Research themes
Our work sits in communication and electrical engineering. The current themes meet in one picture: an indoor space lit by LEDs that also carry data, where devices know their position, and a software-defined network decides which link each device should use. Earlier work on mobile networks and on smart power grids still informs it.
Light as a data link
LED lights can send data by changing their brightness faster than the eye can see. Because an LED can only send a positive brightness, standard radio OFDM does not work directly. We study optical OFDM schemes such as ACO-, DCO- and layered ACO-OFDM, the distortion that LEDs add (clipping, nonlinearity and memory), and how transmitter processing and deep-learning receivers can recover the data.
- Layered ACO-OFDM: inter-layer clipping and transmitter-side overlap suppression
- Neural receivers (MLP, CNN, CNN-LSTM, wavelet features) and their computing cost
- Hybrid LiFi-WiFi networks
Knowing where things are indoors
GPS does not work well inside buildings. Ultra-wideband (UWB) radios measure distance very precisely, but walls and people block the direct path and bias the result. We use machine learning to detect these non-line-of-sight conditions, choose the best anchors in real time, and combine UWB, LiFi and WiFi fingerprints for industrial IoT.
Networks that steer themselves
Software-defined networking (SDN) separates the network's decisions from its hardware, so traffic can be moved when a link or controller is overloaded. We work on switch migration and load balancing in fog computing, context-aware multipath TCP offloading in hybrid LiFi-WiFi networks, and traffic in the Internet of Vehicles.
Sensors that work in the field
We build wireless sensor networks and IoT systems that run outside the lab: bio-inspired routing for sensor networks, lightweight middleware, stingless-bee hive monitoring, water-quality and irrigation systems, a self-sustaining weatherproof monitoring system (SWIMS), and an off-grid solar-powered IoT system for an aquaculture farm in Johor.
Staying connected on the move
My PhD work at Shibaura Institute of Technology studied how phones and vehicles keep their connection while moving: velocity-aware handover using ant colony optimisation and neural networks, multipath transport (SCTP) path selection, Host Identity Protocol mobility, and phone-based traffic-state estimation for intelligent transportation systems.
Communication for the power grid
With power-engineering colleagues we worked on distribution networks with distributed generation: optimal sizing and placement by swarm and gravitational search algorithms, loss minimisation, battery switching stations, microgrid tariffs, and the communication a smart grid needs for frequency control.