A VANET Geographic Routing Method Based on an Improved EKF Position Prediction Algorithm

Abstract

A VANET geographical routing method based on an improved EKF position prediction algorithm first uses the improved EKF positioning algorithm to predict the positions of neighbor nodes in the next time slot, and executes the neighbor table update algorithm to ensure the reliability of the positions of neighbor nodes when performing greedy forwarding; it judges whether it is locally optimal. If the current node is not trapped in local optimality, it uses greedy forwarding to select the best relay node; if the node is trapped in local optimality, it uses perimeter forwarding to select relay nodes, improving the routing reliability. When the destination node is in the neighbor table of the relay node, it forwards the data; otherwise, it repeats the above operations. Compared with the existing VANET geographical routing methods, this method has better performance in terms of packet delivery ratio, end-to-end delay, and throughput.

Description

Technical Field

[0001] The present invention relates to mobile communication technology, and more particularly to a VANET geographic routing method based on an improved EKF position prediction algorithm. Background Art

[0002] With the sharp increase in the number of vehicles on the road, traffic accidents and congestion often occur. How to reduce casualties and traffic congestion has become the focus of researchers. As an important part of the intelligent transportation system [1,2] Vehicle Ad Hoc Network (VANET) is mainly responsible for transmitting important traffic information such as accident information and road congestion to reduce casualties and maintain traffic order. According to the objects participating in message transmission, the communication methods of VANET can be divided into: vehicle-to-vehicle communication (V2V), vehicle-to-infrastructure communication (V2I), and vehicle-to-everything communication (V2X). [2,3] However, the high-speed movement of vehicles leads to frequent changes in the network topology, easy interruption of links, and increased network latency. Therefore, it is of great significance to study the high-reliability and low-latency transmission of data packet information in a complex urban environment.

[0003] Summarizing the previous research, the routing protocols of VANET can be mainly divided into two types: topology-based routing protocols and location-based routing protocols. [4] Topology-based routing protocols need to establish a routing table for packet transmission and then forward packets according to the routing table. However, each node in this type of protocol needs to maintain a routing table, resulting in a large network overhead. Location-based routing protocols do not need to establish and maintain a route before packet transmission. When forwarding packets, the optimal relay node is selected according to the node location information to forward the data packet, which can adapt to the highly dynamic topology of VANET.

[0004] Currently, most of the research on location-based routing protocols is based on the greedy forwarding idea of GPSR. [4,5] The greedy forwarding principle is to select the neighbor node closest to the destination node as the forwarding node. When the neighbor node is at the edge of the communication range of the current node, it may leave the communication range of the current node in the next time slot, resulting in packet transmission failure. When the greedy forwarding falls into a local optimum, it will switch to the perimeter forwarding mode. However, perimeter forwarding has routing redundancy, resulting in an increase in end-to-end latency.

[0005] For this reason, Reference [6] proposed an MM-GPSR routing protocol. In the greedy forwarding process, this protocol selects the node with the maximum link duration as the relay node, and in the perimeter forwarding process, it selects the node with the minimum angle as the relay node. However, in the greedy forwarding process, this protocol does not consider the location information of neighbor nodes in the next time slot, resulting in the interruption of the communication link. Reference [7] proposed a GPSR-WG routing protocol. In the greedy forwarding process, this protocol establishes a weight gradient criterion based on the node movement direction, node distance, link risk degree, and node speed to select the best relay node, but it does not optimize the perimeter forwarding criterion. Reference [8] proposed an E-GPSR routing protocol. This protocol uses a dual-path to forward data packets, improving the reliability of forwarding data packets, but it does not consider the link stability problem, and there is still a possibility of data packet loss. Reference [9] proposed a GPSR routing protocol forwarding method based on link lifetime position prediction. This method updates the node location information through the link lifetime to improve the accuracy of node location; by adding a hole avoidance list to the neighbor list and considering two factors, the distance between the node and the destination node and the link lifetime, it reduces the occurrence probability of the routing hole phenomenon. However, this method does not solve the routing redundancy problem existing in perimeter forwarding. Reference

[10] proposed a PA-GPSR routing protocol. This protocol adds an additional extended table to the neighbor table to select the best path, bypasses the nodes that have already sent such packets in the perimeter forwarding mode, reducing the routing redundancy of the packets, but it does not consider the link stability. Reference

