Optical buffer utilizing a cell pointer

Details Optical buffer utilizing a cell pointer Optical buffer utilizing a cell pointer

1999-01-19

2002-07-16

Pascal, Leslie (Department: 2633)

Optical: systems and elements

Deflection using a moving element

Using a periodically moving element

C359S199200, C359S199200

Reexamination Certificate

active

06421152

ABSTRACT:

FIELD OF INVENTION
The present invention relates to an apparatus for buffering data; and, more particularly, to an apparatus for buffering data by using an optical buffer so that optical cells of asynchronous transfer mode can be stored and processed without the restriction of processing time for a high speed network.
DESCRIPTION OF THE PRIOR ART
Wavelength division multiplexing (WDM) is conventionally used for a high speed network. Recently, because of a limitation of flat-gain bandwidth of an optical fiber amplifier and a limitation of a wavelength selectivity of an optical filter for choosing a needed signal at a receiving end, a time division multiplexing (TDM) is combined with the Wavelength Division Multiplexing to increase total transmission capacity of a network. At this high speed network, a high speed optical buffer is required to avoid the collision of cells transmitted on the same channel.
With a usual optical buffer design, the information of each cell should be read by the optical buffer controllers to make decisions on buffering periods for each cell, so that the decision making electronic circuit is required indispensably. However, a limitation of processing of the electronic circuit makes such buffering difficult to use at the high speed network.
As shown in
FIG. 1
, wavelength of asynchronous transfer mode cells are converted in sequence at a buffer input part in order to have delays of needed periods. A cell should be converted to &lgr;
2
at the buffer input part to be stored for the amount of one cell period. This wavelength converted cell is demultiplexed by a wavelength division demultiplexer (Wave length Division Demultiplexer). This cell, which is demultiplexed in wavelength is routed to a wavelength converter for the wavelength conversion to &lgr;
1
. For longer delay, this cell should be converted to &lgr;
3
at the buffer input part to be stored during the amount of the two cell periods. This cell is converted to &lgr;
2
and &lgr;
1
by wavelength conversion for each looping, and is output after the total two cell periods (T).
As can be seen in
FIG. 1
, the first wavelength is output without passing the delay line, the second wavelength is output through the delay of one cell period, and the Nth wavelength is output through the delay of the N−1 cell periods.
If two cells arrive simultaneously at the said buffer input part, one cell is converted to &lgr;
1
, and the other cell is converted to &lgr;
2
. The cell converted to &lgr;
1
, does not experience the delay, however the other cell converted to &lgr;
2
experiences the delay of one cell period.
With the buffering scheme in
FIG. 1
, the wavelength should be assigned sequentially on incoming packets which are wavelength multiplexed. Otherwise, there is a problem of time utilization. The reason is that time slots of the number in association associated with the vacant wavelengths cannot be used in time domain, if there are vacant wavelengths in the wavelength span of &lgr;
1
~&lgr;n. When &lgr;
2
does not exist between &lgr;
1
and &lgr;
3
, &lgr;
1
cell is output without a delay. However, &lgr;
3
is converted to &lgr;
2
and experiences the delay of one cell period, then &lgr;
2
is converted to &lgr;
1
to be output. That is, &lgr;
2
experiences unnecessary delay of one cell period, which causes a vacant time slot of one cell period between &lgr;
1
and &lgr;
3
. This causes a waste for total network transmission capacity. Therefore, in order to avoid the vacant time slots of the buffer, the wavelength should be reallotted sequentially at the buffer input part. To do so, it should be known how many cells are injected through the all inlets before they arrive at the wavelength converter
1
. In order to acquire this information, an electronic detection component and a processing component should be integrated in a buffer, which becomes the restriction factor of buffering capacity for every channel.
Because the buffer consists essentially of the wavelength converter utilizing a cross gain modulation (XGM), there are severe restrictions on the maximum number of wavelength conversion. Because the wavelength converter utilizing XGM has no good contrast ratio, a cell which is allotted to &lgr;n should undergo wavelength conversion by N−1 times, so because of the influence of an amplifier spontaneous emission (ASE) noise of a semiconductor optical amplifier (SOA) and because of the degradation of the contrast ratio, it is difficult to get a satisfied packet signal at a final output.
As shown in
FIG. 2
, the optical buffer utilizing a frequency division multiplexing (Frequency Division Multiplexing) method consists essentially of an optical delay loop
4
, a quasi 2*2 switch
5
, and a fast frequency-selective filter
6
. This buffer has the function of multi-input and one-output. Multiple input packets are injected to the vacant optical delay line loop
4
at the same time. The fast frequency-selective filter
6
chooses one of the wavelength multiplexed packets which are copied for every looping in the optical delay line. Frequency division multiplexed Packets, “&lgr;A, &lgr;B and &lgr;C” are entered to the optical delay line loop
4
, and are produced at the quasi 2*2 switch
5
through the optical gate Gr
8
and the optical coupler (transfer mode: Gt=open, Gc=close)
9
. At the same time, the packets are injected to the optical delay line loop
4
. And they are inserted to the optical delay loop through the optical gate Gc
7
and the coupler (circulate mode: Gt=close, Gc=open) at the next time slot. During a circulate mode (Circulate Mode), the loop makes two copies of the packets and stores one copy in the loop, and send the other copy to a fast frequency-selective filter.
As shown in
FIG. 3
, the optical delay line loop
4
transmits copies of the original packets, first. And, it produces the frequency division multiplexed packets “&lgr;A, &lgr;B and &lgr;C” which are copied two times during the three time slots. As the filter
6
continuously selects “&lgr;A, &lgr;B and &lgr;C”, packets with the same wavelength sequence transmits the frequency division multiplexing buffer.
As described above, in order to select one packet out of the copies of wavelength division multiplexed packets, one of the optical gates of the fast frequency-selective filter should be opened for one time slot, then be closed promptly. However, in order to control this, it is difficulty to acquire the information that what wavelengths are multiplexed and injected. Furthermore, in order to detect the injected wavelengths, good amount of electronic signal processing component should be integrated to a buffer, so the processing speed per channel is restricted.
When N packets multiplexed with N wavelengths come to the buffer, they should be copied by N times through the optical delay line loop
4
, so there are loss related problem. Even the semiconductor optical amplifier (SOA) and the filter are inserted at optical delay line loop
4
to reduce the noise accumulation, it can not store packets indefinitely because of the amplifier spontaneous emission (ASE) noise.
SUMMARY OF THE INVENTION
It is, therefore, an object of the present invention to provide an apparatus for buffering data by using an optical buffer utilizing a cell pointer so that optical cells of asynchronous transfer mode can be stored and processed at high speed without the restriction of processing time for a high speed network, wherein an internal switch of the apparatus would be controlled by extracting optically the cell pointer indicating a starting point of a cell by using terahertz optical asymmetric demultiplex (Terahertz Optical Asymmetric Demultiplexer) in order to avoid a collision of each cell which is transmitted on a same channel especially at the high speed network.
In accordance with the present invention, there is provided an apparatus comprises: a 1*N arrayed waveguide grating (AWG) demultiplexer that separates the simultaneously incoming wavelengths at the input part of the high speed optical buf

Affiliated with

Also associated with

Rating

Optical buffer utilizing a cell pointer 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 buffer utilizing a cell pointer, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Optical buffer utilizing a cell pointer will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2865698