A predictive anomaly algorithms on spatio-temporal traffic flow-enabled internet of things

Abstract

The transportation infrastructure has advanced significantly in the last few decades, yet traffic issues persist since more people are
living in metropolitan areas, necessitating the usage of various modes of transportation. Due to this, there are now more challenges with  traffic control that directly affect the public, such as air pollution, traffic rule violations, and accidents. In this regard, intelligent transport  systems integrate intelligent algorithms and the internet of things as an alternative for improving the traffic environment. In this study, a  thorough analysis of traffic anomaly prediction involving the transition from spatiotemporal data flow is presented. It consists of a  comprehensive analysis of various techniques applied for anomaly prediction on spatiotemporal data traffic. The various benchmark  algorithms and models adopted to validate the performance of the proposed techniques are presented. Metrics adopted to evaluate the  performance of proposed techniques are highlighted and briefly discussed. Limitations of the proposed techniques during and after the prediction phase are documented. The outcome of this study shows that Convolutional Neural Network techniques were majorly proposed and applied to predict anomalies in spatiotemporal data traffic flow, while Classification algorithms were mostly adopted as benchmarks for performance validation of the proposed techniques. It was also observed that Root Mean Squared Error (RMSE) was  majorly adopted to evaluate the performance of the proposed techniques. Also, Computation Complexity was discovered as the most  prevalent challenge bedeviling the proposed techniques, paving the way for future research directions in this field.