Optical Switching

Author: Justin Luu (00047629)

Warning: This assignment has been written by an undergraduate student of Communication Networks of UTS Engineering

Statement Of Originality:

I hereby declare that all work contained within this website is that of my own, unless stated otherwise.

Acknowledgements

I acknowledge the authors any images which were used for this webpage. Their respective references are under each image

Contents:

    1.Evolution of Optical Switching

    2. Technical Aspects of Optical Switching

2.1 Types of Optical Switches

2.2 Protocols and Standards

2.3 Architecture and operations

    3. Applications

    4. Technologies

4.1 MEMs

4.2 Liquid Crystals

4.3 Bubbles

4.4 Thermal Optics

        5. Issues

5.1  Advantages of Optical Switching

5.2  Alternatives and Problems of Optical Switching

5.3  Companies producing Optical Switching

    6. Future of the technology

Key Learning Points

Review Questions

    Multiple Choice Questions

    Answers

References

Glossary

Acronyms

Appendix

 

Body of Report

1. Evolution of the technology

There has been an ever increasing demand for higher transfer rates, and larger bandwidths. This demand has been met with an increasing interest and development of optical networking technologies, and the setup of optical networks.

An optical network consists of a network where optical fiber serves as the fundamental medium of transmission. However, switching, signaling and processing functions are accomplished electronically. Hence, two types of optical switches have emerged, as described later.

Optical switches evolved naturally from the development of optical fiber technology. As the technology grew, new methods were required to channel, switch and multiplex these high speed signals. Hence in 1996, the first all-optical switch was created by Corvis Corporation. Other companies have tried to emulate and enhance this design and have come up with many innovative ideas, as will be discussed later.

2. Technological Aspects of Optical Switching

2.1 Types of Optical Switches

An optical switch is simply a switch which accepts a photonic signal at one of its ports and send it out through another port based on the routing decision made.

There are two fundamental types of optical switching methods. There is the all-optical switch (OOO) which the switching fabric is done purely through photonic means, and the more intelligent electro-optical switch (OEO), which requires the analogue light signal is converted to a digital means first to be processed and routed, before being converted back to an analogue light signal. An illustration of these switches can be seen below.

Obtained from [4]

. Obtained from [4]

A future is envisioned where an All Optical Network (AON) can become a reality. An AON will be transported, switched, and managed all at an optical level. Hence, this will result in it being cheaper and faster than our existing optical network with electrical parts. However, our current technology is not sufficiently advanced for this to occur. While we wait for our network to evolve, we must continue using OEO technology, for both OEO and OOO switching have their own place in today’s network. Both technologies have their own benefits and competition exists in how they both can interoperate.

The OEO switch is a technology already used in networks today. The problem with this technology is that may not be able to keep up with the speed of optical transmission in the future. Although it is not a problem today, the amount of data transmitted today is growing twice as fast as the ability to switch it. OEO switches are also bit rate and protocol dependant, which means that protocols must be used and bits from the transmitted frames must still be processed (unlike with OOO swiitches). They have the ability to process header information, and are able to make routing decisions based on this.

The OOO switch holds the promise of the future AON, but it is still an emerging technology and cannot adequately function without the intelligence that OEO switches currently hold. The different implementations and of the OOO fabric are discussed later in Technologies.

2.2 Protocols and Standards

Optical networks use the standard protocol of traditional networks, such as IP and Ethernet, while also requiring new protocols for its operation.

OEO switches are already being used within the confines of SONET (North America, Japan and Korea) and SDH (rest of the world). SONET and SDH are multiplexing protocol standards which were setup to support the very fast data rates required in optical networks. With data rates of 10Gb/s (and more than 40Gb/s in the future), more sophisticated reliability and network monitoring measures must be introduced. In terms of the OSI model, SDH/SONET protocols relate to the physical layer (layer 1), in which frames must be encased in before they are able to recognized or processed by SONET/SDH equipment. An explanation of an SDH frame is described below.

 

                 Obtained from [4]

An SDH frame is called a STM1 (Synchronous Transport Module) or an OC3 (Optical Container) for SONET. The first nine bytes of the SDH  is the Section Overhead, which contains information used for error detection, network management and protection switching (explained in later in Applications) of frames during tranfer within the SDH network. The rest of the frame is called the VC4 (Virtual container) which includes a Path Overhead byte. This byte is used for quality monitoring as well as information about the data type of the frame. The frams's data is contained in the rest of VC4.

