The localization algorithm is essential to support location-aware or geographical-based communication protocols to perform properly. However, recently proposed localization algorithms generally designed with the focus on achieving high localization accuracy without the consideration of the real implementation of the system. Computation capability and processing time are mostly neglected in the existing proposed localization algorithms, which causes the incompatibility of these algorithms to be implemented in the wireless sensor network system. In this thesis, a novel localization system is proposed to solve the mentioned problem through the use of the unmanned aerial vehicle as the anchor node and a single hidden layer feedforward neural network to estimate the position information. The proposed neural network architecture is trained using a fast learning algorithm called extreme learning machine algorithm. Simulation results show good performance of the proposed scheme in the term of localization accuracy and computing time compared to other learning algorithms and previously proposed similar localization systems.