Coded Computation Against Straggling Channel Decoders in the Cloud for Gaussian Channels

Abstract

The uplink of a Cloud Radio Access Network (C-RAN) architecture is studied, where decoding in the cloud takes place at distributed decoding processors. To mitigate the impact of straggling decoders in the cloud, the cloud re-encodes the received frames via a linear code before distributing them to the decoding processors, which estimate linear combinations of the codewords. Focusing on Gaussian channels, and assuming the use of lattice codes at the users, we derive the computational rates and frame error probabilities at the cloud. The approach differs from Compute-and-Forward in that the combination of codewords is not caused by the channel but purposefully created in the cloud by encoding the received signals to reduce the decoding delay.