In this paper, we propose a hybrid iterated local search (ILS) heuristic, named GPP4G-ILS, to solve the global planning problem of survivable wireless networks. The planning problem of wireless networks is to determine a set of sites among potential sites to install the various network devices in order to cover a given geographical area. It should also make the connections between the devices in accordance with well-defined constraints. The global planning consists in solving this problem without dividing it into several subproblems. The objective is to minimize the cost of the network while maximizing its survivability. The GPP4G-ILS algorithm is a new form of hybridization between the ILS algorithm and the integer linear programing (ILP) method. We propose a configuration that allows to reuse a previously developed ILP algorithm by integrating it in the ILS algorithm. This allows to benefit from the advantages of both methods. The ILS algorithm is used to effectively explore the search space, while the ILP algorithm is used to intensify the solutions obtained. The performance of the algorithm was evaluated using an exact method that generates optimal solutions for small instances. For larger instances, lower bounds have been calculated using a relaxation of the problem. The results show that the proposed algorithm is able to reach solutions that are, on average, within 0.06% of the optimal solutions and 2.43% from the lower bounds for the instances that cannot be solved optimally, within a reduced computation time.
Index Terms-Fourth-generation (4G) networks design, global approach, hybrid methods, iterated local search (ILS), metaheuristics, reliability.
1063-6692 His research interests deal with the solution oflarge-scale instances of difficult (NP-hard) combinatorial problems by using metaheuristics, such as neighborhood search techniques and evolutionary algorithms, in particular memetic algorithms. The problems he studies include both academic problems (notably graph-related problems such as graph coloring, graph partitioning, graph matching, etc.) and various real-world applications, notably in the areas of telecommunication networks and software engineering.