无人机自组织网络多包接收智能信号检测算法

Intelligent Signal Detection Algorithm for Multi-Packet Reception in UAV Ad Hoc Network

在无人机自组织网络(UAV Ad Hoc Network, UANET)中,传统的基于单包接收的信号检测算法极大限制了多路传输共享的并发通信性能。针对此问题,利用迭代并行干扰消除技术和多输入多输出技术并联合机器学习设计出一种UANET多包接收智能信号检测算法。该算法保留了迭代并行干扰消除算法的整体结构,采用最合适的深度神经网络来代替传统的基于信道模型的复杂计算,使得分簇UANET的簇头节点不仅可以对任意无记忆固定信道进行处理,而且也不需要去获取准确的信道状态信息便可以同时正确接收来自多个发送节点并发传输过来的数据包。仿真结果表明,该算法可以在不同场景下有效降低系统误码率(Symbol Error Rate, SER),从而有效增加UANET的通信并发度。在线性信道多节点通信场景下,所提出的算法相比于最优MAP(Maximum A Posteriori,最大后验概率)检测算法,系统误码率可以降低约25%。

In the UAV Ad Hoc Network(UANET),the traditional signal detection algorithm based on single-packet reception greatly limits the concurrent communication performance of multiplex transmission sharing.In order to solve this problem, an intelligent signal detection algorithm for multi-packet reception in UANET is designed by using iterative parallel interference cancellation technology and multiple-input multiple-output technology combined with machine learning.The algorithm retains the overall structure of the iterative parallel interference cancellation algorithm, and adopts the most suitable deep neural network to replace the traditional complex calculation based on the channel model.The cluster head node of the clustered UANET can not only process any fixed channel without memory, but also correctly receive the data packets concurrently transmitted from multiple sending nodes without acquiring accurate channel state information.The simulation results show that the algorithm can effectively reduce symbol error rate(SER) of the system in different scenarios, thereby effectively increasing the communication concurrency of UANET.The proposed algorithm can reduce the system SER by about 25% compared to the optimal maximum a posteriori(MAP) detection algorithm in the scenario of linear channel multi-node communication.