UCLA

In Dynamic Spectrum Access (DSA) networks, secondary users (SUs) scavenge for unused frequency spectrum by performing spectrum sensing. Multi-stage sensing is a novel concept that refers to a general class of DSA spectrum sensing algorithms that divide the sensing process into a number of sequential stages. The number of sensing stages and the sensing technique per stage can be used to optimize performance with respect to SU throughput, energy consumption per bit, and the probability of collision between licensed (primary) and SUs. In this work, we present the first analytical framework which enables performance evaluation of different multi-channel multi-stage spectrum sensing algorithms for DSA networks. The contribution of our work lies in studying the effect of the following parameters on performance: number of sensing stages, physical layer sensing technique and duration per stage, single and parallel channel sensing and access, number of available channels, primary user (PU) and SU traffic, as well as MAC layer sensing algorithms. Our results emphasize the fact that the performance of multi-stage sensing is superior to the single stage sensing counterpart, where the optimal number of sensing stages and sensing duration per stage depend on the network traffic.

For DSA spectrum sensing algorithms in general, and multi-stage spectrum sensing algorithms in specific, DSA systems can assign PU resources to its subscribers more efficiently if knowledge of PU temporal statistics is available. In this work, we explore the benefit of incorporating PU traffic statistics in spectrum sensing algorithms. Moreover, we study the impact of PU traffic parameter estimation errors on algorithm performance. We extend our study to investigate the accuracy bounds on PU traffic parameter estimation. We present a mathematical analysis of the accuracy of estimating the mean PU duty cycle, and the PU mean off- and on-times, where the estimation accuracy is expressed in terms of the mean squared estimation error. The analysis applies for the traffic model assuming exponentially distributed PU off- and on-times. We derive the Cramer-Rao bounds on the estimation error of the mean PU duty cycle, and the mean PU off- and on-times. The bounds are derived for the case when perfect knowledge is assumed for one of the parameters, and the case where all parameters are jointly estimated. Besides, the impact of spectrum sensing errors on the estimation accuracy is studied analytically. Furthermore, we propose a number of estimators for the traffic parameters and quantify their estimation accuracy. Finally, we develop algorithms for the blind estimation of the traffic parameters based on the derived theoretical estimation accuracy expressions. We show that, for a fixed observation window length, the estimation error for all traffic parameters is lower bounded due to the correlation between the traffic samples, while on the other hand, the impact of spectrum sensing errors on the estimation error of the mean PU duty cycle can be eliminated by increasing the number of traffic samples.