Enhanced detection methods for compressive spectrum sensing based on antieigenvalues

Abstract

Spectrum sensing is a crucial process in devices that dynamically and opportunistically attempt to access the spectrum based on its availability, including secondary users in a cognitive radio network and internet-of-things (IoT) devices with cognitive radio capabilities. Eigenvalue-based detection methods have recently gained considerable attention over classical detection techniques due to their efficiency. In this paper, we propose two new detection methods based on antieigenvalues, which have proven to be more efficient than existing eigenvalue-based methods. The first method uses the smallest antieigenvalue of the incoming signals’ covariance matrix as a test statistic, and the second method uses the slope of the antieigenvalues, whose behavior and probabilistic distribution is studied and presented in this paper. The two methods are also applied in a compressive sensing framework, where the samples are acquired at rates below the Nyquist rate. Both these methods have also proven to be more efficient than the only other existing method based on antieigenvalues, which uses the sum of the smallest P antieigenvalues — where P is the number of primary user signals — as a test statistic.