fonnance of the DSD light source and successful timeslot interchange between the four packets are well understood from Fig. 2. Polarisation guiding in ultralong distance soliton transmission T. Widdowson, A. Lord and D.J. Malyon Indexing terms: Opticalfibre polarirarion, Soliton transmission ~ Signal polarisation rotation due to polarisation dependent loss has been observed for the first time in an optical communications transmission system. In particular its effect as a limiting factor in solitonic polarisation multiplexed systems is discussed. An analytic expression for the degree of rotation is presented and experimentally verified. Additionally, SGbit/s solitons have been transmitted a distance of 1 I OOO km without transmission control. -52 - 48 -40 -44 received optical power,dBm Fig. 3 Bit eror rate curves of each packet 0 baseline A packet A 0 packet B U packet C 0 packet D The measured bit error rate curves taken for each packet are shown in Fig. 3. The bit-error free operations of each packet are accomplished under switching. Power penalties for each packet from the baseline are distributed in the range 4 6dB at a bit error rate of lW9. The power penalties are mainly caused by the imperfect extinction ratio performance of the 2x2 optical switches in the DSD light source and the optical TSI fabric. The power penalty variation of -2dB is due to the difference of the TSI switching pattern and the transmission imbalance of the arrayedwaveguide demultiplexer. In conclusion, it is experimentally revealed that the crossconnect switching fabric examined here successfully operates under optical switching in both the frequency and time domains. Bit-error free operation is demonstrated for four channels of 3OGHz-spaced 622 Mbit/s data in a 53 byte packet. ~ Introduction: Polarisation division multiplexing (PDM) has been proposed as a possible means of doubling the capacity of soliton transmission systems [I]. Demultiplexing a PDM signal may he achieved with polarisers provided that the two channels remain largely orthogonal on propagation. The advantage of this technique is that it offers greater immunity to Gordon-Haus (GH) jitter and soliton-soliton interactions (SSI) than optical time division multiplexing (OTDM). However, theoretical work to date [l] has assumed that there is no polarisation dependent loss (PDL) in the transmission path and, therefore, the two soliton data streams remain orthogonal. When a signal propagates through a PDL element the vector component of the electric field in the low-loss polarisation state experiences less attenuation than the component orthogonal to it. This has the effect of rotating the field very slightly towards the lower loss polarisation state. Thus, in a PDM system both channels will he ‘guided’ towards the same state of polarisation (SOP). Consequently, orthogonality is lost and demultiplexing with simple polarisers may not be possible. Indeed, in the limit an OTDM demultiplexer would be required, therefore losing the advatages offered by PDM. 90 I\ I I I 0 IEE 1994 25 March 1994 Electronics Letters Online No: 19940563 S. Kuwano, 0 . lshida and N. Shibata (NTT Transmission Systems Laboratories. 1-2356 Take, Yokosuka. Kanagawa, 238-03 Japan) H. Ishu and T. Kitoh (NTT Opto-Electronics Laboratories, 1-2356 Take. Yokosuka, Kanagawa. 238-03 Japan) o 50 References 100 150 200 (3300) (4950) (6600) section number m (distance, k m ) Fig. 1 Systematic variation of f3 assuming all low loss states aligned at 45” to launch fwld vectors ( P D L = 0.165dB). and 60 variation of p allowing PDL elements to be randomly orientated (PDL = 0.165dB) (1650) 1 KOBRINSKI, H., VECCHI, M.P., GOODMAN, M S., GOLDSTEIN, E L . , c , and MENOCAL, s 0.: ‘Fast wavelength-switching of laser transmitters and amplifiers’, IEEE J. Se]. Areas Commun., 1990, SAC-8, (6), pp. 1190-1201 2 GLANCE, B , KOREN, U., BURRUS, C.A., and EVANKOW, I.D.: ‘Discretelytuned N-frequency laser for packet switching applications based on WDM’, Electron. Lett., 1991, 27, (15), pp. 1381-1383 3 SHIBATA, N., ISHIDA, o., TADA, Y , KUWANO, s., TOHMORI, Y., and SUZUKI, s.: ‘Performance of four-channel FDM crossconnect switching without bit loss’, Electron. Lett., 1993, 29, (9), pp. 800CHAPLJRAN, T E., COOPER, I.M., TUR, M., ZAH, 801 s., TADA, Y., and SHIBATA, N.: ‘IO0 ps frequency switching without bit loss for a lOGb/s ASK modulated signal’, IEEE Photonics Technol. Lent., 1993, 5, (3), pp. 35&356 5 HUNTER, D K., ANDONOVIC, I , CULSHAW, E., and BARNSLEY, P.E.: ‘Experimental test-bed for optical time-domain switching fabrics’. Dig. Conf. Optical Fiber Communicationht. Conf. on Integrated Optics and Optical Fiber Communication, 1993 Tech. Dig. Series, 1993,4, Tu02, pp. 71-72 6 ISHIDA, o., TADA,Y., SHIBATA,N., and ISHII,H.: ‘Fast and stable frequency switching employing a delayed self-duples (DSD) light sonrce’, to be published in IEEE Photonics Technol. Lett. 7 TAKAHASHI. H., NISHI, I., and HIBINO, Y.: ’IOGHZ spacing optical frequency division multiplexer based on arrayed-waveguide grating’, Electron. Lett., 1992, 28, (4), pp. 380-381 4 KUWANO. ELECTRONICS LE77ERS 26th May 1994 Vol. 30 (i) variation of t3 assuming low loss stat= aligned at 45” (ii) 60 variation of f3 Theory: The direction of an electrical field is rotated by a PDL element. Two initially orthogonal fields are rotated towards each other causing a reduction in the angle p between them. Assuming all i PDL elements are aligned with the low loss axis bisecting the two fields, then by summing the rotations of the fields caused by every element it can be shown that Fig. 1 shows this curve for P D L = 0.165dB