The Optical Networking and Communication
Conference & Exhibition

San Diego Convention Center,
San Diego, California, USA

What Will Drive the Transition to Coherent Intra-Data-Center Optics?


Fotini Karinou, Microsoft Research Ltd, UK; Clint Schow, University of California Santa Barbara, USA; Joe Khan, Stanford University, USA


Everybody seems to agree that coherent optics will eventually pervade data centers, but nobody seems to agree on when, why and how this will occur. This workshop studies the transition from direct detection (DD) to coherent detection (CD) in the context of intra-data-center network evolution and tries to answer some important questions.

  • Which factors will force the transition? Will it be the emergence of in-package (aka co-packaged) optics? The introduction of optical switching? A need to scale pluggable transceivers beyond 1 Tb/s per fiber? Something else entirely?
  • Which advantages of CD will be decisive in driving the transition? Improved receiver sensitivity? Increased throughput per fiber, per wavelength or per unit area? Temperature-insensitive demultiplexing? Electronic compensation of distortion and dispersion?
  • Will the transition be gradual or abrupt? DC operators often want links to be backward-compatible by at least one generation. What migration strategies can introduce CD links that are backward compatible (to DD links) and forward compatible (to future generations of CD links)?
  • Should external light sources replace integrated or in-package lasers to help facilitate CD and dense wavelength-division multiplexing (DWDM)? What about resulting increases in loss and number of fiber attachments? What about the impact of light source failure on reliability?
  • What role will optical amplification, whether transceiver-based or network-based, play in the transition? If we introduce DWDM and optical amplification, do we even need CD?
  • Considering trade-offs between power, area, fabrication complexity and functionality, how should optical, analog or digital signal processing be used for CD to perform carrier recovery? Polarization and phase-quadrature demultiplexing? Data recovery? Impairment compensation?
  • How will advances in photonic integration and considerations of yield, reliability and packaging enable or favor DD or CD optics?
  • How will an expected stalling of electrical interfaces at 100 Gb/s per port affect the transition from DD to CD?

Sponsored by: