Markov State Space Analysis of IEEE Standard MAC Protocols

In recent years, the standardized Media Access Control (MAC) protocol plays an important role in wireless local area networks (WLANs) and wireless sensor networks (WSNs). The IEEE 802.11 protocol with distributed coordination function (DCF) is the most popular standard in WLANs that includes specifications for both MAC and physical layers, whereas the IEEE 802.15.4 PHY-MAC specifications represents a significant milestone in promoting deployment of WSNs for a variety of commercial uses. The core of the 802.11 DCF and 802.15.4 MAC protocols is the Carrier-Sense Multiple-Access protocol with Collision Avoidance (CSMA/CA).;Although the studies of such kinds of MAC protocols have been lasted for several decades, a thorough network performance analysis of these wireless networks still cannot be tackled in the existing works.;In light of this concern, we propose a generic Markov state space model of the MAC protocols with CSMA/CA for contention resolution in this thesis. The input buffer of each node is modeled as a Geo/G/1 queue, and the service time distribution is derived from a Markov chain describing the state transitions of head-of-line (HOL) packets. This Markov model is well demonstrated by the IEEE 802.11 DCF networks in either ideal channels or imperfect channels, and IEEE 802.15.4 networks, with probabilistic exponential backoff scheduling algorithm under non-saturated condition.;With this queueing model, we obtain the steady state characteristic equation of network throughput as well as the mean packet access and queueing delays of packets. Moreover, for the IEEE 802.15.4 networks, the accurate expressions of energy consumptions for each node can also be obtained through this Markov model.;In this dissertation, we specify the stability conditions in terms of throughput and queueing delay for MAC networks. These two stable conditions enable us to derive two kinds of regions: the stable throughput region and the bounded delay region, which is dependent on the backoff scheduling algorithm and the aggregate input traffic. We prove that the stable regions still exist even for an infinite population with exponential backoff.;For the IEEE 802.11 DCF networks, it depicts that the network performance of RTS/CTS access scheme is less dependent on the aggregate input rate and retransmission factor than that of the Basic access mechanism. Additionally, with the comparison of the networks performance under ideal and imperfect channels, we also show that the transmission errors have a significant impact on both throughput and delay of networks. For the IEEE 802.15.4 networks, our results confirm that the energy consumption of a single node is kept small within its stable throughput region.;Last but not least, we extend our approach to the contention-window-based backoff model, and depict that the probabilistic backoff model can serve as a good analytical model for the practical contention window mechanism.