In-band OSNR monitoring using a pair of Michelson fiber interferometers

Size: px

Start display at page:

Download "In-band OSNR monitoring using a pair of Michelson fiber interferometers"

Logan Miller
5 years ago
Views:

1 In-ban OSNR monitoring using a air of Michelson fiber interferometers E. Floo 1*, W. H. Guo 1, D. Rei, M. Lynch 1, A. L. Braley 1, L. P. Barry, an J. F. Donegan 1 1 Semiconuctor Photonics Grou, School of Physics, Trinity College, Dublin, Irelan Research Institute for Networks an Communications Engineering, Dublin City University, Dublin 9, Irelan *flooea@tc.ie Abstract: Two olarization-ineenent Michelson fiber interferometers with ifferent otical elays were use to measure the in-ban OSNR of an otical signal from 5 to 30B within an accuracy of 0.5B. Using an exansion of the amlitue autocorrelation function of the signal aroun zero elay, it was ossible to erform measurements without any rior knowlege of the signal. The system is shown to be immune to the effects of moulation frequency u to 10G), artially an fully olarize noise, chromatic isersion an oorly biase moulators. 010 Otical Society of America OCIS coes: ) Fiber otics communications; ) Interferometry. References an links 1. D. C. Kiler, R. Bach, D. J. Blumenthal, D. Einstein, T. Lanolsi, L. Ostar, M. Preiss, an A. E. Willner, Otical Performance Monitoring, J. Lightwave Technol. 1), ).. H. Suzuki, an N. Takachio, Otical signal quality monitor built into WDM linear reeaters using semiconuctor arraye waveguie grating filter monolithically integrate with eight hotoioes, Electron. Lett. 3510), ). 3. J. H. Lee, H. Y. Choi, S. K. Shin, an Y. C. Chung, A Review of the Polarization-Nulling Technique for Monitoring Otical-Signal-to-Noise Ratio in Dynamic WDM Networks, J. Lightwave Technol. 411), ). 4. C. Xie, D. C. Kiler, L. Moller, an R. Ryf, Orthogonal-Polarization Heteroyne OSNR Monitoring Insensitive to Polarization-Moe Disersion an Nonlinear Polarization Scattering, J. Lightwave Technol. 51), ). 5. Z. Tao, Z. Chen, L. Fu, D. Wu, an A. Xu, Monitoring of OSNR by using a Mach-Zehner interferometer, Microw. Ot. Technol. Lett. 301), ). 6. Y. K. Lizé, J. Y. Yang, L. Christen, X. Wu, S. Nuccio, T. Wu, A. E. Willner, R. Kashya, an F. Séguin, Simultaneous an Ineenent Monitoring of OSNR, Chromatic an Polarization Moe Disersion for NRZ- OOK, DPSK an Duobinary, in Otical Fiber Communication Conference an Exosition an The National Fiber Otic Engineers Conference, OSA Technical Digest Series CD) Otical Society of America, 007), aer OThN. 7. X. Liu, Y.-H. Kao, S. Chanrasekhar, I. Kang, S. Cabot, an L. L. Buhl, OSNR Monitoring Metho for OOK an DPSK Base on Otical Delay Interferometer, IEEE Photon. Technol. Lett. 1915), ). 8. J. M. Oh, M. Brosky, L. E. Nelson, G. Caena, an M. D. Feuer, Interferometric otical signal-to-noise ratio measurements of telecom signals with egrae extinction ratio, Ot. Lett. 3318), ). 9. A. D. Kersey, M. J. Marrone, an M. A. Davis, Polarisation-insensitive fibre otic Michelson interferometer, Electron. Lett. 76), ). 1. Introuction In orer to realize next generation communications networks, it will be essential to imlement erformance monitoring in the otical layer. Currently, monitoring is realize through use of otical/electrical/otical evices, which will be imossible for the case of a transarent DWDM network. Several iagnostics of the otical signal will be require to meet this monitoring nee, one of which is otical signal to noise ratio OSNR) monitoring. The cascaing of otical amlifiers in communications networks causes the accumulation of amlifie sontaneous emission ASE) noise, which is one of the main imairments in otical networks. ASE noise is tyically quantifie by the otical-signal-to-noise-ratio OSNR), which is measure within an otical banwith of 0.1 nm. Due to its influence on # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 3618

