–
December 1, 2016
In the future large-scale wireless Internet of Things, spectrum sharing calls for interference management techniques with low-complexity, low signaling overhead and flexible scalability. A pragmatic strategy is to focus on the underlying Gaussian interference networks, exploiting the structural property of network topologies and the optimality of the simplest techniques of practical interest. In this talk, we focus on the single-antenna Gaussian interference channels, and consider the information-theoretic optimality of power control and treating interference as noise. By formulating this problem from a combinatorial optimization perspective, we are able to cast the power control problem into an assignment problem and discover the fundamental structural properties for the optimality of treating interference as noise in terms of Generalized Degrees-of-Freedom (GDoF). The insights are also translated into spectrum sharing mechanism design for device-to-device communications. We propose a low-complexity distributed link scheduling and power control mechanism (ITLinQ+) that improves upon previously proposed schemes, including the powerful heuristic approach of Qualcomm’s FlashLinQ. More notably, the energy efficiency of the newly proposed ITLinQ+ scheme is substantially larger than that of FlashLinQ.
