This paper addresses the problem of accurately modeling the spectrum occupancy patterns of real radio communication systems, which is an essential aspect in the study of cognitive radio (CR) networks. The main drawbacks and limitations of previous works are identified, and the methodological procedures on which they rely are improved and extended. Two sophisticated measurement platforms, providing low and high time resolutions, are used to obtain extensive real-world data from a multiband spectrum measurement campaign, embracing a wide variety of spectrum bands of practical interest for CR applications. A comprehensive, systematical, and rigorous analysis of the statistical properties observed in the measurement data is then performed to find accurate models capable of capturing and reproducing, within reasonable complexity limits, the statistical properties of temporal patterns, at both short and long timescales, in real wireless systems. Innovative modeling approaches capable of simultaneously describing statistical properties at both timescales are developed as well. In summary, this paper contributes realistic and accurate time-dimension spectrum usage models for their application to the study and development of CR

Peer Reviewed