The effect of node connectivity control on the throughput of distributed MAC networks

Abstract

We investigate the impact on the overall throughput of ad-hoc networks using two node connectivity control schemes. We propose a varying power constant connectivity (VPCC) scheme, where each node in a network regulates its transmission power until the number of its neighbors (node connectivity) is maintained at some desired number, whose optimal value is a design parameter to be determined. The achievable throughput is compared with the commonly adopted constant power varying connectivity (CPVC) scheme, where each node transmits at a constant power and hence has fixed transmission coverage but the connectivity changes from node to node. We derive the theoretical probability mass function of the number of neighbors of a given node for the CPVC scheme, and the probability density function of the coverage radius for the VPCC scheme. Simulation results are made to study the achievable throughput of the two schemes under the Poisson packet arrival assumption and when carriersense multiple access with collision avoidance and exponential back-off timer are used. The results from the simulation show that even if VPCC may result in a higher average number of hops per transmission under the same traffic condition, it can still achieve a much higher throughput than CPVC scheme. To have a better insight, we postulate and verify via simulations that the overall system throughput depends only on a parameter, which is equal to the frequency reuse factor divided by the average number of hops regardless of the schemes and the nodal density. This relationship provides a useful guide to designers, in which the parameter can be computed to predict the throughput before the most suitable scheme is selected. © 2008 IEEE.