Experiments on spectrum sensing algorithms of pilot-added OFDM signals with a cognitive LTE-A system

The reliability of spectrum sensing is a challenging issue in cognitive radio (CR) systems. In this paper, we validate the reliability of two spectrum sensing algorithms for pilot-added OFDM signals: time-domain symbol cross-correlation (TDSC) and periodical peaks of autocorrelation (PPA) with a real system in real environments. To validate, these two algorithms carry out detection function for real signals captured by a test-bed of cognitive Long Term Evolution Advanced (LTE-A) systems. Moreover, we use a transmission between a vector generator and a spectrum analyzer to cross-check with results by the test-bed. The experimental results agree with each other and with simulated results in previous works. Two algorithms work well in real-environments and are insensitive to noise-uncertainty. The results show that PPA algorithms outperform TDSC algorithms by 1 dB - 2.5 dB with the observation durations in experiments. Additionally, PPA algorithms are suitable for short observations. For example, PPA algorithms can work with a 5 ms duration of 8K mode Digital Video Broadcasting Terrestrial (DVB-T) signals, but TDSC algorithms cannot. The results also show the performance of TDSC and PPA algorithms by a test-bed of cognitive LTE-A systems. They give clues to apply suitable algorithms for different operations such as in-band and out-band sensing modes in cognitive cellular systems.