2 the bit-error-rate of an otical signal, its role in fault iagnosis an as a measure of general network health, it is imortant to have a reliable means of OSNR monitoring over a large range of signal an noise [1] Current methos involve measuring the noise at some offset from the centre frequency of a channel an erforming a linear interolation in orer to make an estimate of the in-ban OSNR []. This technique becomes unreliable with the use of reconfigurable otical a-ro multilexers ROADM). Due to the resence of channels with vastly ifferent transmission histories, an inter-channel OSNR measurement becomes unreliable. In orer to eluciate any meaningful result, a irect measurement of the in-ban OSNR is necessary. Several in-ban OSNR measurement techniques have been roose, one examle of which is the olarization nulling metho. This technique relies on the assumtion that the signal is comletely olarize while the noise is comletely unolarize [3]. In ractice, the resence of olarize noise ue to olarization eenent loss in the fiber an eolarize signal as a result of olarization moe isersion PMD) an nonlinear olarization scattering makes the measurement less reliable. The use of orthogonal olarization heteroyne mixing has also been roose an, while it is imervious to the effects of PMD, it relies on the assumtion of comletely unolarize noise [4]. It is ossible to overcome these issues by emloying a Mach-Zehner MZ) interferometer to exloit the coherence ifference between signal an noise by moulating the hase of one arm of the interferometer an using the extinction ratio ue to the interference in the two arms to calculate the OSNR. Such a system has been shown to be immune to the effects of artially olarize noise, PMD-inuce signal eolarization, chromatic isersion, bit-rate, moulation format an caable of measuring OSNR values exceeing 30 B for OOK an PSK signals [5 7]. These systems have been shown to be relatively chea an reliable but the roose single-interferometer schemes are not without isavantages. In orer to oerate such a evice one must have rior knowlege of the amlitue autocorrelation AAC) function of the signal. In the case of the above metho, this quantity can only be measure by turning off the noise, scanning the signal interference an using the resultant value as the signal AAC in subsequent calculations. Such a roceure is unrealistic for ractical usage ue to the effect of moulator bias on the signal AAC [8]. In this aer, we resent a metho which, like the aforementione interferometer-base examles, is base on exloiting the coherence ifference between signal an ASE. This retains the avantages of such systems but isenses with the requirement of knowing the AAC of the signal. The technique is insensitive to the inut olarization, evoi of free-sace elements an can measure the OSNR over a similar ynamic range. This can be achieve by using a ouble Michelson fiber interferometer scheme with ajustable hase elay an ifferent otical elays that are much smaller than the bit erio of the signal an may be still within the coherence time of the noise, which is etermine by the transmission function of the channel filter. Oerating rincile Figure 1a) shows the OSNR monitor setu. The interferometers each consist of two 3 B coulers an two Faraay rotator mirrors. The effect of the Faraay mirrors is to imose a 45 egree single ass olarization rotation, effectively unwining the olarization change in the fiber [9]. This remove the necessity for ynamic olarization control in the evice an shoul allow accurate OSNR measurements to be erforme on olarization multilexe signals. One arm of each interferometer has a fixe otical elay of the orer of s. These elays were measure by scanning the interferometers transmission as a function of wavelength an calculating the elay from the free-sectral range. The elays can be set by using ifferent combinations of fiber coulers as their magnitues are of the orer of the manufacturing tolerance of the fibers arox 1mm) an were characterize as 8 s an 17 s. The other arm of each interferometer ha a variable hase element basically another elay element but of the orer of fs), which ajuste the hase in each arm over multiles of π by stretching the fibers using a iezoelectric PZ) tube. High sensitivity, low see hotoioes # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 3619

