Simulation of an all Optical Time Division Multiplexing Router Employing TOADs. Razali Ngaha, Zabih Ghassemlooya, Graham Swifta, Tahir Ahmadb and Peter Ballc a
Optical Communications Research Group, Sheffield Hallam University, S1 1WB, UK Phone: +44 (0)114 2253301. Fax: +44 (0)114 2253433. E-mail:
[email protected] b
Faculty of Science, University Technology of Malaysia
c
Fujitsu Europe Telecom R&D Centre Ltd, UB11 1AB, UK
Abstract Photonic packet switches can offer high data rates and format transparency as required by future communication system. Although electronic technology can achieve high switching speed, it is not compatible with the transmission bandwidths of the fiber-optic links. This paper is concerned with alloptical time division multiplexing router based on all-optical switching device. Three terahertz optical asymmetric demultiplexers (TOAD) have been used to model a simple 1X2 router. Simulation results show that synchronization between clock and data packet is achieved and the payload has successfully been switched to the correct destination port. Keywords: Clock recovery, Communication System, Optical Time Division Multiplexing, Optical Router, Terahertz Optical Asymmetric Demultiplexer, Synchronization. 1. Introduction Optical time-division multiplexing (OTDM) technology is a promising technique to realize the ultraspeed optical transmission system and to take full advantage of the wide bandwidth of the fiber. So far, 640 Gbit/s single carrier OTDM transmission system has been demonstrated [1]. One of the key challenges to be solved in ultra-fast all optical router for OTDM based data format is the clock recovery and consequently the synchronization between the ultra-high-speed packets with the clock signal. Conventional electrical clock recovery schemes are unable to cope with high-speed networks, with throughput of hundreds of gigabits per second. Therefore, there is no choice but to perform clock recovery in all optical domains. Several optical clock recovery techniques for ultrafast system have been reported among them are; the inject-locking of a mode locked laser [2], optical tank circuit by use of stimulated Brillouin scattering [3], and optical phase lock loop [4,5]. However, all these techniques are applicable only to synchronous OTDM networks. For asynchronous packetswitched OTDM networks, clock and data separation can be achieved by means of all-optical switching device combined with optical feedback [6]. This paper presents clock and data synchronization in OTDM router based on the ultrafast switching of semiconductor-based nonlinear interferometer with optical feedback.
2. OTDM Router A block diagram of a simple 1X2 OTDM router composed of three TOADs is shown in Fig. 1. The first TOAD is to extract the clock pulse from incoming OTDM packet for synchronisation purposes. This is achieved by employing a TOAD with feedback configuration, as shown in Fig. 2. The TOAD consists of a short length of optical fibre loop with a semiconductor optical amplifier (SOA) located at the centre of the loop. Polarization beam splitters (PBS) are used to couple the clock signal into and out of the loop. The clock pulse is the first to enter the loop through a 50:50 coupler, resulting in two counter-propagating pulses travelling in opposite directions within the loop. Since initially there is no control pulse present within the loop, the two pulses experience the same relative phase shift during propagation. They recombine at the 50:50 coupler and are then reflected back to the input port. The
reflected clock pulse is amplified and then passed through a polarisation controller (PC) before being fed back into the control port of the loop at orthogonal polarization to the data packet. The function of the control pulse is to create a switching window for the data packet (address bit and payload bit) to pass through to the output port of the loop. The clock pulse can be accessed via the PBS2 for synchronisation purpose at the later stages. With this approach, synchronization is achieved by locating only a single clock pulse, which has the same wavelength, polarisation and amplitude as the pulses in the payload, separated in time by fixed amount (width of clock pulse required) ahead of the data packet pulses. Unlike in [7], control pulses with different width are applied to vary the size of the switching window to cater for different packet length. The clock extracted from TOAD1 is fed to control port of TOAD2 in order to read the payload destination address bit. The clock signal is set so that the address bit and the payload is transmitted through the reflected and transmitted ports of the TOAD2, respectively. Address bit of TOAD2 is fed into the control port of TOAD3 for payload routing. In a single bit routing scheme, the packet with address bit of “1” are routed to the output port 2 (transmitted port), while packet with an address bit of “0” are routed to output port 1 (reflected port). Hence using this approach, photonic packets are selfrouted through an all-optical ultra fast switch without the need for optoelectronic conversion.
clo ck
D a ta p a ck e t
D a ta P a ck e t
D a ta P a ck e t
TOAD1 (C lock e x tra .)
