] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can implement the method for allocating synaesthesia resources in a mixed traffic scenario.
[0087] Compared with the prior art, the present invention has the following beneficial effects:
[0088] 1. The impact of HDVs on the communication between CAVs in mixed traffic scenarios is clarified;
[0089] When studying the impact of traditional vehicles HDVs in mixed traffic scenarios on the communication between intelligent connected vehicles CAVs, the present invention proves the impact of penetration rate on the distance between intelligent connected vehicles CAVs. Based on the relationship between distance and transmission delay and transmission energy consumption, the influence of penetration rate on transmission delay and transmission energy consumption is proved, and the conclusion is drawn that transmission delay and transmission energy consumption are inversely proportional to penetration rate.
[0090] 2. Alleviate the contradiction between resource supply and demand to a greater extent;
[0091] The present invention gives the intelligent networked vehicles CAVs the function of mobile edge servers, which can alleviate the contradiction between resource supply and demand to a greater extent. The number of roadside units as mobile edge servers is limited, and vehicles are easy to leave their working range. In contrast, intelligent networked vehicles CAVs have a longer companionship time, and there are more intelligent networked vehicles CAVs with computing resources. Using intelligent networked vehicles CAVs with idle computing resources to offload tasks can bring greater benefits, significantly alleviate the contradiction between resource supply and demand, and improve communication quality.
[0092] 3. Compared with the existing methods, the present invention takes into account that one party in vehicle-to-vehicle V2V communication can serve as a mobile edge server, and considers the impact of traditional vehicle HDVs on transmission delay and transmission energy consumption, thereby realizing a vehicle-to-vehicle V2V resource allocation strategy in mixed traffic scenarios and improving the efficiency of vehicle task offloading.
[0093] To sum up, compared with the prior art, the present invention implements a resource allocation strategy based on mixed traffic scenarios, which can improve the efficiency of vehicle task unloading in mixed traffic scenarios, reduce vehicle cost consumption, and has the advantages of quickly deriving unloading actions and reducing the total energy consumption and total delay of vehicle task processing. BRIEF DESCRIPTION OF THE DRAWINGS
[0094] Figure 1 It is a flowchart of the implementation process of the present invention.
[0095] Figure 2 It is a scene schematic diagram of the present invention.
[0096] Figure 3 This is a diagram that proves the relationship between transmission energy consumption, transmission delay and penetration rate in the present invention.
[0097] Figure 4 This is a simulation diagram comparing the gain ratios under different permeabilities of the present invention.
[0098] Figure 5 It is a simulation diagram comparing the system utility of different task processing methods under different permeabilities of the present invention. DETAILED DESCRIPTION
[0099] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.
[0100] A method for allocating synaesthesia resources in a mixed traffic scenario comprises the following steps:
[0101] like Figure 1 As shown, step 1, establishing a mixed traffic scenario vehicle-to-vehicle V2V communication model;
[0102] like Figure 2 As shown, the hybrid traffic scenario used in the present invention includes intelligent network-connected vehicles CAVs and traditional vehicles HDVs. The intelligent network-connected vehicles CAVs communicate with each other through a vehicle-to-vehicle V2V link. The intelligent network-connected vehicles CAVs can be divided into customer vehicles CVs and service vehicles SVs. When executing an offloading decision, the customer vehicle CVs offloads the task to the service vehicle SVs or processes it locally;
[0103] Use the Internet of Vehicles simulation software to randomly generate N vehicles, and randomly select N of them according to the penetration rate p. CAVThe number of CAVs involved in the calculation is 1, and the rest are traditional HDVs. CAV It can be obtained by the following formula:
[0104] N CAV =pN
[0105] Among them, N CAV represents the number of intelligent connected vehicles (CAVs), p represents the penetration rate, and N represents all vehicles;
[0106] Assume that a customer vehicle CV has M independent tasks to be processed, expressed as: Use d m Denotes the workload of the mth task. The customer vehicle CV can decide whether to offload its task m to the service vehicle SV. This process is represented by a set of binary offloading decisions. The offloading decision is represented by the variable x m To express, x m represents the offloading decision of the mth task, where x m ∈{0,1}, specifically, x m =1 means that the customer vehicle CV decides to offload task m to the service vehicle SV, x m =0 means that the client vehicle CV decides to execute task m locally.
[0107] Step 2, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the energy consumption and delay of the intelligent connected vehicle (CAV) serving as the edge server in the mixed traffic scenario, i.e., the service vehicle (SV);
[0108] 2.1, calculate the total energy consumption of the client vehicle CV to offload its task m to the service vehicle SV for processing:
[0109]
[0110] Among them, d m represents the workload of the mth task, represents the energy consumed by the client vehicle CV to upload its workload m to the service vehicle SV, where α is the weight of the server energy consumption. When α = 0, only the energy consumption of the client vehicle CV is considered. This includes the energy consumption of sending the task and the energy consumption of the server utility that executes this task;
[0111] 2.2, calculate the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, which is given by:
[0112]
[0113] Among them, d mrepresents the workload of the mth task, c represents the bandwidth of communication between the customer vehicle CV and the service vehicle SV;
[0114] 2.3, calculate the service vehicle SV processing delay, which is given by the following formula:
[0115]
[0116] Among them, d m represents the workload of the mth task, f c Indicates the server processing rate;
[0117] 2.4, according to the uninstall decision {x m}, calculate the total delay of the client vehicle CV offloading the task to the service vehicle SV for processing, which is given by the following formula:
[0118]
[0119] in, represents the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, represents the service vehicle SV processing delay, x m represents the offloading decision of task m.
[0120] Step 3, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the local energy consumption and delay;
[0121] 3.1, calculate the energy consumption of the client vehicle CV to perform its task m locally, which is given by:
[0122]
[0123] Among them, d m represents the workload of the mth task, e l Represents the local energy consumption of each data bit of the client vehicle CV;
[0124] 3.2, calculate the total processing time of the client vehicle CV to perform its task m, which is given by:
[0125]
[0126] Among them, d m represents the workload of the mth task, t l The local processing time for each data bit representing the client vehicle CV;
[0127] 3.3, according to the unloading decision of the customer vehicle CV {x m}, calculate the total delay for the client vehicle CV to complete its task locally, which is given by:
[0128]
[0129] in, represents the total processing time of the client vehicle CV to perform its task m, x m represents the offloading decision of task m.
[0130] Step 4: According to the energy consumption and delay formulas of vehicle-to-vehicle V2V communication in mixed traffic scenarios in steps 2 and 3, a resource optimization problem R related to delay and energy consumption is established;
[0131] 4.1 Assuming that the vehicle-to-vehicle V2V communication speed and energy loss per kilometer remain unchanged, the relationship between the transmission delay and transmission energy consumption of vehicle-to-vehicle V2V communication in mixed traffic scenarios and the penetration rate p, that is, the CAV occupancy rate is:
[0132] The permeability p is defined as follows:
[0133]
[0134] Where p represents permeability, N CAV Represents the number of intelligent connected vehicles, N HDV represents the number of conventional vehicles;
[0135] like Figure 3 As shown, it is the reference model for proving the relationship between transmission delay and transmission energy consumption and penetration rate under ideal conditions in step 4.
[0136] Define a one-way road of length S0. On this road, the communication vehicles include CAVs and HDVs. We assume an ideal situation where CAVs are evenly distributed on this road and the traffic density on this road remains unchanged, that is, the distance between any two adjacent vehicles is fixed to D0. In this case, the number of vehicles that this road can accommodate is fixed to N0. Assume that the distance between two adjacent CAVs is D CAV , the following relationship exists:
[0137] (N0-1)D0=S0
[0138] N CAV +N HDV =N0
[0139] (N CAV +N HDV -1) D0 = S0
[0140] When S0>>D CAV , the following formula holds:
[0141] (NCAV -1)D CAV =S0
[0142] Right now:
[0143] (N CAV -1)D CAV =(N CAV +N HDV -1)D0
[0144] When N CAV >>1, N0>>1, the following equation is true:
[0145]
[0146] When the communication speed remains unchanged, the communication delay is proportional to the communication distance. So the communication delay is inversely proportional to the penetration rate. Assuming that the energy consumption is lost by the same amount per kilometer during the transmission process, the transmission energy consumption is also proportional to the distance and inversely proportional to the penetration rate.
