The trend in oil companies nowadays is to integrate the entire well sensors (modern and legacy sensors) with wireless sensor network (WSN). In this work, we introduced a new framework from such sensors using a heterogeneous network of sensors taking in our consideration the WSN's constraints. The framework combined two modules: a Wireless Sensor Data Acquisition (WSDA) module and a Central Data Fusion (CDF) module. A test bed was established from ten acoustic sensors mounted on a closed loop pipeline. The flow rate and the differential pressure were monitored as well. The CDF module was implemented in the gateway using four fusion methods; Fuzzy Art (FA), Maximum Likelihood Estimator (MLE), Moving Average Filter (MAF) and Kalman Filter (KF). The results show that the KF fusion method is the most accurate method. Unlike the other methods, Kalman filter algorithm does not lent itself for easy implementation; this is because it involves many matrix multiplication, division and inversion. Among these 17 matrix operations, there are 10 matrix multiplications, 2 matrix inversions, 4 matrix additions and 1 matrix subtraction. Moreover, these tasks are computationally intensive and strain the energy resources of any single computational node in a WSN. In other words, most sensor nodes do not have the computational resources to complete a central KF task repeatedly. Furthermore, the computational complexity of the centralized KF is not scalable in terms of the network size.