An expected transmission number ETX assisted unmanned aerial vehicle routing method based on a k-means algorithm

Abstract

The application discloses a kind of ETX auxiliary unmanned aerial vehicle routing methods based on k-means algorithm, belong to the routing technical field of self-organizing network network layer.The application uses transmission expectation technology ETX value to replace the minimum principle of hop number in AODV routing protocol, effectively improve link quality, simultaneously, by the distance of node to destination, ETX value, node current free buffer space three features as classification standard, the network node is classified and handled, finally, the optimality cluster is selected to carry out the purpose of routing request message RREQ message forwarding.The application can effectively reduce network overhead, reduce end-to-end transmission delay, while well handle network congestion, compared with ND-AODV-ETX, AODV-ETX, AODV protocol, the application K-AODV-ETX routing protocol has better performance in routing overhead, network delay, throughput, is more suitable for topology change faster, higher delay requirement FANET.

Description

Technical Field

[0001] This invention belongs to the field of routing technology for the network layer of self-organizing networks, and specifically relates to a routing method for UAV communication. Background Technology

[0002] Mobile Ad Hoc Networks (MANETs) are a network paradigm that interconnects communication devices in single-hop or multi-hop configurations without requiring any prior infrastructure or management support. These networks are autonomous, dynamic, infrastructure-free, and self-organizing. Participating nodes in the network are automatically configured, have wireless connectivity, and can move arbitrarily. These networks face many challenges due to the wireless characteristics of the nodes and resource constraints. The dynamic topology caused by node mobility makes these networks more challenging from a routing perspective. In recent decades, MANETs have seen widespread development in many fields due to their unique advantages. Network dynamism has become a fundamental theme in networking, accompanied by frequent network link failures, making the discovery of the optimal path between the source and destination nodes within a limited time a challenging problem.

[0003] Flying Ad-Hoc Networks (FANETs) are an important area of ​​mobile ad hoc networks. The rapid and frequent changes in network nodes give FANETs unique characteristics, namely higher mobility and spatial dimension. One of its main requirements is a low-overhead, high-efficiency routing protocol.

[0004] On-Demand Distance Vector (AODV) is a leading responsive routing protocol designed for mobile ad hoc networks. As one of the earliest developed routing protocols, it boasts excellent performance in terms of hop count, packet delivery ratio, and network control overhead. AODV has been extensively studied, with researchers proposing numerous variants. Its convenient and efficient performance has led to its widespread application in mobile ad hoc networks. However, traditional AODV protocols frequently face challenges in intensive, highly dynamic, and rapidly changing unmanned aerial vehicle (UAV) ad hoc networks, such as communication link breaks and insufficient data transmission. Therefore, designing a more efficient and stable routing protocol is crucial.

[0005] ETX metric is one of the most widely used routing metrics. Taking the AODV protocol as an example, ETX metric can effectively balance the link quality issues brought about by the minimum hop count path, but it also brings the problem of high network routing overhead. The flooding broadcast of the AODV protocol further aggravates this problem, which is a waste of limited wireless resources.

[0006] The k-means algorithm is one of the most widely used clustering algorithms. It is widely applied in data mining, sensor networks, and ad hoc networks. It is a simple unsupervised learning algorithm that classifies a dataset into a set of K clusters specified at the start of the process, aiming to increase the inter-class distance and decrease the intra-class distance. In the AODV routing protocol, route discovery is accomplished through flooding, which involves broadcasting Route Request (RREQ) packets to all nodes within the sender's range. This often leads to unnecessary RREQ packets and retransmissions of Reply (RREP) packets generated as responses, resulting in packet collisions and congestion in the network. Summary of the Invention

[0007] To overcome the shortcomings and defects of the traditional AODV routing protocol in FANET applications, this invention proposes an ETX-assisted UAV routing method based on the k-means algorithm. This method can effectively reduce routing overhead, decrease end-to-end latency, and achieve greater network throughput, thereby effectively improving the performance of UAV networks.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A UAV routing method based on the k-means algorithm and assisted by ETX (Expected Transmission Count) is proposed for multi-UAV network communication scenarios. The multi-UAV network adopts a decentralized structure, meaning each UAV node has an equal status in the network. The method includes the following steps:

[0010] Step 1: Expand the message structure of the route request message RREQ, route reply message RREP, and route maintenance message Hello by adding three features to the three control message messages: node-to-destination distance, expected number of transmissions, and node free buffer space.

[0011] Step 2: Initiate the route discovery process and control the propagation of message packets between drone nodes. At this time, information about each drone node is obtained based on the three features in Step 1.

[0012] Step 3: Based on the collected information on the distance from each node to the destination, the expected number of transmissions, and the size of the node's free buffer space, the k-means algorithm is used to classify the nodes, and the optimal cluster is selected as the forwarding cluster for the Route Request Message (RREQ).

