Flight ad hoc network AODV routing method based on multi-metric weight fusion

Abstract

The invention discloses a flight ad hoc network AODV routing method based on multi-metric weight fusion. Path criterion parameters are introduced on the basis of an AODV routing protocol. On the basis of three aspects of node prediction residual energy ratio, link congestion degree and link stability, performance criterion parameters are obtained, and then the performance criterion parameters and normalized path hop count parameters are subjected to weighted fusion to obtain path criterion parameters. In the route selection stage, a theoretical optimal path is selected according to path criterion parameters, a parameter normalization method is used, it is guaranteed that the influence effects of all the criterion parameters are similar, and path selection is more reasonable. According to the method, a multi-metric normalized weight fusion method is introduced, optimization is carried out aiming at the characteristics that the flight self-organizing network nodes are high in moving speed, relatively short in path duration time, too high in consumption of part of high-load nodes and the like, and the routing performance in a flight self-organizing network scene is improved. According to the method, the FANET operation time can be effectively prolonged, and stable and reliable data transmission is provided on the whole.

Description

Technical Field

[0001] The invention relates to the technical field of flight self-organizing network routing protocols, in particular to a flight self-organizing network AODV routing method based on multi-metric weight fusion. Background Art

[0002] Mobile Ad Hoc Network (MANET) is a self-organizing, self-managing, decentralized multi-hop wireless autonomous network that does not require other infrastructure support. Each mobile node can act as a router to communicate directly with another node within its communication range without the need for other auxiliary facilities. At present, MANET has many applications in a large number of scenarios such as vehicles, fishing boats, drones, satellites, etc. With the rapid development of communication technology and drone technology, the application of unmanned aerial vehicles (UAVs) has expanded to many military and civilian fields. A single drone has certain limitations when dealing with complex tasks. Its working range is limited and it is easily affected by many aspects. In order to avoid the above situation as much as possible, the concept of drone clusters came into being. Drone clusters can greatly expand the working range of drones and have higher reliability than single drones. For the use of a large number of drones, the communication and control between each drone node is a key issue that needs to be dealt with. In order to solve this problem, researchers designed a self-organizing network composed of multiple drones, namely the Flying Ad hoc Network (FANET). FANET is a special scenario of MANET in airspace, with the characteristics of distributed nodes, fast mobility, high latency requirements, and limited energy.

[0003] At present, FANET routing protocols are mainly improved from MANET routing protocols. Ad hoc On-Demand Distance Vector Routing (AODV) is one of the most classic protocols and has been widely used. AODV is an on-demand routing protocol that dynamically establishes routes when needed, saving network resources. However, the AODV protocol lacks the ability to perceive the network environment and does not consider the impact of node energy and load on routing. It may cause some nodes to fail due to excessive load, affecting the network structure and reducing network communication performance.

[0004] An existing routing protocol based on the number of neighbor nodes and node location information is improved. It screens the neighbor nodes of drone nodes according to their number, node location information and remaining energy value, limits the number of forwarding nodes and forwarding areas of data packets, and improves the stability and reliability of routing. However, in working scenarios with heavy data transmission tasks, the AODV-C algorithm may cause some nodes to be overloaded and unable to process data in a short period of time, resulting in network performance degradation and packet loss. In addition, the AODV-C algorithm has poor adaptability to high-speed mobile scenarios. When drone nodes move at high speed, neighbor nodes change relatively quickly, and the probability of routing path failure is high.

[0005] In order to solve the above problems, the present invention proposes a routing selection method for selecting a relatively stable link with relatively high predicted residual energy for data transmission. The network perception capability of the AODV protocol is improved by a multi-metric weight fusion method, and a better routing path is selected to avoid frequent routing requests. Summary of the invention

[0006] The purpose of the present invention is to provide a multi-metric weight fusion flight self-organizing network AODV routing protocol to solve the problems of poor research on energy consumption balance and network life cycle in the existing routing protocol during the route selection stage, and the adverse effects caused by abnormal nodes in the network. At the same time, a normalization method is designed to solve the problem that some metric criteria that may appear when multi-parameters are fused have too little influence on the fusion result.