SONET and SDH are the main protocols used within the optical network.

2.3 Architecture and Operations

Optical switching networks makes use of Dense Wave Division Multiplexing (DWDM). This is a multiplexing technique where multiple signals can be shared on a single optical fiber, with each signal sharing a different wavelength (ie. a different spectrum of light). DWDM systems today carry up to 160 different signals on a single fiber. This greatly enhances bandwidth of networks today.

The Optical switch will soon be operating within a network architecture known as GMPLS (Generalised MPLS). GMPLS is an extension to MPLS (a protocol which enhances efficiency of data being sent) to add control capabilities to SONET/SDH and DWDM equipment. According to the GMPLS draft (currently being developed by the Internet Engineering Task Force (IETF)) , there will be four types of interfaces on a switch namely PSC, TDM, LSC and FSC.

PSC (Packet-Switch Capable) include interfaces which are able to recognize bits, packet boundaries and can process control information. They are used for routers and ATM switches. TDM (Time Division Multiplex Capable) is an interface which can recognize repeating and synchronous frames, and can forward frames on the basis of its information position within the timeslot. It will be used for digital multiplexers and cross connects. LSC (Lambda Switch Capable) are for devices which cannot recognize bits or frames, such as OOO switches and multiplexers, and forwards information on the basis of wavelength. FSC (Fiber Switch Capable) interfaces cannot recognize bits and do not have access to wavelengths. They switch or forward information on the basis of the physical position of the fiber itself.

3. Applications

There are a variety of applications optical switching devices are typically used for.

As described earlier, switches are used in high speed optical networks. They require very high switching speeds (a 53-byte packet at 10 Gbs/s is 42ns long), to keep up. They will also require very large switches to handle the amount of traffic. Switches for this function tend to be used within optical cross-connects (OXC). Optical Cross Connects are similar of electronic routers which forward data using switches. An OXC may contain a whole series of Optical Switches.

Optical Switches are also used for Protection Switching. If a fiber fails, the switch allows the signal to be rerouted to another fiber before the problem occurs. However, this operation must take milliseconds, including the time required to detect the failure, inform the network and the actual switching time. Normally this operation requires a 1x2 switch but with complicated cross-connects hundreds may be required. 

Optical Add – Drop Multiplexers (OADM) are used to remove some wavelengths from a multiwavelength signal without requiring it to be completely demuxed. Most OADMs use a number of optical switches (namely 2x2) to switch signals from a DWDM stream. These OADMs allow carriers to selectively remove wavelengths from a signal using software as opposed to OADMs without optical switches.

Optical switches are also useful for many other applications including, external modulators, network monitors and fiber optic component testing. It represents an industry of significant growth potential.

 

4. Technologies

4.1 MEMS

The Micro Electrical Mechanical System (MEMS) was the first all optical device to be developed into a physically feasible product and is now the most common wavelength switching technique without initial electronic conversion.

These devices are normally miniscule mechanisms made from silicon, with many moving mirrors ranging from a few hundred micrometers to a few millimeters. These mirrors exist on a silicon wafer and are packed as an array. The switch works by deflecting light waves from one port to another through these mirrors.

There are two main types of MEMS mirror structures. The simplest type is the called the 2D mirror. It has two states, one where the light pass over the mirror without deflecting it, and another where the mirror pops up and the beam is deflected into another output port (refer to diagram 1).  

                                    Figure 1 (Obtained from [9])

The more scaleable (able to be used for large applications) mirror structure however is the 3D MEMS. Refer to Figure xxx to understand how the process works. The switch works by using two arrays of beam steering mirrors.  Each mirror is fixed to flexures within 2 frames which allow the mirror to rotate in any direction as seen in diagram 2.

Figure 2 (Obtained from [10])

Figure 3 describes how a MEMS switch works. If a signal needs to be sent from port i to port j, mirrors i from the first array and mirror j, from the second array will point to each other. Light sent through port i will then have a direct path to port j, as can be seen in the diagram. However, if the signal needs to be switched to port k, the beam is scanned from mirror j to mirror k, passing over the other mirrors in between. Crosstalk is avoided in these circumstances, as a connection is only established when the mirrors are facing each other. An animation of this process can be seen here.

Figure 3 (Obtained from [9])

3D MEMS technology has generated the most interest in large scale switching industry. Lucent Bell Technology has developed the WaveStar LambdaRouter, which uses 256 miniature mirrors, which it believes can direct optical traffic16 times faster than technology of today. Xros Inc, believes it has developed "the world's highest capacity optical cross-connect system for open optical networks". Its system uses 1152 silicon mirrors on each array and, through computer controlled signals, is able to move the mirrors to direct signal in more than 1000 directions.

However the problem which arises with MEMS switching technology, as with all OOO switches, is the inability to effectively determine which direction the incoming light paths should take. This requires the switch to read and store header information contained in the signal itself. At the present time only OEO switches have this ability. Light signals can be contained in buffer for only a limited amount of time. However, there is no doubt that technological advances can enable a truly intelligent OOO switch in the future.

The other problems of MEMS technology is its switching speed (10ms for 6x6 devices, and 20ms for larger than 16x16 devices). Also the number of moving parts inside the switch may influence its reliability over time. There also a power loss of 7db for 16x16 devices, which becomes significant for many devices linked together.

 

4.2 Liquid Crystal Switches

Figure 4. (Obtained from [9])

The Liquid Crystal Switch makes uses of the polarisation effects of light in liquid crystals (similar to the type used for laptop screens) to switch light. The switch works in three steps. In the first stage, the light is filtered through a polarisation beam splitter where it can be split into two or more paths. In the second stage, the light is put through a liquid crystal cell and its polarisation properties may be changed. In the third stage the light is put through the polarisation beam combiner where the beams will be steered into whichever output port depending on its new polarisation property. This is illustrated in Figure 4. In (a), the light wishes to pass out of port 1, and the beam is split, passed through the liquid crystal cell and recombined out to port1. No rotation was necessary. However, in (b) the light wishes to pass out of port 2, so when the split beam passes through the liquid cell, a voltage is applied to the cell altering the polarisation properties of the beam. When the beams are recombined, it is steered out into port 2 due to its new polarisation properties.

The advantages of the liquid crystal switch lies in its low power consumption. Crystals that only need two states require only  a change in voltage applied for it to change state. It has significantly low power consumption compared to MEMS. However, a liquid crystal switch is slower than most switches. To overcome this, the crystals can be heated to make them less viscous, but this will undermines its low power consumption. However, Chorum Technologies Inc has claimed to have developed a switch which has millisecond switching times without any heating applied.

4.3 Bubble Based Switching

Figure 5 (Obtained from [9])

A bubble based switch, named the Photonic Switching Platform has been developed by Agilent Technologies Inc, using technology similar to that which is used in inkjet printers. This switch is capable of using 32x32 switches without the moving parts of MEMS. Figure 5 shows how the switch works. The switch consists of 32 ink filled micro trenches, called wave-guards, aligned vertically and horizontally. If no switching is required, a stream of light will pass through these trenches uninterrupted. When switching is necessary, heat is applied at the intersection of these trenches and a bubble is formed. This will effectively redirect the signal to the other output port.

The advantages that this technology lies in its low cost (reusing current inkjet technology) and fast switching times (10ms). It is also reliable, as inkjet is a used and tested technology. However, it may become problematic when the bubble must be maintained for long periods of time. Agilent technology refutes this by stating that the unit operates at a temperature where gas and liquid may effectively co-exist.

4.4 Thermo-Optic Switches

These switches are normally small in scalability, from 1x2 to 6x6 switches. There are two main types of switches: digital optical switches (DOS) and interferometric switches. The DOS works by changing the refractive index of light. Using a 1x2 Y switch, light travels through both arms of the switch. One of the arms is heated and the light will be blocked in the switch.

The Interferometric switch is smaller but is more wave-length sensitive. The switch works sending down light down a waveguard. The light is then split into different wave guards. If a switching command is issued, one of the waveguard arms is heated and the light within the waveguard will have it optical path length is changed. The light is then recombined and the path lengths of the lights are measured. If the lengths are different then the beam is switched into one output port and into another if they are the same.

