An exact formulation and a hybrid heuristic method for a Hop-constrained WSN using a Delay-constrained Mobile Agent

Wireless Sensor Networks (WSNs) have emerged as an attractive and challenging research field. One of the main challenges lies in the constrained energy resources available to sensor nodes. A possible way to save energy is to use a mobile agent to collect the data moving through the WSN, but this approach increases the messages delivery delay. In this work we propose a Mixed-Integer Linear Programming (MILP) formulation to model the problem as a communication forest linked by a mobile agent route, where the roots of the trees are cluster heads sensors visited by the mobile agent; the other sensors transmit their information to the cluster heads using one or more hops. We also introduce a hybrid approach using GRASP to select the cluster heads, and constructive heuristics with local search to define the communication forest and the mobile agent route, in order to minimize the total energy consumption within a limited delay. Results are presented for WSN with up to 80 sensors with different limits for the mobile agent route length, to control the delivery delay of messages. The optimality of instances with 20 and 30 nodes were confirmed by solving the MILP formulation.