A Soft-Input/Soft-Output Dequantizer for Cloud-Based Mobile Networks

The use of soft information instead of hard bits has been widely adapted in signal processing for digital communication to improve the reliability of data transmission by techniques like turbo equalization, turbo decoding, or LDPC codes. For future mobile networks employing a centralized architecture in which the antenna and the baseband processing are separated by an additional fronthaul channel that forwards digitalized samples, we have identified another process that can be redesigned to embrace the concept of soft information: the dequantizer. A dequantizer transforms a vector of bits into amplitudes representing digitalized samples. However, if the bits were transmitted through a lossy fronthaul channel, the soft information that can be extracted from this channel's detection process is lost in a classical dequantizer. In this work, we propose a soft-input/soft-output dequantizer that passes this information through to the subsequent signal processing steps and can thereby improve the overall reliability of centralized, cloud-based mobile networks.