A Novel Scalable Optical Packet Compression/decompression Scheme

A novel optical packet compression/decompression scheme is proposed allowing for high compression rates and large packet sizes thereby reducing the rate conversion latency. Extensive simulations are performed in order to investigate the feasibility of the proposed scheme. An OSNR of more than 22 dB is estimated for very large packets. Introduction Ultra-fast access to the optical medium can be particularly obtained by using optical packet compression/decompression technique. Recently, several methods for optical compression and decompression have been proposed. Most of these optical rate conversion schemes are based on an optical recirculating loop and a sampling technique /1//2/. However, this method leads to some restrictions concerning bit-rate and packet size. A further technique for optical packet rate conversion is a feed-forward delay-line structure consisting of q = log2 (N) stages reported in /3/. This technique allows simultaneous compression and decompression of N-bit large optical packets using the same device. Since only passive components are required it is easy to implement, but for larger packet lengths the splitting losses (3 dB per stage) must be compensated by an optical amplifier. However, the maximum packet size is here limited by the compression rate K as: Lp,max = K – 1, where K = T0/τ0. Consequently, for an 100:1 compression (K = 100) Lp,max is limited to 99 bits. Such short packets are usually impractical in many applications. Therefore, we propose a scalable optical packet compression/decompression scheme using a parallel arrangement of Optical Delay Line Lattices (ODLLs). This scheme allows simultaneous compression and decompression of optical packets not limited in size and compression rate. Packet Compression/Decompression Units The proposed scheme depicted in Fig. 1 consists of M parallel compression/decompression units, each of them responsible for the rate-conversion of a part of the optical packet. The principle operation of the device can be described as follows /5/: N–bit large high-speed input packets (at H-S In) are divided into M separate n-bit sequences using a splitter and M bi-directional optical gates. Those sequences are then copied n times in an n-bit ODLL (n = N/M). Each copy is delayed by (T0 – τ0) with respect to the next copy of the packet. A fast OTDM demultiplexer selects bits separated by the bit period T0 within very narrow switching window, thereby down-converting the highspeed sequence. Finally, all M sequences are delayed by an appropriate optical delay line DL2,m and combined by a M × 1 combiner. Thus, the whole high-speed input packet is down-converted. Moreover, by using the same device in the reverse direction, a low-speed packet (with bit period T0) can be compressed. The low-speed optical packet (at LS In) is first divided into M sequences by a 1 × M distributor. Each sequence is then delayed by an appropriate delay line DL2,m and compressed by an n-bit ODLL. The fully compressed sequences are finally selected by the gates and combined by a M x 1 combiner. The result is a high-speed output signal (with bit period τ0) that has the same bit pattern sequence as the low-speed packet. Note that both compression and decompression can occur simultaneously within the same device. M x 1 Combiner M x 1 Combiner 1 x M Splitter 1 x M Distributor OTDM DEMUX