[11] proposed an LRGR routing protocol. This protocol uses information such as road and intersection locations to establish a road network model, makes a section decision at intersections to select a suitable section, and gives a carry-and-forward mechanism to adapt to the low node density scenario, but it does not consider the real-time nature of node location information. Reference

[12] proposed a BOD-KF-GPSR routing protocol. This protocol proposed the concept of on-demand beacons, reducing the control overhead, and using Kalman filtering to improve the accuracy of node location information, but it does not consider the node movement direction and may fall into a routing hole.

[0006] [1]Kayarga T,Kumar S A.A Study on Various Technologies to Solve theRouting Problem in Internet of Vehicles(IoV)[J].Wireless PersonalCommunications,2021,119(1):459-487;

[0007] [2]Lee M, Atkison T. VANET applications: Past, present, and future[J]. Vehicular Communications, 2021, 28: 100310 - 100323;

[0008] [3]Tripp - Barba C, Zaldívar - Colado A, Urquiza - Aguiar L, et al. Survey on Routing Protocols for Vehicular Ad Hoc Networks Based on Multimetrics[J]. Electronics, 2019, 8(10): 1177 - 1208;

[0009] [4]Ayyub M, Oracevic A, Hussain R, et al. A comprehensive survey on clustering in vehicular networks: Current solutions and future challenges[J]. Ad Hoc Networks, 2022, 124: 102729 - 102763;

[0010] [5]Jianming Cui, Liang Ma, Ruirui Wang, Ming Liu. Research and Optimization of GPSR Routing Protocol for Vehicular Ad - Hoc Network[J]. China Communications, 2022, 19(10): 194 - 206.

[0011] [6]Yang X, Li M, Qian Z, et al. Improvement of GPSR Protocol in Vehicular Ad Hoc Network[J]. IEEE Access, 2018, 6: 39515 - 39524;

[0012] [7] Singh P, Raw R S, Khan S A. Link risk degree aided routing protocol based on weight gradient for health monitoring applications in vehicular ad-hoc networks[J]. Journal of Ambient Intelligence and Humanized Computing, 2022, 13(12):5779-5801;

[0013] [8] A B, Bengag A, Boukhari M E. Enhancing GPSR routing protocol based on Velocity and Density for real-time urban scenario[C] / / 2020 International Conference on Intelligent Systems and Computer Vision(ISCV). Fez, Morocco: IEEE, 2020:1-5;

[0014] [9] South China University of Technology. GPSR routing protocol forwarding method based on link survival time position prediction: CN202210300999.6[P]. 2022-07-29;

[0015]

[10] Silva A, Reza N, Oliveira A. Improvement and Performance Evaluation of GPSR-Based Routing Techniques for Vehicular Ad Hoc Networks[J]. IEEE Access, 2019, 7:21722-21733.

[0016]

[11] Gao Tianxiang, Shi Ying, Liu Ziwei, et al. Improved GPSR protocol based on road network and QoS model[J]. Computer Engineering, 2019, 45(2):7-12;

[0017]

[12] Houssaini Z S,Zaimi I,Drissi M,et al.Trade-off between accuracy,cost,and QoS using abeacon-on-demand strategy and Kalman filtering over a VANET[J].Digital Communications and Networks,2018,4(1):13-26。 Summary of the Invention

[0018] To solve the above technical problems,the present invention provides a VANET geographical routing method based on an improved EKF position prediction algorithm.

[0019] To achieve the above technical objectives,the technical solution adopted is:a VANET geographical routing method based on an improved EKF position prediction algorithm,comprising the following steps:

[0020] Step 1:The current node runs an improved EKF positioning algorithm,updates the position information of neighbor nodes,and updates the neighbor table.The improved EKF positioning algorithm first uses the position and speed information of neighbor nodes in the neighbor table as the input of the input layer of the DNN,then runs the EKF algorithm on the hidden layer,and finally obtains the future position information of neighbor nodes at the output layer;

[0021] Step 2:Determine whether it is locally optimal.If the current node is not trapped in local optimality,perform greedy forwarding to select a relay node as the new current node.If the node is trapped in local optimality,perform perimeter forwarding to select a relay node as the new current node.When the destination node is within the neighbor table of the relay node,forward the data;otherwise,execute Step 1.

