Designing an IoT-cloud solution for precision aquaculture

• Main Author: Liew, Wei Zheng
• Format: Final Year Project / Dissertation / Thesis
• Published: 2023
• Subjects: S Agriculture (General); T Technology (General); TA Engineering (General). Civil engineering (General)
• Online Access: http://eprints.utar.edu.my/5993/1/fyp_IA_2023_LWZ.pdf; http://eprints.utar.edu.my/5993/

In recent years, the demand of aquaculture product is increasing due to the growing population of the world and the depletion of wild fish population. To address the depleting fish stocks caused by overfishing and environment pollution, aquaculture will become an alternate source to supply fish stocks to fulfil the demand. Aquaculture has been evolved throughout times by utilizing different technology to improve efficiency and increase production. But, even with the implementation of technology, aquaculture industry is still vulnerable to factor such as labour shortage and inaccurate data analysis. In this case, water quality needs to be monitor periodically because it plays an important role in determine the health and growth of fish. Measuring water quality involve fish farmer collect sample from ponds and pass them to an examiner to perform water testing. The accuracy of the analyse result are determine by the experience of the examiner and prone to human error. Thus, this project designs an IoT-Cloud Solution for Precision Aquaculture in predicting water quality to improve the efficiency and effectiveness of aquaculture operations to increase the overall production. In this paper, an aquaculture farm will be referred as an edge computing environment consisting of IoT sensors and devices used to obtain water parameters from the farm, processed them and store them locally as well as send to cloud for other uses. Sensors of various purpose are used to collect real time water parameter so that the data can be transmit to cloud for analysis via edge device using MQTT protocol. Edge computing can be used to process data locally, reducing the bandwidth requirements for transmitting large amounts of data to the cloud and also reduces the latency in receiving results from the analysis from cloud. While the cloud can store the data as a backup in unstructured format and later serves as data for visualise graphically on a dashboard. The federated learning framework also involves, employing a client-server architecture. In this setup, the edge environment functions as the client, which trains a local machine learning model using local dataset. Meanwhile, the server aggregates the model weights from multiple clients to create a more powerful, global model without accessing the individual client's data. The trained model from the server is then redistributed back to the clients. This process allows clients to have an improved model while ensuring the privacy of their data remains intact.