[0007] Specifically, the method comprises the following steps:

[0008] S1: When the source node needs to send data to the destination node, if there is no existing path in its own routing table, it is processed according to the routing protocol algorithm: the source node creates an RREQ message and broadcasts it to all nodes within the communication radius;

[0009] S2: After receiving the RREQ message, the node determines whether it is the source node, the destination node, or has received the RREQ message repeatedly. If it is the destination node, the path determination phase can be started. If it is the source node, the RREQ message is discarded. If it is received repeatedly, the message is discarded. Otherwise, the subsequent steps are continued.

[0010] S3: The node obtains relevant data through the received RREQ message, including the load of the previous hop node, the coordinates and predicted coordinates of the previous hop node, performance criterion parameters and other information, determines whether the propagation conditions are met, calculates the performance criterion parameters based on the above information combined with the data of this node, updates the RREQ message, and continues broadcasting.

[0011] The path determination phase includes the following steps:

[0012] 1) After the destination node receives the RREQ message for the first time, it starts the timer and continues to receive RREQ messages within a certain period of time;

[0013] 2) During the timer operation, the path criterion parameter is calculated based on the newly received RREQ message and updated to a larger value;

[0014] When calculating the path criterion parameters, we first need to calculate the normalized path hop count parameter H normalized :

[0015]

[0016] Among them, hops max Indicates the maximum number of hops of the path received by the destination node, hops i Indicates the current path hop count, which is obtained from the RREQ received by the destination node.

[0017] Path criteria parameter path i The calculation formula is as follows:

[0018] path i =β1×w accumulated +β2×H normalized

[0019] Among them, w accumulated is the cumulative performance criterion parameter, H normalized is the normalized path hop count criterion parameter, β1+β2=1.

[0020] 3) When the timer ends, a route is established based on the path corresponding to the saved path criteria parameters.

[0021] Preferably, in S1, the RREQ message adds the following content: Node load status Load i , node coordinates and predicted coordinates, cumulative performance criterion parameter w accumulated ;

[0022] Preferably, S3 is specifically:

[0023] S3.1: Check whether the node currently meets the following conditions: the node's remaining energy ratio is less than 5%, the node's congestion level is greater than 97%, and the node's current speed exceeds the maximum speed by 50%. If the above conditions are met, the RREQ message is discarded;

[0024] S3.2: Compute the node's predicted remaining energy ratio, link congestion level, and link stability.

[0025] 1) Node predicted remaining energy ratio:

[0026] This patent uses a cross-layer method to obtain the predicted remaining energy ratio of the node. By obtaining cross-layer information, the length of the node's current data packet queue from the MAC layer is obtained. queue , calculate the energy consumed by processing the data in the current queue, and get the predicted energy loss E prec The total node energy is E total , the current energy of the node is E current , the average energy consumption of a node processing each request is E iconsum , then the node predicts the remaining energy ratio E er It can be expressed as:

[0027]

[0028] 2) Link congestion level:

[0029] Let the queue length be length queue , the total length of the queue is recorded as length total , the node load degree is recorded as Load i , then the node load degree can be expressed as:

[0030]

[0031] The link congestion degree works better when the link status is balanced and the congestion degree is low. Therefore, the link congestion degree can be expressed as follows:

[0032]

[0033] Normalize the above formula and scale it to [0, 1] to achieve the balanced effect of multiple criterion parameters:

[0034]

[0035] 3) Link stability

[0036] FANET nodes are in high-speed motion and the topology changes dramatically. The Kalman filter algorithm can predict the node location at the next moment, and the two nodes can obtain the link change trend based on the predicted location.