[0147] 4.2, according to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission delay in the fully intelligent connected vehicle (CAV) scenario, the transmission delay under the penetration rate is calculated, which is expressed as follows:
[0148]
[0149] in, Transmission delay in the fully intelligent connected vehicle (CAV) scenario;
[0150] According to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission energy consumption in the fully intelligent connected vehicle (CAV) scenario, the transmission energy consumption under this penetration rate is calculated, which is expressed as follows:
[0151]
[0152] in, It represents the transmission energy consumption in the scenario of fully intelligent connected vehicle (CAV);
[0153] 4.3, using a system utility Q(d,x), defined as the weighted sum of energy consumption and task completion delay:
[0154]
[0155] in, β represents the weight of energy consumption and task completion time;
[0156] 4.4, we formulate a resource optimization problem R, for the task transfer of the client vehicle CV, by optimizing its unloading decision {x m}, in order to minimize the Q value, it can be expressed as follows:
[0157] R:
[0158] stx m ∈{0,1}
[0159] Among them, Q(d,x) represents the system utility, x m represents the offloading decision of task m.
[0160] Step 5: By minimizing the objective function Q(d,x) of the resource optimization problem R, the optimal offloading action {x m}, complete the allocation of vehicle-to-vehicle V2V resources in the intelligent connected vehicle (CAVs) scenario.
[0161] 5.1, using multiple parallel deep neural networks DNN to generate binary offloading decisions, and selecting a set of offloading decisions x with the minimum system utility * As output, store (d,x * ) and used to train the DNN later, repeating the above process until the system utility Q(d,x) reaches the minimum value in this scenario;
[0162] 5.2, set the gain ratio to the ratio of the historical minimum system utility to the current system utility;
[0163] 5.3. According to the above settings, when the gain ratio converges, the unloading strategy corresponding to minimizing the system utility is obtained, and the resource allocation between the intelligent connected vehicles (CAVs) in the mixed traffic scenario is completed.
[0164] like Figure 4 As shown, it is the convergence performance of the gain ratio in step 5, which includes curves under four penetration rates. Compared with the existing technology, its application in mixed traffic scenarios can bring better benefits.
[0165] like Figure 5 As shown, it is a comparison chart of the system utilities obtained by processing the task only locally, only by processing the task on the edge server, and by processing the task using the offloading decision obtained by the present invention. It can be seen that when the penetration rate is high, using the offloading decision of the present invention can bring better benefits, that is, smaller system utility.
[0166] The present invention also provides a hybrid traffic scenario synaesthesia resource allocation system, comprising:
[0167] Transmission delay calculation module: Calculate the transmission delay in step 2, and use the penetration rate p to calculate the delay of the intelligent connected vehicle CAV, i.e., the service vehicle SV, as the edge server in the mixed traffic scenario;
[0168] Module for calculating transmission energy consumption: Calculate the transmission energy consumption in step 2, and use the penetration rate p to calculate the energy consumption of the intelligent connected vehicle CAV, i.e., the service vehicle SV, as the edge server in the mixed traffic scenario;
[0169] Calculate local energy consumption module: calculate the local energy consumption in step 3, and calculate the energy consumption required for local processing tasks according to the task size;
[0170] Module for calculating local delay: Calculates the local delay in step 3 and the delay required for local processing of tasks based on the size of the task;
[0171] Unloading decision generation module: implements the resource optimization problem related to energy consumption and delay in step 4 and the unloading decision generation in step 5, and obtains the binary unloading decision by minimizing the system utility to complete resource allocation.
[0172] The present invention also provides a hybrid traffic scenario synaesthesia computing resource allocation device, comprising:
[0173] Memory: a computer program storing the above-mentioned method for allocating synaesthesia resources in a mixed traffic scenario, which is a computer-readable device;
[0174] Processor: used to implement the method for allocating synaesthesia resources in a mixed traffic scenario when executing the computer program.
[0175] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can implement the method for allocating synaesthesia resources in a mixed traffic scenario.
