Research on Multipath Interference Preprocessing Algorithm for Fjord Underwater Acoustic Communication Based on SRE Topological Operators

Shallow fjord underwater acoustic (UWA) channels pose severe challenges, including strong multipath reflections, rapidly time-varying fading, and weak direct-path signals that are easily submerged by clutter. Traditional amplitude-threshold time-frequency filtering struggles to balance multipath suppression with weak communication symbol preservation, and it relies on dense pilot symbols for channel estimation, thereby consuming valuable communication bandwidth. Leveraging an overnight measured UWA dataset from a Nordic fjord (https://doi.org/10.1109/IEEEDATA.2025.3577998), this paper proposes a UWA time-frequency preprocessing algorithm based on a ten-operator topological pipeline driven by State-Relational Entropy (SRE). By using graph-theoretic topological impedance and path-interference discrimination to distinguish direct-path signals from multipath clutter, the algorithm constructs a three-layer mutually exclusive mask to achieve differential energy attenuation. Simultaneously, an energy-complex dual-path architecture is designed to fully preserve the phase information required for OFDM demodulation. This is integrated with a complete post-processing chain featuring channel equalization, symbol-level Maximum Ratio Combining (MRC) diversity, and communication-prior false-alarm filtering. Full-pipeline validation was conducted using real-world dual-channel hydrophone waveforms across three frequency bands: 4–8 kHz (LF), 9–14 kHz (MF), and 24–32 kHz (HF). The average multipath suppression rates reached 85.8% and 93.8% in the low- and medium-frequency bands (LF/MF), respectively, and the QPSK bit error rate (BER) post-preprocessing was reduced by an average of 17%–18%. In the high-frequency (HF) band, strong noise interference caused topological recognition to fail, yielding a multipath suppression rate of only 22.2%. Experimental results demonstrate that without explicit channel estimation and with lightweight computation, the proposed algorithm effectively balances weak signal preservation and multipath clutter suppression in multipath-dominated low- and medium-frequency shallow-water environments. It also outputs quantitative channel time-variation metrics, making it well-suited for batch offline data processing on underwater embedded communication hardware.