Generalized Degrees of Freedom for Network-Coded Cognitive Interference Channel

TL;DR

This paper characterizes the generalized degrees of freedom for a Gaussian cognitive interference channel where one transmitter has linear combinations of messages, revealing significant gains from network-coded cognition using advanced coding schemes.

Contribution

It introduces the GDoF characterization for Network-Coded CIC and proposes novel coding schemes like PCoF and DPC based on nested lattice codes.

Findings

Linear combinations of messages provide multiplicative SNR gain.

Network-coded cognition yields up to 100% gain when SNR equals INR.

Achievability achieved through novel lattice-based coding schemes.

Abstract

We study a two-user cognitive interference channel (CIC) where one of the transmitters (primary) has knowledge of a linear combination (over an appropriate finite field) of the two information messages. We refer to this channel model as Network-Coded CIC, since the linear combination may be the result of some linear network coding scheme implemented in the backbone wired network.In this paper, we characterize the generalized degrees of freedom (GDoF) for the Gaussian Network-Coded CIC. For achievability, we use the novel Precoded Compute-and-Forward (PCoF) and Dirty Paper Coding (DPC), based on nested lattice codes. As a consequence of the GDoF characterization, we show that knowing "mixed data" (linear combinations of the information messages) provides a {\em multiplicative} gain for the Gaussian CIC, if the power ratio of signal-to-noise (SNR) to interference-to-noise (INR) is larger than certain threshold. For example, when $\SNR=\INR$, the Network-Coded cognition yields a 100% gain over the classical Gaussian CIC.