Maximum entropy quantization for link-state adaptation in two-way relaying

In this paper, we investigate how to adapt a composite decode-forward (DF) two-way relaying scheme to fading links. We develop a maximum entropy (ME) quantization algorithm to feed back the link magnitudes such that the transmitters are able to perform link adaptation. The ME algorithm outperforms that of Lloyd-Max in mid and high range SNR, and performs comparably at low SNR. With just one bit of ME quantization feedback, rate performance of the composite DF scheme is better than that of long-term channel state information (CSI). With three quantization bits, performance is close to that of perfect CSI. We further evaluate the sensitivity of the ME quantizer to errors in the estimated channel distribution. The developed ME quantization algorithm can also be applied in any compression scenario with known data distribution, including real-time applications because of the simplicity of the quantizer.