Paper on Deep-unfolded Interleaved Detection and Decoding for MIMO Wireless Systems accepted at the Asilomar Conference on Signals, Systems, and Computers 2022!
Congratulations to Reinhard Wiesmayr for the acceptance of his paper at Asilomar 2022.
The paper “DUIDD: Deep-unfolded Interleaved Detection and Decoding for MIMO Wireless Systems” proposes a novel iterative detection and decoding (IDD) paradigm called deep-unfolded interleaved detection and decoding (DUIDD). The algorithm is applied in a practical 5G MU-MIMO-OFDM wireless system, where a novel low-complexity soft-input soft-output data detection algorithm is interleaved with a low-density parity-check (LDPC) decoder. The method furthermore utilizes deep unfolding to automatically tune the key parameters of the entire DUIDD pipeline. In contrast to classical outer IDD iterations, interleaving breaks the strict separation of detector and decoder, which expedites convergence while maintaining the same complexity. The paper demonstrates the efficacy of DUIDD using the NVIDIA Sionna link-level simulator with channel vectors from a commercial ray-tracer. The results show that DUIDD outperforms classical IDD in terms of error rate performance and computational complexity.
The paper was co-authored with Chris Dick (NVIDIA), Jakob Hoydis (NVIDIA), and Prof. Christoph Studer (IIP Group). The paper will be presented at the 2022 Asilomar Conference on Signals, Systems, and Computers in Pacific Grove, USA. A preprint will be released soon to arXiv.
