Digitalization in the Aquaculture Industry: Validation Trials over a Commercial 5G Network

The aquaculture industry has a goal of automating as much as possible to minimize cost and improve product quality. Cameras and environmental sensors are extensively used to monitor the fish farming sites, and generate huge amounts of data. 5G technology is seen as an enabler to further improve efficiency and digitize the fish farming industry. In this work, the combination of 5G, Device Edge, Cloud and Artificial Intelligence (AI) has been tested to evaluate the benefits and limitations of 5G technology in aquaculture. By emulating a typical Norwe-gian Atlantic salmon farm, remote monitoring, feeding decision support using AI for pellet detection, and 5G performance has been assessed. Peak uplink data rate is the most important key performance indicator, due to the large amount of data produced in the farm itself. To reduce the uplink requirements, a Device Edge has been deployed for running AI-driven pellet detection. Results show that operating full video coverage both underwater and for surveillance clearly exceeds the offered uplink data rate of a typical 5G base station operating in the C-band. Video compression can only be used to a mild extent, due to early deterioration of the pellet detection precision. Therefore, the use of a Device Edge to avoid uplink transmission of the video streams seems to be a better solution. Latency has not been critical in the scenario investigated, however introduction of remote control of cameras and feed provision might change this.