Noise floor dependent data fusion for reliable REM generation with a spectrum sensing grid

For several years the cognitive radio (CR) technology is under research as promising solution supporting the efficient utilization of the scarce radio frequency spectrum. The cognitive cycle enables the adaptation of operating parameters according to observations made from the environment. A radio environment map (REM) which contains data from spectrum sensing devices was identified to be an adequate tool to realize CR systems. In this contribution the challenges in generating a REM from sensor nodes with limited dynamic range is pointed out. Afterward, the noise floor dependent data fusion (NDDF) algorithm is proposed. It is able to generate a reliable REM in a fusion center of a sensing grid. The algorithm merges spectrum measurements from collocated sensor nodes to reduce the amount of data whilst preserving all advantages gained from macro diversity by taking into account the noise floor of each sensor. The algorithm has been verified with measurements in the laboratory and a measurement study at the fairground of Berlin shows the applicability of the NDDF algorithm.