projects

Research projects as PI or Co-PI

May 2023 - : Fair-RL-CC: Towards Fair Congestion Control with Reinforcement Learning, funded by GÉANT (€30,000).
Reinforcement Learning-based Congestion Control is still in its infancy, and substantial research is required to yield deployable solutions. Here, we identify two key limitations that we aim to tackle in this project; (1) existing approaches cannot guarantee fairness amongst competing network flows; the stability of said CC policies with flow churn is problematic and performance is acceptable only in fixed and static network scenarios that cannot deviate much from the learning environment; (2) crucially, the research community lacks the tools to train and evaluate RL-based CC policies in a reproducible way, at scale. In this project, we develop RL-based CC approaches that provide fairness amongst competing network flows, without sacrificing on performance or overall network utilisation. We also develop a learning playground with simulation models for existing RL-based CC and compare them with approaches developed in this project.

April 2023 - : Rethinking large-scale network management through the lens of neuroscience, funded by EPSRC (£265,000).
Much of our lives today relies on the continuous and reliable provision of ICT infrastructure. With networks, services, traffic workloads and user needs ever-growing in size and complexity, a new paradigm in network management and control is required. This project is in collaboration with Prof Berthouze (as PI) and aims to transform the way ICT networks are being conceptualised for management, by developing a purely data-driven characterisation of emerging dependencies between ICT components, inspired by recent neuroscientific paradigms used to study the brain.

June 2022 - : Developing an industry 4.0-ready system for real-time management and monitoring of motor controllers, funded by Innovate UK and Sprint Electric (£289,000).
In this project we develop software that will enhance the motor controllers manufactured by SEL by adding Industry 4.0 and cloud-based services, enabling the following features; a secure web interface, embedded in the product, that enables service personnel to configure and commission a motor controller, remotely, through a standard web browser and, on-site, through a mobile handheld device; a web service to connect to single or multiple devices over the Internet, enabling remote diagnostics and fault-finding; a telemetry service for streaming operating data from the motor controller to the cloud for on-line and off-line viewing, aggregation and storage.

2021: RaptorQ-based data transport for low earth orbit Satellite Constellations, funded by GÉANT Innovation Programme (€30,000).
Very large constellations of Low Earth Orbit (LEO) satellites are currently being deployed; private companies (e.g., SpaceX, OneWeb) lead the race and billion-dollar investments are already in place. This will give rise to unprecedented wide-area network deployments that will provide 100% geographic coverage and exhibit a unique combination of characteristics. This project will be a first step towards rethinking data transport for LEO satellite networks.

2020 - 2024: Effective and efficient operation, management and monitoring of large-scale networks, funded by Moogsoft and the School of Engineering and Informatics (£264,000).
This project is in collaboration with Prof Berthouze (as co-PI) and will fund PhD studentships in the broader research areas of large-scale network management, monitoring and operation. These include the design of machine learning and data science approaches for identifying and preventing network and service outages, and the development of machine-learned network protocols.

2019 - 2020: Reinforcement learning in next generation data transport mechanisms, funded by Amazon ($27,200 in AWS credits).
This project is about the development of computer-generated, machine-learned congestion and flow control algorithms for next-generation computer network systems. The focus is on Reinforcement Learning using Actor-Critic approaches with deep neural networks. Amazon is funding this project through the provision of AWS services, including CPU and GPU resources.

2019 - 2021: Efficient and scalable computation of metrics for logical and physical IT infrastructure networks, funded by InnovateUK KTP and Moogsoft (£276,803).
This project is in collaboration with Prof Berthouze (as co-PI) and aims at the design and implementation of distributed algorithms for computing graph metrics and context-specific approximations for large scale computer networks.

2015 - 2016: Developing a simulation framework for data transport protocols in data centre networks, funded by the Sussex Research Development Fund (£10,000).
This project built on existing, community-driven work on packet-level network simulations, in order to develop a platform for experimenting with network protocols in simulated data centre topologies.

Research projects as Research Associate

2012 - 2013: FP7 Trilogy 2, funded by the EU.
A liquid system should ideally allow resources including bandwidth, storage and processing to be used by any application, whether they are contributed by network operators, data centre operators or end systems. In today’s Internet resource pools have limited scope and the net result is inefficient utilization of resources. Enabling Internet liquidity effectively creates “communicating vessels”, enabling larger resource pools and allowing applications to discover idle resources and use them. The main objective of Trilogy 2 is to unlock the value inherent in joining up the pools of liquidity in the Internet.

2010 - 2013: FP7 PURSUIT - Pursuing a pub/sub Internet, funded by the EU.
As a research associate, in the context of PURSUIT, I have developed an information-centric network (ICN) stack. The prototype has been deployed in a worldwide testbed. I have also researched architectural aspects of the ICN, such as information structuring, realisation of intra- and inter-domain rendezvous, forwarding and topology management. PURSUIT, was led by our team at the University of Cambridge and won the EU Future Internet Award. Groups from UC Berkeley, UCL, Huawei and Ericsson have cited my research. The open-source ICN network stack that I have developed has been used by partners of the PURSUIT project as well as by researchers at MIT and Ericsson. It has also been the main development platform of the EU H2020 research project Point.

2010 - 2012: EIFFEL Support Action, funded by the EU.
The EIFFEL initiative was a Support Action (SA) proposed for the 7th Framework Programme (FP7). The EIFFEL SA has been about mobilising European researchers to discuss and debate on the future of the Internet towards the development of the future networked society.

2006 - 2009: “PENED 2003”, funded by the Greek Secretariat of Research and Technology (GSRT).
This project funded my research during my Ph.D.

Consultancy projects

2004 - 2009: Technical Support Consultant for the Development of Metropolitan Optical Fibre Networks in the Region of Northern Aegean, Greek Information Society, 3rd CSF, Measure 4.2.
This project involved the design and supervision of the development of optical fibre metropolitan networks in three Greek islands (Chios, Samos and Mitilini). The MANs provide interconnection for the local networks of various public institutions (healthcare, administration, cultural). I was also involved in project management activities and coordination tasks. The project was funded by the Greek Information Society.

2006 - 2008: Member of a Consulting Committee for the Greek General Secretariat of Sports, Greek Ministry of Culture.
I was member of a consulting committee for the Greek General Secretariat of Sports. Our goal was the design of a nationwide access management and ticketing system for the Greek football league. The proposed design involves the installation of a local network, turnstiles (with integrated ticket readers and Ethernet connectivity) and surveillance cameras in all Greek football stadiums.