A linear predictive coding based compression algorithm for fronthaul link in C-RAN

This paper focuses on the data compression for fron-thaul link in cloud radio access network (C-RAN). We introduce a linear predictive coding (LPC) based compression algorithm, which includes three main modules, linear predictor, piecewise quantizer and arithmetic coding. The linear predictor is used to eliminate redundancies caused by the correlations of LTE signals. According to the distribution characteristics of the output from the linear predictor, a piecewise quantizer is designed to decrease the quantization noise. To get extra coding gain, arithmetic coding is applied based on the output codebook from the piecewise quantizer. Furthermore, an alphabet reconstruction scheme is proposed to reduce the complexity of arithmetic encoding and decoding. Omitting the up/down-sampling and low-pass filtering operation in traditional schemes, our solution has simpler structure, but demonstrates better compression performance with comparable computational complexity. Remarkably, our scheme presents strong flexibility to different load ratios. Taking 40% load ratio as an example, within 2% error vector magnitude (EVM) distortion, our algorithm can achieve 23% compression ratio for uplink and 17% compression ratio for downlink, which has nearly 15% gain compared with the traditional algorithm.