[0022] The specific implementation method of the greedy forwarding described in the present invention is to construct an evaluation function through the node relative angle,the node relative distance,and the link stability factor,and select the node with the largest evaluation function value as the relay node

[0023]

[0024] In the formula:cosθ represents the cosine value of the node relative angle;d ND represents the node relative distance;T S&N represents the link stability factor;ω1,ω2,and ω3 are the weighting values of the node relative angle,the node relative distance,and the link stability factor respectively,and ω1+ω2+ω3 = 1.

[0025] The calculation method of the link stability factor described in the present invention is

[0026]

[0027] V S&N = |V S - V N |

[0028] Where: R represents the transmission range of the current node; D S&N represents the distance between the current node and its neighbor node; V S&N represents the relative speed between the current node and its neighbor node, V S represents the speed of the current node; V N represents the speed of the neighbor node.

[0029] The specific method of perimeter forwarding in the present invention is to back up data packets during perimeter forwarding, then perform routing forwarding according to the left - hand and right - hand rules simultaneously, and select the neighbor node with the smallest relative angle of the node as the relay node.

[0030] The beneficial effects of the present invention are as follows: When selecting the relay node, this method uses the DNN - DKF position prediction algorithm to update the position information of neighbor nodes, reducing the possibility of link interruption; during greedy forwarding, an evaluation function is constructed based on the relative angle, relative distance, and link stability factor, increasing the reliability of the link; when perimeter forwarding, introducing the relative angle reduces routing redundancy, and at the same time using the left - hand and right - hand rules improves the reliability of data packet transmission. The NS3 simulation results show that compared with the GPSR and MM - GPSR protocols, LP - GPSR has better performance in terms of packet loss rate, end - to - end delay, and throughput. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is the schematic diagram of the relative angle of the present invention;

[0032] Figure 2 is the relationship diagram of neighbor nodes of the present invention;

[0033] Figure 3 is the coordinate system diagram of the node positions of the present invention;

[0034] Figure 4 is the flow chart of the present invention;

[0035] Figure 5 is the schematic diagram of greedy forwarding of the present invention;

[0036] Figure 6 is the schematic diagram of perimeter forwarding of the present invention;

[0037] Figure 7 is the experimental simulation scenario diagram;

[0038] Figure 8 is the comparison diagram of packet loss rates under different numbers of nodes;

[0039] Figure 9 Packet loss rate comparison chart for different node speeds;

[0040] Figure 10 End-to-end delay comparison chart for different numbers of nodes;

[0041] Figure 11 End-to-end delay comparison chart for different node speeds;

[0042] Figure 12 Throughput comparison chart for different numbers of nodes;

[0043] Figure 13 Throughput comparison chart for different node speeds. Specific implementation manners

[0044] Although the geographical location routing based on the existing technology GPSR can adapt to the changing topological structure of VANET, there are the following several problems with the greedy forwarding of GPSR and its recovery strategy - perimeter forwarding:

[0045] (1) When a neighbor node is located at the boundary of the communication range of the current node, the neighbor node may break away from the current node at any time, resulting in the interruption of the transmission link and the failure of packet forwarding;

[0046] (2) When performing greedy forwarding, the movement direction of the node is not considered, resulting in the data packet being transmitted in a direction away from the destination node, generating routing redundancy and even causing a routing hole;

[0047] (3) There is routing redundancy in the perimeter forwarding mode, and even a routing loop is generated, resulting in the loss of data packets.

[0048] To solve these problems, the following improvements are made to the existing technology GPSR in this paper:

[0049] (1) Introduce an improved EKF positioning algorithm. When a node has a data packet to forward, run this algorithm to update the node positions in the neighbor table to ensure the reliability of the data transmission link;

[0050] (2) Introduce the relative angle of the node, the relative distance of the node, and the link stability factor. Based on the relative angle of the node, the relative distance of the node, and the link stability factor, establish an evaluation function, and select the neighbor node with the largest evaluation function value as the relay node;

[0051] (3) Before perimeter forwarding, copy the data packet, forward it simultaneously according to the left - hand and right - hand rules, and consider the relative angle during forwarding. Select the neighbor node with the smallest relative angle of the node as the relay node.

