Multi-Group Multicast Beamforming for Multi-User Two-Way Relaying

Authors

Aditya Umbu Tana Amah, Anja Klein, Yuri C. B. Silva, and Andreas Fernekess

Reference

International ITG Workshop on Smart Antennas , Berlin, Feb. 2009.

[BibTeX, LaTeX, and HTML Reference]

Abstract

In this work, we consider a multi-user two-way relaying protocol. A multi-antenna relay station (RS) serves 2K nodes where K pairs of nodes would like to perform bidirectional communication. In the first phase, the 2K nodes transmit simultaneously and the RS spatially separates and decodes all the 2K bit sequences. In the second phase, after performing bit-wise XOR network coding for each bidirectional pair, which results in only K bit sequences, the RS sends all the K XOR-ed bit sequences simultaneously. At each node, having received its intended XOR-ed bit sequence, it cancels the self-interference by XOR-ing its received bit sequence with its transmitted one to obtain the bit sequence sent by its partner. Considering the second phase, as each node in a specific pair expects the same XOR-ed bit sequence from the RS, while seeing other pairs’ XOR-ed bit sequences as interference, the RS has to separate the different pairs’ XORed bit sequences spatially. Thus, in the second phase, multigroup multicast beamforming can be applied with the node pairs being the multicast group. With this new perspective on the second phase, we propose to apply multi-group multicast beamforming algorithms which provide fairness to all pairs and have low computational complexity. We investigate several multi-group multicast beamforming algorithms, namely Zero Forcing (ZF), Multicast Aware ZF (MAZF) and SINR Balancing with Bisection Search (SINRB-Bisec), for the second phase transmission of the multi-user two-way relaying. We consider also two different XOR network coding approaches, namely with and without zero padding (ZP). The overall two-phase sum rate analysis is given which shows that both MAZF and SINRB-Bisec have similar performance and outperform ZF. Nevertheless, MAZF has lower computational complexity compared to SINRB-Bisec.

This publication is currently not available for download.