[0037] The Kalman filter algorithm is mainly divided into two parts: prediction and update:

[0038] Prediction part:

[0039] X k|k-1 =FX k-1 +u

[0040] P k|k-1 =FP k-1 F T+Q

[0041] Updated part:

[0042] K k =P k|k-1 H T (HP k|k-1 H T +R) -1

[0043] X k =X k|k-1 +K k (Z k -HX k|k-1 )

[0044] P k =(IK k H)P k|k-1

[0045] Among them, P k and P k-1 Represent the covariance matrix at time k and time k-1 respectively. k-1 Based on the predicted value at time k and the optimal value at time k-1, P can be iterated k The value of .

[0046] In a short period of time, the motion state of the drone can be regarded as uniformly accelerated motion, and its three-dimensional position p and velocity v at time k can be expressed by the following formula:

[0047]

[0048] v k =v k-1 +a k-1 ×t

[0049] The Kalman filter algorithm is used to predict the node location X at the next moment. k After that, the predicted distance d between the two nodes can be obtained. pre At this time, the link stability can be written as follows:

[0050]

[0051] Among them, d current is the distance between the two nodes at the current moment, v max represents the maximum moving speed of the UAV network, and Δt represents the data interaction time interval of the UAV system. pre and d current The calculation formula is as follows:

[0052]

[0053] Among them, (x i+1 ,y i+1 ) is the two-dimensional position of the node at the next moment predicted by the Kalman filter algorithm, (x i ,y i ) is the two-dimensional position of the node at the current moment.

[0054] S3.3: Update the hop count and node load in the RREQ message i , current node coordinates and predicted coordinates, cumulative performance criterion parameter w accumulated .

[0055] Performance criterion parameter w i The calculation formula is as follows:

[0056] w i =α1×E er +α2×Link n_ij +α3×Stable ij

[0057] Among them, α1+α2+α3=1, E er To predict the remaining energy ratio, Link n_ij To normalize the link congestion level, Stable ij Link stability.

[0058] The performance criterion parameters obtained by the above formula are accumulated to obtain the cumulative criterion parameter w accumulated , accumulated hop by hop in RREQ to record the link performance:

[0059] w accumulated =w1+w2+…+w i ,i=hops

[0060] The present invention has the following beneficial effects:

[0061] 1. By introducing three measurement methods, namely node prediction remaining energy ratio, link congestion degree and link stability, and by normalizing the fusion of multi-metric weights, the system performance is analyzed more comprehensively, providing reliable criteria for the selection of routing paths.

[0062] 2. It effectively reduces the load level of some busy nodes, avoids their rapid energy consumption, prolongs the FANET operation time to a certain extent, and improves the system throughput and packet delivery rate of data packets.

[0063] 3. It limits the communication work of nodes in bad state, avoiding problems such as excessive communication delay and short routing duration that may be caused by selecting nodes in bad state for communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Overall structure diagram of path criterion parameters;

[0065] Figure 2 Schematic diagram of the route establishment process;

[0066] Figure 3 Route discovery flow chart. DETAILED DESCRIPTION

[0067] The technical solution of the present invention is further described below with reference to the accompanying drawings.

[0068] Figure 1 It is the overall structure diagram of the path criterion parameters. The path criterion parameters are calculated through the performance criterion parameters and the normalized hop count parameters. The performance criterion parameters are obtained by fusion calculation based on the normalized weights of the node's predicted remaining energy ratio, link congestion level and link stability. They are accumulated hop by hop during the route discovery process, and finally the path criterion parameters are calculated at the destination node to provide a judgment basis for subsequent route selection and other tasks.

[0069] The calculation of the predicted remaining energy ratio, link congestion level and link stability is as follows:

[0070]

[0071] Among them, E er Represents the predicted remaining energy ratio, E current Indicates the remaining energy of the current node, E prec represents the predicted energy loss, E iconsum represents the average energy consumption of each request processed by the node, length queue Indicates the current packet queue length information of the node, E total Represents the total energy of the node.

[0072] The link congestion level is calculated as follows:

[0073]

[0074] Among them, Load i is the node load level, length queue is the queue length, length total is the total queue length, Link ij Link congestion degree, Link n_ij is the normalized link congestion level.

