Compact header processing circuit for optical DPSK packets N. Calabretta, M. Presi, G. Contestabile and E. Ciaramella By means of only one SOA a compact all-optical circuit that simultaneously extracts the header information and produces a sequence of multiwavelength coloured pulses from the DPSK packets is demonstrated. Combined with an optical demultiplexer, this realises a highly scalable all-optical header processor based on time-to- wavelength conversion. Experimental results demonstrate operation on packets with eight different DPSK headers at 10 Gbit=s. The system requires low optical power, can potentially operate at high speed, is asynchronous and suitable for photonic integration in a single chip. Introduction: An all-optical header processor (AOHP) providing the packet’s forwarding information is an essential function to realise an all-optical packet switch (OPS). Critical features for practical imple- mentations of the AOHP are low power consumption, a limited number of active components and compactness. Most AOHPs are based on optical correlators [1, 2] each one recognising a distinct header. However, this technique is unpractical because it requires a high number of correlators, equal to the number of possible headers, and thus leads to an unscalable and power-consuming architecture. PG PM OG compact all-optical circuit SOA AWG AWG DI 100 ps WDL 20 km DFB p DFB 1 DFB 8 output 1 output N t 1 t i t N optical demultiplexer NRZ PRBS 2ˆ31-1 (10 Gbit/s) header 1 header i header N NRZ PRBS 2ˆ31-1 (10 Gbit/s) NRZ PRBS 2ˆ31-1 (10 Gbit/s) a b c pattern generator phase modulator optical gate header information extractor seed-pulse extractor multiwavelength converter SOA SOA SOA EDFA multi-l source PC 2 PC 2 PC 1 t PC 2 PC 1 PC 1 FBGs 550 ps optical demultiplexer l 1 l 2 l 3 l 4 BPF PBS PBS PBF PBS output 1 output 2 output 3 output 4 AWG EDFA FBG PC Fig. 1 Scheme of AOHP proposed in [3]; experimental setup for novel all-optical circuit; packet format to demonstrate AOHP a Scheme of AOHP proposed in [3] b Experimental setup for novel all-optical circuit PG: pattern generator; PM: phase modulator; OG: optical gate; SOA: semicon- ductor optical amplifier; WLD: wavelength dependent delay; DI: one-bit delay interferometer; AWG: array waveguide grating c Packet format to demonstrate the AOHP To overcome these limitations, we have recently demonstrated a novel scalable AOHP based on time-to-wavelength conversion, i.e. the forwarding information of each header is converted to a pulse at a distinct wavelength [3]. Fig. 1a shows the scheme presented in [3]. The DPSK packet format (shown in Fig. 1c) consisted of two header pulses and a data payload. The forwarding information is coded in the time- delay (t i , with i ¼ 1, ... , N) between the two header pulses [2, 3] (see Fig. 1c). The input DPSK packets are split into two arms. In the lower arm a header information extractor (HIE) extracts the two header pulses. In the upper arm, a seed-pulse extractor (SPE) extracts a pulse at the beginning of the packet and then a multiwavelength converter (MWC) generates N coloured copies of the seed-pulse. This results into a sequence of N coloured pulses properly delayed so that each pulse matches only one position of the second header pulse. When both the coloured pulses and the header pulses are coupled into an optical demultiplexer (OD), the header pulse demultiplexes only one of the coloured pulses. This pulse, at distinct wavelength, univocally represents the packet’s forwarding information [3]. Although this architecture requires a number of optical components by far lower than the other solutions, careful control of the polarisation state and optical power of the signals at the input of each of the four blocks make the overall system still complex. In this Letter we demonstrate a single compact all-optical circuit based on only one semiconductor optical amplifier (SOA) that can simultaneously implement three of the above operations in one shot. Namely it can effectively replace the HIE, the SPE and the MWC. This largely decreases the complexity and the number of active components in the AOHP [3]. Principle of operation: The proposed scheme is shown in Fig. 1b. The optical circuit consists of an SOA and a one-bit delay inter- ferometer (DI), followed by a wavelength dependent delay line (WDL). The incoming DPSK packets are coupled by an AWG with N local CW lightwaves (probes) and fed into the optical circuit. The SOA combined with the DI acts as an optical switch [4], where the packet acting as the pump has an optical power higher than the probes. In the SOA, the constant-envelope DPSK packet induces a cross-phase modulation on all the CW probes proportional to the packet intensity. Passing through the DI, the phase modulated CW probes are converted to amplitude modulation. As a pulse emerges at the DI output only if the signal and the one-bit delayed signal have significantly different phase [5], coloured pulses occur only at the beginning and at the end of the packet. These N coloured pulses at the DI output are fed into the WDL to obtain a sequence of pulses delayed so that each of the coloured pulses matches only one position of the second header pulse. We note that simultaneously the DI demodulates the two DPSK header pulses. As a result, the output of the all-optical circuit consists of both the demodulated header pulses and the sequence of N coloured pulses, which can be directly sent to an OD. As in [3], the second header pulse demultiplexes only a coloured pulse, representing the packet forwarding information. As mentioned, the same N-pulses sequence is produced at the packet end (see Fig. 2c). However, by setting a guard-time between the packets longer than the header section, those pulses fall within the packet’s guard time (see Figs. 2b and c), and therefore they can be neglected in the OD process. Experiments: The experimental setup is shown in Fig. 1b. A CW laser at 1540.54 nm was modulated by a LiNbO 3 phase modulator (PM) driven by a pattern generator (PG) at 10 Gbit=s. A programmed sequence of eight sequential packets with different headers (t i ¼ i  260 ps) was periodically produced by the PG according to the format shown in Fig. 1c. An optical gate (OG) was used to generate a packet guard-time of 3.2 ns. Eight CW lasers (from l 1 ¼ 1542.92 nm to l 8 ¼ 1548.9 nm, spaced by 0.8 nm) were coupled with the packet and fed into the all-optical circuit. The total power of the eight probes and the power of the DPSK packets at the SOA input were 3 and 2.2 dBm, respectively. The SOA had a 22 dB small-signal gain, saturation output power of  8 dBm at 180 mA and a polarisation dependent gain < 1 dB. The bit-delay in the DI was 100 ps to demodulate the DPSK 10 Gbit=s packets. The WDL was realised by 20 km of singlemode fibre (SMF). This introduced a wavelength dependent delay of 325 ps=nm in order to space the coloured pulses of 260 ps and thus to match the positions of the second header pulse of the headers. In other implementations, the SMF may be replaced by an array of Bragg gratings as in [3]. The input constant-envelope DPSK packets are shown in Fig. 2a. The demodulated input packets are shown in Fig. 2b, where the eight different headers with t i ¼ i  260 ps can be seen. Fig. 2c shows ELECTRONICS LETTERS 20th July 2006 Vol. 42 No. 15