Tomlinson-harashima precoder with tilted constellation for reducing transmission power

Abstract

Tomlinson-Harashima precoder (THP) is being noted for an efficient alternative to dirty paper coder. THP increases the transmission power due to the use of the modulo operator. In this paper, we propose a method that reduces the transmission power by tilting the constellations of the transmit data symbols. The symbol error rate (SER) is improved for a fixed SNR by reducing the transmission power. Average transmission power is calculated asymptotically by using an extreme value distribution. We have shown that the transmission power can be significantly reduced when we use the tilted constellation THP. As the transmission power determines the SER slope, at high SNRs we achieve lower SER than in the case of a transmission over an AWGN channel. The tilted angle can be estimated at the receiver. An angle estimation algorithm is proposed that optimizes angle estimation error and SER.We use the approximation of the angle estimation error and SER to find an optimum number of tilted angles. By using the proposed angle estimation algorithm and the optimum number of tilted angles, the tilted constellation THP always outperforms the conventional THP algorithm.