Mohammed Bou Rabee
Public Authority for Applied Education and Training, Kuwait
Abstract Title:
A Battery-Free, Energy-Autonomous LoRaWAN Smart Bin with TinyML-Driven Route Optimization for Sustainable Municipal Waste Management
Biography:
Mohammed A. Bou-Rabee (Senior Member, IEEE) received his B.E. in electrical engineering from Wichita State University, USA (1984), his M.Sc. from North Carolina A&T State University, USA (1986), and his Ph.D. in electrical engineering from the University of New South Wales, Australia (1992). He is currently an Assistant Professor with the Department of Electrical Engineering, College of Technological Studies, PAAET, Kuwait. His research interests include resonant converters, PWM inverters, and power conditioning systems for new and renewable energy sources, including PV, fuel cells, wind, and solar energy. He led the PAAET-funded Sus-Bin project (Grant TS-24-01).
Research Interests:
Overflowing municipal garbage bins remain a persistent problem in urban environments, where the absence of real-time monitoring leads to environmental pollution, public health hazards, and inefficient collection logistics. While IoT-based smart bins have been widely explored, their dependence on batteries introduces recurring maintenance costs and electronic waste, undermining their sustainability. This work presents Sus-Bin, an energy-autonomous, battery-free smart garbage bin addressing both the sensing and logistics dimensions of solid waste management. The node harvests ambient solar energy through a thin-film photovoltaic harvester coupled with an ultra-low-power management circuit and a Lithium-ion Capacitor storage element, achieving an average harvesting rate of 10.44 mW under clear-sky conditions. Energy budget-based duty cycling reduces mean power consumption to 131 µW, enabling perpetual operation solely from harvested energy—sustaining over 15 days of operation even without sunlight. The system transmits fill-level data over LoRaWAN at 867 MHz, achieving a range of up to 361 m with over 90% packet delivery in a densely populated urban area. At the intelligence layer, a hybrid multi-objective framework combines a TinyML stacked-ensemble model predicting garbage fullness with 96.22% accuracy and a Tiny Genetic Algorithm tuned by lightweight Reinforcement Learning, reducing collection travel distance from 7.49 km to 4.41 km and CO₂ emissions from 5.02 kg to 2.95 kg in real-world testing. Extensively validated on a deployed testbed, the system offers a maintenance-free, environmentally responsible alternative to conventional smart waste management, directly supporting United Nations Sustainable Development Goal 7