[0052] 1. Specific implementation process

[0053] Such as Figure 4As shown in the figure, this paper proposes a VANET geographical location routing method (LP-GPSR) based on an improved EKF location prediction algorithm. LP-GPSR mainly consists of three parts: neighbor table update, restricted greedy forwarding, and restricted perimeter forwarding.

[0054] A VANET geographical location routing method based on an improved EKF location prediction algorithm includes the following steps:

[0055] Step 1: The current node runs the improved EKF positioning algorithm to update the location information of neighbor nodes and update the neighbor table. The improved EKF positioning algorithm first takes the location and speed information of neighbor nodes in the neighbor table as the input of the input layer of the DNN, then runs the EKF algorithm on the hidden layer, and finally obtains the future location information of neighbor nodes on the output layer.

[0056] Step 1.1: Improved EKF positioning algorithm

[0057] The Extended Kalman Filter (EKF) is a prediction algorithm widely used in various fields such as signal processing, radar, and image processing. The Deep Neural Network (DNN) is a branch of machine learning, and its essence is to obtain the desired result through multiple linear regression. This application combines the Extended Kalman Filter and the Deep Neural Network to improve the efficiency of the positioning algorithm.

[0058] This algorithm first takes the location and speed information of neighbor nodes in the neighbor table as the input of the input layer of the DNN, then runs the EKF algorithm on the hidden layer, and finally can obtain the future location information of neighbor nodes on the output layer. Taking the current node as the origin and the direction of the line connecting the current node and the destination node as the vertical axis, a two-dimensional coordinate system is established as Figure 3 shown in the figure. Then the state matrix of the neighbor node can be defined as:

[0059]

[0060] In formula (1): x i (u), y i (u) are the horizontal and vertical coordinates of the vehicle node respectively; are the velocity components of the horizontal and vertical axes respectively, and they can be calculated as follows.

[0061]

[0062] In formulas (2) and (3): v i represents the speed of neighbor node i; θ1 represents the angle of neighbor node deviating from the vertical axis, and the calculation process is shown in formula (9).

[0063] According to the location of the current node, the location of the neighbor node can be estimated as follows:

[0064]

[0065] In Equation (4): Pos x (S), Pos y (S) are the horizontal and vertical coordinates of the current node respectively; D S&N represents the distance between the current node and its neighbor node, and is calculated as shown in Equation (11); θ1 represents the angle by which the neighbor node deviates from the vertical axis.

[0066] According to EKF, the position of the neighbor node in the next time slot is calculated as follows:

[0067]

[0068] In Equations (5) and (6): x i (u), y i (u), are the horizontal and vertical coordinates of the vehicle node defined in the state matrix in Equation (1) and the velocity components on the horizontal and vertical coordinates respectively; T is the time slot interval; ρ x (n), ρ y (n) are the acceleration variables on the horizontal and vertical axes respectively.

[0069] Finally, the state matrix of the neighbor node in the next time slot can be obtained at the output layer:

[0070]

[0071] In Equation (7): x i (u + 1), y i (u + 1) are the horizontal and vertical coordinates of the vehicle node in the next time slot respectively; are the velocity components on the horizontal and vertical axes in the next time slot respectively.

[0072] Step 1.2, Neighbor Table Update

[0073] In this paper, the neighbor table update is divided into active update and passive update. When a node needs to forward a data packet, it first runs the improved EKF positioning algorithm to update the position information of the neighbor nodes in the neighbor table and deletes the neighbor nodes that are not within the node's transmission range. This is the active update. Every 2 time slots, the nodes exchange Hello data packets (as shown in Table 1), and maintain the neighbor table by checking the timestamps of the nodes in the neighbor table (as shown in Table 2), and delete the neighbor nodes that are not within the node's transmission range. This is the passive update. The neighbor table update process is shown in Table 3.