[0075] The calculation of link stability mainly uses the Kalman filter algorithm, which is mainly divided into two parts: prediction and update.

[0076] Prediction part:

[0077] Xklk-1 =FX k-1 +u

[0078] P k|k-1 =FP k-1 F T +Q

[0079] Updated part:

[0080] K k =P k|k-1 H T (HP k｜k-1 H T +R) -1

[0081] X k =X k|k-1 +K k (Z k -HX k|k-1 )

[0082] P k =(IK k H)P k|k-1

[0083] Among them, P k and P k-1 Represent the covariance matrix at time k and time k-1 respectively. k-1 Based on the predicted value at time k and the optimal value at time k-1, P can be iterated k The value of .

[0084] In a short period of time, the motion state of the drone can be regarded as uniformly accelerated motion, and its three-dimensional position p and velocity v at time k can be expressed by the following formula:

[0085]

[0086] v k =v k-1 +a k-1 ×t

[0087] The three-dimensional position p and speed of the drone are recorded as the drone state X, and the above two equations are substituted into the state X at time k: k , which can be written as follows:

[0088]

[0089] Among them, F is the state transfer matrix, and u is the system noise error.

[0090] The covariance matrix of the system noise error u can be written as:

[0091]

[0092] In the Kalman filter algorithm, it is necessary to obtain relevant data information at the current moment through sensors and other means. In the FANET system, the required parameters can be obtained through satellite positioning, and the observed state quantity at time k is recorded as Z k ,but:

[0093]

[0094] Among them, H is the state transfer matrix, and R is the error between the satellite positioning and the actual position.

[0095] The Kalman filter algorithm is used to predict the node location X at the next moment. k After that, the predicted distance d between the two nodes can be obtained. pre At this time, the link stability can be written as follows:

[0096]

[0097] Among them, d current is the distance between the two nodes at the current moment, v max represents the maximum moving speed of the UAV network, and t represents the data interaction time interval of the UAV system. pre and d current The calculation formula is as follows:

[0098]

[0099] Among them, (x i+1 ,y i+1 ) is the two-dimensional position of the node at the next moment predicted by the Kalman filter algorithm, (x i ,y i ) is the two-dimensional position of the node at the current moment.

[0100] According to the above formula, three measurement parameters can be obtained, which can be substituted into the following formula to obtain the performance criterion parameter w i :

[0101] w i =α1×E er +α2×Link n_ij +α3×Stable ij

[0102] Among them, α1+α2+α3=1.

[0103] The path criterion parameters are accumulated during the RREQ propagation process to obtain the cumulative path criterion parameter w accumulated , and the normalized hop count parameter can be combined to obtain the path judgment parameter path i :

[0104] path i =β1×w accumulated +β2×H normalized

[0105] Figure 2 This is a schematic diagram of the routing establishment process. In order to more concisely represent the routing establishment process, assuming that the current nodes are at the same height, the node topology distribution is as follows: Figure 2 As shown in the figure, the existing node A needs to communicate with node I, and there is no route to I in the routing table of source node A, so path finding needs to be started:

[0106] (1) Source node A creates a modified RREQ message and broadcasts it to all nodes within the communication radius. In this figure, the nodes within the communication radius of node A include node B and node E.

[0107] (2) After receiving the RREQ message, neighboring nodes B and E determine that they are not source node A or destination node I and have not received the RREQ message repeatedly, so they proceed according to the subsequent process;

[0108] (3) Nodes B and E obtain relevant data from the received RREQ message, including the load of the previous hop node, the coordinates and predicted coordinates of the previous hop node, performance criterion parameters, etc., to determine whether the propagation conditions are met. Based on the above information and the data of the local node, the performance criterion parameters are calculated, and the information is updated to the RREQ message, and the broadcast continues;

[0109] (4) At this time, nodes A and C will receive the RREQ message from node B. After receiving the RREQ message, node A confirms that it is the source node and discards the message. After receiving the message, node C processes it according to (2) and (3). Nodes A and F will receive the message from node E. After receiving the RREQ message, node A confirms that it is the source node and discards the message. After receiving the message, node F processes it according to (2) and (3).

