Niched ant colony optimization with colony guides for QoS multicast routing

Abstract

Quality-of-service (QoS) multicast routing is essential to many network applications such as IPTV, Internet radio, multimedia broadcasting, and real-time telecommunication. Recently, the minimum-cost multicast tree with the delay-and-bandwidth constraint (MinC/DB) problem in QoS multicast routing has caused the most attention. As the MinC/DB problem can be reduced to the constrained Steiner tree problem which has been shown to be NP-complete, it is very unlikely that a polynomial time algorithm would exist. In this paper, we propose a niched ant colony optimization with colony guides (NACOg) algorithm to tackle the MinC/DB problem. The NACOg algorithm first deliberates a constrained tree traversal (CTT) strategy that guarantees the search to any feasible trees with respect to the QoS constraints. The proposed CTT strategy employs adaptive memory structure as contemplated in the tabu search and the strategy is more effective in constraint-handling than both the penalty-function and the produce-and-repair strategies which were broadly used in the literature. The evolutionary optimization of the NACOg algorithm is empowered by the search balance between diversification and intensification. The diversification search is practiced by the individual niche-colony with its own pheromone matrix, and the intensification search is facilitated by the learning scheme which refers to the experience obtained by the niche-colony guide and the entire-colony guide. The experimental results on 100 problem instances have shown that the proposed NACOg algorithm produces the least cost QoS multicast trees compared to those obtained by the Haghighat genetic algorithm and the well-known KPP heuristic.