Pilot-Based Decision Feedback Signal Combining for Cell-Free Massive MIMO Networks

The distributed network architecture allows cell-free massive MIMO(mMIMO) with multiple ways to combine uplink signals. The decision feedback scheme uses the detected signal of the distributed combiner to estimate the equivalent channel at the central processing unit (CPU), and iteratively updates the symbols using the estimated channel state information (CSI). The performance of this algorithm greatly depends on the accuracy of initial signal detection. In this paper, we first propose a pilot-based signal combining (PC) scheme at the CPU to improve the estimation accuracy of the symbols. We use pilot signals instead of detected signal to estimate the equivalent channel at the CPU. The estimated data symbols is then used as the training input in the decision feedback scheme, enabling iterative refinement of the equivalent CSI. Simulation results show that the proposed PC scheme can significantly reduce the system bit error rate compared to the decision feedback scheme with lower computational complexity and less decoding latency, albeit with slightly increase fronthaul signaling load. Furthermore, continuing the iterative process with the decision feedback scheme after initial PC further reduces the bit error rate, albeit with added decoding latency and computational complexity similar to the standard decision feedback scheme.