[0110] (5) Broadcast the RREQ message according to the above process until the RREQ message reaches the destination node. Assume that the RREQ message from node D first reaches node I at this time. Node I calculates the path criterion parameter path1 according to the following formula and saves it in the destination node I:

[0111] path1=β1×w accumulated +β2×H normalized

[0112] Subsequently, node I starts a timer and continues to receive RREQ messages within a certain period of time;

[0113] Assume that during the timer timing, node I receives a RREQ message from node H, calculates the path criterion parameter path2 of the path, and if path2 is higher than the original path criterion parameter path1, then the path criterion parameter saved by the destination node I is updated. If path2 is lower than the original path criterion parameter path1, the original path criterion parameter path1 is saved in the destination node I, and no parameter update operation is performed on it. Assuming that path2 is lower than path1 at this time, and no other RREQ message arrives during the timer timing, then according to Figure 2 The RREP message is returned in the order of I→D→C→B→A, and the route is established to start transmitting data.

Claims

1. A method for routing in an AODV self-organizing network based on multi-metric weight fusion, characterized in that: In the route discovery phase, a multi-metric weight fusion method is used to obtain path criterion parameters by introducing the predicted node residual energy ratio, link congestion level, link stability and route hop count for normalized weight fusion, and select the theoretically optimal routing path; the method steps include: S1: When the source node needs to send data to the destination node, if there is no existing path in its own routing table, it is processed according to the routing protocol method: the source node creates an RREQ message and broadcasts it to all nodes within the communication radius; S2: After receiving the RREQ message, the node determines whether it is the source node, the destination node, or has received the RREQ message repeatedly. If it is the destination node, the path determination phase can be started. If it is the source node, the RREQ message is discarded. If it is received repeatedly, the message is discarded. Otherwise, the subsequent steps are continued. S3: The node obtains relevant data through the received RREQ message, including the load of the previous hop node, the coordinates of the previous hop node, the predicted coordinates of the previous hop node, performance criterion parameters and other information, and determines whether the propagation conditions are met. Based on the above information and the data of this node, the performance criterion parameters are calculated, the RREQ message information is updated, and the broadcast continues.

2. According to claim 1, an AODV routing method based on multi-metric weight fusion is characterized in that: The following modifications are made to the RREQ message format: Add node load condition Load i , node coordinates and predicted coordinates, cumulative performance criterion parameter w accumulated .

3. The AODV routing method based on multi-metric weight fusion according to claim 1, characterized in that: The path determination stage of S2 is specifically as follows: (1) After the destination node receives the RREQ message for the first time, it starts a timer and continues to receive RREQ messages within a certain period of time; (2) During the timer operation, the path criterion parameter is calculated based on the newly received RREQ message and updated to a larger value; When calculating the path criterion parameters, it is necessary to calculate the normalized path hop count parameter H normalized And the cumulative performance criterion parameter w accumulated ; The calculation formula for the normalized path hop count parameter is as follows: Among them, hops max Indicates the maximum number of hops for a path, which is determined by the number of RREQ messages received by the destination node. i Indicates the current path hop count, obtained from the RREQ received by the destination node; The cumulative performance criterion parameter is obtained by adding up the performance criterion parameters, and the formula is as follows: w accumulated =w1+w2+…+w i ,i=hops Among them, hops is the number of path hops; The performance criterion parameter calculation formula is as follows: In i =α1×E er +α2×Link n_ij +α3×Stable ij Among them, E er To predict the remaining energy ratio, Link n_ij To normalize the link congestion level, Stable ij is the link stability, α1+α2+α3=1; According to the above formula, the path judgment parameter path is calculated according to the following formula i : path i =β1×w accumulated +β2×H normalized Among them, w accumulated is the cumulative performance criterion parameter, H normalized is the normalized path hop count criterion parameter, β1+β2=1; (3) When the timer expires, a route is established based on the path corresponding to the saved path criteria parameters.

Images