3 PD) recor the transmitte intensity an the information is acquire using the analogue inut of a ata acquisition car DAQ) which measures a eenence as shown in Fig. 1b). The extinction ratio of this lot is the ratio of the maximum to the minimum transmitte voltage, ρ = V max /V min an may be use to calculate the OSNR by methos escribe in the next section. Theory Fig. 1. a) Interferometer setu, PD: hotoioe, PZ: Piezo crystal; Int1 an : interferometers; b) Interferometer transmission. The signal after the channel filter can be exresse as e ) )ex ) in t = E t j π f0t, where f 0 is the carrier frequency. After the interferometer the signal is escribe by Eq. 1). 0 ) 0 ) e t) = Κ E t+ )ex j π f t+ ) + Κ E t+ )ex j π f t+ ) 1) out where Κ reresents the ower of the signal assing through the interferometer arm with the otical elay an Κ reresents the ower of the signal assing through the interferometer arm with the variable otical hase, is the elay which is of the orer of s an is for the hase ajustment which of the orer of the otical cycle ~5 fs). For noise we can use similar exressions. The noise immeiately after the channel filter can be exresse as n ) )ex ) in t = N t j π f0t. After the interferometer it will be exresse as Eq. ) ) n t) = Κ N t+ )ex j π f t+ ) out 0 0 ) + Κ N t+ )ex j π f t+ ) We assume the interferometer is olarization insensitive so we aot a scalar escrition of the signal an the noise. The low see hotoioe will etect a voltage as in Eq. 3) ) V = C e t) + n t) 3) out where C is a constant relating to the etector efficiency. The noise may be exresse as in Eq. 4) N t) = 0, N t) N t) = P N t) N t+ ) = F f )ex j π f ) f N t) N t+ ) = F f )ex j π f ) f F f ) + F f ) ex j π f ) f γ n ) Pn where Ff) is the noise ower sectrum, P n is the noise average ower. If the noise ower sectrum is simly etermine by the ower transmission function of the channel filter which can be accurately measure, γ n, the autocorrelation of the noise amlitue, can be etermine through its efinition in Eq. 4). In the event that the signal has asse through many n out 4) # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 360

4 ROADMs before reaching the channel filter, the noise ower sectrum will not simly be etermine by the ower transmission function of the channel filter. In this situation we think it is ossible to turn off the signal an measure the interferogram visibility of the noise which will still be resent ue to the cascae fibre amlifiers. γ n can then be obtaine from the measure visibility through the following Eq. 1) when r, γ n = M). In the following analysis we assume γ n has been known. As seen from 4), γ n will be a real an even function of the elay an will have a eak value of 1 at = 0. The signal after the channel filter may be efine as in Eq. 5) E t) E t) = P s E t) E t+ ) = S f )ex j π f ) f E t) E t+ ) = S f )ex j π f ) f S f ) + S f ) ex j π f ) f γ s ) Ps where Sf) is the signal ower sectrum, P s is the signal average ower. As seen from 5), γ s ), the autocorrelation of the signal amlitue will be a real an even function of as well an will always have a eak value of 1 at = 0. With the above efinition, the voltage etecte can be simlifie as in Eq. 6) 5) Κ ) Ps + Pn ) + Κ Κ V = C cos π f0 )) Psγ s ) + Pnγ n )) Because <<, the Eq. 6) can be further simlifie to give Eq. 7) 6) V = C[ Κ ) P + P ) + Κ Κ s n π f0 ) Psγ s Pnγ n ) cos ) ) + ) ] For each with changing, the current will follow a sine curve which is the interferogram of the noisy signal. The interferogram has maximum an minimum value as in Eqs. 8) an 9) resectively γ γ ) Vmax = C Κ ) Ps + Pn ) + Κ Κ Ps s ) + Pn n ) γ γ ) Vmin = C Κ ) Ps + Pn ) Κ Κ Ps s ) + Pn n ) 9) The extinction of the interferogram is ρ = V max /V min which is the main quantity we nee to measure in the roose scheme. In ractice the extinction is always exresse on the log scale because it is generally a large value. From the maximum an minimum values of the interferogram we can efine the quantities V ave an V iff as in Eqs. 10) an 11) resectively Vmax + Vmin Vave = C Κ ) Ps + Pn ) 10) Vmax Vmin Viff = C Κ Κ Psγ s ) + Pnγ n )) 11) The visibility of the interferogram is efine as µ = V iff /V ave. The visibility an extinction of an interferogram are relate to each other simly through µ = ρ 1)/ρ + 1) an ρ = 1 + µ)/1 µ). So we can notice that if the visibility is close to 1 for examle 0.998, the extinction woul be very high for examle close to B). We efine r Pn Ps which leas to Eq. 1) 7) 8) # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 361

