PCA based Spatial Spectrum Sensing for MIMO Cognitive Radios

In the recent years spatial spectrum sensing become a promising approach due to the convergence of almost all wireless standards to incorporate spatial dimensions and use of multiple antennas at both transmitter and receiver. Keeping in consideration such wireless environment, we proposed a spectrum sensing algorithm based on principal component (PC) of spatially received signals. The proposed algorithm is analyzed under SISO (single input single output), SIMO (single input multiple output) and MIMO (multiple input multiple output) (employing stream multiplexing and Alamouti space time coding) scenario. Performance comparison was done by receiver operating curve (ROC) with other proposed algorithms in literature i. e. maximum minimum eigenvalue (MME). No prior information about the channel or primary user's (PU) signal is assumed. Simulations show the improved performance when info about spatial diversity of PU is incorporated in the proposed PCA. All the algorithms were tested using experimental data while using USRP (universal software radio peripheral) test bed that was controlled by GNU radio software.