[0013] Step 4: The expected number of transmissions (ETX) mechanism is used to replace the shortest hop count mechanism of the On-Demand Distance Vector Routing (AODV) protocol for path selection, and the optimal path is selected for data transmission.

[0014] Furthermore, in step 3, the specific method for classifying nodes using the k-means algorithm is as follows:

[0015] Step 3-1: Select the initial cluster size K;

[0016] Step 3-2: Randomly select three centroid nodes;

[0017] Step 3-3: Cluster the nodes according to the three clustering features, and assign each node to the nearest centroid until all nodes are assigned.

[0018] Furthermore, in step 3, the evaluation of the cluster is accomplished by comparing it with the characteristics of an ideal repeater. The optimal cluster has the following characteristics: node-to-destination distance = 0; expected number of transmissions = 0; node free buffer space = maximum buffer size.

[0019] The beneficial effects of this invention are as follows:

[0020] (1) Compared with the traditional AODV routing protocol, this invention is based on the k-means algorithm and combined with the expected transmission number ETX mechanism. It focuses on improving the flooding broadcast mechanism and path selection mechanism of routing requests, so that the routing request message selects the best cluster for forwarding, which greatly reduces network overhead and saves limited wireless network resources.

[0021] (2) The present invention uses the k-means clustering algorithm and classifies network nodes using three features: distance to destination node, ETX value, and node free buffer space. Compared with the traditional AODV routing protocol, the present invention (which can be called K-AODV-ETX routing protocol) can better guarantee the quality of the link, effectively prevent network congestion, and improve network throughput.

[0022] (3) Compared with the current advanced protocols such as P-AODV, AODV-ETX, and AODV-NLS-ETX, the K-AODV-ETX routing protocol of the present invention has better performance and is more suitable for highly dynamic UAV networks.

[0023] In summary, this invention is based on the classic reactive AODV routing protocol, combined with the k-means clustering algorithm, and adopts ETX metric instead of the traditional hop count metric to ensure the transmission quality of the link. It designs a low-overhead K-AODV-ETX routing protocol suitable for high-dynamic, high-latency UAV networks. Attached Figure Description

[0024] Figure 1 , 2 Figures 3 and 4 are schematic diagrams of the frame structure of the extended RREQ message, RREP message, and HELLO message in the embodiments of the present invention.

[0025] Figure 4 This is a flowchart of the k-means algorithm in an embodiment of the present invention.

[0026] Figure 5 , 6 Figures 7 and 8 are performance simulation results of embodiments of the present invention. Detailed Implementation

[0027] The present invention will now be described in detail with reference to the accompanying drawings.

[0028] This paper proposes an ETX-assisted UAV routing method based on the k-means algorithm, which is one of the most widely used clustering algorithms. It is applied in various fields such as data mining, sensor networks, and ad hoc networks. This is a simple unsupervised learning algorithm that classifies a dataset into a set of K clusters specified at the start of the process, aiming to increase inter-class distance and decrease intra-class distance. In the AODV routing protocol, route discovery is accomplished through flooding, which broadcasts Route Request (RREQ) packets to all nodes within the sender's range. This often leads to unnecessary RREQ packets and retransmissions of Response Echo (RREP) packets generated as responses, resulting in packet collisions and congestion in the network. This method proposes an optimized route discovery method based on the ETX AODV routing protocol. The key idea is to use the k-means clustering algorithm to select the optimal cluster of RREQ packet repeaters instead of broadcasting. The aim of this method is to reduce unnecessary control packet transmissions in the network, thereby reducing network congestion, overhead, and end-to-end latency.

[0029] Specifically, the method includes the following steps:

[0030] Step 1: Expand the Route Request (RREQ), Route Recovery (RREP), and Route Maintenance (Hello) messages by adding clustering feature parameters to each message and collect information.

[0031] Step 2: Execute the k-means clustering algorithm to classify network nodes using three features: distance from the node to the destination, expected number of transmissions, and node free buffer space.

[0032] Step 3: Change the flooding broadcast mode of route discovery, and select the optimal cluster as the forwarding cluster for route request RREQ based on k-means feature classification.

[0033] Step 4: Perform the route discovery process and select the best path based on the ETX value.

[0034] Furthermore, the k-means algorithm executed in step 2 randomly selects a K value to classify network nodes based on the three features mentioned above (distance from node to destination, expected number of transmissions, and node free buffer space).

[0035] Furthermore, based on the clustering results in step 2, the optimal cluster is selected for the purposeful forwarding of routing requests, and the protocol routing discovery process is carried out to reduce the resource waste caused by flooding and thus reduce network overhead.

