Optimal configuration for nodes in mixed cellular and mobile ad hoc networks

This dissertation presents a novel approach to combining cellular and mobile ad hoc wireless networks. These technologies have evolved separately to address different applications. Cellular networks are well known to support mobile users in a fixed geographic area with near optimum capacity using centralized infrastructure. Mobile ad hoc networks support mobile users with limited capacity over geography that is limited only by the location of the users with a self organizing network that requires no central infrastructures. We propose a mixed cellular and mobile ad hoc network (MCMN) to extend coverage and capacity to nodes in both regions. We present an architecture for MCMN and a performance analysis for the mixed network. For this architecture the problem of organizing nodes in this mixed network is a key. Optimal node organization is critical to the performance of the mixed network and to improve communication. We evaluate the configuration of nodes required to optimally organize the nodes in MCMN. The problem of evaluating configuration parameters for nodes in a mixed network is a nonlinear and complex one. This is due to the number of nodes, geographical location, signal strength, mobility, and connectivity. A clustering technique is shown to provide the best practical solution for this problem. We demonstrate a clustering scheme based on the k-means algorithm to partition the MCMN into subnets or clusters. The scheme includes the establishment of the optimal boundary between cellular and MANET, and the optimal number of cluster partitions for the network. We also develop a method of selecting gateway nodes for each of the clusters in the network. Finally, we showed that when the network is well organized, both coverage and performance can be improved.