The advantages of Thermo-Optical switches are in its speed (Lynx Photonics Networks claimed to have achieved 2ms) and  low costs (as it is mass produced already) and reliability (however, constant cooling and heating may limit its life). The disadvantages lie in its low scalability and its high power consumption.

 

5. Issues

5.1 Advantages of Optical Switching

There are many distinct advantages to the use of all- optical switches in networks today, compared with electrical switches.

All-Optical Switches reduce both floor space and power consumption significantly. Compared to an electrical switch, there is an 92 percent reduction in floor space requirements as well as a 96 percent saving in power requirements. These power savings translate into significant cost reductions with a 3 kW reduction for each rack. This saves the carrier from expensive diesel power generators, rectifiers and batteries, the monthly maintenance costs for these devices and the purchasing and maintenance of cooling equipment for these devices.

Optical Switches are a lot more scaleable and faster than current electric switches, as all-optical switches are protocol and bit rate independent. Of course this may also be problematic with the inability to interpret header as seen later. However, due to its immense scalability and flexibility all-optical switches are pretty much future proof, where transfer rates will not be affected bit rate limitations of switching equipment.

5.2 Alternatives and Problems with Optical Switching

Despite the remarkable benefits that Optical Switches companies can offer today, there are still many functions traditional technologies offer that all- optical switches are not yet capable of. Electronic switches allow for buffering and statistical multiplexing when there are excess packets waiting in line to be switched. ATM switches also allow different queues for different queuing priorities to offer guaranteed bandwidth. These functions can be achieved through a simple electronic chip which is inexpensive to produce.

However, current all – optical switching technology have not advanced to a level of technology where photonic signals can be as stored as easily as electrical signals. It is possible to store them using fiber delay lines, as light take a certain time to travel through lengths fiber (200,000 km p/second in silica). This would mean a 10000 bit frame traveling at 10G b/s will require 200m of fiber. This is both expensive and impractical. Another problem with this is that once a signal is put through a delay line, it cannot be processed until it comes back out. A solution to this is through adding switches within the lines, but this will only increase the costs of optical switching.

The other problem with all- optical switching is it cannot process header information of packets, especially at such high speed which the signals travel at. The maximum speed electronic routers currently can operate is at 10 Gb/s while optical signals today can travel up to 40 Gb/s. Hence the routers will not be able to process the signals as fast as the transmission.

5.3 Companies Producing Optical Switches

Many companies have chosen to take advantages of the potential that optical switches hold. The market of optical switches is dominated by four companies, namely Lucent, Nortel, Alcatel and Cisco, with the first two holding 50% of the market share. Besides the companies mentioned above and within the Technologies section, here are some other companies worth noting.

Ciena Corporation developed one of the first DWDM systems, and produces switches for both regional, metropolitan and long haul carrier networks. Its recent advances in its "MultiWave CoreStream" products has gained industry attention by enabling optical systems to perform at 16 Tb/s.

The first all-optical switch was produced by Corvis Corportion, and was called the "Corvis Optical Switch". The company has been able to gain other technological breakthroughs such as the ability to transmit signals over 20,000 miles without the need to regenerate signals.

Trellis Photonics has come up with a "Intelligent Lambda Switch" which uses hologram technology to reflect specific wavelengths of light, using an array of tiny crystals. The technology is immensely scalable, with Trellis proposing a to develop 3840 x 3840 port switch, with no moving parts and a crystallite structure which is 95 percent efficient. However, the downside of the  technology is that there is too  much power consumption (100 volts) and difficulty in cross connecting wavelengths.

 

6. Future of the Technology

MOST [6] believes that internet traffic is currently overwhelmed by the immense number of people who try to access it everyday. A growing demand for video conferencing and audio will also challenge the data capacities and bandwidth of networks in the future. Optical Networks is the clear solution, with experiments demonstrating that data rates of 400 Gb/s can be achieved using DWDM technologies.

The future of Optical Switching is bright for the industry, which is set to grow to $6bn in a few years time. With the evolution of an AON, equipment and managerial costs will be lowered and efficiency will be increased in the network.

