A Prediction-Based Solution for Improving the Performance of CSMA/CA Networks Under the Hidden Collision Effect
Marouane Sebgui, Slimane Bah, Belhaj Elgraini · International Review on Computers and Software (IRECOS) · 2015
In Dynamic Spectrum Access (DSA) policy, a secondary user (SU) is allowed to access the primary user (PU) channel when it is idle. This access has to be disruption free for the PU. Literature on DSA focuses on PU and SU signal interferences as the main source of disruption. However the idle state may result from a non-transmitting activity such as the case where the primary user is in the Backoff Window in CSMA/CA networks. This particular idle state can be wrongly seen as an opportunity for SU to access the channel. However, accessing the channel in this context will cause Hidden Collision (HC) and will decrease the primary user performance. In this paper, we extend our previous work on HC and propose to distinguish the PU behaviour, modeled as a three states process, from the channel evolution modeled as an ON/OFF process. To combine these two interdependent processes we use the Hidden Markov Model (HMM) and we propose a solution to predict the PU state and therefore reduce the hidden collision effect. As a proof of concept, our approach accuracy is evaluated using a set of inputs obtained from simulation. We demonstrate the ability of the model to predict accurately enough the PU state and reduce hidden collision.