On joint topology design and load balancing in free-space optical networks

Abstract

Free-space optical networks have emerged as a viable technology for broadband wireless backbone networks of the next generation. In this paper, we investigate the challenging problem of joint topology design and load balancing in FSO networks. We consider FSO link characteristics, cost constraints, traffic characteristics, traffic demand, and QoS requirements in the formulation, along with various objective functions including network-wide average load and delay. We apply the reformulation–linearization technique (RLT) to obtain linear programming (LP) relaxations of the original complex problem, and then incorporate the LP relaxations into a branch-and-bound framework. The proposed algorithm can produce highly competitive solutions with the performance guarantees in the form of bounded optimality gap. For reducing computation complexity, we also develop a fast heuristic algorithm to provide highly competitive solutions. The heuristic algorithm iteratively perturbs the current topology and computes network flows for the new topology, thus progressively improving the configuration and load balancing of the FSO network. The proposed algorithms are complementary to each other, since jointly applying the algorithms can make the FSO network dynamically adaptive to events occurring at both large and small timescales. The proposed algorithms are evaluated with extensive simulations. Our simulation results show that the heuristic algorithm can achieve an optimality gap close to that of the branch-and-bound algorithm, with significantly reduced computation time.