Context-Aware Services in Aquaculture: FiFaMoS - Fish Farm Monitoring System

Abstract

This thesis focuses on context-aware services that make decisions based on the situation (i.e. context) of the involved entities. Context may in general be based on user input, sensed or derived (e.g. combination of multiple context entities). The type of such services is vast, but in this thesis the system is aimed towards the aquaculture industry. During the last years, aquaculture quality has become more and more important in the fish farming industry. But this importance has not been reflected yet in using information and communication technologies (ICT). The main problem in a fish farm is that most of them are without supervision for a long time while they are exposed to changing weather conditions. This problem gets even bigger when fish farms are established far from land and often becomes exposed to extreme weather conditions. In order to minimize the consequences caused by lack of information on the fish farm (such as weather conditions and other variables) when there is no workers around, ICT surveillance systems should be used. Context-aware services are perfectly suited for this type of application, and the task of this thesis is to specify, design and implement a context-aware application for the aquaculture industry. This includes a context source application, a context consumer application and a service to be deployed on a context management system. Our solution is named FiFaMoS (Fish Farm Monitoring System) and is based on the APMS context manager. This is a context management system that provides easy service deployment due to built-in support of multiple binding types as well as persistent storing of context. As a context source, an application for an M2M module is developed. There will be one module situated at each sea cage, which collects information like feed level, temperature, pH, oxygen level from the sensors. In addition, the module gets the positioning information for the sea cage from a connected GPS receiver. This information is periodically sent to the context manager that interprets the context and makes it available to users of the system (context consumers). In addition alarms will be triggered if values are out of bounds. It has been developed two different context consumers; one for personal computers and one for mobile phones. In these applications it is possible to view both current and historical sensor values, and receive alarms. It is also possible to alter the fish farm configuration via the PC client. Aspects that will be discussed are the use of different binding types, representation of data when transmitting and storing, hardware choices and various implementation choices. The implemented FiFaMoS system uses web services as binding to get a loosely coupled system, and objects are represented in XML which makes the system easy to alter. Detailed testing has been performed, and the system works as intended.