Computationally efficient stokes-domain phase demodulation for polarization-diversity coherent φ-OTDR with apache flink stream processing

Distributed acoustic sensing (DAS) based on phase-sensitive optical time-domain reflectometry ( \(φ\) -OTDR) enables distributed vibration detection over long optical fibres. We investigate a computationally efficient Stokes-domain demodulation approach for polarization-diversity coherent \(φ\) -OTDR. The Stokes representation provides access to the polarization phase, orientation, and ellipticity, which depend on different combinations of the Stokes parameters and therefore exhibit different conditioning under polarization variations. The state-of-polarization (SOP) evolution is further characterized using the degree of polarization (DOP) and Poincaré-sphere representation, enabling visualization of the polarization behaviour under vibration-affected, silent-fibre, and noise-floor conditions. The polarization phase, derived from the ratio of two Stokes parameters, is demonstrated to be equivalent to the differential phase between orthogonal polarization components while requiring only a single Stokes-domain phase computation, thereby reducing computational complexity. The method is experimentally evaluated over a 10.15-km fibre using 100-ns pulses with a 120- \(μ s\) repetition period, providing approximately 10-m spatial resolution and a Nyquist-limited maximum detectable vibration frequency of 4.17 kHz. Experimentally investigated vibrations from 400 Hz to 3 kHz yielded spectral SNRs of 29.50 dB and 23.98 dB, respectively, while a 2-kHz vibration was localized near 10.11 km with a peak SNR of 27.14 dB. A proof-of-concept Apache Flink implementation demonstrates reduced processing time for Stokes-domain phase extraction relative to differential-phase processing under the same local execution environment. The three Stokes-derived angular observables exhibit different conditioning under polarization imbalance, while deep Rayleigh interference fading can degrade all of them as the total backscattered signal approaches the receiver-noise level.