TOAD2 (read ad d ress)
C lock
P ort 1
TOAD3 ( r o u te p a ylo a d )
P a yloa d
A d d ress
P ort 2
Fig 1: Block diagram of 1X2 OTDM router
SOA
PBS1
SOA
Fibre loop
PBS2
Clock out
PC
Reflected clock pulse
Clock + data packet in
50:50
Data packet out Reflected Port (Port 1)
Transmitted Port (Port 2)
Fig. 2: Clock recover module
4. Simulation Results and Discussion The above model was simulated using VPI Virtual Photonics simulation package. Relevant simulation parameters are shown in Table 1. Figure 3 shows an OTDM packet is composed of a clock pulse (first bit), followed by one bit of an address and two bits of payloads. All the signals have same amplitude, polarization and wavelength. The clock pulse has been chosen to have relatively wider width compared with address and payload bits to ensure that the switching window is wide enough to switch data packet. As discussed above the clock signal is extracted from the incoming OTDM signal using TOAD1, and the result is shown in Fig. 4a. The extracted clock pulse profile is the same as the input clock except for higher intensity due to TOAD switching gain. Similarly the signal at the transmitted output port of TOAD1 shows no change in the pulse width but an increase in the intensity. As expected the reflected port displayed higher gain compared to the transmitted port. TOAD2 separates the address bit from the payload bits with the former and latter emerging from the reflected and transmitted output ports, respectively, see Fig. 5. Notice that the address bit peak is non-flat, which is due to SOA amplification response profile. The payload bits from the TOAD2 (Fig. 5b) are routed to the transmitted output ports of the TOAD3 depending on the address bit. For address bit of “1” the payloads are routed to transmitted port (Port 2) of the TOAD3, as shown in Fig. 6.
Parameters Data bit rate per channel Clock pulse FWHM width Address bit FWHM width SOA injection current SOA length SOA active area SOA transparent carrier density SOA confinement factor SOA differential gain SOA linewidth enhancement factor SOA recombination coeff. A SOA recombination coeff. B SOA recombination coeff. C SOA initial carrier density Table 1: Simulation parameters
Fig. 3: OTDM input signal
Values 2.5 Gb/s 1 ns 0.5 ns 0.15 A 0.5 mm 3.0x10-13m2 1.4x1024m-3 0.15 2.78x1020m2 5.0 1.43x108 1/s 1.0x10-16m3/s 3.0x10-41m6/s 3.0x1024m-3
(a)
(b) Fig. 4: (a) Clock signal extracted of OTDM signal, and (b) transmitted output of TOAD1
(a)
(b) Fig.5: (a) Address bit, reflected output of TOAD2, and (b) payload, transmitted output of TOAD2
Fig. 6: Payload at the Port 2 of TOAD3
5. Conclusion In this paper a node model for an OTDM router (1X2) for asynchronous packet routing is presented. The switching devices employed for clock recovery and payload routing are carried out in optical domain using TOADs. Simulation results given demonstrate that clock recovery, address recognition and payload routing has been achieved successfully with very little crosstalk. Further work on crosstalk and noise analysis is currently underway for multiple input and output networks. References [1] M.Makazawa, E. Yoshida, T. Yamamoto, E. Yamada, A. Sahara, OFC’98, 1998, p.PD14 [2] L. Adams, E. Kintzer, J. Fujimoto, Electron. Lett. 3 (1994) 1696 [3] D.L. Butler, J.S. Wey, M.W. Chbat, G.L. Burdge, J. Goldhar, Opt. Lett. 20 (1995) 560 [4] S. Kawanishi, M. Saruwatari, J. Lightwave Technol. 11(1993) 2123 [5] O. Kamatani, S. Kawanishi, J. Lightwave Technol. 14(1996) 1757 [6] P. Toliver, I. Glesk, P.R. Prucnal, Opt. Commun. 173(2000) 101-106 [7] J.P. Sokoloff, P.R. Prucnal, I. Glesk, M. Kane, IEEE Photon. Technol. Lett. 5(1993) 787
