Sub-optical components for a new optical digital signal processor

Development of sub-optical components and optical materials that will be the building blocks for a new optical digital signal processor (dsp) and that will be the key parts of new components and applications.

The objective of this project is to develop sub-optical components and optical materials that will be the building blocks for a new optical Digital Signal Processor (DSP) called EnLight256. Furthermore, these building blocks can be and will be the key parts of new components and applications. LENSLET will integrate the building blocks with the Optical DSP. LENSLET LABS, with its unique technology and business background, has developed a very powerful Optical Digital Signal Processing Engine (ODSPE) for a wide range of applications such as Telecom/Datacom, Industrial automation, Voice and speech recognition, Imaging and pattern recognition, high density/ high-speed modems, Compression and decompression products and more. LENSLET's pioneering Tera (10 12) operations-per-second ODSPEs utilize a fully re-configurable state-of-the-art optical signal-processing core. The ODSPEs provide system architects and manufacturers with greater speed, lower cost, greater power density, smaller footprint and quicker time-to-market over current electronic silicon-based DSPs, Field Programmable Gate Arrays (FPGAs) and Application-Specific Integrated Circuits (ASICs). * The Architecture (in brief): This unique technology takes electronic digital input, converts it into optical signals (photons), performs the computation at light speed in the programmable optical core, and then converts the optical output signals back into digital electronic form. The native operation of the optical core of the ODSPE is Vector-Matrix Multiplication (VMM) rather than the Multiply-Accumulate (MAC) operation. This allows computation-intensive algorithms to be mapped onto the ODSPE at conceptually high levels of abstraction, and allows massive mathematical operations to be performed at the speed of light. The heart of the ODSPE architecture is an optical Vector-Matrix Multiplication core. It is the key to massive data parallelism and on-the-fly reconfigurability. The optical VMM core converts electronic inputs into light by using Vertical Cavity Surface Emitting Lasers (VCSELs) and then performs a VMM operation by directing this light through programmable internal optics. The light that emerges is sensed by an array of detectors and translated back into electronic signals. By processing incoming signals at the photon level, Vector-Matrix Multiplication is performed at the speed of light. The first ODSPE in the ENLIGHT family will be EnLight64(TM) (input vector of 64 elements) and the second will be EnLight256(TM) (input vector of 256 elements). During each cycle, an input vector of 256 (or 64 in EnLight64(TM)) one-byte wide elements is multiplied by a matrix of 256x256 (or 64 x 64 in EnLight64(TM)) one-byte wide matrix elements (64K elements in the matrix in EnLight256 or 4K in EnLight64(TM)), generating an output vector of 256 (or 64 in EnLight64(TM)) one-byte wide elements. This is the key to the massive data parallelism of the ODSPE. * EnLight Building Blocks Layers: The reconfigurability of the ODSPE is provided by changing the values stored in the SLM (Supported Liquid Membrane). Changing applications or changing tasks within an application is equivalent to replacing the matrix in the SLM and thus providing a new transform. The ease with which matrices can be changed provides the on-the-fly reconfigurability needed for a commercially viable device. A second building block is the Vector-Vector engine, which LENSLET calls the Vector Processing Unit (VPU). It is part of the Application Processing Layer (APL), an implementation layer that may change from one Enlight256(TM) / Enlight64(TM) derivative to another. The VPU is responsible for processing the results of the Vector-Matrix Multiplication. For example, the VPU integrates the results of the VMM (vectors) via the register file that holds intermediate results. A third layer is used mainly for scalar operations and control operations. These are done by an embedded, off-the-shelf DSP that is connected to the same register file for scalar and control operations. Keywords: optical DSP.
Project ID: 
3 010
Start date: 
Project Duration: 
Project costs: 
8 600 000.00€
Technological Area: 
Optical Networks and Systems
Market Area: 

Raising the productivity and competitiveness of European businesses through technology. Boosting national economies on the international market, and strengthening the basis for sustainable prosperity and employment.