Server Optimization Using Heuristic Algorithms for Dynamic-Split-and-Merge Scheme in Wireless Multicast

Abstract

Problem statement: In order to minimize the overall network traffic in a multiserver system, the number of users served by each server (and hence the group size) should remain constant. As the underlying traffic fluctuates, a split and merge scheme is implemented in a physical server to achieve load balancing. Approach: Minimizing the number of servers during the merge operation is NP hard and to achieve these two algorithms namely FFD bin packing algorithm and LL algorithm were proposed to find the near optimal values of destination servers. Results: The performance of these algorithms were analyzed and compared based on several parameters. Conclusion: Results showed that LL algorithm outperforms FFD algorithm.