Average Channel Utilization of CSMA With Geometric Distribution Under Varying Workload

Performance analysis of fixed-size contention window carrier sense multiple-access (CSMA) protocol with geometric distribution for slot selection probability is addressed in the paper. This MAC scheme, called Sift, was initially proposed for large-scale event-driven wireless sensor networks. The goal of the present paper is to evaluate the standard performance measures (throughput, protocol capacity, collision rate, and mean access delay) for geometrically distributed CSMA both in the context of data-centric dense sensor networks and node-centric industrial automation systems. The analytical approach based on the stochastic analysis has been applied. To demonstrate how the protocol is able to cope with bursty traffic, the average throughput defined over a specified workload range has been introduced and examined. Using the average throughput as the performance criterion, the geometric CSMA has been compared to conventional CSMA schemes with uniform distribution. The latter are represented by the classical p-persistent CSMA and the predictive p-persistent CSMA used in LonWorks control networks. It is shown that G-CSMA is overload-tolerant event-driven MAC protocol since the average throughput may be kept on high level in wide range of workload if the shape of geometric distribution is well chosen.