Gaussian Two-way Relay Channel with Private Information for the Relay

TL;DR

This paper proposes a near-optimal decode-and-forward scheme for a generalized Gaussian two-way relay channel with private information, combining computation over MAC and BC with side information, achieving rates close to capacity.

Contribution

It introduces a novel DF scheme for the Gaussian two-way relay channel with private info, combining MAC and BC strategies with lattice coding, near achieving capacity.

Findings

Scheme achieves rates within half a bit of capacity

Combines computation over MAC with BC with side info

Near-optimal performance across channel conditions

Abstract

We introduce a generalized two-way relay channel where two sources exchange information (not necessarily of the same rate) with help from a relay, and each source additionally sends private information to the relay. We consider the Gaussian setting where all point-to-point links are Gaussian channels. For this channel, we consider a two-phase protocol consisting of a multiple access channel (MAC) phase and a broadcast channel (BC) phase. We propose a general decode-and-forward (DF) scheme where the MAC phase is related to computation over MAC, while the BC phase is related to BC with receiver side information. In the MAC phase, we time share a capacity-achieving code for the MAC and a superposition code with a lattice code as its component code. We show that the proposed DF scheme is near optimal for any channel conditions, in that it achieves rates within half bit of the capacity region of the two-phase protocol.