Optical packet processing

Details Optical packet processing Optical packet processing

1996-12-24

1998-11-03

Negash, Kinfe-Michael

Optical: systems and elements

Deflection using a moving element

Using a periodically moving element

370500, 370520, 375354, 359158, H04J 1408

Patent

active

058317520

DESCRIPTION:

BRIEF SUMMARY
BACKGROUND TO THE INVENTION

1. Field of the Invention
The present invention relates to a method and apparatus for processing a multi-bit packet carried on an optical network. Typically the package might comprise an ultrafast binary data signal having a bit rate of 10 Gbit/s or more and carrying e.g., telecommunications traffic.
2. Related Art
The term packet as used herein encompasses both asynchronous packets as transmitted, for example, in an ATM system, and packets of synchronous data, such as frames in a synchronous OTDM system.
Ultrafast binary data signals can be generated quite readily in return-to-zero (RZ) format by using a source of ultrashort e.g. (picosecond) pulses at a repetition frequency which is a sub-multiple of the required bit rate. These pulses are then each split into a number of separate optical paths, modulated using electro-optic devices, time-delayed and then recombined so as to interleave them to achieve an capability of electronics, any subsequent signal processing (such as retiming, regeneration or demultiplexing) must use all-optical technologies. This in turn in general will require the provision of an optical clock signal in precise (picosecond) bit synchronisation with the data.
Whilst in ultrafast circuit-switched or synchronous transmission systems, clock recovery can be carried out using phase-locked loops, this is not possible for asynchronous transmission systems and especially asynchronous packet-switched systems because clock recovery has to be performed on a packet-by-packet basis in a small fraction of the packet duration. This would require a phased-locked loop with an extremely short acquisition time (possibly as short as a few tens or hundreds of picoseconds). Even if the loop round-trip delay in an optical phase-lock loop could be reduced by making the optical path lengths very short by using e.g. some kind of integrated optical device, for the phase-lock loop, to work robustly it would still require a large number of signal pulses to acquire lock with sufficient phase-error signal quality and with sufficient lock-in range. The acquisition times would therefore still be much longer than the packet duration in a packet-switch system. Moreover even if locking could be achieved in time, the short acquisition time would then make the phase-locked loop more sensitive to disturbance by noise. a comb of clock pulses before each data packet. Then at each position in the network where signal processing of the packet is to be performed, the comb of clock pulses is separated from the data using optical couplers, fast photodetectors, electronic sense circuitry and an optoelectronic space switch. However this approach has serious disadvantages. Firstly, the payload traffic capacity of the network is reduced by the additional overhead associated with the transmission of the comb of clock pulses. Although this might be minimised by transmitting only a relatively short segment of the clock comb there would nonetheless have to be a time guard band between the clock comb and data packet long enough to accommodate the time uncertainty in the sense circuitry and the commuting time of the optoelectronic switch. This guard band may need to be as long as 0.5-1 ns, equivalent to 50-100 bit periods at a data packet bit rate of 100 Gbit/s. A second disadvantage is that the clock comb, or segments of it, will be subject to the same transmission impairments as the data packet, suffering, for example, amplitude noise and timing jitter arising from This would limit the uses to which the clock pulse train could be put.
Another proposal disclosed in Zhang et al., Electronics Letters, Vol. 29, no. 21, 14 Oct. 1993, pp 1871-1873, uses clock pulses transmitted on a separate wavelength channel from the OTDM packet. Since the clock pulses are at a different wavelength, their phase relationship to the frame varies as a result of dispersion over the transmission path.
A paper published in Journal of Lightwave Technology, vol. 11, no. 5/6, May 1993, pp 829-835 discloses a system in which a synchrono

REFERENCES:
patent: 5172258 (1992-12-01), Verber
patent: 5457559 (1995-10-01), Saito et al.
Driessen, "Robust Frame Synchronisation Using Standard VLSI Synchronous Input/Output Devices" Electronics Letters, 27th Oct. 1988, vol. 24, No. 22. pp. 1381-1383.
Hermes et al., "LOCNET--an Experimental Broadband LAN With Optical Switching", Optical Communication, ECOC, '84, Tenth European Conference, Sep. 3-6, 1984, pp. 274-275.

Affiliated with

Also associated with

Rating

Optical packet processing does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Optical packet processing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Optical packet processing will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-696059