Model-based Service Selection for Reliable Service Access in Manet

Due to the dynamic infrastructure in Mobile Ad-hoc Networks (MANET), supporting a reliable and timely service access is one of the main requirements of applications deployed over MANETs. In this paper, a reliability-aware service selection scheme is proposed in which a client node collects and ranks the available service providers based on their estimated response time and reliability. However, estimating the reliability of a service provider in MANET is a challenging issue because it is strongly dependent on the current conditions of the network (dynamic or stable). To estimate the network state based on the local observations, Hidden Markov Model (HMM) is applied prior to the service selection.. The experimental results show that by using our method, not only the unreliable service providers are filtered but also, comparing to the existing service selection methods, lower violations of the client response time requirements (defined in the client SLA) are observed.