Wideband spectrum sensing techniques for wireless sensors

Abstract

The limited availability of radio frequency spectrum demands for more efficient ways to utilize it in future wireless networks. Spectrum sharing radios are an interesting solution to the spectral scarcity problem, where the available resources are adaptively used across time and frequency without affecting other user's transmissions. In this context, sensing the spectrum for its occupancy is needed to increase the awareness among technologies that share the same spectrum. In a typical wireless sensor network, each node senses and transmits data constrained by a very low power budget. At the same time, they should be capable of finding a free frequency channel with minimal latency. A solution to this problem is to make radios capable of sensing multiple frequency bands, in the order of a few hundred MHz, all at once. The technical challenge lies in the design of low-complexity wideband spectrum sensing techniques that increase context awareness at the wireless node. In this thesis, we address this problem with two approaches. The first approach is based on Compressed Sampling (CS) theory, where a new perspective is taken, different to conventional methods that estimate the spectrum and perform detection on the reconstructed spectrum. Instead a direct detection is performed on the sub-Nyquist rate sampled wideband signal. In the second part of this thesis, an alternative approach to reduce the power at an architectural level is proposed, by avoiding the Nyquist rate wideband Analog-to-Digital Converter (ADC) and pushing the conventional digital processing to the analog domain.