A vehicle networking on-demand scheduling method based on information value in typical scenarios of autonomous driving

Abstract

The present invention discloses a vehicle networking on-demand scheduling method in typical scenarios of autonomous driving based on information value, which specifically includes the following steps: Step 1: Establish an information value model based on the AV control system, aiming to evaluate the urgency of AV transmission packets; Step 2: Establish a communication model of the vehicle networking to obtain the channel capacity and the receiving probability of packets; Step 3: Based on the information value and the communication model, establish a problem of maximizing the system information value and allocate network resources on demand; Step 4: Obtain a sub-optimal time slot scheduling at the packet level based on the greedy algorithm and allocate network resources on demand. The technical effect of this invention is to realize the consideration of the value of the information of the packets transmitted in specific scenarios, allocate network resources based on the value, reduce redundant resources, and at the same time ensure that the system will not linearly decline even in the case of insufficient resources.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a method for on-demand scheduling of a vehicle-to-everything (V2X) network in typical scenarios of autonomous driving based on information value. Background Art

[0002] With the bottleneck encountered in the intelligent driving of individual vehicles, the development of wireless sensor networks and information and communication technologies has promoted the application of intelligent connected vehicles. Vehicles are no longer isolated individuals but part of the entire vehicle-to-everything (V2X) network. The V2X network is a complex system that deeply integrates sensing, communication, computing, and control. Communication is the prerequisite and foundation of the V2X network. However, under the constraint of cost, communication resources are often limited. In the face of high-load scenarios such as intersections with many service users, complex traffic conditions, and large information throughput, the quality of service often cannot be satisfied, and it may even affect the driving safety of networked autonomous vehicles. Therefore, more network resources can be allocated to the V2X network, but a scheduling method is still needed to ensure that the performance of the traffic system will not drop precipitously when communication resources are insufficient.

[0003] In the traditional V2X network, the network has no intelligence and relies on a large amount of redundant resources to ensure that the network always reaches the required highest performance, which is a waste of resources. When resources are insufficient, the system performance will decline. All data packets are transmitted equally without considering the specific situation of the data packets, such as whether the vehicle is moving at a high speed or a low speed, whether it is in a straight cruise or an emergency turn, and whether the data packet has timeliness. This indiscriminate transmission causes the problem of waste of network resources.

[0004] However, the present application realizes considering the value of the information of the packets transmitted in a specific scenario, allocating network resources based on the value, reducing redundant resources, and at the same time ensuring that the system will not linearly decline when resources are insufficient. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for on-demand scheduling of a vehicle-to-everything (V2X) network in typical scenarios of autonomous driving based on information value to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the invention provides the following technical solution: A method for on-demand scheduling of a vehicle-to-everything (V2X) network in typical scenarios of autonomous driving based on information value, specifically including the following steps:

[0007] Step 1: Establish an information value model based on the AV control system:

[0008] Let the single-vehicle control system of the AV be as follows:

[0009] The state variables of the vehicle are: , that is, x(t) changes with time t;

[0010] Controlled variable , i.e., u(t), changes with time t;

[0011] where ( ) is the position coordinate of the vehicle, is the lateral velocity, is the longitudinal velocity, is the lateral acceleration, is the longitudinal acceleration;

[0012] If a controller is established by means of model predictive control (MPC), the state equation of the single-vehicle control problem is as follows:

[0013] ;

[0014] where A and B are coefficient matrices. After discretization, the discrete state equation is:

[0015] ;

[0016] where , T is the control period. According to the principle of MPC control, the single-vehicle control problem is modeled as follows:

[0017] ;

[0018] ;

[0019] ;

[0020] where is the path obtained by model prediction in MPC, Q and P are weight matrices, N is the optimization window, is a quadratic programming problem and is solved by a quadratic programming solver;

[0021] Based on the MPC single-vehicle control model, the information value of the packets generated by wireless control in connected and autonomous driving is defined as follows:

[0022] ;

[0023] ;

[0024] ;

[0025] where is the sampling time of the vehicle, is the vehicle ID, is the information value function, which is determined by the state deviation of the vehicle and the information age is composed of, and the information value function will be selected according to the specific application scenario. For the intersection scenario, it is defined as a linear combination of the two:

[0026] ;

[0027] The purpose of the definitions of the information value function formulas (4) and (7) is to make a quantitative definition of the states of all vehicles in the network according to the urgency;

[0028] Step 2: Establish a communication model for the vehicle network;

[0029] Let the communication bandwidth of the connected autonomous driving be F, which is divided into three sub-channels, namely , , corresponding to the state information channel (State information), the uplink channel, and the downlink channel respectively; is used for the scheduling unit to obtain the information value of the participating traffic vehicles, the channel is used to upload the physical information of the vehicle's position or state, the channel is used for the controller to send control commands to the vehicle; for each channel, it is divided into n links, and each link adopts a slot structure. According to the sampling period of the control system, the slot is divided into a virtual frame structure. At the beginning of each virtual frame, all packets to be transmitted need to arrive, and the packets that fail to be transmitted are discarded at the end of the frame; the delivery probability of each packet is represented by p, and the channel capacity C of the virtual frame is given by the Shannon formula (8):

[0030] ;

[0031] Let the channel noise power of link n in the uplink channel at the beginning of k frames be N, and the transmission power be S,

[0032] ;

[0033] T is the standard slot length, then the channel capacity is represented by the normalized number of slots ;

[0034] Step 3: Based on the information value and the communication model:

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] Among them, for the kth frame, there are m packets to be transmitted at the beginning, and the information value of packet i is , the reception probability is , the channel capacity is , from the perspective of the receiving end, the expected value of the packet is , is a scheduling variable;

[0040] As a time-slot scheduling binary variable, its value is limited to two states: 0 or 1, and the specific definition is:

[0041] When , it means that the th transmission time slot is allocated to data packet ;

[0042] When , it means that the th transmission time slot is not allocated to data packet ;

[0043] Step 4: Obtain the sub-optimal time-slot scheduling at the packet level based on the greedy algorithm, solve the above formula (10), and allocate network resources as needed;

[0044] Based on the greedy algorithm of information value, perform sub-optimal scheduling at the packet level. The steps are as follows:

[0045] ① First, obtain the information value of the packets arriving at the beginning of the virtual frame;

[0046] ② Sort the packets according to the information value;

[0047] ③ Evaluate the channel state and obtain the channel capacity of the current link;

[0048] ④ According to the channel capacity of the current link, discard the packets with low value.

[0049] Preferably, connected autonomous driving is based on a connected autonomous driving vehicle system, including autonomous driving vehicles, communication base stations, and a control unit (Controller); AVs and base stations (Base station, BS) communicate through a wireless link. AVs upload their own status information x(t) to the control unit through the base station via the uplink, and the control unit issues control commands u(t) and a window path to the AVs through the base station via the downlink. Among them, the control unit consists of a solver (MPC solver) and a path prediction model (Path predict model).

[0050] Preferably, the quadratic programming solver used in Step 1 focuses on solving a special type of non-linear programming problem, where the objective function is in the form of a quadratic function and the constraints are all linear constraints.

[0051] Preferably, the MPC in Step 1 is a commonly used control model in autonomous driving, which has the ability to explicitly handle constraints. This ability comes from its prediction of the future dynamic behavior of the system based on the model. By adding constraints to future input, output, or state variables, the constraints can be explicitly represented in a quadratic programming or non-linear programming problem solved online.

[0052] Compared with the prior art, the beneficial effects of the invention are as follows:

[0053] 1. The vehicle-to-everything (V2X) on-demand scheduling method in the typical scenario of autonomous driving based on information value well addresses the traffic scenario with a large number of connected autonomous vehicles by Step 2: establishing a time slot model in the communication model of the vehicle-to-everything network and Step 3: the scheduling mechanism based on information value and the communication model.

[0054] 2. The vehicle-to-everything (V2X) on-demand scheduling method in the typical scenario of autonomous driving based on information value reduces the impact on connected autonomous vehicles caused by communication resource shortage through the combined action of Step 1: establishing an information value model based on the AV control system, Step 2: establishing the communication model of the vehicle-to-everything network, and Step 3: based on information value and the communication model.

