Performance Analysis of Optimal Perturbing Vector with Double Decomposition Precoding in Massive MIMO Systems

In a multi-user MIMO system, the application of the Vector Precoder is aimed at improving the system capacity by minimising the issues of power scaling. However, there is difficulty in acquiring the power scaling factor over a slow-fading environment, and this leads to increased transmission overhead. The complex signal selection from the lattice grid requires minimal optimisation and the highest precoding matrix vector selection requires minimal precoding matrix computation. In this research, an improved Vector Perturbation (VP) precoding is provided by introducing an MMSE solution to find the optimal perturbing complex signal vector. The complex signal in the lattice space is optimally obtained based on inaccurate channel state information (CSI). In the precoding matrix operation, we apply Hessenberg reduction to a QR Decomposition (HQRD) algorithm as it balances the trade-off between the vector precoding and computational complexity. The simulation is conducted to estimate the bit error rate (BER) using efficient fixed sphere detection based on varying channel conditions. The results show that the proposed vector precoder achieves a BER of 10
⁻³
better than the state-of-the-art precoders.