to allow comparison with experiment. Fig. 2 shows the variation of total power of the two fields measured both parallel and orthogonal to the low loss state. In a real system the PDL elements will be randomly oriented to the incoming signal polarisation. Because there is no preferred No. 11 879 direction, the mean relative rotation of the fields will be zero. However, there will be a significant standard deviation about this mean caused by a random walk effect. The second curve of Fig. 1 shows the 60 condition for a system with elements having a mean PDL of 0.165dB. Clearly random alignment cannot be relied on to eliminate the effect. L . s 0 ,- 23 42 32 564 loss state IOW distance.2000 km/division a b Fig. 4 Evolution of power in the SGHz spectral component a Low loss state b High loss state high 1055 state -1 5 I \, I I 50 200 (1650) (6600) 100 150 (3300) (4950) section number (distance , km 1 Fig. 2 Theoretical power in high loss and low loss states 0 33 km 33 km 33 km DSF DSF DSF with the theoretical plot given in Fig. 2 in which timing jitter is not modelled. Fig. 5a shows the pulse pattern detected through the polariser after lOOOkm of transmission with the polarisation at the receiver adjusted such that maximum signal power was obtained from one of the channels (channel 1). It can be seen that channel 2 time .ZOO ps/division time, 500 ps/division a b rn Fig. 5 Pulse pattern ai IDODkrn and pulse pattern at 156Wkm a Pattern at IOOOkm (bandlimited to 32GHz) b Pattern at 15600km (bandlimited to 1.76GHz) Fig. 3 Recirculating loop confipuration Experiment: The experimental configuration is illustrated in Fig. 3. Details of the soliton source and recirculating loop transmission path can he found in [2]. The ring laser pulse stream was modulated with a 2’-1 PPBS pattern and hit interleaved to form a 2 x 2.5Gbitk PDM/OTDM transmitted signal. At the receiver the signal was optically time division demultiplexed to 2.5Gbith via a 20GHz LiNbO, modulator, providing a 170ps acceptance window to soliton jitter. A polariser with 30dB extinction was situated in front of the demultiplexer to ensure that solitons from only one polarisation state were detected. The mean transmission fibre dispersion at the operating wavelength was 0.5psinmlkm and the mean PDL per section was 0.165dB. Polarisation controllers within the loop allowed accurate control of the signal SOPS. In particular, the eigenstate that corresponds to the signal expenencing the same polarisation state on every recirculation could be found. The transmitter data stream was then configured such that both signal SOPs were at 45” to the low loss state of the loop. Results and discussion: Fig. 4 details the evolution of power in the 5GHz spectral component of the pulses in both the low loss and high loss eigenstates of the loop. With the polariser aligned to the low loss state (Fig. 4a) an increase in power of 3dB is observed as the two channels rotate from initial orthogonality to being coincident in the low loss state at 3000km. With further transmission the spectral component diminishes due to the accumulation of timing jitter. Power in the high loss state (Fig. 46) falls rapidly as the signal rotates into the low loss SOP. Fig. 4 is in excellent agreement 880 is no longer orthogonal, 66% of its power is in the same polarisation state as channel 1. To ascertain the validity of the data, BER measurements were performed on both channels after the signal had passed through the polariser. Error free operation was possiBER was measured ble to a distance of IlOOOkm where a this compares well with other work in which significantly stronger wavelength guiding was used [3]. Further transmission resulted in errors due to soliton timing jitter. This was due to the combined effects of GH and SSI. Because transmission control is not used, GH builds rapidly over the first few thousands of kilometres and results in the possibility of adjacent solitons being jittered much closer together. Also over this distance the two channels rotate into the same polarisation state. Consequently, on subsequent transmission, far stronger interaction forces are experienced than would be expected for unperturbed solitons. Fig. 56 shows the received pulse pattern at 15600 km revealing SSI on adjacent ‘1’s with isolated ‘1’s appearing to be unaffected. To reduce the interaction forces one of the channels was disconnected at the transmitter and BER measurements performed on the other through the OTDM demultiplexer. Error free operation was possible to a distance of 15600km which is in agreement with analytic theory as being the G H limit for this system. Thus, SSI was limiting the maximum transmission distance in the two channel experiment. 0 IEE 1994 Electronics Letters Online No: 19940618 8 April 1994 T. Widdowson, A. Lord and D. J. Malyon (ET Laboratories. Martlesham Heath. Ipswich, Suffolk IPS 7RE. United Kingdom) References 1 EVANGELIDES, S G., MOLLENAUER, L.F., GORDON, I.P., and BERGANO, N.s.: ‘Polarisation multiplexing with solitons’, J. 2 3 Lightwave Technol., 1992, LT-10, (I), pp. 28-35 WIDDOWSON, T., MALYON. D.I., SHAN, x., and WATKINSON, P.J.: ‘Soliton propagation without transmission control using a phase locked erbium fibre ring laser’, Electron. Lett., 1994, 30, (8), pp. 661463 SUZUKI, M., TAGA, H , EDAGAWA, N., TANAKA, H I YAMAMOTO, S., and AKIBA, A.: ‘Experimental investigation of Gordon-Haus limit on soliton transmission by using optical short pulses generated by an InGaAsP electroabsorption modulator’, Electron. Lett., 1993, 29, (18), pp. 1643-1644 ELECTRONICS LEl7ERS 26th May 1994 Vol. 30 No. 1 I
1/--страниц