Autoregressive moving average based anycast with support vector machine clustering in mobile ad‐hoc networks

Abstract

AbstractNodes in mobile ad‐hoc networks (MANETs) are battery‐powered, therefore, need energy‐efficient techniques to cut down the cost of energy consumption. Energy consumption is directly related to route discovery, which could be reduced through anycasting. Anycast is an operation in MANETs, where the source sends messages to one particular destination node of anycast region. In general, the message transmitted is divided into certain packets, and those are transmitted continuously one after the other through selected routes. If the link from anycast source to destination breaks in between or an anycast destination goes out of the anycast region, in that case, a new anycast destination will have to be selected on a new route to the destination. This will incur huge message costs and energy. In this work, we select desirable anycast destinations based on their eligibility by using autoregressive moving average and support vector machine. Eligibility is modeled as a function of residual energy, location, the relative velocity concerning neighbors, and the number of packets successfully forwarded earlier by those neighbors' locations. Reducing the number of anycast receivers reduces the message cost of protocols significantly. We have compared our protocol's performance with the same of anycast‐AODV, anycast‐DSR, anycast‐FAIR, and MQAR (mobility and QoS‐aware anycast routing protocol) results show significant improvement. Collective improvement amounts to 89% more energy preservation and 39% delay on average.