Adaptive Deployment of Directory for Service Discovery in MANET

Abstract Ensuring service discovery adaptability is a key issue for mobile ad-hoc network (MANET). The protocol should adapt to the dynamics of service discovery queries pattern and network topology. In this paper, we propose an adaptive directory deployment service discovery protocol (ADDSDP). ADDSDP decides whether to relocate the directory to an appropriate node, based on a heuristic approach, which is formulated as Markov Decision Process (MDP). We solve this problem with Q-Learning. The simulations results show that, compared with similar works, ADDSDP reduces packets cost and discovery delay, and it works well when MANET environment changes.