I The market

I.1 The needs

The current needs of optical switch are mainly located in the telecom companies, in the companies manufacturing test equipment for optical devices and in TV station. The main applications that have been identified are the following:

• High speed LAN management (remote reconfiguration of network on campus)
• Redundancy (switch between two optical fibres linking two points in the case that a fibre is broken)
• Disaster recovery (switch the signal to an alternative path in the case of node failure)
• Test access monitoring (enable to monitor the signal in a fibre without physically ‘breaking’ the fibre)
• Optical hardware sharing (different fibre can share the same optical device without physical manipulations)

I.2 The current market

Conclusion : The current market for optical switch is 33M$ with 40% grow per year and is held at 97% by the opto-mechanical and fibre switches (3% by integrated optic devices). The domination of the mechanical-based switches is an interesting feature of the existing market for the MEMS. The introduction of the MEMS switch will be seen more as an evolution than as a revolution… This characteristic will enable the MEMS approach to directly benefit from the existing technology, solving the egg-and-chicken problem, provided a good compatibility exists between the two types of devices.

II The Industry

I.1 The products

Here are the results of a survey on the manufacturers of optical-switch, conducted during February 1998, where ‘MO’ denotes a mechanical based switch while ‘IO’ stands for an integrated optics based switch. The other parameters are the overall loss of the switch (in the state where it is the highest), the isolation between two neighbour chanel, the switching speed, the size and the cost:
 

Company name	Switch principle		Loss (dB)	Isolation (dB)	Speed (ms)	Size (mm)	Price (US$)
Astarte	MO	movable fibre	2.4		150		30000 (8x8 MM)
Dicon	MO MO	movable prism movable fibre	1.0 2.4	80 80	15 750	67x32x16	970 (2x2) 44192 (8x8)
E-Tek	MO	mechanical	1		10		1060 (1x2) 2150 (2x2)
JDS-Fitel	MO	movable prism	0.8	60	20	40x17x70	2300 (1x2)
NTT	MO	movable fibre	0.5				~1.200 (2x2)
Bookham	IO						available (?)
Intercom	IO	opto-optic					not available
Optivision	IO	semiconductor optical amplifier	0		0.001		not available
PIRI	IO	thermo-optic	2.1	16.5	2	75x10x6	750 (2X2)

Conclusion : This survey confirms that almost all the available optical switches are of the opto-mechanical type (open air or fibre). The quality of the proposed switch is very high and they have almost perfect characteristics (except for speed). The price is still high for large matrix (not so high if we consider the market volume…). The companies listed here can also be found in the next paragraph about patents, demonstrating their constant interest in the field… However, most of the current mechanical designs are based on fine-mechanics and most of the companies involved in the field seem to lack the development facilities to leap into the MEMS technology. New companies will have to be created…

I.2 The patents

A search among the patents related to the optical switching filled in the USA from January 1997 to March 1998 gives 167 hits. Here is a short list of the company that have filled patents relating the realization of an optical switch (the number of patents filled and the type of switch is described in the parenthesis):

• IBM (~ MEMS stack of detectors)
• HP (~ MEMS TIR frustrated with bubble)
• Boeing (FS with bulk electrooptic effect)
• Whitaker Corp. (2, MO moving prism with a reflector)
• NEC (IO with electrooptic effect)
• Alcatel (2, IO in semi-conductor III-V)
• Lucent (3, MO using movable fibre, MO using movable waveguide, MEMS using movable plate reflector)
• JDS Fitel (2, MO using moving fibre, MO using movable prism)
• E-Tek (MO using movable fibre)

II.1 Evolution

The theme of the optical switch has drained a lot of research effort recently. Of course not all the paper present the realisation of a working device. Many of them merely show physical principle that could be applied to optical switching in a more or less foreseeable future…

Conclusion: the field is very active and does not show any slow down (decrease in the number of published papers could be the sign of a mature technology).

II.2 Current status

For the years 1995, 1996, and 1997 we have also studied in more detail the proposed architectures and classified them according to the type of the developed switch. In this study we have rejected all the paper that did not present a real switch. This has reduced the number of paper to 45 from more than 150 that had ‘optical switch’ as keywords…

IV The technology

IV.1 Switch classification

Switch can be classified according to many criteria.

• The size of the switch: they are single 1X2 or 2X2 switch or larger matrix switch (up to 8X8). Some switching principles are easily scalable and can be used either for building individual switch or for large NXN switch array, some can not, or need cumbersome procedure;
• The principle of switching: mainly either induced by a change of the optical index or by the redirection of the light
• The kind of power needed to switch: mechanical, electrical, magnetic, …
• The medium that transmits light inside the switch: a waveguide (integrated optics), an optical fibre, air (free-space).

<<To be continued>>