[0036] The following is a simulation implementation process of the method of the present invention:

[0037] Step 1: Construct a drone network communication scenario based on the NS3 simulation platform. The number of drones is set to 10, 20, 30, 40, and 50; the node packet sending speed is set to 10 packets / s, 20 packets / s, 30 packets / s, 40 packets / s, and 50 packets / s; and the node speed is set to 10 m / s, 15 m / s, 20 m / s, 25 m / s, and 30 m / s.

[0038] Step 2: Based on the NS3 simulation platform, extend the relevant message messages. The extended message format is shown below. Figure 1-3 The specific method is as follows:

[0039] In the NS3 simulation software, a UAV network simulation environment was built. New fields were added to the "RREPHeader" class: m_ETX, m_freeSpace, m_positionX, and m_positionY; and new fields were added to the "RoutingTableEntry" class: m_ETX, m_freeSpace, m_positionX, and m_positionY.

[0040] m_position: A vector containing the position of this node.

[0041] m_lastKnowPosition: A map containing the physical location of the last known IP address.

[0042] m_lastKnownCluster: A map containing the last known IP address forwarded to the cluster (periodically cleaned).

[0043] This completes the expansion of the message message.

[0044] Step 3: See Figure 4 The k-means algorithm is executed, and the initial number of clusters is selected at the beginning of the algorithm. The goal is to set of points Reorganized into There are several clusters. Therefore, k-means randomly selects data from the dataset. point As centroids, each centroid belongs to a cluster. The algorithm then assigns each point in the dataset to its nearest centroid. This process is based on an objective function that calculates the sum of squared distances across all clusters.

[0045]

[0046] in This is the distance from the point to the cluster centroid. The location of this point. for The position of the centroid at time It represents the number of clusters. For example... Figure 1 As shown, after assigning points to each cluster, the k-means algorithm updates the position of each centroid using the following formula:

[0047]

[0048] In this method, the k-means algorithm uses three features to select the optimal cluster based on these features: (1) distance to the destination node (2) available buffer space of the node (3) link ETX value.

[0049] ETX is defined as follows:

[0050]

[0051] In the formula, For routing between two nodes in the network, For routing A jump. Forward-to-receive ratio, i.e., the ratio of the number of hops. The probability that a sent data packet is successfully received. This is the reverse reception ratio, i.e., the probability that the corresponding ACK packet is successfully received. In other words, ETX estimates the expected number of transmissions (including retransmissions) required to send the data packet from the source to the final destination. The AODV protocol then selects the route with the minimum ETX, not necessarily the route with the fewest hops. If formed All hop counts are correct, that is:

[0052]

[0053] but equal The number of jumps.

[0054] AODV Link Quality Extensions uses a control message called the LPP message for link probing.

[0055] The ETX metric for a link is represented by the predicted number of data transfers required to send packets on that link, including retransmissions. If This indicates the probability of a message being successfully transmitted. Let represent the probability of successfully receiving an ACK message. Then, the probability of successfully sending and acknowledging a message is: .link The ETX metric is represented as:

[0056]

[0057] Nodes in the network have an average period Broadcast fixed-size Link Probe Packets (LPPs) with probability. and It is measured using dedicated link probe packets, periodicity. The maximum jitter is set to 10% to avoid accidental synchronization. Each node remembers the most recent... The probability of receiving LPPs within a second at any time t is calculated using the following formula:

[0058]

[0059] in It is in the window The actual number of LPPs received during the period, This refers to the number of LPPs that should be received during this window. In the case of link X→Y, this technique allows X to easily measure the number of LPPs successfully received from Y. However, due to the lack of confirmation, the probability of node X cannot be determined. Because calculating the ETX metric for link X→Y requires... and Therefore, each LPP sent by node Y contains the most recent The number of LPPs received from X within seconds. This allows node X to calculate... Experiments have found that taking It performs well and is usually used The value. Routing. The metric is the sum of the ETX values ​​of each link l in the route:

[0060]

[0061] The cluster was evaluated by comparing it to an ideal repeater with the following capabilities:

[0062] (1) The distance to the destination node is 0 (i.e., the current node is the destination node);

[0063] (2) Available buffer size = maximum buffer size;

[0064] (3) The link where the node is located has the largest ETX value.

[0065] Step 3: The protocol begins searching for a path, forwarding the corresponding routing request messages, and the nodes reply with the corresponding routing reply messages (RREP) to find the destination node.

[0066] Step 4: Select the optimal path based on the link's ETX value.

[0067] Step 5: Perform data transmission.

[0068] Simulation results: Figure 5 The figure shows the PDR (Progression Direction) of the four protocols as the number of nodes changes when the node speed is 20 m / s. It can be seen from the figure that using the ETX transmission mechanism can significantly improve the packet delivery rate of the protocols. Compared to AODV, AODV-ETX can improve PDR by up to 7%. ND-AODV-ETX does not show a significant advantage in PDR compared to AODV-ETX. The PDR of the proposed K-AODV-ETX is slightly lower than that of AODV-ETX, which is due to the selective forwarding during the route discovery process of the k-means mechanism.

