A new modelling approach to represent the DCF mechanism of the CSMA/CA protocol

In this paper, a Markovian agent model is used to represent the behavior of wireless nodes based on CSMA/CA access method. This kind of network was usually modeled by means of bidimensional Markov Chains and more recently using semi-Markov process based models. Both these approaches are based on the assumptions of both full load network and independence of collision probability with respect to retransmission count of each packet. Our model inherently releases the latter hypothesis since it is not necessary to establish a constant collision probability at steady state. Here, we investigate the correctness of our approach analyzing the throughput of a network based on two IEEE 802.11g nodes when the amount of traffic sent by each one varies. Results have been compared with Omnet++ simulations and show the validity of the proposed model. © Springer International Publishing AG 2017.