Deep Learning and Internet of Things (IOT) Integration in Smart Livestock Farming for Climate Adaptation

Abstract

As the world enters the fourth revolution, the entire ecosystem gradually revolves around the evolving technologies led by Artificial Intelligence and its offspring, of which Internet of Things is making the world smarter than expected. AI and IoT-based technologies have revolutionised the ecosystem with the advent of smart systems capable of positively transforming the agricultural sector with the aim of a better agricultural system. This research harnessed the strengths of Deep Learning and an IoT-based model aimed at increasing productivity, improving livestock welfare, reducing disease outbreaks and primarily reducing the response time of the previous systems. Data were gathered using the TeSLF embedded system, which contains the sensors for sound, weight, humidity and temperature. The TeSLF embedded system was attached to both the ruminants and their environments (pen) for purposes of data collection. The harvested data were transmitted through an edge gateway to reduce the latency of the existing system. These data were then analysed by the Arduino Nano BLE Sense microcontroller using the data science PCA methodology. Having preprocessed the inputs using Python programming language, the trained model was then sent to the Mathworks platform for farmers, veterinarians and researchers to monitor the animals’ activities remotely and in real-time on their mobile and handheld devices using the TeSLF app. This system provided an automated data collection and analysis, real-time monitoring for timely disease detection and intervention, resulting in improved health of the entire herd. The system’s result was experimented on the farm; it produced an accuracy of about 98% and real-time information on the livestock, as opposed to that of the existing system, which produced an accuracy of 92%.