The binary exponential back-off mechanism is one of the basic elements that constitute the IEEE 802.11 protocol. The models of the back-off mechanism have been developed with the assumption that collisions occur only due to nodes within the carrier sensing range and the collision probability is constant in steady-state. However, the transmission collisions can occur due to hidden nodes and these tend to occur consecutively, contrary to the collisions due to nodes within the carrier sensing range. Consecutive collisions increase the back-off time exponentially, resulting in less frequent transmission attempts. Ignoring this collision characteristic in modeling the back-off mechanism can produce large errors in the performance analysis of networks. In this paper, we model the back-off process as a Markov renewal process by taking into account such consecutive collisions due to hidden nodes, and then compare this result with NS2 simulation results. According to the simulation results, the proposed model reduces the relative error in the attempt probability by more than 90% in the grid topology. We also propose a new collision model for a simple network considering consecutive collisions due to hidden nodes, and analyze the network under saturated traffic condition using the proposed models. The attempt and collision probabilities are estimated with high accuracy.

The channel characteristics of IEEE 802.11 WLAN vary with time and this can affect packet transmission performance. For achieving robust and efficient transmission, the transmission rate is controlled by exploiting the multi-rate capability of the IEEE 802.11 physical layer (PHY) to respond to the time-varying channel condition. In this paper, we propose a novel rate adaptation scheme, called RA-MCE, in which the transmitter estimates channel quality in the MAC layer to enhance throughput performance without the need to use the RTS-CTS mechanism nor to modify the IEEE 802.11 standard. RA-MCE adaptively controls the transmission rate according to the estimated channel quality by the MAC layer channel quality estimator (MCE) that uses only local MAC layer measurements. Through extensive simulations, we validate the accuracy of MCE and evaluate the performance of RA-MCE to show that it achieves higher throughput performance than other rate adaptation schemes under various circumstances.