Congestion Prediction in Internet of Things Network using Temporal Convolutional Network A Centralized Approach

Abstract

The unprecedented ballooning of network traffic flow, specifically, Internet of Things (IoT) network traffic, has big stressed of congestion on todays Internet. Non-recurring network traffic flow may be caused by temporary disruptions, such as packet drop, poor quality of services, delay, etc. Hence, the network traffic flow estimation is important in IoT networks to predict congestion. As the data in IoT networks is collected from a large number of diversified devices which have unlike format of data and also manifest complex correlations, so the generated data is heterogeneous and nonlinear in nature. Conventional machine learning approaches unable to deal with nonlinear datasets and suffer from misclassification of real network traffic due to overfitting. Therefore, it also becomes really hard for conventional machine learning tools like shallow neural networks to predict the congestion accurately. Accuracy of congestion prediction algorithms play an important role to control the congestion by regulating the send rate of the source. Various deeplearning methods (LSTM, CNN, GRU, etc.) are considered in designing network traffic flow predictors, which have shown promising results. In this work, we propose a novel congestion predictor for IoT, that uses Temporal Convolutional Network (TCN). Furthermore, we use Taguchi method to optimize the TCN model that reduces the number of runs of the experiments. We compare TCN with other four deep learning-based models concerning Mean Absolute Error (MAE) and Mean Relative Error (MRE). The experimental results show that TCN based deep learning framework achieves improved performance with 95.52% accuracy in predicting network congestion. Further, we design the Home IoT network testbed to capture the real network traffic flows as no standard dataset is available.