Project: Multi-wavelength Optical Switch Networks
The objective of this collaboration with Intel Research Cambridge is to investigate data coding and synchronization schemes to enable efficient multi-wavelength optical packet communication. Optical interconnects have proven themselves for long distance communication due to low power loss and high data density. The cost of low power, highly integrated optical components is reducing which is starting to make them competitive with copper interconnects over shorter distances.
This research is in the anticipation that the cost of active optical components will diminish and that we need to resolve a number electro-optical issues to take advantage of them.
One important scaling issue is how multi-wavelengths can be effectively used to aggregate bandwidth. This leads us to the following research questions:
- How can a receiver clock be efficiently recovered from multi-wavelength optical stream whose source changes rapidly due to the optical packet switch?
- How should the multi-wavelength data be encoded efficiently to ensure that the optical power changes through the switch stay within operating bounds?
- How can skew between multiple wavelengths be tolerated?
Our hypothesis is that delay insensitive codes based on experience in the asynchronous circuits community can be used to efficiently recover data from multi-wavelength transmissions whilst balancing power. Initially we will verify the hypothesis through simulation, but will then move to prototype silicon with the objective of integrating these test chips with electro-optical test rigs at Intel Research Cambridge.
Solving the problem of high speed bit level clock recovery by the use of coding techniques from the circuits community would be an enabler for packet switched systems. Also, applying coding techniques to multiwavelength devices for optical balancing and/or to overcome other problems such as unequal errors across the wavelengths would enhance the system performance of multiwavelength devices.