A VLSI architecture for multiple antenna eigenvalue-based spectrum sensing

An Eigenvalue-based detection (EBD) scheme, is proposed as an efficient method to overcome the noise uncertainty and the SNR wall problem in conventional energy detection (ED) schemes. Despite remarkable efforts made to analyze the EBD performance, a VLSI implementation is missing in literature. In this paper, a new FFT-based EBD algorithm is introduced, which eliminates the need for filter banks and discrete wavelet packet transform to channelize the input signal. The proposed method enables the utilization of the EBD algorithm in high-resolution spectrum sensing approaches. Moreover, it enables the detection of signals with SNRs as low as -10 dB. A low-power, area-efficient yet real-time VLSI architecture is developed for the algorithm, which is implemented in a 0.18μm CMOS technology. The implemented design occupies a total area of 3.4mm
²
and dissipates 78mW for a 40 MHz sensing bandwidth consisting of 32 sub-channels.