Rate Allocation for Simultaneous Multi-Class Service Transmission Over Dynamic Spectrum Access-Based Networks

In this paper, we study rate allocation mechanisms that allocate available transmission rate of a particular secondary user (SU) among its different classes of services assuming that a specific channel assignment algorithm across active SUs is in place. In particular, we formulate the rate allocation mechanism of a SU between its two different classes of services namely, delay sensitive (DS) and best effort (BE) services as a Markov decision process. Then the optimal rate allocation mechanism that minimizes the average queuing delay of DS packets while guaranteeing the packet loss probabilities of both classes of packets is obtained using a linear programming technique. Since the optimal rate allocation mechanism can be complex for implementation in a SU device due to its probabilistic nature, we also study a low-complexity suboptimal rate allocation mechanism. For this suboptimal scheme, we develop a queuing analytic model to measure and compare different data-link layer performance parameters e.g., packet loss probability and queuing delay distribution. The developed analytic model for the suboptimal scheme is further useful for a call admission decision. We compare the performances of the suboptimal rate allocation mechanism with its optimal counterpart. Presented simulation results show that the suboptimal mechanism has a quite similar performance as the optimal mechanism.