MPBC: A Mobility Prediction-Based Clustering Scheme for Ad Hoc Networks

Creating a hierarchical structure by clustering has been considered an effective method to improve the performance of ad hoc networks, such as scalability and stability. This is particularly important for networks with mobile nodes, where the mobility can cause randomly and dynamically changed network topology. In this paper, we propose a mobility prediction-based clustering (MPBC) scheme for ad hoc networks with high mobility nodes, where a node may change the associated cluster head (CH) several times during the lifetime of its connection. The proposed clustering scheme includes an initial clustering stage and a cluster maintaining stage. The Doppler shifts associated with periodically exchanged Hello packets between neighboring nodes are used to estimate their relative speeds, and the estimation results are utilized as the basic information in MPBC. In the initial clustering stage, the nodes having the smallest relative mobility in their neighborhoods are selected as the CHs. In the cluster maintaining stage, mobility prediction strategies are introduced to handle the various problems caused by node movements, such as possible association losses to current CHs and CH role changes, for extending the connection lifetime and providing more stable clusters. An analytical model is developed to find the upper and lower bounds of the average connection lifetime and to find the average association change rate of MPBC. Numerical results verify the analysis and further show that the proposed clustering scheme outperforms the existing clustering schemes in ad hoc networks with high mobility nodes.