A multicast reinforcement learning algorithm for WDM optical networks

The popularity of multimedia applications has resulted in increased interest in efficient implementation of multicasting. Wavelength division multiplexing (WDM) is a promising technology for backbone networks, because they provide high bandwidth and methods for control and management. Since not every switches are equipped with wavelength conversion facility, it is a really important issue to design multicast trees, which result in less blocking-probability in the networks. In this paper we propose a reinforcement learning algorithm for the multicast routing and wavelength assignment (MRWA) problem. This distributed algorithm is compared with a multicast algorithm used by a well-known multicast protocol (MOSPF). The proposed algorithm is a good solution to the (MRWA) problem, since it reduces the blocking probability of the multicast communication requests.