The Wireless Sensor Networks (WSNs) deployment introduces many issues and challenges mainly in terms of energy independence. In this context, we adopted the IBM control loop which is composed of four steps (Monitor, Analyze, Plan and Execute) to manage Quality of Service (QoS) 1. This paper focuses on the first step which consists in monitoring and sending periodically QoS values such as the value of power remaining in the battery of each sensor. We notice that the transmission process is very costly in terms of energy and reduces the battery lifetime. In this work, we propose a probabilistic approach that estimates a part of these QoS monitoring values and therefore economizes their transmission energy and extends the sensor battery lifetime. Our approach is based on the hidden Markov chain and the fuzzy logic. It is composed of two steps: (i) learning which allows apprehending the WSNs behavior and (ii) prediction which estimates QoS monitoring values. A WSN application deployed in a datacenter is studied as an illustration. The carried out experiments over AZEM WSN simulator show that the gain varies from 25% to 75% of the battery energy.