Claims
1. A method for allocating synaesthesia resources in mixed traffic scenarios, characterized by: The following steps are involved: Step 1, establish a mixed traffic scenario vehicle-to-vehicle (V2V) communication model; The hybrid traffic scenario includes CAVs and HDVs. CAVs communicate with each other via a vehicle-to-vehicle (V2V) link. CAVs are divided into client vehicles (CVs) and service vehicles (SVs). When executing an offloading decision, client vehicles (CVs) offload tasks to service vehicles (SVs) or process them locally. Use the Internet of Vehicles simulation software to randomly generate N vehicles, and randomly select N of them according to the penetration rate p. CAV The number of CAVs involved in the calculation is 1, and the rest are traditional HDVs. CAV It is derived from the following formula: N CAV =pN Among them, N CAV represents the number of intelligent connected vehicles (CAVs), p represents the penetration rate, and N represents all vehicles; Step 2, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the energy consumption and delay of the intelligent connected vehicle (CAV) serving as the edge server in the mixed traffic scenario, i.e., the service vehicle (SV); Step 3, based on the mixed traffic scenario vehicle-to-vehicle (V2V) communication model established in step 1, calculate the local energy consumption and delay; Step 4: According to the energy consumption and delay formulas of vehicle-to-vehicle V2V communication in mixed traffic scenarios in steps 2 and 3, a resource optimization problem R related to delay and energy consumption is established; The specific method of step 4 is: 4.1 Assuming that the vehicle-to-vehicle V2V communication speed and energy loss per kilometer remain unchanged, the relationship between the transmission delay and transmission energy consumption of vehicle-to-vehicle V2V communication in mixed traffic scenarios and the penetration rate p, that is, the CAV occupancy rate is: The permeability p is defined as follows: Where p represents permeability, N CAV Represents the number of intelligent connected vehicles, N HDV represents the number of conventional vehicles; Define a one-way road of length S0. On this road, the communication vehicles include CAVs and HDVs. Assume that CAVs are evenly distributed on this road. The traffic density on this road is constant. The distance between any two adjacent vehicles is fixed to D0. The number of vehicles that this road can accommodate is fixed to N0. Assume that the distance between two adjacent CAVs is D CAV , the following relationship exists: (N0-1)D0=S0 A CAV +N HDV =N0 (N CAV +N HDV -1)D0=S0 When S0>>D CAV , the following formula holds: (N CAV -1)D CAV =S0 Right now: (N CAV -1)D CAV =(N CAV +N HDV -1)D0 When N CAV >>1, N0>>1, the following equation is true: When the communication speed remains unchanged, the communication delay is proportional to the communication distance. Proportional to the distance, communication delay is inversely proportional to the penetration rate. Assuming that the energy consumption is lost by the same amount per kilometer during the transmission process, the transmission energy consumption is also proportional to the distance and inversely proportional to the penetration rate. 4.2, according to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission delay in the fully intelligent connected vehicle (CAV) scenario, the transmission delay under the penetration rate is calculated, which is expressed as follows: in, Transmission delay in the fully intelligent connected vehicle (CAV) scenario; According to the penetration rate p of vehicle-to-vehicle (V2V) communication in mixed traffic scenarios and the transmission energy consumption in the fully intelligent connected vehicle (CAV) scenario, the transmission energy consumption under this penetration rate is calculated, which is expressed as follows: in, It represents the transmission energy consumption in the scenario of fully intelligent connected vehicle (CAV); 4.3, using the objective function Q(d,x), defined as the weighted sum of energy consumption and task completion delay: Where d = {d m |m∈M},x={x m |m∈M}, β represents the weight of energy consumption and task completion time; M represents the number of tasks that the customer vehicle CV needs to handle, and d m represents the workload of the mth task; 4.4, formulate the resource optimization problem R, for the task transfer of the client vehicle CV, by optimizing its unloading decision {x m }, in order to minimize the Q value, which is expressed as follows: s.t.x m ∈{0,1} Among them, Q(d,x) represents the system utility, x m represents the offloading decision of task m; Step 5: By minimizing the objective function Q(d,x) of the resource optimization problem R, the optimal offloading action {x m }, complete the allocation of vehicle-to-vehicle V2V resources in the intelligent connected vehicle CAVs scenario; The specific method of step 5 is: 5.1, using multiple parallel deep neural networks DNN to generate binary offloading decisions, and selecting a set of offloading decisions x with the minimum system utility * As output, store (d,x * ) and used to train the DNN later, repeating the above process until the system utility Q(d,x) reaches the minimum value in this scenario; 5.2, set the gain ratio to the ratio of the historical minimum system utility to the current system utility; 5.
