Performance analysis of resource allocation schemes for cellular networks with multimedia services

Abstract

In this dissertation, we study resource allocation problems in cellular networks with multimedia services. In particular, we focus on the performance analysis of various resource allocation techniques for cellular network with multimedia services. We extend resource allocation technique proposed for voice-oriented cellular networks to one for cellular networks with multimedia services. First, we develop an efficient approximation method (to be called reduced load approximation) for performance analysis of cellular networks with multimedia services and mixed platform types. To maintain hand-off priority and access regulation, we assume that system operates under trunk reservation policy. The correlation among cells due to hand-off traffic is expressed by rate equations, and the distribution of the number of busy channels for each cell in the network is approximated by one-dimensional recursion. This leads to a set of nonlinear fixed point equations which can be solved by repeated substitution. The computational complexity and memory requirements of proposed algorithm are linear so that we can use this algorithm to estimate performance measures of the cellular mobile network even for large size. We test the proposed approximations on a number of examples. Numerical results for two cells and five cells networks are given. The comparison between approximation and simulation results shows that the proposed fixed point approximation for blocking probability is quite accurate. Second, we investigate the traffic performance of Channel Borrowing Without Locking (CBWL) schemes with multimedia services on non-homogeneous cellular network in which cut-off priority is given to handoff calls. We take into account two generic routing schemes analyzed: one is the randomized routing and the other is the least loaded routing. The performance measures are the new call blocking probabilities and the handoff failure probabilities. To evaluate blocking probabilities, we constr...