An efficient distributed QoS-based multicast routing algorithm

This paper proposes an efficient distributed multicast routing algorithm that provides quality of service (QoS) guarantees for real-time applications. It focuses on the delay-constrained minimum cost tree (or constrained Steiner tree) problem; that is, it constructs a multicast tree that not only meets the constrained end-to-end delay requirements but also is the minimum cost multicast tree. This problem has been proved to be NP-complete. Several distributed multicast routing algorithms that are based on heuristics have been proposed in the literature. They can be classified either as a minimum spanning tree (MST) heuristic or as a shortest path (SP) heuristic. Two representative algorithms DKPP and DSPH are MST and SP based heuristics, respectively. The MST based algorithms like DKPP run with a high message and time complexity of O(n(3)). Furthermore, the MST based algorithms have a very low success rate in constructing a constrained multicast tree especially under tight delay constraints. On the other hand, the SP based algorithms like DSPH have a fatal deficiency; that is, these algorithms will not be able to find a solution if the delay constraint is less than the maximum delay of a cost based shortest path tree. Both types of algorithms try to construct the multicast tree sequentially without taking advantage of the concurrency in a distributed algorithm This paper proposes an efficient distributed multicast routing algorithm which builds the multicast tree in a concurrent manner rather than in a sequential manner. As a result, the proposed algorithm runs with a very low message and time complexity and has a near zero failure rate in constructing a multicast tree even under tight delay constraints without any limitation on the value of delay constraints.