Proposal and comparative analysis of a voting-based election algorithm for managing service replication in MANETs

Novel approaches are needed to better facilitate dynamic service replication management in mobile ad-hoc networks (MANETs) and to use and apply them within current and emerging autonomous intelligent systems and the Internet of Things (IoT) paradigm. Such approaches should address the context-awareness and self-adaptation of service replication, while paying special attention to quality attributes (e.g. availability, reliability, etc.) under specific runtime changes and adverse conditions with unstable communications and network partitions. The dynamic election for a node to host a service replica in MANETs can be based on the use of leader election (LE) algorithms. In this research work, a new voting-based election algorithm for managing dynamic service replication in MANETs (namely, VOELA) is proposed. This algorithm is based upon a utility function to score node resources and features (i.e., battery level and topology position) to decide where the service replica will be activated. VOELA is compared to a previously proposed consensus-based algorithm and three other well-known leader election algorithms in terms of service availability, election algorithm reliability, coordination message usage, and network lifetime. For this comparative analysis, the ns-3 network simulator is used together with three different mobility models, namely Manhattan Grid Mobility (MGM), Random Walk Mobility (RWM) and Reference Point Group Mobility (RPGM). The VOELA algorithm demonstrates in balance the most promising results.