Listen-Before-Talk Versus Treating Interference as Noise for Spectrum Sharing

The spectrum sharing problem between heterogeneous networks that are not interoperable is considered. Two strategies for interference management are studied. First, by treating the possible interference from adjacent transmitters as noise, each transmitter can achieve a certain information rate which merely depends upon the channel quality, but does not depend upon the burstiness of packets arrival. Second, in an ideal listen-before-talk (LBT) strategy, where perfect sensing between transmitters is assumed, the stable achievable rates of the pairs of transceivers are analyzed, and shown to exhibit interaction and dependence upon the bursty arrivals of packets. It is revealed that, for receivers in certain regions which are exposed to strong interference and low traffic burstiness, LBT performs better; while in the other regions with decreased interference level or higher traffic loads, treating interference as noise leads to better performance.