Leveraging Service Discovery in MANETs with Mobile Directories

This paper advances a novel approach that facilitates the location of services and/or digital assets advertised by directories in a Mobile Ad hoc Network. The proposed Service Directory Placement Protocol (SDPP) improves scalability and reduces packet traffic overhead by advancing a multi-directory extension of an earlier approach that relied on the migration of a single directory through the network. This investigation demonstrates that modelling the directory replication problem as a Semi-Markov Decision Problem solved by means of a Reinforcement Learning technique known as Q-learning improves the performance of SDPP. Computer simulations validate the feasibility of the proposed scheme that enables packet overhead reductions between 15 and 75 %, whereas the directory location success rate improves by up to 11% when compared with pure broadcast flooding and other existing approaches in wireless networks where hosts move at walking speeds.