5 γ ) ) iff ) s + rγ n V Κ µ = = M 1+ r Vave Κ Κ Κ Κ Κ ) where M is the interferogram visibility of the noisy signal when the otical ower imbalance in each arm of the interferometer has been aresse. r is then given by Eq. 13) γ s ) M r = 13) M γ n ) As state above γ n ) is known, so if γ s ) is known then the signal to noise ratio can be foun an the OSNR can therefore be erive as in Eq. 14) NEBnm) NEBnm) OSNR = 10 log10 = 10 log 10 r) + 10 log10 r 0.1nm) 0.1nm) where NEB is the noise equivalent banwith of the channel filter. For the methos that are alreay eveloe base on a single interferometer with a fixe otical elay, the elay between the two arms is always mae large enough to ensure γ n ) is aroximately zero. γ s ) is obtaine by turning off the noise source, in which case only signal is resent so r = 0 an γ s ) = M can be measure irectly as from Eq. 1). After the noise is ae, the same γ s ) is use to obtain the OSNR. This metho is shown to be immune to various imairments such as eolarization of the signal cause by olarization moe isersion PMD) an artial olarization of the noise cause by olarization eenent loss PDL) of otical elements installe in the otical link. One roblem is that ractically it is ifficult to turn off the noise because in a long otical link the fibre amlifiers have to be there so the signal will always accumulate aitional noise. For this reason it is ifficult to obtain γ s ). Furthermore, γ s ) coul change because the DC bias or the rive signal of the moulator use to generate the signal coul change. Aitional filtering an self cross) hase moulation which change the signal sectrum coul also change γ s ). So the signal amlitue autocorrelation shoul be erioically monitore, which is ifficult for a single interferometer with a fixe otical elay. Observing the function of γ s ), we can fin that it always has a nearly arabolic shae aroun the zero elay. So if we emloy otical elays much less than the bit erio 1/ν, where ν is the frequency of the signal, we can exan γ s ) aroun = 0 as in Eq. 15) K k s ) 1 c. k 1 k = γ 1) 14) = 15) We just have terms with owers of k in 15) because as ointe out above γ s ) is an even function of. Using this relation we can change Eq. 13) into Eq. 16). γ ) K k c 1 k, q + r M q n, q ) = 1 M k= q 16) where the subscrit q means we have mae a series of measurements with ifferent elays reresente by, q where q is from 1 to Q. So Eq. 16) reresents a series of linear equations with unknown values of r an c k which can be solve if Q K + 1. Generally selecting k = 1 an oing two measurements at two ifferent elays, i.e. Q =, can achieve results within the aroriate range an accuracy 30B ± 0.5B OSNR). In the following escrition we generally assume that we measure two interferograms of the noisy signal at two ifferent elays which are small comare with the bit erio 100 s for a 10G signal). # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 36

