New Kick Operators in Iterated Local Search Based Metaheuristic for Solving the Node Placement Problem in Multihop Networks

We consider a problem of finding an optimal node placement that minimizes the amount of traffic by reducing the weighted hop distances in multihop networks. The problem is called Node Placement Problem (NPP) and is one of the most important issues in multihop networks. NPP is known to be NP-hard. Therefore, several heuristic and metaheuristic algorithms have been proposed for optimizing NPP. Recently we proposed Iterated k-swap Local Search (IKLS) algorithm, which showed better performance than previous metaheuristic algorithms proposed by other researchers. IKLS simply consists of k-swap local search and a kick (mutation or perturbation) operator called Cross-Kick, a method that aims to escape from local optima. In this paper we focus on the kick operators in order to improve the performance of IKLS for NPP. New kick operators are presented and their effectivities are shown through computational experiments on the benchmark instances of NPP. The results show that IV-Kick with Rhombus is more effective than Cross-Kick and other kick operators, particularly for large-scaled instances.