Practical implementation of time covariance based spectrum sensing methods using WARP

Cognitive radio is a promising technology to deal with the spectrum scarcity problem. One important application of cognitive radio is spectrum reuse. To reuse a licensed spectrum, secondary users (unlicensed users) must ensure the frequency bands are free and no primary users (licensed users) exist. Thus, accurate spectrum sensing is a fundamental requirement of cognitive radio for avoiding signal collision and several methods have been proposed. In this paper, we examine the performance of three spectrum sensing methods with Wireless Open Access Research Platform (WARP) which is developed by RICE University. Considered first is the conventional energy detection (ED) method which requires perfect knowledge of noise power and is very sensitive to noise uncertainty. Many approaches have been proposed to mitigate the noise uncertainty problem. We mainly focus on two methods based on the time covariance matrix of the received signal, called covariance method (COV) and maximum to minimum eigenvalue (MME) method. COV method utilizes ratios of time correlations with different time separations. MME method computes maximum and minimum eignvalues from the sample covariance matrix. Both COV and MME methods are insensitive to noise uncertainty and do not need any prior knowledge of signal properties of licensed users. Furthermore, these two methods are not affected theoretically by noise uncertainty. To examine the three above mentioned spectrum sensing methods in low SNR environment, we implement them by adjusting the setting in physical and network layer protocols of WARP. As predicted from theory, experimental and numerical results show that COV and MME outperform ED in large margins.