[0069] Figure 6 The graph shows the latency variations of four protocols with different numbers of nodes. It can be seen that the latency of ND-AODV-ETX and AODV-ETX protocols is significantly higher than that of AODV and K-AODV-ETX protocols. This is due to the retransmission mechanism of ETX. The K-AODV-ETX protocol uses the K-means clustering algorithm to purposefully forward RREQ packets, achieving fast path discovery and effectively balancing the high latency issue brought by the ETX mechanism. Its latency performance is comparable to that of the AODV protocol.

[0070] Figure 7 The throughput performance of four protocols is shown under different numbers of nodes. The figures demonstrate that the ETX mechanism effectively improves the throughput of the AODV protocol. In the K-AODV-ETX protocol, the node buffer term in the clustering features effectively reduces network congestion and increases the protocol's throughput, achieving a maximum improvement of 11.3% compared to the AODV protocol.

[0071] Figure 8The routing overhead of four protocols with different numbers of nodes is shown. Since the ETX mechanism itself occupies bytes in the routing control packet, it will generate certain routing overhead. Therefore, the routing overhead of the AODV protocol is the lowest. The K-AODV-ETX protocol proposed in this paper selects the best cluster for forwarding in the route discovery broadcast stage, effectively controlling invalid route broadcasts, and its overhead is less than that of the AODV-ETX protocol and the ND-AODV-ETX protocol.

[0072] This invention employs the Transmission Expectation Value (ETX) technique to replace the minimum hop count principle in the AODV routing protocol, effectively improving link quality. Simultaneously, it combines the k-means clustering algorithm, using three features—distance to the destination, ETX value, and the node's current free buffer space—as classification criteria to categorize network nodes. Finally, it selects the optimal cluster for targeted forwarding of Route Request Messages (RREQ). This invention effectively reduces network overhead and end-to-end transmission latency, while also handling network congestion well. Compared to ND-AODV-ETX, AODV-ETX, and AODV protocols, this K-AODV-ETX routing protocol exhibits better performance in terms of routing overhead, network latency, and throughput, making it more suitable for FANETs with rapid topology changes and high latency requirements.

[0073] In summary, this invention addresses the problems of traditional AODV routing protocols, such as poor adaptability to highly dynamic UAV networks, high latency, and large routing overhead. It employs the k-means clustering algorithm to classify network nodes based on three node characteristics, selecting the optimal cluster for forwarding Route Request Messages (RREQs). This targeted forwarding of RREQs replaces the resource consumption and waste caused by the indiscriminate flooding of the original AODV routing protocol, effectively alleviating network congestion. In path selection, it uses the Expected Transmission Count (ETX) metric, replacing the minimum hop count principle in the original AODV routing protocol, ensuring link quality and improving network throughput.

Claims

1. A UAV routing method based on k-means algorithm with expected transmission count (ETX) assisted by the algorithm, characterized in that, This technology is applied to multi-drone networking communication scenarios. The multi-drone network adopts a decentralized structure, meaning each drone node has an equal status in the network; it includes the following steps: Step 1: Expand the message structure of the route request message RREQ, route reply message RREP, and route maintenance message Hello by adding three features to the three control message messages: node-to-destination distance, expected number of transmissions, and node free buffer space. Step 2: Initiate the route discovery process and control the propagation of message packets between drone nodes. At this time, information about each drone node is obtained based on the three features in Step 1. Step 3: Based on the collected information on the distance from each node to the destination, the expected number of transmissions, and the size of the node's free buffer space, the k-means algorithm is used to classify the nodes, and the optimal cluster is selected as the forwarding cluster for the Route Request Message (RREQ). Step 4: The expected number of transmissions (ETX) mechanism is used to replace the shortest hop count mechanism of the On-Demand Distance Vector Routing (AODV) protocol for path selection, and the optimal path is selected for data transmission.

2. The method for UAV routing based on the k-means algorithm with expected transmission count (ETX) according to claim 1, characterized in that, In step 3, the specific method for classifying nodes using the k-means algorithm is as follows: Step 3-1: Select the initial cluster size K; Step 3-2: Randomly select three centroid nodes; Step 3-3: Cluster the nodes according to the three clustering features, and assign each node to the nearest centroid until all nodes are assigned.

3. The method for UAV routing based on the k-means algorithm with expected transmission count (ETX) according to claim 1, characterized in that, In step 3, the evaluation of the cluster is accomplished by comparing it with the characteristics of an ideal repeater. The optimal cluster has the following characteristics: node-to-destination distance = 0; expected number of transmissions = 0; node free buffer space = maximum buffer size.

Images