[0055] 3. The vehicle-to-everything (V2X) on-demand scheduling method in the typical scenario of autonomous driving based on information value saves communication and computing resources through the scheduling mechanism in Step 3: based on information value and the communication model.

[0056] 4. The vehicle-to-everything (V2X) on-demand scheduling method in the typical scenario of autonomous driving based on information value improves traffic efficiency and reduces carbon emissions through the greedy algorithm in Step 4. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the connected autonomous vehicle system of the invention;

[0058] Figure 2 It is a schematic diagram of the communication model of the vehicle-to-everything network of the invention;

[0059] Figure 3 It is a schematic diagram of the greedy algorithm of the invention. DETAILED DESCRIPTION OF THE INVENTION

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0061] Please refer to Figures 1-3 , the invention provides a technical solution: a vehicle networking on-demand scheduling method in a typical scenario of autonomous driving based on information value, specifically including the following steps:

[0062] Step 1: Establish an information value model based on the AV control system, aiming to evaluate the urgency of the AV transmission packet:

[0063] Suppose the single-vehicle control system of AV is as follows:

[0064] The state variables of the vehicle are: , that is, x(t) changes with time t;

[0065] Control variable , that is, u(t), changes with time t;

[0066] Among them, ( ) is the position coordinate of the vehicle, is the lateral speed, is the longitudinal speed, is the lateral acceleration, is the longitudinal acceleration;

[0067] Adopt the method of model predictive control (MPC) to establish a controller, then the state equation of the single-vehicle control problem is as follows:

[0068] ;

[0069] Among them, A and B are coefficient matrices. After discretization, the discrete state equation is:

[0070] ;

[0071] Among them , T is the control period. According to the principle of MPC control, the single-vehicle control problem is modeled as follows:

[0072] ;

[0073] ;

[0074] ;

[0075] Among them, is the path obtained by model prediction in MPC, Q and P are weight matrices, and N is the optimization window. is a quadratic programming problem and is solved by a quadratic programming solver.

[0076] Based on the MPC single-vehicle control model, the information value of the packets generated by wireless control in connected autonomous driving is defined as follows:

[0077] ;

[0078] ;

[0079] ;

[0080] Among them, is the sampling time of the vehicle, represents the vehicle ID, is the information value function, which is composed of the state deviation of the vehicle and the age of information . The selection of the information value function will be selected according to the specific application scenario. For the intersection scenario, it is defined as a linear combination of the two:

[0081] ;

[0082] The purpose of the definitions of the information value function in formulas (4) and (7) is to make a quantitative definition of the states of all vehicles in the network according to the urgency, so as to analyze the urgency of vehicle requirements based on the information value. The purpose is to allocate network resources on demand, reduce the transmission of redundant information, and relieve the dilemma of insufficient communication resources.

[0083] Step 2: Establish a communication model for the vehicle network to obtain the channel capacity and the packet reception probability.

[0084] Suppose the communication bandwidth of connected autonomous driving is F. As Figure 2 shown, it is divided into three sub-channels, namely , , which respectively correspond to the state information channel (State information), the uplink channel, and the downlink channel. is used for the scheduling unit to obtain the information value of the participating traffic vehicles. The channel is used to upload the physical information of the vehicle's position or state. The channel is used for the controller to send control commands to the vehicle; for each channel, it is divided into n links, and a slot structure is adopted on each link. According to the sampling period of the control system, the slot is divided into a virtual frame structure. At the beginning of each virtual frame, all packets to be transmitted need to arrive, and the packets that fail to be transmitted are discarded at the end of the frame; the delivery probability of each packet is represented by p, and the channel capacity C of the virtual frame is given by the Shannon formula (8):

[0085] ;

[0086] Let the channel noise power of link n in the uplink channel at the beginning of k frames be N, and the transmit power be S,

[0087] ;

[0088] If T is the standard slot length, then the channel capacity is represented by the normalized number of slots ;

[0089] Step 3: Based on the information value and the communication model, establish the problem of maximizing the system information value and allocate network resources as needed:

