Emerging with the Internet of Things, new applications are requiring more intelligence on these things, for them to be able to learn about their environment or other connected objects. One such domain of application is for livestock monitoring, in which farmers need to learn about animals, such as percentage of time they spend feeding, the occurrence of diseases, or the percentage of fat on their milk. Furthermore, it is also important to learn about group patterns, such as flocking behaviors, and individual deviations to group dynamics. This work addresses these problems, by collection and processing each animal location and selecting appropriate metrics on the data, so that behaviors can be learned afterwards using machine learning techniques running on the cloud.

Hence, the following main components were implemented:

• A collar with sensing devices implementing energy efficient functions such as performing oportunistic routing of data and selectively selecting a single gateway to the internet among all the animals' collars

• A user interface to the farmer - intuitive, mobile support

• A cloud solution, cost effective for farmers, that storages all collected information and further process it

• Smart algorithms to learn patterns from data