Periodic Reporting for period 1 - PROMETHEUS (PROgraMmable integrated photonic nEuromorphic and quanTum networks for High-speed imaging, communications and sEcUrity applicationS)
Période du rapport: 2022-09-01 au 2024-02-29
Bio-inspired computational models and quantum computing redefine computation beyond traditional architectures. PROMETHEUS aims to unite quantum and neuromorphic photonics, breaking barriers for enhanced performance. The heart of this endeavor is a Photonic Integrated Circuit (PIC) on silicon-on-insulator (SOI) featuring a Field Programmable Photonic Gate Array (FPPGA) synaptic layer. Fortified by energy-efficient non-volatile barium titanate (BTO) phase shifters and co-integrated III-V lasers, it forms an ultra-fast spiking neural layer. Advanced packaging techniques create a robust, adaptable device with unmatched low power consumption and rapid processing. PROMETHEUS' versatile platform explores spiking networks, reservoir computing, and quantum neural networks, catering to industry-driven applications like high-speed image processing for biomedical and industrial sectors. The PIC's functions extend to quantum random number generation (QRNG) and photonic physical unclonable function (PUF) for cybersecurity, enhancing both neuromorphic processing and hardware-based authentication.
Work on the reconfigurable photonic chip includes: completion of RUN1 PIC design; miniaturized BTO weighting elements design; quantum/neuromorphic laser source design; ongoing driving electronics development; ongoing flip-chip approach analysis for RUN2; draft design and layout for RUN2 PIC; RUN2 PIC distribution revision; mapping of RUN1 PIC; and a packaging workshop. Fabrication and characterization include: tapeout of the first Run; ordering and receiving of epitaxial wafers; completion of Scintil Foundry fabrication; start of fabrication at CEA-Leti; and validation of fabrication flow. Packaging and testing progress: identification and preliminary testing of DACs and OpAmps; completion of driving electronics schematic design; ongoing Altium PCB design software use; identification and testing of TIAs; initiation of electronic test-bench creation at IZM; and basic requirements setting for software development. Implementation of computing models: simulation and investigation of neuromorphic networks; theoretical analysis of spiking time delay reservoir computing and extreme learning machine models; experimental work on event-based imaging and SNNs; theoretical and experimental study of hardware accelerators for edge computing; theoretical investigation of Quantum Reservoir Computing processors and crypto-neuromorphic schemes; and design of multi-layer interferometer for QRNG. System evaluation progress: focus on system integration for flow cytometry application; development of low-cost IFC system with neuromorphic camera; setup for PIC pumping and quantum source performance characterization; and experimental evaluation of OSS scheme for edge computing applications on photonic integrated hardware.entation on photonic integrated hardware by relying on IPRONICs’ programmable photonic processor. The performance improvement provided by the OSS is confirmed by the experimental validations.
The first period of the PROMETHEUS project achieved significant milestones, including the validation of the first-generation PIC and the commencement of manufacturing for silicon passive levels. Contamination and implantation tests are in progress.
The project demonstrated the concept of optical spectrum slicing in edge computing and image classification, showcasing promising performance in mitigating transmission impairments and image classification. Design convergence for fabrication run1 is nearing completion, with exploration for higher dimensionality in run2. A Python implementation of the NIST test suite for randomness is under development.
In terms of impact, PROMETHEUS perfectly aligns with climate neutrality objectives, emphasizing low-power, high-performance aspects, and a less demanding integration strategy to reduce the environmental footprint in digital supply chains. It also adheres to climate-neutral production mandates, mitigating manufacturing-related carbon footprints.
PROMETHEUS holds relevance for clean water and sanitation efforts, reducing the climate impact of medical processing and indirectly benefiting public health by reducing pollution and enhancing water supply quality.
The project significantly impacts good health and well-being by drastically improving sample processing time and accuracy through Spiking Neural Networks, reducing false positives/negatives in critical medical examinations, enhancing health management within the European Union.
Economically, PROMETHEUS contributes by revolutionizing processing speed and power consumption, enhancing productivity across sectors and fostering competitiveness.
In terms of industry, innovation, and infrastructure, PROMETHEUS develops a versatile photonic platform accommodating emerging concepts like spiking networks, reservoir computing, and quantum neural networks, driving technological innovation. It addresses critical industry applications, revolutionizing image processing and optical signal processing, thereby advancing industries and infrastructure.
The project demonstrated the concept of optical spectrum slicing in edge computing and image classification, showcasing promising performance in mitigating transmission impairments and image classification. Design convergence for fabrication run1 is nearing completion, with exploration for higher dimensionality in run2. A Python implementation of the NIST test suite for randomness is under development.
In terms of impact, PROMETHEUS perfectly aligns with climate neutrality objectives, emphasizing low-power, high-performance aspects, and a less demanding integration strategy to reduce the environmental footprint in digital supply chains. It also adheres to climate-neutral production mandates, mitigating manufacturing-related carbon footprints.
PROMETHEUS holds relevance for clean water and sanitation efforts, reducing the climate impact of medical processing and indirectly benefiting public health by reducing pollution and enhancing water supply quality.
The project significantly impacts good health and well-being by drastically improving sample processing time and accuracy through Spiking Neural Networks, reducing false positives/negatives in critical medical examinations, enhancing health management within the European Union.
Economically, PROMETHEUS contributes by revolutionizing processing speed and power consumption, enhancing productivity across sectors and fostering competitiveness.
In terms of industry, innovation, and infrastructure, PROMETHEUS develops a versatile photonic platform accommodating emerging concepts like spiking networks, reservoir computing, and quantum neural networks, driving technological innovation. It addresses critical industry applications, revolutionizing image processing and optical signal processing, thereby advancing industries and infrastructure.