Throughput Computation in CSMA Wireless Networks with Collision Effects

TL;DR

This paper extends the ideal CSMA network model to include collision effects caused by discrete backoff countdowns, providing a more accurate method to compute link throughputs in practical wireless networks.

Contribution

It introduces the GICN model that accounts for collision states and link interactions, improving throughput calculations for real-world CSMA protocols.

Findings

GICN accurately predicts link throughputs and collision probabilities.

Original ICN model remains fairly accurate despite ignoring collisions.

Simulation results validate the GICN model's effectiveness.

Abstract

It is known that link throughputs of CSMA wireless networks can be computed from a time-reversible Markov chain arising from an ideal CSMA network model (ICN). In particular, this model yields general closed-form equations of link throughputs. However, an idealized and important assumption made in ICN is that the backoff countdown process is in "contiuous-time" and carrier sensing is instantaneous. As a result, there is no collision in ICN. In practical CSMA protocols such as IEEE 802.11, the stations count down in "mini-timeslot" and the process is therefore a "discrete-time" process. In particular, two stations may end their backoff process in the same mini-timeslot and then transmit simultaneously, resulting in a packet collision. This paper is an attempt to study how to compute link throughputs after taking such backoff collision effects into account. We propose a generalized ideal CSMA network model (GICN) to characterize the collision states as well as the interactions and dependency among links in the network. We show that link throughputs and collision probability can be computed from GICN. Simulation results validate GICN's accuracy. Interestingly, we also find that the original ICN model yields fairly accurate results despite the fact that collisions are not modeled.