Prof. Georgia Sakellari

Energy Efficiency

Edge IoT Computing

Cloud Computing

Large-scale Network Infrastructures

Investigating ways to reduce the ICT sector’s CO2 emissions by leveraging the environmental benefits of edge and cloud computing for IoT devices and individual PC users. We are looking into the development of a multi-criteria decision mechanism to guide the choice of execution location. According to criteria such as the processing intensity and the data collected by a device, the energy consumed and the conditions of the network, there can be three potential choices: executing everything locally on the device, or on the edge device or offloading the execution to a more powerful but remote cloud infrastructure which would incur additional networking overhead.

We have also investigated the effect that deep sleep based power savings have on the quality of service of the backbone network. We investigated the tradeoff between power consumption and network performance by introducing enhanced neighbour discovery, where special queues at each node store and delay forwarding packets to next hop nodes that are in sleep mode.

For more information, please see my publications.

Research in

IoT and Emergency Response

Data orchestration

Infrastruuctureless IoT Environments

Inertial-based localisation

Investigating ways to regulate IoT/augmented IoT information by determining what data should be shared with what entities and at what times. Depending on the environment and case study, multiple parameters need to be considered when regulating IoT data information, such as the priority of the information, the network conditions and the energy levels of the devices. Furthermore, for emergency response situations, where first responders receiving the information might be under stress, aspects such as the cognitive load and ability of the FR to process the information received should also be considered in data orchestration.

Another topic we have looked at in emergency response situations, is indoor localisation. We have investigated ways to provide more accurate indoor location by a sensor fusion approach primarily utilising the Inertial Measurement Unit (IMU) of Commercial Off the Shelf (COTS) modern smartphone devices, consisting of 3D accelerometer, 3D gyroscope, and 3D magnetometer, that can be easily carried by an FR, and using collaborative and opportunistic approaches to improve their accuracy and reduce drift.

For more information, please see my publications.

Research in

Self-Aware Networks

Quality of Service

Admission Control

Resilience to Worms

A Self-Aware Network can probe itself through special packets, collect data at different distributed points, and adaptively take corrective action based on the Quality-of-Service objective that it is pursuing. Self-Aware Networks use adaptive packet routing protocols, such as the Cognitive Packet Network, which is designed to perform self-improvement by using random neural networks with reinforcement learning, and genetic algorithms.

In a Self-Aware Network we have introduced an admission control mechanism that controls traffic congestion and improves Quality of Service throughout the lifetime of all connections by proactively probing the network. Uniquely, it enables network users to specify their own individual QoS requirements and maximises multiple QoS criteria using a novel algebra of QoS metrics. The algorithm accommodates new flows after identifying paths with acceptable QoS values. This work was carried out within the award-winning BAE/EPSRC/SELEX-funded ALADDIN project and has led to several more publications in high-level conferences, a best-paper award in Chinacom 2008 and a patent filed in collaboration with BAE Systems.

For more information, please see my publications.

Dr. Georgia Sakellari

Research in

Energy Efficiency

Research in

IoT and Emergency Response

Research in

Self-Aware Networks

Teaching in

Computer Networks

Research Projects

RESCUER: First Responder Centered Support Toolkit for Operating in Adverse and Infrastructure-less Environments

C4IIoT: Cybersecurity 4.0 - Protecting the Industrial Internet of Things

SUN: Social and hUman ceNtered XR

EUNOMIA: User-oriented, secure, trustful and decentralised social media

ENSURESEC: Safeguarding the Digital Single Market’s E-Commerce Ecosystem

Research Project Participations

EPSRC (GR/S52360/01) on Self-Aware Networks and QoS: Developed the first network admission control mechanism able to take into account multiple criteria set by individual users of the network, resulting in markedly reduced congestion and improved quality of service.

DIF DTC HYPERION on Automated and Adaptive Information Management: Developed a smart networking protocol that was naturally resilient to failures and worms through self adaptation, using Random Neural Networks

BAE/EPSRC ALADDIN on Dynamic Distributed and Decentralised Intelligent Systems: Extended previous work to develop a distributed multi-criteria admission control mechanism for communication networks

TSB/EPSRC SATURN on Resilience of the UK Critical National Infrastructure: Carried out research on self-aware networking for critical national infrastructures.

FP7 FIT4Green on energy-aware data centre technologies: Developed a distributed mechanism for minimising a large network’s energy cost by regulating user admission in an automated manner, and experimentally evaluated the tradeoff between performance and energy cost

Contact info