Wetlands provide a huge amount of goods and services to humankind. Monitoring their health status is paramount in decision-making. Many tools have been tried but most of them are expensive and time-consuming. In this research, anuran is used as an accurate indicator species to determine the health of Rwandan wetlands. Anuran monitoring using traditional methods turned out to be expensive because of the time spent in data collection and sometimes weather conditions are limiting factors for sampling. As such, we propose an intelligent system based on Internet of Things technologies to monitor the health of Rwanda’s wetlands ecosystem.

Firstly, battery-powered sensor nodes containing the acoustic sensor with an external microphone, recording module, sensitive sound sensor, display unit, ambient sensor and, ESP8266 Node MCU were intended to be deployed in sites such as the lower Akagera Wetlands complex, upper Akagera wetlands both in undisturbed and cultivated swamps, Lake Rweru and Nyabugogo wetland for permanent recordings without ecologists’ intervention. Due to the difficulties imposed on the world by the covid-19 pandemic, this activity was not successful, instead, we managed to successfully develop a prototype and perform in-house testing at a small scale. Secondly, those data without anuran calls are filtered out and those with anuran calls are sent via WI-FI connectivity to ThingSpeak for monitoring and analysis. Next, anuran calls recorded that help in anuran identification is accessed from ThingSpeak without human attending the wetlands. Finally, the health of wetlands is determined based on the comparison of this system’s output and the findings of the aforementioned researchers.

Experimental results of the developed prototype for the proposed intelligent system demonstrates that our intelligent system significantly eradicate the time ecologists spend in the process of data collection as data is automatically collected by sensors. The result of the prototype also indicates the state of the wetland health can be estimated by automatically comparing this system’s output and the finding of aforementioned researchers.