[0074] Table 1 Hello Data Packet Format

[0075] Vehicle ID V(t) X(t) Y(t) Timestamp

[0076] Table 2 Neighbor Table Format

[0077]

[0078] Step 2: Determine whether it is a local optimum. If the current node has not fallen into a local optimum, perform greedy forwarding to select a relay node as the new current node. If the node has fallen into a local optimum, perform perimeter forwarding to select a relay node as the new current node. When the destination node is in the neighbor table of this relay node, forward the data; otherwise, execute Step 1.

[0079] Determining the local optimum actually follows the ordinary method: that is, comparing the distances between the current node and the destination node and between the neighbor nodes and the destination node. If no neighbor node is closer to the destination node than the current node, it falls into a local optimum.

[0080] Step 2.1: Constrained greedy forwarding

[0081] Due to the problem of unstable communication links in the traditional greedy criterion, this paper proposes a constrained greedy criterion. This criterion constructs an evaluation function based on the relative angle of nodes, the relative distance of nodes, and the link stability factor as shown in Equation (8), and selects the node with the largest evaluation function value as the relay node. An example of constrained greedy forwarding is as Figure 5 shown.

[0082]

[0083] As in Equation (8): cosθ represents the cosine value of the relative angle of nodes; d ND represents the relative distance of nodes; T S&N represents the link stability factor; ω1, ω2, and ω3 are the weighting values of the relative angle of nodes, the relative distance of nodes, and the link stability factor respectively, and ω1 + ω2 + ω3 = 1. In this paper, they are respectively taken as 0.3, 0.3, and 0.4.

[0084] The relative angle of nodes is the angle formed by the current node, the destination node, and the neighbor node, as Figure 1 shown. The neighbor nodes N1, N2, N3, and N4 are located in the first to fourth quadrants respectively. The relative angles of the neighbor nodes in the first and second quadrants are calculated as follows.

[0085]

[0086] The relative angles of the neighbor nodes in the third and fourth quadrants are calculated as follows.

[0087]

[0088] In Equations (9) and (10): (x s , y s ) represents the coordinate of the current node position; (x n , y n) Represents the coordinates of neighbor nodes.

[0089] The relative distance of a node is the distance between a neighbor node and the destination node, which can be calculated according to the Euclidean distance formula as follows.

[0090]

[0091] In Equation (11): (x n , y n ) represents the position coordinates of the neighbor node; (x d , y d ) represents the coordinates of the destination node.

[0092] The link stability factor is an index to evaluate the reliability of the link between two nodes. In this paper, it refers to the link survival time between the neighbor node and the current node. The relationship between the current node and the neighbor node is as Figure 2 shown. The link stability factor (T S&N ) is calculated as follows:

[0093]

[0094] V S&N = |V S - V N | (13)

[0095] In Equation (12): R represents the transmission range of the current node; D S&N represents the distance between the current node and the neighbor node; V S&N represents the relative speed between the current node and the neighbor node. In Equation (13): V S represents the speed of the current node; V N represents the speed of the neighbor node.

[0096] As Figure 5 shown, S is the current node, D is the destination node, N1, N2, N3, and N4 are the neighbor nodes of the current node, and A, C, and N4 are the neighbor nodes of N2. The routing generation process from the current node S to the destination node D is described as follows. ① The node S runs the improved EKF positioning algorithm to update the position information of the neighbor nodes N1, N2, N3, and N4; ② The node S calculates the evaluation function value of each neighbor node according to formula (8); ③ The node S selects the relay node according to the evaluation function value. Since the relative distance between the node N2 and D is small, the relative angle of the node is small, the link stability factor is large, and the evaluation function value is the largest, so N2 is selected as the relay node; ④ Only considering greedy forwarding, the relay node repeats steps ① - ③ to select the next relay node until the data packet reaches the destination node; ⑤ Finally, the route S -> N2 -> A -> D is generated.

[0097] Step 2.2, Restricted Perimeter Forwarding

[0098] When a node falls into a local optimum, GPSR will use perimeter forwarding to bypass the routing hole area. However, there are problems of routing redundancy and routing loops in perimeter forwarding. Therefore, this paper makes the following improvements to the traditional perimeter forwarding rules.

[0099] (1) To solve the packet loss problem caused by routing loops, backup the data packet during perimeter forwarding, and then perform routing forwarding according to the left - hand and right - hand rules simultaneously.

