Optimal Time-Frequency Diversity Exploitation for Multichannel System under Rayleigh Fading

In multichannel system, user could keep transmitting over an instantaneous "on peak" channel by opportunistically accessing and switching among channels, so as to exploit link layer time-frequency diversity. In this paper, we consider an opportunistic channel accessing/releasing scheme for maximizing system throughput in multichannel system under Rayleigh fading environment. The time-dependence of Rayleigh fading is accurately characterized by finite-state Markov channel (FSMC) model. The main goal of this paper is to devise throughput-optimal strategy for determining when to access (which) channel and when to release it. The chanllenge of the problem comes from the fact that user can never know the instantaneous quality of all channels. In fact, it has to make real time decisions purely depending on the quality of current channel and the statistics of candidate channels. We formulate this real time decision making process as a two dimension optimal stopping problem. We prove that the complexity of the two dimensional optimal stopping rule can be reduced to a simple threshold-based policy. The dynamic data transmission process under threshold-based opportunistic channel access/release strategy is then analyzed by properly constructing absorbing Markov chain. Leveraging the absorbing Markov chain theory, we attain the optimal threshold as well as maximum achievable throughput with computational efficiency. Numerical results show that our proposed channel utilization scheme achieves up to 120% throughput gain over opportunistic transmission with a single channel and up to 70% throughput gain over opportunistic channel access with constant transmission duration.