The rapid and irregular movements can be limiting in observing UAV behaviour. Thus, accurate coordinate estimation is a key component for precise UAV trajectory recognition. In this thesis, a generic model for predict future movement of UAV is proposed by using a bi-directional long-short term memory (Bi-LSTM) based approach. First, a representative dynamic 3D space UAV movement model by random walk mobility is created using MATLAB simulation. Based on the irregular mobility pattern from object trajectories, the propose system will try to forecast the future movement of an object using previous and subsequent action using Bi-LSTM neural network model. Performance evaluation analyzed by comparing the method with other time-series algorithm; long-short term memory (LSTM) and gated recurrent unit (GRU) by considering how small the system creates errors values and also how fast the prediction system. The results show that the proposed work effectively to produce excellent performance in prediction as well as computing time, even though in random behaviour.