[0090] ;

[0091] ;

[0092] ;

[0093] ;

[0094] Among them, there are m packets to be transmitted at the beginning of k frames. The information value of packet i is , the reception probability is , the channel capacity is . From the perspective of the receiving end, the expected value of the packet is , is the scheduling variable;

[0095] As a slot scheduling binary variable, its value is limited to two states, 0 or 1. The specific definition is:

[0096] When , it means that the th transmission slot is allocated to data packet ;

[0097] When , it means that the th transmission slot is not allocated to data packet ;

[0098] Step 4: Obtain the sub-optimal slot scheduling at the packet level based on the greedy algorithm, solve the above formula (10), and allocate network resources as needed;

[0099] Based on the greedy algorithm of information value, as Figure 3 shown, perform sub-optimal scheduling at the packet level, and the steps are as follows:

[0100] ① First, obtain the packets arriving at the start time of the virtual frame for information value;

[0101] ② Sort the packets according to the information value;

[0102] ③ Evaluate the channel state and obtain the channel capacity of the current link;

[0103] ④ According to the channel capacity of the current link, discard the packets with low value.

[0104] In this embodiment, the connected autonomous driving is based on the connected autonomous driving vehicle system, as Figure 1 shown, including autonomous driving vehicles, communication base stations, and a control unit (Controller); AVs and base stations (Base station, BS) communicate through wireless links. AVs upload their own state information x(t) to the control unit through the base station via the uplink, and the control unit issues control commands u(t) and window paths (windowpath) to the AVs through the base station via the downlink. The control unit consists of a solver (MPC solver) and a path prediction model (Path predict model).

[0105] In this embodiment, the quadratic programming solver used in Step 1 has its core in solving a class of special non-linear programming problems, where the objective function is in the form of a quadratic function, and the constraint conditions are all linear constraints. For quadratic programming problems, there are many mature open-source solvers available to solve them within polynomial time. For example, Matlab integrates solvers for solving various quadratic programming problems.

[0106] In this embodiment, MPC in Step 1 is a commonly used control model in autonomous driving. MPC originated in the 1970s. Its greatest attraction lies in its ability to explicitly handle constraints, which comes from its prediction of the future dynamic behavior of the system based on the model. By adding constraints to future input, output, or state variables, the constraints can be explicitly represented in an online-solved quadratic programming or non-linear programming problem. The most significant feature of MPC is having a model that can predict future states, and this model is closely related to the definition of information value later.

[0107] Technical effects: The on-demand scheduling method for vehicle-to-everything (V2X) in typical scenarios of autonomous driving based on information value well addresses the traffic scenarios with a large number of connected autonomous vehicles through Step 2: establishing a time slot model in the communication model of V2X and Step 3: the scheduling mechanism based on information value and the communication model. The on-demand scheduling method for V2X in typical scenarios of autonomous driving based on information value reduces the impact on connected autonomous vehicles caused by communication resource shortage through the combined action of Step 1: establishing an information value model based on the AV control system, Step 2: establishing the communication model of V2X, and Step 3: based on information value and the communication model. The on-demand scheduling method for V2X in typical scenarios of autonomous driving based on information value saves communication and computing resources through the scheduling mechanism in Step 3: based on information value and the communication model. The on-demand scheduling method for V2X in typical scenarios of autonomous driving based on information value improves traffic efficiency and reduces carbon emissions through the greedy algorithm in Step 4.

