We present a distributed dynamic reservation (DDR) scheme that supports mobility in wireless multimedia communications, noting that low handover blocking and high channel utilization are possible if channel reservation can be done dynamically. As the traditional reservation schemes do not fit to handling time-varying multiclass multimedia traffic due to the involved computational complexity, we employ an elaborate two-regional approximation scheme that can reduce the computation dramatically. We approximate the channel occupancy distribution based on the observation of arrival rates, means, and variances of total calls and handover calls, which can be easily measured locally at each base station or switch in a distributed manner. The approximation is made by arranging the distribution into two regions such that a simple distribution model can be applied in each region and then joining the two regions using legitimate boundary conditions. This approximation enables us to estimate the relevant number of reservation channels very quickly, with the computational complexity reduced to the order of O(log C) for the channel capacity C. Nevertheless, the estimation turns out very close to the exact solution determined by applying the multidimensional Markov chain approach. Simulation results reveal that the proposed DDR scheme can adapt itself well to time-varying multiclass multimedia traffic and achieve high channel utilization, yet maintain a very low handover blocking probability.

We present distributed dynamic reservation (DDR) scheme that supports mobility in the wireless multimedia communications. As the traditional reservation schemes do not fit to handling time-varying multi-class multimedia traffic due to the involved computational complexity, we employ an elaborate two-regional approximation scheme that can reduce the computation dramatically. We approximate the channel occupancy distribution based on the observation of arrival rates, means, and variances of total calls and handover calls, which can be easily measured at each base station or node in a distributed manner. This approximation enables us to estimate the relevant number of reservation channels very fast, with the computational complexity reduced to the order of O(logC) for the channel capacity C. Nevertheless, the estimation turns out very close to the exact solution determined by applying the multi-dimensional Markov chain approach. Simulation results reveal that the proposed DDR scheme can adapt itself well to time-varying multi-class multimedia traffic and achieve high channel utilization, yet maintaining a very low handover blocking probability.