With the new GMPLS standard coming into place, companies are able to truly design networks and systems that are able to fully utilize the power of optical networks while consumers will be able to enjoy high data rates, without bottlenecks.

However, until all-optical switches develop a level of intelligence matching that of an electro-optical switch, both technologies will continue to be used by all sectors of the industry.

Key Learning Points

        Optical switching technology developed naturally from the introduction of optical fiber networks.

        There are two types of optical switches: The all-optical switch (OOO) and the electronic- optical switch (OEO).

        The OOO switch has a photonic fabric making it bit-rate and protocol transparent.

        The OEO switch requires analogue signals to be translated into digital format to be processed for header information.

        Optical networks use the SONET/SDH standard as a physical layer protocol

        GMPLS is the new standard used to define interfaces which SONET/SDH equipment are supported.

        DWDM is used to store multiple signals on a single fiber

        Optical switching devices can be used within OXCs, used for protection switching and OADMs

        MEMS technology involves the use tiny mirrors to switch signals to different ports

        The liquid crystal switch changes the polarization of light in order to switch sigls

        Bubble Switching is a technology which switches lights by deflecting signals using bubbles

        Thermo – optics technology, relies on heating light to change its refractive index

        Optical Switches provide carriers with significant power, space and cost savings

        Optical switching technology currently cannot effectively buffer packets, with fiber delay lines being an inefficient solution

        It is not possible for optical switches to process header and routing information transmitted at high speeds

        The major companies producing optical device technologies include Lucent, Nortel, Alcatel and Cisco

        Ciena, Corvis and Trellium also have had significant impact on the industry

        With the increase of traffic over networks in the future, the demand for Optical switching technologies will follow

 - AONs can only be achieved when OOO technology reaches the same level of intelligence as OEO technology today

 - The future holds alot of potential for Optical networks

Review Questions

Multiple Choice

1. Which of the following OOO switching technologies is the most widely deployed:

(a) MEMS

(b)  Thermal – Optical

(c)    Bubble Switching

(d)   Liquid Crystal

 

2. An all-optical switch requires:

(a) analogue to digital conversion

(b) digital to analogue conversion

(c) digital to digital conversion

(d) none of the above

 

3. An optical cross connect will most likely use which type of GMPLS interface:

(a) PSC

(b) TDM

(c) LSC

(d) FSC

 

4. In country which is the physical layer protocol SDH implemented?

(a) Canada

(b) USA

(c) UK

(d) Japan

 

5. Which of the following is false:

(a) Optical switching give carriers power and space savings

(b) Buffering within all-optical switches is a problem

(c) It is possible for all-optical switches to process header information

(d) Optical switching allow for higher bandwidth ultiliation

 

6. Dense Wave Division Multiplexing (DWDM) involves:

(a) Dividing a signal up according to frequency

(b) Dividing a signal up according to time allocation

(c) Dividing a signal up according to spectrum

(d) none of the above

 

7. An OADM stands for:

(a) Optical Analogue to Digital Multiplexing

(b) Optical Add-Drop Multiplexer

(c) Optical Analogue Deflecting Machine

(d) none of the above

 

8. The all-optical switch has advantage of being:

(a) Protocol and bit-rate independent

(b) wave length convertible

(c) well established technology

(d) all of the above

 

9. Which company is famous four its Bubble Based switching technology

(a) Lucent Technologies

(b) Ciena Corporation

(c) Panasonic

(d) Agilent Technologies

 

10. Within a MEMS optical switching system, switching occurs when:

(a) Mirrors of adjacent ports are aligned

(b) Connection is established

(c) A signal has been received

(d) all of the above

 

11. Which of the following is NOT an application of an all-optical switch:

(a) Protection switching

(b) OXC

(c) OADM

(d) Interpreting header information

 

12. The OEO switch is called an intelligent optical switch because:

(a) It can interpret and process information contained in frames

(b) It switches optical signals with greater precision

(c) It is less restrictive than an OOO switch

(d) It uses more sophisticated technology

 

