Variable depth search and iterated local search for the node placement problem in multihop WDM lightwave networks

Abstract

Abstract-We address a problem of finding an optimal node placement that minimizes the amount of traffics by reducing the weighted hop distances in multihop lightwave networks. The problem is called Node Placement Problem (NPP). NPP is known to be NP-hard and one of the most important problems in wavelength division multiplexing (WDM) based networks.In this paper we propose a new local search algorithm for the NPP based on variable depth search, and show its extension to an iterated local search algorithm. To evaluate the performance of the proposed methods, we provided the benchmark instances with known optimal solutions, and performed extensive experiments on the instances. The computational results showed that our iterated local search outperformed multistart local search methods and the best available metaheuristic for the problem.

An iterated greedy algorithm for the node placement problem in bidirectional Manhattan street networks

“…To evaluate the performance of IGKLS, we compare the performance of IGKLS with that of IKLS proposed by Katayama, et al [5]. IKLS is currently the best metaheuristic algorithm for the node placement problem.…”

“…IKLS is currently the best metaheuristic algorithm for the node placement problem. For the comparisons, the benchmark instances provided by Katayama [5] In the benchmark set, the traffic matrices are generated by assigning two types of traffic flow, heavy traffic flow,tH , and light traffic flow, tL. tH and tL are set to 1 and 0 respectively.…”

“…By adding this termination condition, the efficiency of KLS increases without large loss of solution qualities. The details of KLS is described in [5]. …”

“…They showed that HIWAS obtains better solutions with shorter running times than the tabu search, but there is a restriction that the size of BMSN is fixed by 2 m ×2 m (where m is a positive integer). Katayama, et al proposed a new local search called variable k-swap local search (KLS) [5], based on variable depth search. In KLS, the neighborhood is defined as sequences of the singleswap moves and the length of the sequence is adaptively decided.…”

“…These phases are applied in [14], and the performance is very competitive to the other metaheuristics. In the proposed method, variable k-swap local search (KLS) by Katayama, et al [5] is applied after the construction phase as a local search algorithm. The computational results showed that our IG algorithm with KLS, called IGKLS, outperformed IKLS which is the best performing algorithm for this problem.…”