Performance improvement in sensing error probability for low SNR scenarios in cognitive radio networks

Cognitive radio (CR) is becoming an effective and promising technique to improve the spectrum utilization. One of the most prominent tasks in the implementation of cognitive radios in communication networks is the spectrum sensing. Various spectrum sensing algorithms are available in the literature among which energy detection technique has been gaining popularity for primary user detection mainly because of its simplicity and low complexity. However, its performance is adversely affected due to noise uncertainty particularly in low signal to noise ratio conditions. In this paper, we present a constant detection scheme and through numerical results it is shown that overall probability of error can be reduced significantly in low SNR regions as compared to conventional energy detection method. The effectiveness of the constant detection scheme is investigated comprehensively under noise uncertainty scenarios. The numerical results obtained through MatLab R.2012a have shown that sensing error probability is typically improved by 0.11 at -15 dB. It is reduced from 0.44 to 0.33, which is very promising for low SNR scenarios.