光储备池通感一体通信噪声抑制方案(封面文章·特邀)

Research on noise suppression in integrated communication and sensing with photonic reservoir computing (cover paper·invited)

通感一体技术可利用已大范围铺设的光纤网实现智能传感,具备智能识别、信息融合优势。在同波长通感共传信道中,通感信号串扰及光纤传输噪声严重影响通信信号质量,需提升发射功率来保证低误码率,但增大了非线性噪声和功耗。提出了基于光储备池计算神经网络的通感共传信道通信失真均衡方法,搭建了信道均衡仿真系统,实现了低发射功率下前向通信信号噪声均衡。周期线性调频光和56 Gb/s四电平脉冲幅度调制信号分别用于分布式瑞利散射传感和数据通信。仿真结果表明,在5~14 dBm入纤功率时,与未均衡信号相比该方法可提供高于3个数量级以上的误码率优化。此方法在短距离模块间15~24 km光纤传输和0.5~2.0 GHz带宽线性调频感测脉冲下均实现了2个数量级以上的误码率下降。依托光储备池计算芯片的低能耗和低延时优势,该方法可为进一步提升通感一体系统中噪声抑制能力提供理论支撑。

Objective Currently,two primary types of integrated communication and sensing systems exist,which are forward optical sensing and backward reflected optical sensing.The former faces limitations in sensing signal accuracy,the latter encounters nonlinear effects like four-wave mixing when communication and sensing signals occupy separate bands in dense wavelength division multiplexing systems.When both communication and sensing signals share the same frequency band,crosstalk becomes a significant issue.To maintain communication quality,higher input power is often utilized,leading to increased power consumption and nonlinear noise generation.To address these challenges,a noise equalizer is proposed,based on optical reservoir computing(PhRC)chips.The majority of computations are efficiently executed using silicon-based integrated optical devices,offering the dual advantages of reduced latency and expanded bandwidth.This innovative approach holds promise in enhancing the performance and reliability of integrated communication and sensing systems in modern optical fiber networks.Methods This article employs numerical simulation techniques to establish an integrated Rayleigh sensing and 56 Gb/s pulse amplitude modulation(PAM4)communication system(Fig.1).Both the communication and sensing signals utilize a common wavelength of 1550 nm.The sensing signal uses linear frequency modulation(LFM)modulation pulse with a pulse duration of 248μs.Initially,the sensing pulse is modulated using a modulator.Subsequently,an optical bandpass filter converts the signal into a single-sideband signal.Then,the PAM4 communication signal is modulated using the modulator and transmitted through an optical fiber.At the optical circulator,sensing signals are detected,while the receiver captures communication signals contaminated by crosstalk and noise.The simulation of the PhRC chip,weight training,as well as the computation of bit error rate(BER)and symbol error rate(SER),were executed using Python(Fig.2).Distorted communication signals are fed into the PhRC to achieve noise equalization.Results and Discussions The weight of the PhRC optical chip noise equalizer is trainable,enabling adaptive equalization for multiple parameters.Under a 20 km fiber condition,the scheme explores the BER equalization capability at incident powers ranging from 5 to 19 dBm.There is a three-order-of-magnitude difference in the BER and SER between the signals before and after equalization with 7 dBm incident power.At an incident power of 10 dBm,the equalization performance is discussed for fiber transmission distances of 15-24 km,a four-order-of-magnitude improvement in BER and SER with 19 km fiber.When using a 20 km fiber and 10 dBm incident power,the scheme is also evaluated for its ability to improve the BER of communication signals affected by 0.5-2.0 GHz LMF sensing pulses.There is a three-order-of-magnitude difference in the BER between the signals before and after equalization at 1.9 G bandwidth of LMF signal.Simulation results indicate that this scheme maintains stable equalization capabilities even when channel conditions change,especially when the input power decreases.Conclusions This article proposes a signal processing scheme for a PhRC chip-based communication and sensing system,utilizing optical computing to achieve noise equalization for a fiber Raman sensing and 56 Gb/s PAM4 communication integrated system.The PhRC chip is capable of equalizing impairments caused by different incident fiber powers,fiber lengths,and LMF bandwidths.It offers advantages such as high integration,large computational bandwidth,low processing delay,and low computational power consumption.Compared to unequalized signals,the equalized signals achieved a three-order-of-magnitude improvement in BER at a lower incident fiber power of 7 dBm,eliminating the need for optimal higher incident power 16 dBm.The scheme supports fiber transmission lengths of 15-24 km and 0.5-2.0 G LFM sensing pulses,achieving over a two-order-of-magnitude improvement in SER and BER.This communication noise processing scheme can achieve low-power and high-quality signal noise recovery,providing a solution for the future development of integrated optical communication and sensing systems.