Robust Spectrum Sensing of OFDM signal without noise variance knowledge

Spectrum sensing is one of the fundamental enabling technologies for cognitive radio (CR) communications. Motivated by the infeasibility of using noise variance dependent signal sensing threshold for CR users, a robust spectrum sensing scheme using a proposed noise parameter cancelation (NPC) method for orthogonal frequency division multiplexing (OFDM) signals is investigated in this paper. Identification of the primary OFDM signal is achieved with the frequency domain correlations between the frequency shifted received signal and the local pilot reference. The impact of noise on the spectrum sensing reliability is significantly reduced with the proposed noise parameter canceled test statistic, which is defined as the ratio of correlation peak when the received signal has no frequency offset with the local pilot sequence, and averaged correlation functions with various frequency offset.