As in 570, a lossy compression can be performed at the UE. In some examples, the lossy compression can be performed by a compression module at the eRRH. As part of the lossy compression (or as a preliminary pre-processing action), each PUSCH I/Q sample in a plurality of PUSCH I/Q samples (e.g., frequency-domain I/Q samples produced by performing an FFT on time-domain I/Q samples corresponding to the PUSCH I/Q symbols) corresponding to the plurality of PUSCH I/Q symbols can be normalized. In some examples, each PUSCH I/Q can be normalized to a value ranging from ?1 to 1 by subtracting a mean value and using a scaling value. The normalized PUSCH I/Q samples can then be quantized based on the number of bits. In some examples, uniform scalar quantization can be applied to each normalized PUSCH I/Q sample (e.g., by a uniform scalar quantizer).

As in 580, a lossless compression can be performed at the UE. In some examples, the lossless compression can be performed by the compression module at the eRRH. In the lossless compression, a prefix-free code can be applied to a plurality of bit subsets, each bit subset in the plurality of bit subsets being associated with a respective PUSCH I/Q sample in the plurality of PUSCH I/Q samples. In some examples, each bit subset in the plurality of bit subsets can comprise the two most significant bits (MSBs) of the PUSCH I/Q sample with which the bit subset is associated. In some examples, the prefix-free code can be a Huffman code in which a value of 00 for the MSBs maps to a single coded bit having a value of 0; a value of 01 for the MSBs maps to two coded bits having a value of 10; a value of 10 for the MSBs maps to three coded bits having a value of 110; and a value of 11 for the MSBs maps to four coded bits having a value of 111.