Switch for optical signals

Details Switch for optical signals Switch for optical signals

2000-02-23

2003-08-12

Healy, Brian (Department: 2874)

Optical waveguides

With optical coupler

Switch

C359S199200

Reexamination Certificate

active

06606427

ABSTRACT:

FIELD OF THE INVENTION
This invention relates to optical switches and is particularly concerned with switches for switching optical signals composed of light of predetermined wavelengths, for example, Dense Wavelength Division Multiplexed (DWDM) optical signals used in optical telecommunications.
BACKGROUND OF THE INVENTION
Optical transmission systems achieve their end-to-end connectivity by concatenating multiple spans between intermediate switching nodes to achieve an overall end-to-end path. When the end-to-end granularity of any given transmission path is a fraction of the capacity of a given optical carrier, time division multiplexing is used to share the overall bandwidth, mandating the use of electronic switching in the intermediate nodes. However, the availability of Dense Wavelength Division Multiplexing (DWDM), combined with the availability of high capacity ports on data switches and routers, has increased the demand for concatenation of individual spans to make end-to-end connections at the wavelength level.
DWDM optical networks transmit multiple channel signals on each optical fiber in the network; each channel signal is modulated light of a predetermined wavelength allocated only to that signal. The result is a plurality of optical carriers on each optical fiber, each optical carrier carrying a channel signal separated from other carriers in optical wavelength. Current DWDM optical networks typically convert channel signals into electrical signals at every switching node in the network because optical switches having sufficiently large enough port counts are not available. To convert the channel signals to electrical signals, transponders are used at every port of the switching node and for every channel wavelength. As DWDM signals become denser, that is, as the number of channels per optical fiber increases, the required accuracy of the transponders, and hence the cost, also increases. Moreover, as the number of ports per switching node increases, the required number of transponders also increases. Consequently, large networks carrying dense DWDM signals require many costly transponders and are therefore costly to build.
To overcome this problem it has been proposed to build large, purely optical switches in various forms, to reduce or eliminate the need for opto-electronic conversion in order to switch channel signals electrically. Some effort has gone into conceiving methods of building very large switches that offer full connectivity between all their ports. However, fabrication of these large optical switches has proven difficult.
Many attempts to create a large non-blocking optical switch use a large number of small switch modules to create a multiple stage switch. One example of this envisages building a 128 port×128 port switch out of three stages of multiple 16×16 crosspoint matrices, or a 512×512 port switch out of three stages of multiple 32×32 crosspoint matrices, in a three stage CLOS architecture. The above is based on the availability of 16×16 or 32×32 switch matrices in the form of Micro-Electro-Mechanical (MEM) switch matrices (e.g. “Free-space Micromachined Optical-Switching Technologies and Architectures”, Lih Y. Lin, AT&T Labs-Research, OFC99 Session W14-1, Feb. 24, 1999). Other multi-stage approaches use smaller matrices and more stages. Even the 3 stage CLOS architecture is limited to 512-1024 switched wavelengths with 32×32 switch matrix modules, which, in today's 160 wavelength per fiber DWDM environment, is only adequate to handle the output/input to 3-6 fiber pairs (480-960 wavelengths). Furthermore, the optical loss through each crosspoint stage (typically ~5 dB with a 16×16 or 32×32 MEMs device) is compounded by the use of three stages, plus a complex interconnect, to provide switch losses in the range of 15-18 dB.
Such multi-stage switches, even at three stages, have significant problems. These problems include high overall optical loss through the switch, since the losses in each stage are additive across the switch, and there is the potential for additional loss in the complex internal interconnect between the stages of the switch. Size limitations in terms of the number of wavelengths switched can be overcome by going to a five stage CLOS switch, but this further increases the loss through the switch as well as adding to its complexity and cost. Using current loss figures, the loss through a 5-stage switch would be in the order of 25-30 dB. This amount of loss is at or beyond the operating link budget of modern high-bandwidth transponders. In addition, one of the major cost-centres is the cost of the MEMs switch modules (or other small matrix modules). Sensitivity of the overall switch cost to the cost of the MEMS modules is exacerbated by the fact that a CLOS switch requires a degree of dilation (i.e. extra switch paths) to be non-blocking and that each optical path has to transit three (or five) individual modules in series.
In U.S. Pat. No. 5,878,177 entitled “Layered Switch Architectures for High-Capacity Optical Transport Networks” and issued to Karasan et al., on Mar. 2, 1999, another approach is disclosed. This approach relies on providing signals received by a switching node with access to any route leaving the node, but not access to every signal path (fiber) on those routes. In this way, Karasan's switching node avoids the large number of switch points that a fully interconnected, or fully non-blocking, switch fabric would require. Although this approach may be adequate at the node level, or even for small networks, it adds further complexity to network planning, which would become increasingly difficult with larger networks.
Some prior art approaches attempt to generate large, general purpose, non-blocking switches, which are then coupled to DWDM multiplexers for coupling into output fibers. This results in substantial waste of the capacity and capability of the non-blocking generic switches, since the DWDM multiplexers are themselves blocking elements on all their ports to any optical carrier except an optical carrier within the specific passband of that port of the multiplexer. Hence the non-blocking switch structure contains many crosspoints that direct specific input ports carrying a given wavelength to output ports that cannot support that wavelength, since it would be blocked in the WDM multiplexer. Such crosspoints cannot be used in operation of the switch, and this wasting of crosspoints makes inefficient use of expensive optical switching matrices.
Optical transmission networks that rely on electrical switching and electrical regeneration at intermediate nodes require one pair of transponders per wavelength channel at each intermediate switching node. Consequently, as the number of wavelength channels per fiber grows, the number of transponders and the resulting costs grow in proportion to the number of wavelength channels.
Optical transmission networks that rely on “opaque” optical switching and electrical regeneration at intermediate nodes experience the same growth in transponder number and cost. (In “opaque” optical switching, incoming optical signals are converted by transponders into different optical signals that are switched optically before being converted by further transponders to different optical signals for further transmission.)
However, in optically switched networks that use cascaded optical amplifiers to compensate for fiber loss on each span and for optical insertion loss of the optical switches, each optical amplifier simultaneously amplifies all wavelength channels on each fiber without the use of transponders. Consequently, the number and cost of the optical amplifiers does not grow with the number of wavelength channels per fiber, and the cost benefits of optically switched and amplified networks relative to electrically switched and regenerated networks increases with the number of wavelength channels per fiber.
Moreover, the cost advantages of optically switched and amplified networks over electrically switched and regene

Affiliated with

Also associated with

Rating

Switch for optical signals does not yet have a rating. At this time, there are no reviews or comments for this patent.

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

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3087582