Abstract
Multimedia applications are expected to guarantee end-to-end quality of service (QoS) and are characterized by stringent constraints on delay, delay-jitter, bandwidth, cost, and so forth. The authors observe that Kruskal's algorithm is limited to minimal (maximal) spanning unconstrained tree. As such, the authors extend Kruskal's algorithm to incorporate the delay bound constraint. Consequently, a novel algorithm is proposed, called EKRUS (Extended Kruskal), for constructing multicast trees. The EKRUS' distinguishing features consists of a better management of Kruskal's priority queues, and in the provision of edge priority aggregation. Preliminary results show that the proposed EKRUS algorithm performs as well as the best-known algorithms (such as the DDMC, DMCTc algorithms) while exhibiting reduced complexity. The authors conducted an intensive analysis and evaluations of different strategies of assigning edges into the classes of the queue as well as edge selection. As a result, the EKRUS algorithm was further extended with different edge assignment and selection strategies. Through extensive simulations, the authors have evaluated various versions of the EKRUS and analyzed their performance under different load conditions.