13. Until the development of an All- Optical Network:

(a) OEO switches should be used

(b) OOO switches should be used

(c) both A & B

(d) neither A or B

14. Which switching method is recognized for low power consumption

(a) MEMS

(b) Bubble switching

(c) Thermal

(d) Liquid Crystal

15. Which switching methods deals which the refractive index of light

(a) MEMS

(b) Bubble switching

(c) Thermal

(d) Liquid Crystal

16. Which GMPLS Interfaces forwards frames on the basis of information position in the timeslot

(a) PSC

(b) TDM

(c) LSC

(d) FSC

17. What is NOT the problem with buffering in all-optical switches:

(a) Photonic signals cannot be adequately stored

(b) It is too expensive

(c) Electrical chips are not fast enough

(d) There is insufficient technology in this area

18. When can an AON be achieved

(a) In ten years time

(b) When OOO have the same intelligence as an OEO

(c) When there is enough capital in the industry

(d) When the demand for bandwidth has sufficiently grown

19. On what level of the OSI model does the SONET/SDH protocol exist?

 (a) Layer 4

 (b) Layer 3

(c) Layer 2

(d) Layer 1

20. What is application are optical switches used for when a signal is transfered to annother fibre when an error is detected

(a) Protection switching

(b) Multiplexing

(c) OXC

(d) Security

Answers

1. a 2. d 3. c  4. c 5. c 6. d 7. b  8. a  9. d 10. d 11. d 12. a 13. c 14. d 15. c 16. b 17. d 18. a 19. d 20.  d

References

[1] Shepard S. 2001 Optical Networking Crash Course

McGraw Hill, United States.

[2] Gamini P.1999 State Of The Art Photonic Packet Switched Networks, Photonic Networks v 40. p 276- 283

[3] 2002 Optical Switches: Making Optical Networks a Brilliant Reality

International Engineering Consortium  avaliable at

http://www.iec.org/online/tutorials/opt_switch/

[4] 2001 Optical Switching

Light Reading – avaliable at

http://www.lightreading.com/document.asp?doc_id=2254

[5] 2002 Ciena Corporation

Avaliable at:

http://www.ciena.com/products/switching/

[6] 2002 Multidisciplinary Optical Switching Technology Center

Avaliable at:

http://www.ece.ucsb.edu/MOST/default.html

[7] S. Yao, B. Mukherjee and S. Dixit Advances in Phototonic Packet Switching: An Overview,

Optical networking , v. 35, p 190- 197

[8] R. Ramaswami, K. Sivarajan 2002 Optical networks: A Practical Perspective,

Academic Press, San Fransciso

[9] Bates R. J 2001 Optical Switching and Networking HandBook

McGraw Hill, New York

[10] Gymnne P.2002 MEMS enables fast, reliable optical switching

OE Reports Avaliable at:

http://www.spie.org/web/oer/august/aug00/cover1.html

 

Glossary

All Optical Network is a network that is transported, switched, and managed all at an optical level.

All-Optical Switch is an optical switch of which the switching fabric is done purely through photonic means

Cross-talk is when unwanted signals are picked up by a reciever

Electro-Optical Switch is an optical switch of which the switching fabric is done through electrical means

GMPLS is the new standard being drafted by the ietf on enhancing the MPLS system to account for SONET/SDH and DWDM equipment

Optical Network consists of a network where optical fiber serves as the fundamental medium of transmission

Optical Cross Connect (OXC) are the equivalent of electronic routers which forward data using switches

Scalable able to be used for large applications

SONET A multiplexing standard used for optical networks within North America, Japan, and Korea

SDH The equivalent of SONET for the rest of the world

 

Acronyms

AON All Optical Network

ATM Asynchronous Tramsfer Mode

DOS Digital Optical Switches

DWDM Dense Wavelength Division Multiplexing

FSC Fiber Switch Capable

GMPLS Generalized Multiprotocol Label Switching

IETF Internet Engineering Task Force

LSC Lambda Switch Capable

MEMS Micro-Electromechanical Systems

MPLS Multiprotocol Label Switching

OADM Optical Add – Drop Multiplexers

OC3 Optical Container

O-E-O Optical-Electrical-Optical

O-O-O Optical-Optical-Optical

OSI Open Systems Interconncetion

OXC. Optical Cross-Connects

PSC Packet-Switch Capable

SDH Standard Digital Hierachy

SONET Synchronous Optical Network

STM1 Synchronous Transport Module

TDM Time Division Multiplex Capable

VC4 Virtual container

Appendix