Associate ProfessorKanchana Thilakarathna

Associate Professor in Distributed Computing

Faculty of Engineering

Research projects & supervision summary

Project Opportunities

Title: Resilient and Secure Federated Learning

Summary of opportunity:

 

Research Area:

Computer Science, Cybersecurity, Distributed Systems.

 

Opportunity synopsis:

 

Billons of intelligent devices (“things”) capable of communicating are being deployed in our physical environment, and embedded in device being worn by humans. This led to the new era of Internet of Things (IoTs) where these lightweight devices - some are intelligent and some only marginally so - collecting and sending vital information to Cloud Data Centres (CDCs) for further processing and decision making. IoT has already begun to transform our industries and many mission critical tasks such as in electricity networks, in crisis response, in factories, in supply chain networks. The data collected from the sensors will enable optimised operations to automate and support various critical tasks such as coordinated defensive actions, ISR operations (Intelligence, Surveillance and Reconnaissance) and rescue missions. The continuous coordination of thousands, if not millions, of heterogenous things will place strict latency bounds and security requirements in dynamically changing often untrusted environments.

 

To address these concerns, it has been proposed to bring cloud-like resources closer to the edge of the network (Edge Computing), as it allows the delivery of delay-sensitive context-aware services by pushing the frontier of applications, data, and services away from centralised models and to distributed extremes of a network. The benefits of EC also come with additional risks: adding more data-generating devices to a network in more locations—particularly those that are physically remote or aren't well monitored—can lead to additional cyber security risks. Security at the edge remains a huge challenge, primarily because most IoT devices do not have software and hardware support for standard security protocols as such the security software updates which are often needed through the lifecycle of the device may not be present. This project focuses on developing resilient and secure distributed edge technologies with an improved performance, cheaper operating cost, and ease of deployment.

 

Further Information:

This scholarship is only available for Australian citizens, Australian Permanent Residents or New Zealand Special Category Visa holders.

 

Successful candidates will have:

  • A bachelor's degree in a relevant discipline

 

How to Apply:

To apply, please email kanchana.thilakarathna@sydney.edu.au, with the subject line"PhD Application” and your name. Include the following:

  • CV and cover letter
  • Transcripts (can be unofficial)

 

RESEARCH PROJECTS & ACTIVITIES

  • RESEARCH-BASED DEGREE SUPERVISION
    Adversarial and Out-of-Distribution Perspectives on Deep Neural Network Robustness
  • RESEARCH-BASED DEGREE SUPERVISION
    AI aid for volumetric video streaming
  • RESEARCH-BASED DEGREE SUPERVISION
    Analytics Over Encrypted Traffic and Defenses
  • RESEARCH-BASED DEGREE SUPERVISION
    Augmented Reality for the Visually Impaired - Sensory Augmentation Navigation System in 3D SceneUnderstanding
  • RESEARCH-BASED DEGREE SUPERVISION
    Automatic Privacy Compliance Checks for Mobile Apps Using Natural Language Processing
  • RESEARCH-BASED DEGREE SUPERVISION
    Composing IoT Wireless Energy Services
  • RESEARCH-BASED DEGREE SUPERVISION
    Crowdsourcing IoT Energy Services
  • RESEARCH-BASED DEGREE SUPERVISION
    Determining the Trust of Social Media images
  • RESEARCH-BASED DEGREE SUPERVISION
    Distributed inference of large neural networks
  • RESEARCH-BASED DEGREE SUPERVISION
    Exploring and Mitigating Privacy and Security Concerns in the Metaverse
  • RESEARCH-BASED DEGREE SUPERVISION
    Improving Privacy and Utility of Spatial Data in Extended Reality
  • RESEARCH-BASED DEGREE SUPERVISION
    IoT Service Recommendation for Multi-resident Smart Homes
  • RESEARCH-BASED DEGREE SUPERVISION
    Local Differential Privacy under Statistical Dependence: Leakage Analysis and Mechanism Design
  • RESEARCH-BASED DEGREE SUPERVISION
    Machine unlearning via hypernetwork
  • RESEARCH-BASED DEGREE SUPERVISION
    MonoDNN: A Deep Learning Compiler for Ultra-Fast Inference with Minimal Runtime Overhead via Holistic Code Optimization and Generation
  • RESEARCH-BASED DEGREE SUPERVISION
    Network and Content Intelligence for 360 Degree Video Streaming Optimization
  • RESEARCH-BASED DEGREE SUPERVISION
    Privacy and Security Preserving Distributed Machine Learning and Data Storage in Large-Scale Internet of Things
  • RESEARCH-BASED DEGREE SUPERVISION
    Reasoning at the Edge: A Compact LLM Agent for Privacy-Preserving Time-Series Computation
  • RESEARCH-BASED DEGREE SUPERVISION
    Resilient Wireless Trust Establishment for Untrusted Distributed Systems
  • RESEARCH-BASED DEGREE SUPERVISION
    Situational Awareness through Passive Monitoring of Network Traffic