基于链路质量预测的UANET改进蚁群路由算法

Improved Ant Colony Optimization routing algorithm for UAV ad-hoc Network based on Link Quality Prediction

摘要:

无人机自组网(UANET)可通过多跳转发增大通信范围，其中路由算法承担数据包传输路径规划的任务。针对高动态网络下，无人机定位偏差带来的定向天线波束对不准所造成的增益衰减问题，提出一种基于链路质量预测的蚁群路由算法(LQP-ACO)。该算法利用双向门控循环单元-全连接神经网络(BiGRU-FCNN)预测无人机节点之间的链路质量，然后根据预测的链路质量，利用蚁群算法寻找最优的2条路径进行业务数据传输。仿真结果表明，提出的路由算法相较于传统的Dijkstra算法，在随机路点(RWP)及随机游走(RW)移动模型下，丢包率分别降低了2.75%、4.5%。

Abstract:

Unmanned Aerial vehicle ad-hoc Network(UANET) can increase the communication range by multi-hop forwarding, in which the routing algorithm undertakes the task of packet transmission path planning. To address the gain attenuation problem caused by inaccurate directional antenna beam pairing due to UAV positioning deviation in highly dynamic networks, an Ant Colony Optimization routing algorithm based on Link Quality Prediction(LQP-ACO) is proposed. The algorithm first predicts the link quality between UAV nodes using Bidirectional Gated Recurrent Unit-Fully Connected Neural Network(BiGRU-FCNN). Then, based on the predicted link quality, ant colony optimization algorithm is employed to find the two optimal paths for business data transmission. Simulation results show that the routing algorithm proposed in this paper reduces the packet loss rate by 2.75% and 4.5% respectively compared to the traditional Dijkstra's algorithm under Random Way Point(RWP) as well as Random Walk(RW) mobile models.