Back to search

University of West Attica

Νευρομορφικές υπολογιστικές τεχνικές σε εφαρμογές τηλεπικοινωνιών και επεξεργασίας εικόνας

Abstract

dc:description

The first quarter of the 21st century was marked by an explosion in the creation, exchange and processing of digital data to such an extent that it has been called the information age. The key technologies that defined this period are the internet, with optical communications as its backbone, and the big data-based artificial intelligence. Two of the most important technological challenges today are the efficient operation of artificial intelligence algorithms and the responsiveness of optical communications to the ever-increasing flow of data. Key challenges are the interconnection of memory and processing over slow electrical channels and the large number of electro-optical conversions at the interconnects. In recent years, the distance between information transfer and information processing has been significantly reduced with the electrical channels being replaced with optical ones at distances as short as 1 m. Also, computing frameworks beyond the traditional Von Neumann architecture have emerged to circumvent the memory bottleneck. Such proposals are called neuro-inspired or neuromorphic computing. With data transfer taking place at the optical domain, it makes sense for neuromorphic computing to also be performed optically. However, at a time where graphics processing units and parallel computing are making huge success, photonic neuromorphic computing, still in its infancy, needs to become more efficient in order to find space in specific applications. The largest category of applications that could directly benefit from the use of photonic neuromorphic techniques is optical communications. Secondly, the processing of other purely optical signals, such as images or radar signals. This thesis attempts to contribute to the efficiency improvement of photonic neuromorphic computing techniques with a focus on their use in optical communication systems and image processing. The following innovations are proposed:1.A bidirectional photonic processor based on two laser neurons. The processor exploits time multiplexing to create tens of virtual neurons. The benchmark application chosen is the compensation of transmission impairments in short-reach links.2.A photonic processor consisting of optical filters for simultaneous spectral decomposition and processing of broadband signals. The most important application for this system is the mitigation of chromatic dispersion in intensity modulation and direct detection optical systems. The proposed system, increases the transmission reach in direct-detection systems from a few km to >100 km, based on high transmission rates (>100 Gbaud). 3.A self-coherent receiver/processor based on neuromorphic spectral decomposition for receiving and processing coherent modulation formats. The receiver is an efficient alternative to classical direct detection and coherent systems, significantly reducing consumption and costs. 4.Two systems based on four-wave mixing effect which can execute linear and nonlinear operations optically. These systems allow the implementation of various activation functions, as well as the multiplication of vectors directly in the optical domain.

Degree

thesis:*
Grantor dc:publisher
University of West Attica
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Sozos, Konstantinos
  • Σώζος, Κωνσταντίνος

Subjects

dc:subject × 20

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/58111

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Sozos, Konstantinos; Σώζος, Κωνσταντίνος. Νευρομορφικές υπολογιστικές τεχνικές σε εφαρμογές τηλεπικοινωνιών και επεξεργασίας εικόνας. University of West Attica, 2024. http://hdl.handle.net/10442/hedi/58111