Analog Photonic Computation

Projektbeschreibung

Rechenmodelle für die programmierbare integrierte Photonik

Seit mehr als 50 Jahren verzeichnet die Rechenleistung in der digitalen Elektronik aufgrund einer regelmäßigen Verdoppelung der Transistordichte in integrierten Schaltkreisen einen Anstieg. Leider stößt dieses Skalierungsgesetz an seine Grenzen, was zu einem Mehr an Anwendungen führt, die von der digitalen Elektronik nicht unterstützt werden können. Die programmierbare integrierte Photonik (PIP) bietet eine effizientere Alternative für diese Anwendungen, es fehlt jedoch an Rechenmodellen, die auf die Nutzung der PIP-Vorteile spezialisiert sind. Das ERC-finanzierte Projekt ANBIT wird eine neue Rechentheorieklasse entwickeln, die als „Analog Photonic Computation“ (analoges photonisches Rechnen) bezeichnet wird. Diese für den Entwurf und die Erstellung von Rechenmodellen spezialisierten Theorien werden die PIP-Technologie effektiv einsetzen.

Ziel

For over 5 decades, digital electronics has covered the increasing demand for computing power thanks to a periodic doubling of transistor density in integrated circuits. Currently, such scaling law is reaching its fundamental limit, leading to the emergence of a large gamut of applications that cannot be supported by digital electronics, specifically, those that involve real-time analog multi-data processing, e.g. medical diagnostic imaging, drug design and robotic control, among others. Here, an analog computing approach implemented in a reconfigurable non-electronic technology such as programmable integrated photonics (PIP) can be more efficient than digital electronics to perform these emerging applications. However, actual computing models were not conceived to extract the benefits of PIP.
The aim of ANBIT is to develop an entirely new class of computation theory – termed Analog Photonic Computation (APC) – specifically designed to unleash the full potential of PIP technology. The core concept revolves around the idea of performing analog operations on a new unit of information, the analog bit or anbit, conceived as a two-dimensional analog function and matched to the building block of PIP circuits.
ANBIT will reach its objectives by: 1) developing the theory of APC based on operations (gates) of anbits, 2) translating the principles of APC to the design of PIP circuits by concatenating single- and multi-anbit gates, 3) fabricating, packaging, testing and validating silicon PIP chips capable of implementing complex APC architectures, 4) designing, coordinating, setting and performing experiments that will prove the unique potential of APC in computational and signal processing applications with huge takeover.
ANBIT will deliver a new computing paradigm that extracts the full potential of PIP technology, which in turn will have a crucial impact on fundamental and applied research and on our information society.