Optical switching fundamentals
Optical switching changes the connection between ports without converting the signal to the electrical domain. An maps an input port to an output port and creates an optical path through the node. The switched unit can be an entire fiber, a waveband, or an individual wavelength. This process is called photonic switching. The photonic element supports flexible, high-capacity networks without the operations required by an electronic switch.
In multi-port optical switches, photonic switching elements form switching fabrics that determine the available connections between input and output ports.
Classifying optical switching systems
Optical switching systems are classified by switching granularity and by how the signal is processed.
By switching granularity
- Fiber switching: The full spectrum occupied by signals in an input fiber is switched as one unit to an output fiber.
- Waveband switching: A defined group of adjacent wavelengths is switched together.
- Wavelength switching: Each optical channel is routed independently. The granularity must match the functions supported by the WDM node. This is a common form in WDM networks.
By processing method
- Opto-electronic switching (O-E-O): The optical signal is converted to an electrical signal, switched electronically, and converted back to optical form. This path can provide 3R regeneration, but its rate and format depend on the electronics.
- All-optical (photonic) switching (O-O-O): The signal remains in the optical domain. The path can be transparent to rate and format, while its transmission parameters depend on the implementation.
Optical switch parameters
Optical component specifications include, among other parameters:
- Switching time: The state-change time measured with a defined output-power criterion, such as reaching 90% of the final value. Depending on the technology, it can range from nanoseconds in electro-optic switches to milliseconds in mechanical designs.
- Insertion loss: The power reduction through the element. A lower value is preferable, and together with the number of elements and ports it contributes to the power budget of the switching fabric.
- Crosstalk: Unwanted signal power that reaches an inactive or unintended port. The parameter is expressed as the ratio of leaked power to wanted output power, usually in dB. A value around -50 dB indicates low leakage and limits interference between channels.
- Port count: The number of input and output ports, for example in a 16x16 switch. Core network applications require fabrics with many ports.
- Wavelength range and input power: The element must operate over a specified spectral range and within its allowed optical input power.
- Polarization, repeatability, and isolation: Polarization and wavelength dependence, stability across repeated switches, and port isolation describe behavior in an installed optical path.
Guided-wave switching elements
Integrated elements guide light at micro-scale in and change their refractive index or coupling to redirect the signal.
1. Electro-Optic Switches
These devices exploit the electro-optic effect, where the of a material changes when an electric field is applied. Lithium niobate is one option, including titanium-doped waveguides (). In a directional coupler, the applied voltage changes coupling between two waveguides and selects a cross or bar state.
Pros: Short switching time, usually in the nanosecond range, and no moving parts. Limitations: cost, loss, and drive requirements depend on the implementation.
2. Acousto-Optic Switches
This technology uses the acousto-optic effect. A sound wave creates a periodic refractive-index change in a crystal, acting as a diffraction grating that deflects the beam. Turning the wave on or off changes the state, and the deflection angle is described by the Bragg angle ().
3. Thermo-Optic Switches
These switches use temperature to change a material's refractive index. One implementation is a bubble switch: a micro-heater at an intersection creates a bubble in the liquid, changing propagation conditions and directing light to an alternate path through total internal reflection.
Pros: Simple structure. Limitations: switching can take milliseconds, and the time and stability depend on thermal inertia and the device design.
4. Semiconductor Optical Amplifier (SOA) Switches
An SOA can act as an optical gate because the drive current changes its gain and transmission. With current applied, the element becomes transparent and amplifies the light; with current reduced, it increases attenuation and absorbs the signal. An SOA matrix can therefore combine fast switching with gain, with noise and nonlinearities remaining design parameters.
Free-space switching: MEMS mirrors
Free-space switches steer beams with arrays of microscopic moving mirrors. In MEMS implementations, small mirrors are integrated on a silicon substrate, and their positions select the connected port pair in the switching fabric.
2D MEMS
In this architecture, mirrors usually have two positions. A mirror can enter the beam path to deflect it or leave the path so light continues to the next mirror in the crossbar. For a switch with ports, the number of switching elements grows approximately as , so array size and accumulated loss limit the fabric scale.
3D MEMS
The design uses two mirror arrays with analog two-axis positioning. An input mirror steers the beam to a selected output mirror, which directs it to the required port. In an idealized model, an switch needs about mirrors, which supports large OXC fabrics, while practical scale depends on geometry, control, and the power budget.
MEMS switch properties
MEMS switches provide optical transparency because they do not convert the signal to the electrical domain. They remain independent of data rate and modulation format, with low loss, low crosstalk, and large-fabric scalability as practical advantages. Insertion loss, crosstalk, switching time, stability, and port count still determine whether an element fits a network node.