Design of an IoT sensor-based network architecture for waste collection in Bukavu city (DRC)

Bukavu, in the Democratic Republic of the Congo, faces a critical waste management crisis due to rapid urbanization. The city produces approximately 898 tons of waste daily, yet only 7\% is collected through an inefficient system, intensifying sanitation and environmental risks. This study designs and validates a comprehensive Internet of Things network architecture to enable real-time monitoring of waste bin fill levels and optimize collection. The methodology involved a 12-week pilot deployment of seven smart bins prototypes across three neighborhood types: residential, dense, and peri-urban. Each 120-liter bin (empty weight: 5 kg, density: 0.20 kg/L) was equipped with an EM310-UDL ultrasonic sensor. Using a two-point calibration for bins with a useful height of 787 mm, a measured distance greater than 881 mm indicates an empty bin (0\% fill), while a distance less than or equal to 94 mm signals a full bin (100\%). Results demonstrate robust system viability: sensor energy autonomy remained above 90\% after 10 months for most units, despite network challenges where 57\% of sensors had a Received-Signal-Strength-Indicator less than or equal to -107 deciBel-milliwatt and 43\% exhibited negative Signal-to-Noise-Ratios values up to -10.25 decibel. The system enabled precise waste quantification, measuring 641 kg over 14 days in one pilot area. This research provides a practical technological and methodological framework for smart waste management in Bukavu. It paves the way for significant resource optimization, improved urban hygiene, and reduced operational costs, with a system design adaptable to bins ranging from 0.787 m to 4 m in height.