Blind MIMO Equalization Using Correlation-Avoidance CMA for Space-Division Multiplexed 3-Mode 54-km Transmission

Space division multiplexing (SDM) transmission systems using coupled spatial channels require mechanisms to compensate for random channel coupling, often implemented as multiple-input, multiple-output (MIMO) digital signal processing (DSP). Efficient MIMO DSP implementation is crucial for SDM fibers with many coupled channels due to increased computational complexity. The constant modulus algorithm (CMA) is a blind adaptive equalizer update strategy that offers significant advantages over data-aided, decision-directed and pilot-based schemes as it is immune to phase and frequency noise. However, the CMA may converge to a state of singularity, where some of the retrieved polarization tributaries are identical, causing an invalid equalization. To address this issue within the context of SDM, we apply an extension of the conventional CMA, the correlation-avoidance (CA)-CMA that prevents singularities by detecting the correlation between the equalizer's outputs. We experimentally demonstrate the singularity-free convergence of the CA-CMA in blind 6×6 adaptive equalization of 32 GBd dual-polarization quadrature phase shift keying (DP-QPSK) over a 54-km few-mode fiber (FMF). For 1500 measurements in the optical signal-to-noise ratio (OSNR) range of 8 dB to 36 dB, the CA-CMA eliminated all singularities, compared to approximately 80% invalid convergences with the conventional CMA. We also propose to apply the CA-CMA with a small tap count as a robust pre-convergence for a computationally-efficient CMA-based blind MIMO equalization with a larger tap count. This enhances the signal-to-noise (SNR) performance by up to 5 dB by better capturing the transmission channel impulse response.