3. According to the above settings, when the gain ratio converges, the unloading strategy corresponding to minimizing the system utility is obtained, and the resource allocation between the intelligent connected vehicles (CAVs) in the mixed traffic scenario is completed.
2. The method for allocating synaesthesia resources in a mixed traffic scenario according to claim 1 is characterized by: The specific method of step 2 is: 2.1, calculate the total energy consumption of the client vehicle CV to offload its task m to the service vehicle SV for processing: Among them, d m represents the workload of the mth task, represents the energy consumed by the client vehicle CV to upload its workload m to the service vehicle SV, where α is the weight of the server energy consumption. When α = 0, only the energy consumption of the client vehicle CV is considered. This includes the energy consumption of sending the task and the energy consumption of the server utility that executes this task; 2.2, calculate the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, which is given by: Among them, d m represents the workload of the mth task, c represents the bandwidth of communication between the customer vehicle CV and the service vehicle SV; 2.3, calculate the service vehicle SV processing delay, which is given by the following formula: Among them, d m represents the workload of the mth task, f c Indicates the server processing rate; 2.4, according to the uninstall decision {x m }, calculate the total delay of the client vehicle CV offloading the task to the service vehicle SV for processing, which is given by the following formula: in, represents the transmission delay when the client vehicle CV offloads its task m to the service vehicle SV, represents the service vehicle SV processing delay, x m represents the offloading decision of task m.
3. The method for allocating synaesthesia resources in a mixed traffic scenario according to claim 1 is characterized by: The specific method of step 3 is: 3.1, calculate the energy consumption of the client vehicle CV to perform its task m locally, which is given by: Among them, d m represents the workload of the mth task, e l Represents the local energy consumption of each data bit of the client vehicle CV; 3.2, calculate the total processing time of the client vehicle CV to perform its task m, which is given by: Among them, d m represents the workload of the mth task, t l The local processing time for each data bit representing the client vehicle CV; 3.3, according to the unloading decision of the customer vehicle CV {x m }, calculate the total delay for the client vehicle CV to complete its task locally, which is given by: in, represents the total processing time of the client vehicle CV to perform its task m, x m represents the offloading decision of task m.
4. The method for allocating synaesthesia resources in a mixed traffic scenario according to claim 2 is characterized by: In step 1, the specific method for the client vehicle CVs to offload the task to the service vehicle SVs or process it locally is: Assume that a customer vehicle CV has M independent tasks to be processed, expressed as: Use d m Representing the workload of the mth task, the customer vehicle CV decides whether to offload its task m to the service vehicle SV. This process is represented by a set of binary offloading decisions. The offloading decision is represented by the variable x m To express, x m represents the offloading decision of the mth task, where x m ∈{0,1},x m =1 means that the customer vehicle CV decides to offload task m to the service vehicle SV, x m =0 means that the client vehicle CV decides to execute task m locally.
5. A hybrid traffic scenario synaesthesia resource allocation system based on the method according to any one of claims 1 to 4, characterized in that: include: Transmission delay calculation module: Using the penetration rate p, calculate the delay of the intelligent connected vehicle (CAV) as the edge server in the mixed traffic scenario, that is, the service vehicle (SV); Module for calculating transmission energy consumption: Using the penetration rate p, calculate the energy consumption of the intelligent connected vehicle (CAV) as the edge server in the mixed traffic scenario, i.e., the service vehicle (SV); Module for calculating local energy consumption: Calculates the energy consumption required for local processing tasks based on the size of the task; Module for calculating local delay: Calculates the delay required for local processing tasks based on the size of the task; Unloading decision generation module: establishes resource optimization problems related to energy consumption and latency and generates unloading decisions, and obtains binary unloading decisions by minimizing system utility to complete resource allocation.
6. A hybrid traffic scene synaesthesia resource allocation device, characterized by: include: Memory: a computer program storing the method for allocating synaesthesia resources in a mixed traffic scenario as described in any one of claims 1 to 4, which is a computer-readable device; Processor: used to implement the hybrid traffic scenario synaesthesia resource allocation method described in any one of claims 1-4 when executing the computer program.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it can implement the method for allocating synaesthesia resources in a mixed traffic scenario as described in any one of claims 1 to 4.
