Software uses and approaches in the Internet of Things (IoT) are very varied and rich. But the data they collect is often gathered from networks of sensors. However, the constraints of Wireless Sensors and Actuators Networks (WSAN) are not always well taken into account by IoT software approaches. Thus, the massive data collection, well managed by Big Data tools in IoT applications, conflicts with the energy constraints to which the sensors are subject. Each transmission is costly for these devices and a massive stream reduces the whole WSAN lifetime. This article presents an approach that meets the data needs in the vision of IoT users while seeking to limit the impact of the massive transmissions required. Edge computing is often used to reduce the amount of data to be treated in this research field. But the reduction is still made outside the WSAN, leaving a significant transmission load for these constrained networks. Energy-aware IoT applications get increased attention from the researchers. Network coding is an appropriate and already well-known tool for saving energy in constrained networks. Then the question is how to combine all the messages to reduce their number while maintaining their accuracy. In this paper, we present an approach focusing on the gain obtained thanks to the semantics of data. The Semantic Network Coding leads to a reduction in the number of messages forwarded, thus leveraging the network and saving energy, while keeping a good quality to the collected data.