Optical Signal-to-Noise Ratio Monitoring Using Uncorrelated Signal-Spontaneous Beat Noise

Abstract—Optical performance monitoring is essential formanaging both current and future optical networks. One im-portant quantity to monitor is the optical signal-to-noiseratio(OSNR). Traditional optical spectrum analyzer (OSA) basedmeasurements become problematic in dynamically reconfigur ablewavelength-division multiplexing (WDM) networks. We prop osea new method based on analyzing the uncorrelated signal-spontaneous beat noise and show through simulation that in theOSNR range from 10 dB to 25 dB, the OSNR estimation errorcan be less than 0.6 dB. I. INTRODUCTIONInterest in optical performance monitoring (OPM) first tookhold in the early 1990s with the introduction of Wavelength-Division Multiplexing (WDM) systems. As people began tothink more about optical networks rather than optical trans-mission systems, it became clear that a solution would beneeded for the monitoring problem [1]. WDM optical networkshave enabled the rapid growth of data traffic in the networkbackbone. Further increases in capacity can be gained bymoving to dense wavelength division multiplexing (DWDM)with large channel counts. Optical performance monitoring isessential for managing such high capacity optical transmissionand switching systems [1]. Optical performance monitoringdescribes an extremely wide range of functionality intendedto be included in a communication network to improve thenetwork’s performance [2]. Both current and future WDMor DWDM optical networks require reliable and economicalmethods for performance monitoring without interrupting theclient connections. For the efficient operation and mainten anceof the network, it is essential to monitor various parametersincluding the wavelength, optical power, optical signal-to-noise ratio (OSNR), and optical paths, etc. [3].The widespread use of optical amplifiers has enabled long-haul optical communication systems [4], but the cascadingof a large number of amplifiers in such systems introducesamplified spontaneous emission (ASE) noise into the opticalsignals which degrades the OSNR [2]. OSNR is a significantfactor for the quality of optical systems and OSNR moni-toring is essential for optical performance monitoring. Forexample, OSNR monitoring is necessary for fault managementincluding fault localization and diagnosis. In order to localizeOSNR degradation faults, network operators require multipletruck rolls along the entire length of the optical segments.In ultra-long-haul networks, diagnosing these faults withoutOSNR monitoring may be challenging and optical add-dropmultiplexers (OADMs) complicate the situation further [5].Quite a few techniques have been developed to monitorOSNR. Current OSNR Monitoring Techniques can be classi-fied into two groups. One is OSNR monitoring by measuringthe ASE noise outside of the channel bandwidth. To simplifythe name, we can call this group of techniques out-of-bandOSNR monitoring. The other is OSNR monitoring by mea-suring the ASE noise inside the channel bandwidth which iscalled in-band or in-channel OSNR monitoring. Out-of-bandOSNR Monitoring includes the traditional optical spectrum an-alyzer (OSA) based OSNR estimation [6], [7] and some tech-niques using arrayed waveguide grating (AWG) circuits [8]–[10]. The traditional optical spectrum analyzer (OSA) basedOSNR estimation [6], [7] relies on interpolation of ASE levelsadjacent to the channel of interest to obtain the approximateASE level in the channel. As the interpolated ASE noise maynot be the real ASE noise in the channels of interest, thoseconventional linear interpolation techniques may not monitorthe OSNR accurately for a dynamically reconfigurable WDMnetwork. This is because in a dynamically reconfigurableWDM network each channel may traverse through differentroute and different number of EDFA’s so that the network hasunflat ASE spectrum. As the principle of the techniques usingarrayed waveguide grating (AWG) circuits [8]–[10] was alsobased on measuring the ASE noise between the channels, theAWG monitoring circuits may not precisely measure OSNR inpresence of multi-path interference effects and OADM/OXC(Optical Cross Connect) networking [1]. Polarization nullingmethod [11]–[14] is one of the in-band OSNR monitoring ap-proaches. But its performance is limited by Polarization ModeDispersion (PMD). Even if PMD effects can be compensated,polarization scattering in ultra long-haul systems results infast bit-to-bit polarization fluctuations that are impractical totrack [1], [15]. In addition, the speed of OSNR monitoringwill be limited by the polarization controller. Another in-bandOSNR monitoring approach is the low frequency beat noisemeasurement [16]. The method is simple to implement by