Research On Load Balancing Algorithm Based On Parallel Multipath Transmission

Wireless Mesh Network (WMN), as a self-organizing multi-hop wireless network, has been one of the extensively used wireless access technology. With the rapid increment of various business data, the network performance is seriously damaged due to the load imbalance caused by local network congestion. The main problems exist in existing load balance mechanisms include:lack of routing metrics, time delay when calling the load balancing algorithms, and inefficient usage of network resources using single path routing protocols. The improvement and perfection of the load balancing algorithm are the key to enhance the network performance.Based on the analysis to the advantages and disadvantages of the existing multipath routing protocols, an improved multipath routing protocol to deal with load balancing problems in wireless mesh networks were proposed. The main contributions are as following.(1) Based on the disadvantages of AOMDV multipath routing protocol, a certain number of available paths were computed during the route discovery phase to transmit the data. And this entitles the newly proposed protocol to have the advantages of multipath concurrency and improve the network resource utilization by taking full advantages of the path diversity.(2) The Multi-Channel Multi-Radio were configured, and a new routing metric is designed to reflect the load status, which is based on the inter-path interference model by considering the factors of inter-link interference and load.(3) Lowest Common Multiple (LCM) is used to get the weight to distribute the load which should be distributed on each path. The ratio of LCM of the load on multiple paths to the current load of each path is calculated as the weighted value. A modified load balancing algorithm based on Parallel Multipath Transmission named LB_PMT is proposed. After applying this algorithm in congested circumstances, the network congestion is alleviated.Through extensive simulations, the proposed algorithm in this paper is compared with AODV and AOMDV algorithms. According to our analysis, our algorithm achieves 62.8% and 37.9% improvements in throughput, compared with AODV and AOMDV, respectively. The end-to-end delay, compared with AODV and AOMDV, decreases 18.9% and 11.2%, respectively. And the packet drop ratio drops 37.9% and 25.1%, respectively.