Real-time blind spectrum sensing using USRP

Efficient spectrum usage is an issue that has been inviting a lot of interest and research in recent times, due to the omnipresence of wireless technologies all around us. Spectrum sensing is a key step towards efficient spectrum usage. Energy detection is a fast and simple method for spectrum sensing, but the sensing precision is limited by the dependency on a threshold value. This paper reports a novel real time energy detection based spectrum sensing technique using a logistic regression classifier. The implementation is done using USRP and GNU-Radio, and achieves a classification accuracy of 98.6% on a dataset that was collected over commercial FM band.