Using Multiple Power Spectrum Measurements to Sense Signals with Partial Spectral Overlap

In this work, we study the wideband sensing problem of detecting intermittently transmitting signals that may have partial spectral overlap. We aim to estimate the frequency bands occupied by transmitters using standards with overlapping frequency bands, such as IEEE 802.11, and those without guard bands, such as LTE-Advanced. Multiple power spectrum measurements are used to distinguish the distinct bands. An extreme ray based non-negative matrix factorization algorithm is proposed to identify measurements where a single transmitter is active. Distinct bands are identified in the presence of frequency-selective fading by using a combinatorial search. In addition, we propose a novel algorithm to automatically estimate the noise power spectrum by identifying measurements that do not have significant signal energy. Its ability to learn colored noise and LO leakage in the signal is demonstrated through USRP measurements. The dependence of the proposed algorithm's performance on the medium access control protocol used by the primary users is discussed. MATLAB simulations are used to verify that the proposed algorithm detects the occupied bands more accurately than existing methods. Over the air USRP measurements in the 2.4 GHz ISM band are used to detect the occupied WiFi channels in a university environment.