6 Results Fig.. OSNR monitoring setu The exerimental setu is shown in Fig.. A signal was rovie by an external cavity tunable laser tune to nm an moulate by a 10G non-return to zero on-off keye NRZ-OOK) seuo-ranom bit sequence PRBS) through a Mach-Zehner moulator MZM). 600s/nm of chromatic isersion was imose by a isersion comensation fiber DCF) after the moulator. Noise was rovie by two Erbium oe fiber amlifiers EDFA1 an EDFA) cascae together, with the outut of EDFA1 filtere to 4 nm an use as the inut for EDFA. The outut of EDFA was filtere using a 100G DWDM filter NEB = 0.6 nm). This was combine with the signal using a 3 B couler an fe into the inut of our OSNR monitor. The OSNR was change by attenuating the noise ower using a variable otical attenuator while keeing the signal ower constant. The reference OSNR was measure initially using a ower meter. The iezoelectric tube was oscillate at a low frequency ~1 Hz) an it was ossible to measure the hotoioe voltage at u to 00 khz through a ata acquisition car DAC). Figure 3a) shows the results of an OSNR measurement using a continuous wave signal source. Figure 3b) shows results obtaine with a 10G NRZ-OOK PRBS moulate signal. Figure 4a) shows the result of aing 600s/nm of isersion onto the signal. Figure 4b) shows the effect of fully olarize an unolarise noise on our measurement. Figure 5a) shows the results obtaine when a oorly biase moulator was use to generate the 10G signal. In Fig. 5b) an 5c) the signal was moulate at G an 5G resectively. Fig. 3. a) OSNR measurement using CW source; b) OSNR measurement using a source moulate at 10G. Outer ashe lines reresent an error of ± 0.5 B. # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 363

7 Fig. 4. a) OSNR measure using a 10G NRZ-OOK signal with 600s/nm isersion with the signal eye-iagram shown inset; b) OSNR results in the case of unolarise noise an fully olarize noise. Outer ashe lines reresent an error of ± 0.5B. Discussion Fig. 5. a) OSNR results when the MZ moulator use to generate the 10G signal is oorly biase; b) OSNR monitoring results with a 5G NRZ-OOK signal an c) a G NRZ-OOK. Outer ashe lines reresent an error of ± 0.5B. It has been shown in Fig. 3a) an 3b) that the monitor was caable of measuring OSNR values of u to 30 B within accuracy of ± 0.5B from a noisy CW signal or a 10G NRZ- OOK signal. When 600 s/nm of isersion was ae, the effect on results was seen to be negligible with a similar measurement range being attaine as shown in Fig. 4a). The effects of fully an artially olarize noise were investigate in Fig. 4b). This can affect measurement accuracy in both the olarization-nulling metho an the orthogonal olarization heteroyne mixing metho. In our case, there was little to no effect on measurement accuracy over the measurement range. A oorly biase Mach-Zehner moulator was emloye in the case of Fig. 5a), the effect of which woul be to change γ s ) an cause measurement errors if a single interferometer was to be use. Owing to the fact that our system has no reliance on the assumtion that γ s ) remains constant, a changing γ s ) # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 364

8 ha a minimal effect on the results obtaine. It was seen that increasing the bit erio to a value much greater than the otical elay by utilizing G an 5G NRZ-OOK signal ha little effect on the scheme as shown in Fig. 5b) an 5c). The otical elay values use in these measurements woul not be suitable for use in a 40G system. However, the use of shorter elays shoul accommoate this with reasonable ease with no further moifications to the setu. The accuracy of the system coul be further imrove by using etectors with lower noise an larger ynamic range because this can hel to measure the extinction ratios of the interferograms more accurately. Another issue is with the arabolic function we have use for γ s. This aroximation is simle an results in linear equations in 16). A more sohisticate aroximation may be more accurate but coul also be more comlex. Conclusions We have shown the oeration of a new tye of OSNR monitor base on two Michelson interferometers with ifferent otical elays uner several ifferent moulation sees u to 10G an uner the influence of numerous signal imairments foun in otical networks. Unlike other interference base systems, our system is ineenent of inut olarization an requires no ynamic olarization control. It is also immune to the effects of chromatic isersion, bit rate, oorly biase moulators an olarize noise u to OSNR values of 30 B an requires no rior knowlege of the signal amlitue autocorrelation function to oerate. # $15.00 USD Receive 15 Dec 009; revise 7 Jan 010; accete 9 Jan 010; ublishe 4 Feb 010 C) 010 OSA 15 February 010 / Vol. 18, No. 4 / OPTICS EXPRESS 365

A novel wavelength shift keying transmitter using a pair of Mach Zehnder modulators

Available online at www.sciencedirect.com Otics Communications 28 (2008) 257 252 www.elsevier.com/locate/otcom A novel wavelength shift keying transmitter using a air of Mach Zehnder modulators Hao Chi