Wideband cognitive sensing from a few bits

University Of Minnesota-Twin Cities, Minneapolis MN

Investigators

Abstract

Objective: Spectrum sensing is the most important element of cognitive radio, as it forms the ba- sis for adaptive spectrum sharing. Wideband spectrum sensing without frequency sweeping or Nyquist-rate sampling is a challenging problem, especially for network sensing from low-end sen- sors with limited communication capabilities. Assume that each sensor can only down-convert, filter, and send out a binary signal when it measures energy above threshold at its filter?s output. Is it possible to form a satisfactory estimate of the ambient power spectrum using just a few such bits? This is the question at the core of this project, and exciting preliminary results suggest that the answer is on the affirmative. A well-rounded research program has been designed to build upon these preliminary findings. The research will evolve along four thrusts: i) core system design and performance analysis; ii) distributed implementation; iii) relevant extensions, such as adaptive sensing; and iv) software radio experiments. Intellectual merit: Power spectrum estimation from few wideband energy detection bits has never been considered - yet is a fundamental extension of spectral analysis. Exciting new applications of convex optimization theory and methods will be found in the area of network-based spectrum estimation. Broader impacts: Wideband spectrum sensing using cheap sensors and few bits is a transformative idea that has the potential to permeate the benefits of cognitive radio down to commodity hardware - something currently far from reality. The topic is highly conducive for involving students and training the next generation of wireless engineers.