[0100] (2) Aiming at the problem that perimeter forwarding is prone to routing redundancy, when executing the right - hand and left - hand rules, consider the relative angle of nodes, and select the neighbor node with the smallest relative angle of nodes as the relay node. An example of restricted perimeter forwarding is Figure 6 shown as follows.

[0101] As Figure 6 shown, the routing generation process for the current node S to forward a data packet to the destination node D is as follows. ① The node S runs the improved EKF positioning algorithm to update the position information of neighbor nodes N1, N2, N3, and N4; ② Backup the data packet, and select the relay node according to the left - hand rule and the right - hand rule; ③ Generate the route S->N1->N4->N5->S… according to the right - hand rule. Since there is a loop in this route, the data packet will be discarded; ④ Generate the route S->N1->N2->N3->A->B->D according to the left - hand rule, and there is routing redundancy. Since restricted perimeter forwarding considers the relative angle of nodes, the data packet is directly forwarded from the source node S to the node N3; ⑤ The finally generated route is S->N3->A->B->D.

[0102] 2. Experimental and Result Analysis

[0103] 2.1 Experimental Environment

[0104] This experiment runs in the ubuntu 16.04 LTS environment with 2 cores. The simulation platforms are NS - 3.23 and SUMO. First, in SUMO, a 1100*1100m 2 urban road environment is created, which contains 9 intersections and 12 road segments, as Figure 7 shown. Vehicle nodes with numbers 30, 50, 70, 90, 110 and maximum speeds 10, 15, 20, 25, 30 are generated. The initial positions of the vehicle nodes are randomly distributed, and their movements on the road are based on the car - following model restricted by the street (Krauss model). Then, the trajectory file generated by SUMO is imported into NS3.23 as the mobile model for network simulation. This experiment conducts simulation analysis on GPSR, MM - GPSR, and LP - GPSR under different numbers of nodes and maximum node speeds. The simulation parameters are shown in Table 4.

[0105] The evaluation metrics of the experiment include packet loss rate, end-to-end delay, and throughput. ① The packet loss rate is defined as the ratio of the number of lost data packets to the total number of data packets sent by the node; ② The end-to-end delay is defined as the average time required for a data packet to reach the destination node from the source node; ③ The throughput is defined as the number of successfully transmitted data packets in the network per unit time.

[0106] Table 4 Experimental simulation parameters

[0107] Parameter Value Emulator NS-3 / SUMO Simulation duration / s 200 Simulation area / m * m 1100*1100 Node transmission range / m 250 Number of nodes 30,50,70,90,110 <![CDATA[Node speed / (m·s -1 )]]> 10,15,20,25,30 Data type CBR Hello packet interval / s 2 Mac protocol 802.11p <![CDATA[ω1,ω2,ω3]]> 0.3,0.3,0.4 MM-GPSR λ value 0.3 Routing protocol GPSR, MM-GPSR, LP-GPSR

[0108] 2.2. Experimental analysis

[0109] In the experimental environment, by changing the number of vehicle nodes and the speed of the nodes in the network, a comparative analysis is carried out on GPSR, MM-GPSR, and LP-GPSR in terms of packet loss rate, end-to-end delay, and throughput.

[0110] 2.2.1. Packet loss rate

[0111] As Figure 8 shown, as the number of nodes increases, the packet loss rates of GPSR, MM-GPSR, and LP-GPSR gradually decrease, and the packet loss rate of LP-GPSR is smaller than that of GPSR and MM-GPSR. This is because as the number of nodes increases, the number of available relay nodes also increases, and a better route can be found. As Figure 9 shown, as the node speed increases, the packet loss rates of GPSR, MM-GPSR, and LP-GPSR gradually increase, and the packet loss rate of LP-GPSR is smaller than that of GPSR and MM-GPSR. This is because as the node speed increases, the link stability between nodes decreases, and the packet loss rate becomes larger. In addition, since LP-GPSR uses the relative angle, relative distance, and link stability factor to establish an evaluation function when performing greedy forwarding and selects the node with the largest evaluation function as the relay node; when performing perimeter forwarding, it considers the relative angle, reduces routing loops, and at the same time uses the left-hand and right-hand rules to forward data, reducing the possibility of packet loss. Therefore, the packet loss rate of LP-GPSR is smaller than that of GPSR and MM-GPSR.

