D-BigBand: sensing GHZ-wide non-sparse spectrum on commodity radios

Abstract

This paper presents D-BigBand, a system that senses a GHz-wide spectrum in real time using ADCs sampling at only tens of MS/s speed. It is an advanced version of our previous work, BigBand, which senses GHz-wide sparse spectrum using commodity radios. However, unlike BigBand that requires the spectrum to be sparse, D-BigBand does not assume sparsity utilization of the spectrum. This is particularly important for the spectrum under 2GHz which is typically not sparse, crowded with different wireless technologies like TV broadcasting, mobile, AM radios, etc. The key idea of D-BigBand is to run sparse recovery on the changes of the spectrum: since only a small fraction of the spectrum is likely to change its occupancy over short intervals of a few milliseconds, the changes of the spectrum is sparse and we can apply sparse recovery to it. Our evaluation shows that D-BigBand works even if 95% of the spectrum is occupied.