A Cross-Layer Routing Metric with Link Prediction in Wireless Mesh Networks

Abstract

Abstract Wireless Mesh Networks (WMNs) is one of the encouraging technologies for future-generation wireless network with diverse applications as it is undergoing rapid progress. WMNs are multi-hop networks which are beneficial in situations where there is little or no network infrastructure. WMNs consist of mesh nodes which interact with each other to establish a network connection. Mesh nodes include mesh clients and mesh routers. Mesh clients can be stationary or move from one place to another, whereas mesh routers have minimum movability and act as a backbone for WMNs. Due to the mobility of mesh nodes, the fundamental problem in WMNs is estimating the link quality. The efficiency of network protocols depends on the accuracy of link quality. Therefore, there is a need for an intelligent mechanism to find solutions to link quality problems in the wireless network. To address this problem, we apply machine learning prediction techniques. We evaluate the performance of prediction techniques like multiple linear regression, support vector regression and Gaussian regression. The experiments were conducted on various scenarios with cross-layer routing metric PCL-IDA to predict the link quality, which is embedded in OLSR routing protocol. Results using NS-2 simulations reveal that the link quality parameters throughput, average delay and packet loss are better when multiple linear regression prediction technique is applied.