Real-time cyclostationary analysis for cognitive radio via software defined radio

Cyclostationary analysis such as spectral correlation function (SCF) and spectral coherence function (SOF) has been accepted as an important tool in signal detection and radio frequency (RF) parameter estimation in cognitive radios. However, cyclostationary analysis requires extremely high resolution to fully observe the cyclic frequency features, leading to very high computational complexity and difficulty in real time implementation, especially on software defined radios where computational capability is limited. In this paper, we implement and demonstrate a real time SCF/SOF RF signal analyzer by adopting previously proposed two-stage dynamic resolution SCF/SOF estimation method using software defined radios. Specifically, high resolution SCF/SOF calculation is only performed near the expected cyclic feature locations obtained through a low complexity coarse estimation employing spectral analysis. This cyclostationary analysis based RF signal analyzer is capable of sensing the existence of primary users and estimating the carrier frequency and symbol rate accurately. Using Universal Software Radio Peripheral (USRP) software defined radio platform and GNU radio software, we implement and demonstrate such a cyclostationary analysis based RF signal analyzer in real time and validate the performance in realistic wireless communication channels.