Maximizing Lifetime of Mobile Actors in Wireless Sensor and Actor Networks
Ka. Selvaradjou and C. Siva Ram Murthy
Maximizing the lifetime of the network is an important problem in Wireless Sensor Networks (WSNs). The presence of special class of nodes called Actors in WSNs enable the network to have a closed loop control. Real-time reporting of events by sensor nodes and timely responses by the actor nodes are important problems in Wireless Sensor and Actor Networks (WSANs). An actor node in WSANs also acts as data sink and it is mainly characterized by its residual energy for mobility and its capability to perform a task. Optimal assignment of actors to the events is essential in order to maximize the lifetime of the actors with a guarantee that the actions are performed at right times. In this work, we consider the problem of residual energy maximization of actors which expend energy due to their mobility by optimally assigning them to the events that have real-time deadline constraints. We formulate the optimization problem as Mixed Integer Non Linear Programming. Given a set of available actors and reported events, finding such an optimal schedule of multiple actors in a centralized manner such that the reported events satisfy timing and resource constraints is an NP-Complete problem. Hence, we propose distributed algorithms based on heuristic functions to find near optimal schedule of actors that scale well in WSANs. As actor nodes are mobile, and their topology tends to be highly dynamic, we propose a suitable architecture in which nodes called agents compute the optimal schedule of actors in a distributed manner. The simulation results show that intra-zone algorithms achieve near-optimal scheduling of actors with minimal message exchange overheads. Results also reveal that the inter-zone scheduling based on Earliest Deadline First (EDF) heuristic performs fairly better than that of other heuristics with respect to minimizing the overall movement required by the actors and reducing the deadline miss ratio. Our work also investigates the impact of optimal positioning of actors at the end of their schedule so as to cover new events that might occur with stringent deadline constraints. It is observed from the simulations that proactive positioning of actors at the end of their schedule such that every zone is guaranteed to have at least one actor, performs better both in terms of increased lifetime and controlled deadline miss ratio.
Keywords: Wireless sensor networks, energy efficient algorithms, scheduling, optimization.