Prediction of Single Object Tracking Based on Learning Approach in Wireless Sensor Networks
Sahar Hamad Ahmed, Ahmed Noori Rashid · 2021 14th International Conference on Developments in eSystems Engineering (DeSE) · 2021
In recent years, there has been a growing interest in wireless sensor networks because of their potential usage in a wide variety of applications such as remote environmental monitoring and target tracking. Target tracking is a typical and substantial application of wireless sensor networks. Generally, target tracking aims basically at estimating the location of the target while it is moving within an area of interest and consequently reporting it to the base station on time. However, achieving high accuracy of tracking together with energy efficiency in target tracking algorithms is extremely challenging. This paper presents an efficient target tracking approach by combining dynamic clustering technique with the predictive tracking technique using the Long Short Term Memory (LSTM) networks which is the extension of Recurrent Neural Networks (RNNs), in this work two models from the LSTM was used, where the first model used for predict the three nodes that will track the object, and the second model was used to classify the direction of movement of the object and then calculate the new position of the object. Through the results obtained, The accuracy was up to 98%, and the error rate was 0.66892, and since the location of the object is random, as well as the distribution of sensors is random, every time the program is executed, the accuracy rate is very high, and this proves that the approach that was proposed, was very successful and the use of two models of LSTM increased the work efficiency.