Neural Networks Based Efficient Multiple Multicast Routing for Mobile Networks
B. Pavan Kumar, S. M. D. Kumar · 2014
Abstract—Mobile multimedia networks that handle multicast communication services require a reliable and efficient point-to-multipoint specific group communications. In mobile networks, this leads to many challenges to provide an efficient multicast routing. In this paper, an extension to our previous work is proposed. In the previous work, the construction of multicast distribution tree (MDT) for multicast routing in mobile networks was proposed. However, in this work, the technique is revised and scalability is adapted to construct an efficient multiple multicast routing in mobile networks. The technique involves the construction of MDTs for multiple multicast groups (MCG), considering the network traffic and available resources. The technique uses Hopfield Neural Network (HNN) and Kohonens Neural Network (KNN) for the construction of efficient multiple MDTs. The computational power of the proposed technique is demonstrated through simulation. The technique is tested for different group size and network topologies along with host mobility. The proposed work facilitates a possible multicast routing technique for future high speed mobile networks and on-demand multicast applications. constructing a reliable multicast tree that connects the participants of a single MCG by considering reliable nodes in a mobile network was proposed. In this work, the technique is modified and scalability is adapted to construct an efficient multiple multicast routing in mobile networks. The technique is aimed at constructing low cost MDTs for the MCGs admitted based on the traffic load on the network and the type of application run by each group. Further, the technique establishes MDTs when there is a change in location of the group members due to mobility. The technique employs two different types of neural networks (KNN and HNN) and their computational power is used as a heuristic approach to tackle the issues in multiple multicast routing. By the massive parallel computation and learning capability of neural networks, we can find a near-optimal multicast route very fast when implemented in hardware (4). The rest of the paper is organized as follows. Some of the related works are highlighted in Section II. The mobile network model is presented in Section III. The proposed multicast routing technique along with algorithms is described in Section IV. The principle of clustering and construction of MDTs by using KNN and HNN are discussed in Section V. Simulation results of the proposed technique and concluding remarks of this work are presented in Sections VI and VII respectively.