Optical interconnect networks

Optical: systems and elements – Optical computing without diffraction – Logic gate

Patent

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

359109, 359115, 359117, G02F 300, G06E 100, H04J 1500

Patent

active

052454584

DESCRIPTION:

BRIEF SUMMARY
Many one dimensional (linear) networks have been proposed for a variety of purposes which have a number of parallel channels which are processed by several distinct stages in a pipelined manner. Each stage can be split into two parts: an interconnect stage where the lines are permuted followed by a layer of two-input two-output processing modules which operates on adjacent pairs of data channels.
Recently, efficient optical computers have been proposed which are networks where the interconnect part is performed using optics and the module part in some other medium, for example a chip of lithium niobate directional couplers (P. Granestrand et al, "Strictly non-blocking 8.times.8 integrated optical switch matrix" Electron. Lett. 22 No. 15 (1986)) or an optoelectronic integrated circuit (J. E. Midwinter, "Novel approach to optically activated wideband switching matrices" IEEE Proc. J 134 261 (1987)). Such machines have the physical layout described above. The use of optics for the interconnect stage has the advantages of high bandwidth, zero time skew and low crosstalk which give the whole processor a high throughout of parallel data. The various interconnection patterns which the networks employ can be generated using bulk or holographic optical components. Such an arrangement does not, however, take full advantage of the parallelism possible with optical systems.
Shing-Hong Lin et al in an article entitled "2-D Optical Multistage Interconnection Networks", SPIE Vol 752 Digital Optical Computing (1987) pp 209-216, describe the use of 2-D networks employing 2-D perfect shuffles interconnects and 2-D four-input, four output processing modules. Lin et al give no indication of the control structure necessary to achieve any particular network interconnection but rather point out that full, 24 cross-bar switches can in principle achieve a desired configuration.
It is an object of the present invention to provide an optical interconnect network having less structural complexity than such known 2-D networks. Accordingly an optical interconnect network comprises at least one stage which has an optical interconnect stage connecting a two-dimensional array of interconnect input ports to a two-dimensional array of interconnect output ports and an array of optical processing modules each having a two-dimensional array of module input ports, optically coupled to a respective interconnect output port, and a two dimensional array of module output ports characterised in that each module is functionally identical to a first and a second pair of two-input, two-output processing sub-modules in which each input of each of the second pair of processing sub-modules is connected to an output of a respective distinct one of the first pair of processing sub-modules.
The two dimensional network can be assembled using optics, with similar performance as before. There is a limit to the number of channels which can be accommodated which is proportional to either the maximum width of he module element or to the distance across which the optics can image faithfully. If this limit is N channels in the case of a one dimensional network then it becomes N.sup.2 for a two dimensional one. A further advantage of a network according to the present invention is that a network of a given size can be built much more compactly in a two dimensional form.
Because the present invention requires only a sufficient number of processing elements as are necessary to carry out the same processing as the four sub-modules, the structural complexity is reduced from that required to provide a full 4.times.4 processing yet, as will be shown later, it can be functionally identical to a concatenation of the processing modules of a one dimensional network and have the same control structure.
The present invention therefore allows the use of one dimensional network control structure with two-dimensional optical networks.
An embodiment of the present invention will now be described, by way of example only, with reference to the accompanying drawings, of which:
FIG. 1 i

REFERENCES:
patent: 4837855 (1989-06-01), Hajikano et al.
patent: 4931959 (1990-06-01), Brenner et al.
patent: 5010542 (1991-04-01), Pfaff
Lin, Shing-Hong, et al., "2-D Optical Multistage Interconnection Networks", SPIE Digital Optical Computing, vol. 752, (1987), pp. 209-216.
Midwinter, J. E., "Novel approach to the design of optically activated wideband switching matrices", IEE Proceedings, vol. 134, No. 5, Oct. 1987, pp. 261-268.
Granestrand et al., "Strictly Nonblocking 8.times.8 Integrated Optical Switch Matrix," Electronics Letters, Jul. 17th, 1986, vol. 22, No. 15, pp. 816-818.

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

Optical interconnect networks 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 interconnect networks, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Optical interconnect networks will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-2031839

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.