
Wavelength Selective Switch
A Wavelength Selective Switch (WSS) is an all-optical device used in fiber-optic communications (lasers and photonics) that independently routes, blocks, or attenuates individual wavelength channels (or flexible spectral slots) of a multi-wavelength Dense Wavelength Division Multiplexing (DWDM) signal to different output ports without optical-electrical-optical (OEO) conversion.
It is the core reconfigurable element of modern Reconfigurable Optical Add-Drop Multiplexers (ROADMs). Unlike fixed filters or demultiplexers, a WSS is software-programmable: any wavelength from a common input can be directed to any of the N outputs (typically configured as 1×N), while others are blocked or passed through, and power levels can be equalized.
Technical Information and Operating Principle:
WSS devices use a disperse-and-switch architecture:
Multi-wavelength light enters a common port, is collimated, and often passes through polarization-diversity optics.
A diffraction grating (or GRISM) spatially disperses the spectrum so different wavelengths (or frequency slots) land on different regions of a switching element.
The switching element independently steers and attenuates each channel.
The light is recombined (via a return pass through the grating) and coupled into the selected output fiber(s).
Dominant switching technologies:
MEMS (Micro-Electro-Mechanical Systems): Tiltable micro-mirrors reflect beams to the desired ports. Early commercial technology; good for fixed grids, relatively low loss, and faster switching, but limited flexibility for variable channel widths due to fixed mirror channelization.
LCoS (Liquid Crystal on Silicon): A high-resolution phase-modulating pixel array creates programmable phase ramps or gratings that diffract/steer beams. Currently dominant for commercial high-performance WSS because it supports flexible-grid operation, fine spectral control, and high port counts with no moving parts.
Typical specifications (commercial modules):
Port counts: 1×4 to 1×20+ (and higher or twin/MxN variants).
Insertion loss: typically 5–7 dB (range ~5–8 dB).
Channel isolation/crosstalk: >30–40 dB.
Per-channel attenuation: commonly 0–15/20 dB (for power equalization).
Reconfiguration time: often <20 ms.
Spectral support: fixed grids (50/100 GHz) or flexible grids (center-frequency steps down to 6.25 GHz, slot widths in 12.5 GHz steps).
Operating bands: primarily C-band; extended to Super-C and C+L for multi-terahertz windows.
LCoS enables “flexgrid” or elastic optical networking, where channel bandwidth can be adjusted on the fly (e.g., allocating ~150 GHz for a high-baud-rate coherent signal).
Applications:
ROADM nodes in DWDM transport networks — The primary use. WSS modules enable multi-degree mesh nodes that can add, drop, or express any wavelength in any direction under remote software control. This supports Colorless, Directionless, Contentionless (CDC) architectures.
Dynamic optical networking / SDN — Software-defined optical networks that reconfigure lightpaths in seconds rather than days, improving capacity utilization and reducing truck rolls.
Optical power equalization — Per-channel attenuation balances power across wavelengths, improving system OSNR and reach.
Flexible spectrum allocation — Supports high-speed coherent transceivers (100G, 400G, 800G+) with variable baud rates and modulation formats by matching channel width to the signal.
Long-haul, metro, and data-center interconnect — Deployed throughout modern optical transport and increasingly in high-capacity interconnect environments.
The WSS turns rigid point-to-point DWDM links into agile, reconfigurable optical networks by performing wavelength-level routing, blocking, and attenuation entirely in the optical domain.