Paper on a Deep-Unfolding-Optimized Massive MIMO Detector accepted by IEEE JSAC

Implementing data detectors for massive MIMO systems faces critical challenges: high throughput requirements, strict area and power constraints, and error-rate performance degradation for realistic mmWave channels. To simultaneously address these issues, we propose a 22nm ASIC for data detection in mmWave massive MU-MIMO-OFDM systems. Our design implements a Gram-domain block coordinate descent algorithm with a deep-unfolding-optimized posterior mean estimate denoiser. It supports 16 UEs transmitting QPSK to 256-QAM symbols to a 128-antenna BS. Measurement results demonstrate a peak throughput of 7.1 Gbps at 367mW with a core area of only 0.97 mm², demonstrating the detector’s best-in-class throughput and area efficiency. The ASIC was designed by Zixiao Li during his MSc Thesis in the IIP Group under the supervision of Seyed Hadi Mirfarshbafan and Dr. Oscar Castañeda.

The corresponding paper "A Deep-Unfolding-Optimized Coordinate-Descent Data-Detector ASIC for mmWave Massive MIMO" will appear in the IEEE Journal on Selected Areas in Communications (JSAC) and an early access version is available external page here.

Note that the detector design was implemented on the Mothra ASIC, which is shared with a matrix preprocessing engine designed by Darja Nonaca.