An EMD-based Spectrum Sensing Technique for Cognitive Radio Networks

Precise noise estimation is essential in enhancing spectrum sensing techniques as estimating the wrong noise power may result in increasing the missed detection or false alarm rates. In this paper, we have created a new technique based on Empirical Mode Decomposition (EMD). Intrinsic Mode Functions (EMD output) of Gaussian noise were proven to follow a distinct characteristic, or behavior. This behavior is exploited to make an occupancy decision for a given bandwidth. Further, a non-parametric threshold is derived and used to determine how many occupied channels are presented. The proposed method exhibits a probability of detection of 96% at -12 dB. Further, a gain of 11 dB is achieved when comparing this method with other EMD-based spectrum sensing approaches.