Interferometric fiber optic hydrophone is a relatively mature solution in the current fiber optic hydrophone system and features high sensitivity, large dynamic range, strong anti-interference ability, and easy array formation. Meanwhile, it is suitable for underwater targets and is widely employed in fields such as detection and underwater energy exploration. In recent years, the application scenarios of fiber optic hydrophones have gradually developed into complex scenarios such as far-reaching seawater acoustic detection and mobile platform deployment. These scenarios pose more challenges to the signal detection performance and noise stability of hydrophones. Phase-generated carrier (PGC) demodulation is a commonly adopted signal detection method for interferometric fiber optic hydrophones. Since the operating point and carrier modulation depth are greatly affected by external environmental changes, the PGC demodulation system has unstable output phase signals. In particular, the system's self-noise stability fluctuates greatly with environmental changes. This problem has become an important factor limiting the performance of fiber optic hydrophone systems.

Centering on the noise stability of interferometric fiber optic hydrophones based on PGC demodulation, we build a noise transfer model of the interferometric fiber optic hydrophone based on PGC demodulation and focus on analyzing changes in the two parameters of the carrier modulation depth and operating point. Meanwhile, the mechanism of influence on the stability of PGC demodulation noise is studied. A new multi-phase PGC demodulation scheme is proposed, where a 3×3 coupler is introduced into the traditional PGC demodulation architecture for multi-phase detection, and the three interference signals are fused by phase shift characteristics of the coupler. The multi-phase PGC demodulation algorithm performs PGC demodulation on the outputs of three 3×3 couplers respectively, and then averages the demodulation results of the three channels. Since the measured phase signals in the three demodulated output signals are the same, the averaging operation has no effect on them, while the noise signals can be suppressed. Additionally, as the initial phases of the three interference signals differ by 2π/3, the noise influence exerted by the initial phase changes can be minimized by averaging regardless of whether the working point of the interference signals changes or not. Therefore, the demodulation noise can be relatively stable. As the working point of the hydrophone changes, this scheme can reduce fluctuations in the noise transfer coefficient of the light source intensity noise.

We conduct simulation experiments to verify the performance of the multi-phase PGC demodulation algorithm. The simulation results show that sound noise stability can be achieved under different carrier modulation depth (C) values. Under different C values, the fluctuation of the noise transfer coefficient is less than 0.5 dB, and compared with the traditional PGC demodulation algorithm, the stability of demodulation noise of multi-phase PGC demodulation algorithm is significantly improved (Figs. 3 and 4). A multi-phase PGC demodulation system based on 3×3 coupler is built, and the demodulation phase noise performance of the system is experimentally verified. A multi-channel synchronous sampling analog-to-digital converter (ADC) is employed to acquire the three outputs of the coupler. The traditional PGC demodulation method and the multi-phase PGC demodulation algorithm are utilized to demodulate the original data collected by the system. Additionally, we calculate the noise spectrum levels of the demodulated signals of the two methods at 1 kHz frequency separately and analyze the noise fluctuation characteristics of the system. The experimental results show that the self-noise fluctuation obtained by demodulating the three outputs of the 3×3 coupler using the traditional PGC demodulation method is greater than 4.5 dB (Fig. 6). The noise spectrum levels obtained by the multi-phase PGC demodulation method are significantly reduced, and the noise fluctuation during the entire test cycle is less than 1.8 dB (Fig. 6). The experimental results verify the effectiveness of the multi-phase PGC demodulation algorithm.

We build a noise transfer model for interferometric fiber optic hydrophones, analyze and derive the noise transfer model of system noise sources on demodulation results, and propose a multi-phase PGC demodulation algorithm. Compared with traditional PGC demodulation algorithm, the proposed algorithm can suppress the fluctuation of light intensity noise transfer coefficient under the changing operating point, and improve the noise stability of demodulation results. Simulation and experimental results are consistent with the theoretical analysis results of the model. In applications such as deep-sea exploration and long-distance target detection which have increasingly stringent noise performance requirements for fiber optic hydrophones, the noise transfer model and the multi-phase PGC demodulation algorithm based on 3×3 coupler proposed in our study have research and practical significance.