- Physical Unclonable Functions (PUFs) and Hardware Security
- Advanced Optical Network Technologies
- Photonic and Optical Devices
- Optical Network Technologies
- Integrated Circuits and Semiconductor Failure Analysis
- Interconnection Networks and Systems
- Neural Networks and Reservoir Computing
- Software-Defined Networks and 5G
- Advanced Optical Imaging Technologies
- Advanced Memory and Neural Computing
- Cryptographic Implementations and Security
- Neuroscience and Neural Engineering
- Digital Media Forensic Detection
- Cell Image Analysis Techniques
- Biometric Identification and Security
- Advanced Steganography and Watermarking Techniques
National and Kapodistrian University of Athens
2018-2021
Innovative Technologies Center (Greece)
2018-2020
Physical unclonable functions are the physical equivalent of one-way mathematical transformations that, upon external excitation, can generate irreversible responses. Exceeding their counterparts, inherent complexity renders them resilient to cloning and reverse engineering. When these features combined with time-invariant deterministic operation, necessity store responses (keys) in non-volatile means be alleviated. This pivotal feature, makes critical components for a wide range...
Data centers are continuously growing in scale and can contain more than one million servers spreading across thousands of racks; requiring a large-scale switching network to provide broadband reconfigurable interconnections low latency. Traditional data center architectures, through the use electrical packet switches multi-tier topology, has fundamental weaknesses such as oversubscription cabling complexity. Wireless intra-data interconnection solutions have been proposed deal with problem...
In an environment where cyber attacks are increasing, both in frequency and complexity, novel ways to shield data, users, procedures have be envisioned. Physical unclonable functions (PUFs) the physical equivalent of one-way mathematical transformations with exception that their inherent complexity renders them resilient cloning. One interesting deployment scenario includes PUFs as random key generators. The deterministic nature operation alleviates necessity store keys non-volatile means....
Beyond pure Von Neumann processing Scalability of AI / HPC models is limited by the bottleneck for accessing massive amounts memory, driving up power consumption.
We introduce an intra-Data Centre, WDM-based, optical packet switching architecture. P-Torus demonstrates low energy footprint, CapEx/OpEx and end-to-end latency relying on the combination of passive schemes Torus architectures that enable single-hop interconnects for 256 ToR switches.
In this work we introduce P-Torus, a wavelength switching architecture for intra-data-center networks that exploits optical-packet-based transport and is based on 2D torus interconnected clusters. P-Torus does not require optical buffers, as all buffering takes place at the end hosts. The proposed combines connectivity richness, time-slotted operation, efficient routing in order to provide single-hop, contentionless, low-latency interconnection among top-of-rack switches. Wavelength...
In this paper we experimentally evaluate a physical unclonable function based on polymer optical waveguide, as time-invariant, replication-resilient, source of entropy. The elevated unclonability our implementation is combined with spatial light modulation and post processing techniques, thus allowing the deterministic generation an exponentially large pool unpredictable responses. quality generated numbers validated through NIST/DIEHARD(ER) suites, whereas overall security scheme...
We introduce LightOn's Optical Processing Unit (OPU), the first photonic AI accelerator chip available on market for at-scale Non von Neumann computations, reaching 1500 TeraOPS. It relies a combination of free-space optics with off-the-shelf components, together software API allowing seamless integration within Python-based processing pipelines. discuss variety use cases and hybrid network architectures, OPU used in CPU/GPU, draw pathway towards "optical advantage".