Estimation-based variable neighborhood search approach for a probabilistic routing problem with deadlines in wireless sensor and actor networks

Mobile actors in a wireless sensor and actor network can patrol along different routes and execute actions according to the random events detected by sensors. It is important to guide the mobile actors to move in an energy-efficient way and to timely respond to the events captured by sensors. This gives rise to a probabilistic multiple salesmen problem with deadlines. In this paper, we present a variable neighborhood search to solve the problem, where the cost change in local search is estimated empirically based on sample average approximation. Numerical experiment results demonstrate that the approach can solve large instances of this problem quickly.