Distributed QoS-sensitive band selection strategies for dynamic spectrum access in unlicensed CDMA networks

Recent studies indicate the presence of a significant amount of idle licensed spectrum, in different time periods and geographic locations. Fixed spectrum allocation lacks efficient mechanisms for sharing spectral resources. Prompted by the recent regulatory changes and radio technology advances, opportunistic use of spectrum bands is becoming increasingly attractive, as it can alleviate the spectrum scarcity problem in the unlicensed bands. We study the dynamic spectrum access for a CDMA wireless environment, where unlicensed users run elastic data applications, i.e., throughput has to be within a range of acceptable values. Unlicensed users may transmit in idle licensed bands (i.e., not used by a licensed user). We focus on the band selection problem of unlicensed users. We propose a prediction-driven and a probabilistic, distributed strategy with which users might efficiently share the available resources. The assessment of the proposed strategies shows that they achieve fast convergence and the attained equilibria are near optimal.