Modeling and characterization of frame loss process in IEEE 802.11 wireless local area networks

The popularity of 802.11-based WLANs is greatly increasing and to keep up with new throughput demands the interaction of some of the protocols (MAC and TCP) is being scrutinized. In order to properly simulate protocol behavior, accurate frame loss models are needed for 802.11 networks. This paper discusses measurements taken to help characterize the frame loss process and compares Markov models with different numbers of states whose parameters have been estimated using the Baum-Welch algorithm. It is shown that higher-order state hidden Markov models provide a better fit to measured data than the traditional 2-state Gilbert models.