A Path Relinking Approach for Delay-Constrained Least-Cost Multicast Routing Problem

A Variable Neighborhood Descent Search Algorithm for Delay-Constrained Least-Cost Multicast Routing

“…Haghighat et al [20] 808.406 Skorin-Kapov and Kos [22] 897.578 TS-based Youssef et al [21] 854.839 Algorithms Wang et al [23] 1214.75 Ghaboosi and Haghighat [24] 739.095 Path relinking Ghaboosi and Haghighat [26] 691.434 VNDMR1 680.067 VNS Algorithms VNDMR2…”

“…This is due to that the time complexity of VNDMR1 is O(l 2 nlog 2 (n)), while BSMA's time complexity is O(kn 3 log(n)) (n: the number of nodes, l: the number of edges, k: the k th shortest path between source and a destination). In [26], Ghaboosi and Haghighat develop a path relinking algorithm and show that it outperforms a number of existing algorithms including KPP, BSMA, GAbased algorithms [19,20], tabu search based algorithms [23,24,22,21] and another path relinking algorithm [38]. In order to compare our VNDMR algorithms with these algorithms in the literature, we generate a group of random graphs with different network sizes (10,20,30,40, 50, 60, 70, 80, 90, 100 nodes).…”

“…In order to compare our VNDMR algorithms with these algorithms in the literature, we generate a group of random graphs with different network sizes (10,20,30,40, 50, 60, 70, 80, 90, 100 nodes). For a fair comparison, three random topologies are generated for each network size, which are the same as the simulations designed in [26]. In these graphs, the link cost depends on the link length, all the link delays are set to 1, the group size is set to 30% of the network size, the delay bounds are set to different values depending on the network sizes (Δ = 7 for network size 10-30, Δ = 8 for network size 40-60, Δ = 10 for network size 70-80 and Δ = 12 for network size 90-100).…”

“…In recent years, metaheuristic algorithms such as simulated annealing [17,18], genetic algorithm [19,20], tabu search [21,22,23,24], GRASP [25] and path relinking [26] have been investigated for various multicast routing problems. In the tabu search algorithm in [24], initial solutions are generated based on Dijkstra's algorithm.…”

“…A modified Prim's algorithm iteratively refines the initial solution by switching edges chosen from a backup path set. In the path relinking algorithm in [26], pairs of solutions in a reference are iteratively improved. A repair procedure is used to repair any infeasible solution.…”