Automatic route shortening based on link quality classification in ad hoc networks

Abstract

The highly dynamic topology of an ad hoc network often causes route redundancy in the network. Several route shortening methods have been proposed to eliminate the redundancy. However, the existing works do not consider intensively the quality of the shortening links; unstable shortening links may degrade the network performance by causing unnecessary control overhead. In this paper, we seek to enhance the performance of route shortening through intensive consideration of the quality of shortening links. In our proposed SVM-ARS, all potential shortening opportunities are classified into preferred shortenings and non-preferred shortenings, and only preferred shortenings are executed in practice. The classification is achieved by Support Vector Machine (SVM). We compared SVM-ARS with a node mobility prediction model UMM and the Geographic Automatic Route Shortening (GARS) protocol. The simulations results confirm that our proposal significantly outperforms UMM and GARS through reducing the control overhead.