Graduate School and Research Center in Digital Sciences

Seminar: A lattice framework for the Gaussian interference channel

Jean-Claude BELFIORE - TelecomParisTech

Communication systems

Date: December 18, 2013

As wireless networks become denser and denser, they are more and more dominated by cochannel interference. Traditional radio resource management solutions such as power control or resource allocation (e.g. through OFDMA) are not enough efficient to cope with the growing increase of the demand. However, in order to mitigate interference, many interesting new techniques have been proposed. Some of them, such as massive MIMO or Network MIMO rely on a very high rate backhaul. When the backhaul has to be reduced or when it does not exist (such as ad hoc networks, for instance), other techniques have to be considered. Interference alignment is one of them. It includes a wide variety of techniques. The most popular one is linear interference alignment which suffers from the idealized assumptions that have to be done, such as global channel knowledge, bandwidth expansion, high signal strengths, etc ... In this presentation, we focus on a different type of interference alignment based on lattices. We first present the use of lattice codes to solve the problem of the 2-user interference channel through the famous "Compute-and-Forward". Then, we present some results concerning many interferers. All these results are mainly based on how good a Diophantine approximation can be. Some insights about how much information has to be sent to the transmitters will also be given.

A lattice framework for the Gaussian interference channel