[0108] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A vehicle networking on-demand scheduling method in typical scenarios of autonomous driving based on information value, Characterized in that: Specifically includes the following steps: Step 1: Establish an information value model based on the AV control system: Suppose the single-vehicle control system of AV is as follows: The state variables of the vehicle are as follows: , that is, x(t) changes with time t; Control variable , i.e., u(t), varies with time t; wherein ( ) is the position coordinate of the vehicle, is the lateral speed, is the longitudinal speed, is the lateral acceleration, is the longitudinal acceleration; Adopt the method of Model Predictive Control (MPC) to establish a controller, then the state equation of the single-vehicle control problem is as follows: ； Among them, A and B are coefficient matrices. After discretization, the discrete state equation is: ； Among them , T is the control period. According to the principle of MPC control, the single-vehicle control problem is modeled as follows: ； ； ； Among them, is the path obtained by model prediction in MPC, Q and P are weight matrices, and N is the optimization window, is a quadratic programming problem and is solved by a quadratic programming solver; Based on the MPC single-vehicle control model, the information value of the packets generated by wireless control in connected autonomous driving is defined as follows: ； ； ； Among them, is the sampling moment of the vehicle, represents the vehicle ID, is the information value function, which is composed of the state deviation and the age of information compounded. The selection of the information value function will be selected according to the specific application scenario. For the intersection scenario, it is defined as a linear combination of the two: ； The purpose of the definition of the information value function formulas (4) and (7) is to make a quantitative definition of the states of all vehicles in the network according to the urgency; Step 2: Establish a communication model for the vehicle networking; Let the communication bandwidth of the networked autonomous driving be F, which is divided into three sub-channels, namely and , corresponding to the state information channel, the uplink channel, and the downlink channel respectively; is used for the scheduling unit to obtain the information value of the participating traffic vehicles, the channel is used to upload the physical information of the vehicle's position or state, the channel is used for the controller to send control commands to the vehicle; for each channel, it is divided into n links, and each link adopts a slot structure. According to the sampling period of the control system, the slot is divided into a virtual frame structure. At the beginning of each virtual frame, all packets to be transmitted need to arrive, and the packets that fail to be transmitted are discarded at the end of the frame; the delivery probability of each packet is represented by p, and the channel capacity C of the virtual frame is given by the Shannon formula (8): ； Suppose the channel noise power of link n in the uplink channel at the beginning of the kth frame is N, and the transmission power is S, ； If T is the standard time slot length, then the channel capacity is represented by the normalized number of time slots ; Step 3: Based on the information value and the communication model: ； ； ； ； Among them, for the k-th frame, there are m packets to be transmitted at the beginning, the information value of packet i is , the reception probability is , the channel capacity is , from the perspective of the receiving end, the expected value of the packet is , is the scheduling variable; As a time slot scheduling binary variable, its value is limited to two states, 0 or 1, and is specifically defined as follows: When it indicates that the th transmission time slot is allocated to data packet ; When it means that the th transmission time slot is not allocated to the data packet ; Step 4: Based on the greedy algorithm, obtain the sub-optimal slot scheduling at the packet level, solve the above formula (10), and allocate network resources on demand; Based on the greedy algorithm of information value, perform sub-optimal scheduling at the packet level. The steps are as follows: ① First, obtain the packets arriving at the beginning of the virtual frame for information value; ② Sort the packets according to the information value; ③ Evaluate the channel state and obtain the channel capacity of the current link; ④ According to the channel capacity of the current link, discard the packets with low value.

2. The vehicle networking on-demand scheduling method in typical scenarios of autonomous driving based on information value as described in claim 1, Characterized in that: The connected autonomous driving is based on a connected autonomous driving vehicle system, including autonomous driving vehicles, communication base stations, and a control unit (Controller); AV and the base station (Base station, BS) communicate through a wireless link. AV uploads its own state information x(t) to the control unit through the base station via the uplink. The control unit issues control commands u(t) and a window path to AV through the base station via the downlink, where the control unit consists of a solver (MPC solver) and a path prediction model (Path predict model).

3. The vehicle networking on-demand scheduling method in typical scenarios of autonomous driving based on information value as described in claim 1, Characterized in that: The quadratic programming solver adopted in Step 1 has its core in solving a special type of non-linear programming problem, where the objective function is in the form of a quadratic function, and the constraint conditions are all linear constraints.

4. The vehicle networking on-demand scheduling method in typical scenarios of autonomous driving based on information value as described in claim 1, Characterized in that: The MPC in Step 1 is a control model in autonomous driving and has the ability to explicitly handle constraints. This ability comes from its prediction of the future dynamic behavior of the system based on the model. By adding constraints to future input, output, or state variables, the constraints are explicitly represented in a quadratic programming or nonlinear programming problem solved online.

Images