Pursuing your PhD ! 

Doctor of philosophy (PhD) is an exciting intellectual journey that will take you to uncharted new dimensions. You will expand your career prospects and propel your understanding of the world around us. In addition to the cutting-edge technical knowledge that you will both create and gain, you will also acquire an unparalleled ability for problem solving and analytic thinking. 

However, PhD is not for you unless you are really ready to undertake this long and difficult journey. If you are a person that enjoys tackling mind-challenging puzzles and dive deep into unknown territories of science and engineering, then this is the way! There is nothing more rewarding than creating new knowledge and feeding back to the immense space of human creation. 

If you are interested in pursuing your PhD with my team, please see the below guiding themes. Pay attention that you will need to meet RMIT University program entry requirements. Also mind that scholarships are quite difficult to get.

Available projects themes

PhD is not a project! it is an exploration journey, so these themes are provided as examples of what I can support you with. There are many other excellent PhD topics listed at RMIT University website.

Artificial Intelligence Methods for Satellite Communications and Sensing

Transferring information over a wireless medium has seen a steep improvement during the last 30 years, taking advantage of the increasing computational power of digital signal processors and their reduced cost. This has allowed the introduction of novel algorithms that operate near the theoretical limits under known channel and noise conditions. However, with more and more wireless devices joining the network, a vast problem in managing the limited radio resources and interference is becoming an increasing hurdle for today’s wireless network. Novel methods in machine learning and neural networks combined with smart signal processing approaches hold a very promising approach to tackle several key problems:

- Interference learning and mitigation in shared spectrum environment
- Radio channel learning and adaptation in dynamic scenarios
- Radio resources sharing methods and distribution using machine learning
- Compensation for RF hardware imperfection
- RF finger printing for increased security
- Adaptive waveform design and optimization
- Cellular-to-UAV communications
- Synthetic aperture radar signal processing
- Interference mitigation in Synthetic aperture radar (SAR)

See related references in the publication page

Internet-of-Things (IoT) over UAVs and Satellites / Satellite Communications and Networks

With the increasing ubiquity of Internet connectivity and cloud services, we are currently witnessing an accelerating pace in connecting more physical things and sensors to the global network. A new concept of the Internet-of-Things (IoT) was born where a vast number of IoT devices, 20-30 billion, will soon be connected to the Internet by 2024- 2030. This vast growth is driven by many emerging business cases such as autonomous vehicles, industry automation, utility metering, precision agriculture, asset tracking and monitoring of livestock and wildlife.

Despite this demand, most rural and remote areas are still outside traditional terrestrial coverage even with the recent enhancements that have occurred in this field, exemplified in the new mobile generation 5G. The lack of affordable remote wireless coverage is due to the expensive and inefficient satellite systems that are currently in place. 

This project aims to develop novel methods for integrating satellite and terrestrial wireless networks into a single seamless continuum. This will enable an extended and reliable wireless coverage across urban and rural environment alike, and will connect remote sensors to the internet in an affordable and reliable manner. 

The project has the following four research areas:

Area 1: Develop novel analytic models to capture and mitigate terrestrial-to-satellite interference (in shared spectrum) using statistical methods and machine learning methods in the 3D geometric space and temporal space.

Area 2: Investigate the use of intermediate relay nodes (or aggregators) such as Unmanned Aerial Vehicles (UAV), low-altitude platforms and high-altitude platforms.

Area 3: Investigate the fundamental performance bounds when using shared radio spectrum resources, aiming to achieve higher spectral efficiency, reduced interference, lower energy consumption and lower latency.

Area 4: Develop practical network access techniques and algorithms that are designed to maximize the set performance metrics of the hybrid satellite terrestrial system

Area 5: Software defined networks and routing techniques for connected satellite constellations including latency analysis and optimization

See related references in the publication page

Machine Learning and Neural Network Methods for IoT Sensing

Internet-connected sensors are increasingly becoming more affordable and widely available. Sensors like radar, infrared, and lidars are getting miniaturised with enhanced performance and accuracy. This theme aims to harvest the recent development in machine learning methods and neural networks to analyse IoT signals and devise new applications of such signals. Fundamental questions in this area include:

• What are the optimal methods for pre-processing sensor signal for improved recognition

• How to optimize sensors design and pre-processing for better feature extraction

• What are the suitable learning algorithms for analysing the signal and how to improve these algorithms

Example research areas under this theme are: 

Area 1: Hand-gesture recognition: One of the promising applications of applying machine learning on sensor signal is hand-gesture recognition. Device control based on hand-gestures is rapidly becoming an important method for human-machine interface with a wide range of applications in various fields of consumer electronics such as: wearable electronic, mobile phones, vehicle control, and medical devices. This theme explores the use of miniature sensors (such as radars and thermal sensors)

Area 2: Objects and materials identification: Sensors and radar signature that reflects from an object is very rich with information. New methods in machine learning and neural networks can open vast possibilities for using embedded sensors to recognize patterns and signatures of different objects, materials or movements.

Area 3: People counting: Knowing the number of people in a certain room is only the start, people counting applications are vital for smart city, smart transport, and smart buildings.

Area 4: Health monitoring: Smart signal processing and machine learning offer non-invasive ways to monitor the people health and provide early alarms if anomaly signature is detected. 

See related references in the publication page