Random Element Quantization for Finite Rate Feedback Systems

Finite rate feedback model based on Random Vector Quantization (RVQ) is efficient in terms of number of feedback bits. However, the required scaling of feedback bits (codebook size) to achieve full multiplexing gain becomes extremely large with high Signal-to-Noise Ratio (SNR) and might not be practical. This paper considers the use of a Random Element Quantization (REQ) approach, which quantizes the channel vector element-wise. A simple statistics is calculated and the presented simulation results validate our calculations. The main primary result is that, the required codebook size per receiver to achieve full multiplexing gain is kept very small, which would avoid high computational complexity and memory usage.