The dairy goat industry has been growing quickly, in Flanders it already houses over 26 000 animals, which produce over 30 million liters of milk annually. As the industry becomes more commercialized the customers become more concerned with the overall welfare of livestock and farmers are looking to further increase the profit margins. This creates a need for wireless monitoring systems that can provide the farmers with real-time insights into the daily activities and social interactions of dairy goats. This information can then be used to further optimize the production process and ensure optimal living conditions for the animals. By equipping each goat with a small wirelessly connected accelerometer and UWB tag we can collect both activity and positional information. By applying novel deep learning techniques on this data we can extract useful behavioral statistics as well as detect key events, such as kidding, and health or welfare issues. To ensure a broad applicability of the proposed system we will focus on reducing both the required infrastructure as the power consumption of the sensors. To this end a significant part of this study will be focused on both the reduction of the required wireless and localization infrastructure as well as on the optimization of the power consumption of the wireless link and on-device processing.