Optical Turing Machine

The Optical Turing Machine (OTM) is an effort to develop optical Tbps in-network processing.

  • It seeks to unify communication and computation using a single encoding format.
  • It provides processing required for in-transit data, e.g., to support:
    • Network packet processing
    • Big data filtering
    • Bulk network security
  • It aims to support serial, high-density encodings for efficient high-speed, long distance communication
  • It leverages the native properties of optical mixing to support transformational, rather than switched processing.

Key observations:

  • Data must be encoded digitally.
    • Analog encodings accumulate noise when composed (cascaded) or recirculated (as for memory).
  • Data must be recirculated.
    • Without recirculation, only combinatorial logic can be supported.
    • Higher levels of computation, such as finite state machines, push-down automata, and Turing machines are required to compute more complex algorithms. These require (respectively) a single state, a set of states in which only one can be modified, and a set of states in which any can be modified – and state requires recirculation.
  • Data must be encoded using M-PSK.
    • Binary encodings are too slow and amplitude-keyed encodings (PAM, QAM) require value-dependent processing.
  • Data must be regenerated while retaining its semantics.
    • Differential regeneration is very effective but prevents composition and recirculation.
    • For M-PSK, this necessitates phase squeezing (phase-sensitive amplification, or PSA).
    • PSA must be supported natively, without needing a pilot, reference, or conjugate to experience the same noise, e.g., “degenerate PSA”.
  • Computation requires a field (for algebra), switching (for reprogrammability), and phase squeezing (for carry generation).
    • Algebra requires a field, i.e., a ring with two operators that satisfy the properties of a group.
  • Data processing requires optical mixing (three-wave or four-wave, i.e., TWM or FWM).
    • Mixing supports Tbps and beyond data rates, whereas switching is limited due to the transmission effects of the drivers needed to modulate fields to vary bulk properties.
  • A demonstration of the above requires hybrid integrated circuit technology.
    • Macroscopic, bench-top devices introduce phase instabilities that cannot be compensated. Integrated approaches allow the use of phase-aligned sources (e.g., combs) and reduce sensitivity to environmental disturbances (thermal, mechanical).
  • Current approaches fall into two categories:
    • “Field of Dreams” (from the movie of the same name) – exploring the design space by building new components. Many of these approaches directly interfere with computation, component composition, or communication using a single format.
    • “The Aluminum Feather” – develop solutions inspired from existing successes, such as electronic computation, by literal translation of concepts. We coined this term to refer to an analogous approach to airplane design: birds fly, birds are made of living matter, birds use feathers — so if we want to build an aluminum bird (an airplane), we clearly need aluminum feathers. This points out the hazard of applying a conceptual homomorphism without context – i.e., translating ideas literally (transliteration) vs. translating the concepts.
    • OTM uses simultaneous consideration of the requirements for computation and communication to avoid these hazards.

Overview:

Complete list of related publications:

OTM architecture and system

Regeneration

Regeneration-related issues:

Amplitude squeezing (multilevel):

Signal generation (high-speed):

Signal reception (high-speed):

Uses of optical computation:

Algorithms for optical computation:

Other related papers

  • A. Mohajerin-Ariaei, M. Ziyadi, M. R. Chitgarha, and A. E. Willner, “All-optical Implementation of Signal Processing Functions,” Invited Paper, Society of Photo-Instrumentation Engineers (SPIE) Photonics West, Optical Metro Networks and Short-Haul Systems VII Conference, paper 9388-7, San Francisco, CA, Feb. 2015.
  • R. Van Meter, S. Suzuki, S. Nagayama, T. Satoh, T. Matsuo, A. Taherkhani, S. Devitt, J. Touch, “Large-Scale Simulation of the Quantum Internet,” Proc., International Conference on Quantum Communication, Measurement and Computing (QCMC), Singapore, 2016.

OTM is supported in part by: