Abstract: The rapid proliferation of wireless devices has led to an increased usage of Wi-Fi. But at the same time, the issue of interference in highly loaded scenarios cannot be neglected. Interference is a major cause of degradation of capacity and thus performance in 802.11 wireless networks. The knowledge, of which links in the network interfere with one another, and to what extent, is important to improve, or even to estimate the performance of these networks. This paper presents a technique to estimate the interference in Wi-Fi networks with the help of hidden Markov model. Wireless traffic traces are captured through sniffer and analysed using a machine learning approach to conclude about the carrier-sense relationship between network nodes. To add to it, an estimation of deferral probabilities helps to understand the interference relationships. The effectiveness of the technique is evaluated using ns2 simulation which shows that this method expresses interference relations with the help of metrics such as probability of deferral, packet delivery ratio etc. in a better manner.

Keywords: 802.11 protocol, interference, hidden Markov model, carrier sense.