Towards a More Accurate Carrier Sensing Model for CSMA Wireless Networks

TL;DR

This paper introduces a probabilistic carrier sensing model for CSMA wireless networks, challenging the traditional binary approach, based on experimental observations of partial sensing relationships that vary over time.

Contribution

It proposes a more realistic, probabilistic carrier sensing model to replace the traditional binary model, supported by detailed experimental evidence.

Findings

Carrier sensing relationships are often probabilistic rather than binary.

Partial sensing occurs over a wide range of distances.

The new model impacts theoretical analysis of wireless networks.

Abstract

This work calls into question a substantial body of past work on CSMA wireless networks. In the majority of studies on CSMA wireless networks, a contention graph is used to model the carrier sensing relationships (CS) among links. This is a 0-1 model in which two links can either sense each other completely or not. In real experiments, we observed that this is generally not the case: the CS relationship between the links are often probabilistic and can vary dynamically over time. This is the case even if the distance between the links is fixed and there is no drastic change in the environment. Furthermore, this partial carrier sensing relationship is prevalent and occurs over a wide range of distances between the links. This observation is not consistent with the 0-1 contention graph and implies that many results and conclusions drawn from previous theoretical studies need to be re-examined. This paper establishes a more accurate CS model with the objective of laying down a foundation for future theoretical studies that reflect reality. Towards that end, we set up detailed experiments to investigate the partial carrier sensing phenomenon. We discuss the implications and the use of our partial carrier sensing model in network analysis.