1: \begin{abstract}

2: In this paper, we develop a unified framework for beamforming designs in

3: non-regenerative multiuser two-way relaying (TWR). The

4: core of our framework is the solution to the max-min

5: signal-to-interference-plus-noise-ratio (SINR) problem for multiuser TWR. We solve this problem

6: using a Dinkelbach-type algorithm with near-optimal performance and superlinear convergence.

7: We show that, using the max-min SINR solution as a corner stone, the beamforming designs under various important criteria, such as

8: weighted sum-rate maximization, weighted sum mean-square-error (MSE) minimization, and average bit-error-rate (BER) or symbol-error-rate (SER) minimization, etc, can be reformulated into a monotonic program. A polyblock outer approximation algorithm is then

9: used to find the desired solutions with guaranteed convergence and optimal performance (provided that the core max-min SINR solver is optimal). Furthermore, the proposed unified approach can provide important insights for tackling the optimal beamforming designs in other

10: emerging network models and settings. For instances, we extend the proposed framework to address the beamforming design in collaborative TWR

11: and multi-pair MIMO TWR. Extensive numerical results are presented to demonstrate the merits of the proposed beamforming

12: solutions.

13: 

14: 

15: \vspace*{-0.1cm}

16: %\par \noindent

17: \textbf{Keywords:} Two-way relaying, beamforming, fractional

18: program, semi-definite program, monotonic optimization.

19: \end{abstract}