MDU team develops system to predict traffic congestion

Prof Preeti Gulia, head, Department of Computer Science at Maharshi Dayanand University (MDU), Rohtak; Dr Nasib Singh Gill, senior professor; and Ayushi Chahal, a research scholar at the department, have developed a system to predict congestion, which would help in improving traffic management in smart cities.

Advertisement

The Patent Office of the Union Government has issued a patent to the team for the system ('A Method to Predict Traffic Congestion in IoT-Enabled Smart City'). "It is an algorithm based on machine learning and deep learning meant for IoT (Internet of Things)-enabled smart cities. The underlying idea is to help residents and commuters by providing real-time data regarding congestion," said Chahal, whose topic of research is 'Review and Design of Predictive Analytics Techniques for IoT-enabled Environment'.

Talking to "The Tribune" on Thursday, the research scholar stated that traffic congestion forecasting also addressed problems such as air and noise pollution and fuel consumption, etc., which would help in maintaining a sustainable environment.

Prof Gulia said the innovation would improve the regulation of traffic by predicting congestion,, thereby leading to significant progress in road-safety management. "The system presents a model through which the functionality of the traffic management mechanism can be easily increased. It helps in reducing traffic on roads, thereby preventing congestion and saving precious time of commuters," she maintained.

The MDU Vice-Chancellor, Prof Rajbir Singh, congratulated the team for the patent.

The City Plus EU FP7 project used in this invention is a freely available dataset. It contains road traffic data of smart cities. The dataset consists of road traffic data collected through sensors at various locations.

The Mean Absolute Error and Mean Squared Error values were reported to be 0.499 and 0.337, respectively, which makes the method of the present invention more error-free and effective than those using the existing models.