Research On Neighbor Discovery Based On Directional Antennas In Wireless Ad Hoc Networks

Wireless ad hoc networks,which have the advantages of no fixed infrastructure,decentralized,and dynamic network topology,have been widely used in special-purpose communication scenarios such as emergency,disaster relief,and battlefields.By using directional antenna technology,the wireless ad hoc networks can effectively improve space utilization,save transmission power,increase transmission distance,and improve the anti-interference ability of network communication.As the networking foundation of wireless ad hoc networks,neighbor discovery technology can generate the table of neighbors’status for each network node,and lay the foundation for subsequent dynamic network topology maintenance and communication transmission services.Different from the traditional omnidirectional antenna neighbor discovery technology,the directional antenna neighbor discovery technology is facing the problem of low efficiency of sector alignment between adjacent network nodes.Therefore,this thesis plans to design an efficient directional antenna state switching algorithm that can complete neighbor discovery in a limited time based on adaptive beam antennas.And uses the Q-learning theory of reinforcement learning to study the node conflict problem in the process of neighbor discovery.The main contributions of this thesis include:Firstly,given the problem of adaptive beam antenna state switching,this thesis respectively designs a deterministic antenna state switching algorithm for clock synchronous and asynchronous wireless ad hoc networks based on the uniqueness of the binary number of network nodes.It can be respectively ensured that two nodes configured with adaptive beam antenna can achieve sector alignment of each other within（n+（?）log₂ n（?）+1）mslots and 4nm slots,thereby laying a physical foundation for discovery.Here n represents the length of the binary number sequence of each network node,and m represents the number of adaptive beam antenna sectors configured by each network node.The algorithm simulation analysis based on MATLAB shows that the clock synchronous and asynchronous wireless ad hoc network directional antenna state switching algorithms proposed in this thesis can obtain a better Maximum Time-to-Discovery than that of the existing algorithms.Secondly,for the problem of node conflict in the process of neighbor discovery,this thesis proposes a neighbor discovery algorithm that can effectively alleviate node conflict.In the early stage of neighbor discovery,the node density information is collected through the node conflict detected in the receiving state,and then dynamically adjusts the parameter of neighbor discovery by Q-learning and the greedy strategy,to select a better communication channel for the directional antenna configured by the node.The simulation analysis of the algorithm based on MATLAB shows that,compared with the random channel switching algorithm,the algorithm proposed in this thesis can better alleviate node conflicts in the antenna sectors,reduce the delay of the entire neighbor discovery process,and provide a relatively stable neighbor discovery performance while the number of channels per sector changing.