A Joint Optimization Design Method for Opportunistic Network Neighbor Discovery and Power Control Based on Q-Learning

Due to the sparsity of nodes in opportunistic networks, data transmission becomes difficult. Moreover, the mobility of nodes makes them unlike static nodes in traditional ad hoc networks, which can replenish energy at any time. Therefore, it is crucial to select an energy-efficient neighbor discovery algorithm. Passive listening constantly monitors channel activity, which saves energy but cannot obtain information about neighbor nodes that are not active on the channel. Active probing discovers neighbors by broadcasting probe packets, which consumes more energy and excessive probe packets in the channel can burden the network. To address the limited node energy in opportunistic networks, this paper proposes a method that combines passive and active neighbor discovery, controls the duration of passive listening, and adopts Q-learning for power control. Simulations are conducted to verify its optimization effect on neighbor discovery performance. Experimental results show that this method improves throughput to a certain extent, especially in the neighbor discovery process, with an energy saving of approximately 28 % at most.