[0112] 2.2.2. End-to-end delay

[0113] As Figure 10 shown, as the number of nodes increases, the end-to-end delays of GPSR, MM-GPSR, and LP-GPSR gradually decrease, especially when the number of nodes is small, this phenomenon is more obvious. Because when the number of nodes is small, the nodes will fall into local optimality and use perimeter forwarding to find the next-hop node. Perimeter forwarding has routing redundancy, resulting in a larger end-to-end delay. As the number of nodes increases, the number of perimeter forwarding times will decrease, and the end-to-end delay will also decrease. As Figure 11As shown in the figure, as the node speed increases, the end-to-end delay of GPSR, MM-GPSR, and LP-GPSR gradually increases. In addition, the end-to-end delay of LP-GPSR is significantly less than that of GPSR and MM-GPSR because LP-GPSR takes into account the relative angle of nodes, the relative distance between nodes, and the link stability factor during greedy forwarding, making the link more stable and reducing the retransmission of data packets. During perimeter forwarding, the left-hand and right-hand rules are used simultaneously for forwarding, ensuring the reliability of data packet transmission.

[0114] 2.2.3. Throughput

[0115] As Figure 12 shown in the figure, the throughput of GPSR, MM-GPSR, and LP-GPSR increases as the number of nodes increases, and the increase is more obvious when the number of nodes is large. Because as the number of nodes increases, the amount of data packets transmitted in the entire network also increases, and the number of data packets received per unit time increases. As Figure 13 shown in the figure, the throughput of GPSR, MM-GPSR, and LP-GPSR decreases as the node speed increases, and the throughput of LP-GPSR is greater than that of GPSR and MM-GPSR. Because as the node speed increases, the stability of the link decreases, the packet loss rate increases, the number of retransmissions of data packets increases, and the end-to-end delay also increases. In addition, since LP-GPSR takes into account the relative angle of nodes, the relative distance between nodes, and the link stability factor during greedy forwarding, the packet loss rate is smaller. During perimeter forwarding, the relative angle of nodes is considered, reducing routing redundancy and the end-to-end delay is smaller. Therefore, the throughput of LP-GPSR is greater than that of GPSR and MM-GPSR.

Claims

1. A VANET geographical routing method based on an improved EKF position prediction algorithm, characterized in that, It includes the following steps: Step 1: The current node runs the improved EKF positioning algorithm to update the position information of neighbor nodes and update the neighbor table. The improved EKF positioning algorithm first takes the position and speed information of neighbor nodes in the neighbor table as the input of the input layer of the DNN, then runs the EKF algorithm on the hidden layer, and finally obtains the future position information of neighbor nodes at the output layer; Step 2: Determine whether it is a local optimum. If the current node has not fallen into a local optimum, execute greedy forwarding to select a relay node as the new current node. If the node has fallen into a local optimum, execute perimeter forwarding to select a relay node as the new current node. When the destination node is within the neighbor table of this relay node, forward the data. Otherwise, execute Step 1; The specific implementation method of the greedy forwarding is to build an evaluation function through the node relative angle, node relative distance, and link stability factor, and select the node with the largest evaluation function value as the relay node Where: cosθ represents the cosine value of the relative angle of the node; d ND represents the relative distance of the node; T S&N represents the link stability factor; ω1, ω2, and ω3 are the weighted values of the relative angle of the node, the relative distance of the node, and the link stability factor respectively, ω1 + ω2 + ω3 = 1, the relative angle of the node is the included angle formed by the current node and the destination node and the neighbor node, and the relative distance of the node is the distance between the neighbor node and the destination node; The calculation method of the link stability factor is V S&N = |V S -V N | Where: R represents the transmission range of the current node; D S&N represents the distance between the current node and its neighbor node; V S&N represents the relative velocity between the current node and its neighbor node, V S represents the velocity of the current node; V N represents the velocity of the neighbor node; The specific method of the perimeter forwarding is to back up the data packet during the perimeter forwarding, then perform routing forwarding according to the left-hand and right-hand rules simultaneously, and select the neighbor node with the smallest node relative angle as the relay node.

Images