Analytical Modeling of IEEE 802.11-Type CSMA/CA Networks With Short Term Unfairness

Abstract

We consider single-hop topologies with saturated transmitting nodes, using carrier-sense multiple access with collision avoidance (CSMA/CA) for medium access, as standardized under the IEEE 802.11 distributed coordination function. We study systems where one or more backoff parameters of the CSMA/CA protocol (the initial backoff, the backoff multiplier, and the number of retries) are different from the standard. It is known that, for several classes of these protocol parameters, such systems exhibit a certain performance anomaly known as short term unfairness. We also find that the phenomenon of short term unfairness is observed in systems where the propagation delays among the participating nodes are not negligible compared with the duration of a backoff slot, even when the nodes use the default backoff parameters of the standard. It also turns out that the standard fixed point analysis technique (and its simple extensions) does not predict the system behavior well in such cases. For systems with large propagation delays, we observe that, as propagation delay increases, the collision probability of a node initially increases, but then flattens out, contrary to what is predicted by the standard fixed point approximation. Our study of several example systems reveals some interesting connections between the protocol parameters, the number of nodes, the propagation delay, and the degree of unfairness. This paper reveals that the inability of the standard fixed point model to capture the performance in such cases is due to its state-independent attempt rate assumption. In this paper, we develop a novel approximate, but accurate, analysis that uses state-dependent attempt rates with a parsimonious state representation for computational tractability. The analytical method is also able to quantify the extent of short term unfairness in the system, something not possible with existing analytical techniques, and can, therefore, be used to tune the protocol parameters to achieve desired throughput and fairness objectives.