The Traffic Flow Prediction Models in Vehicular Networks
- Monday, April 29, 2019 from 3:00pm to 4:00pm
- Barnard Hall, 347 - view map
Traffic flow prediction is a challenging, yet important process due to an increase in the number of vehicles on the roads each year. More dense traffic results in higher probabilities of collisions and less optimal traffic flow. However, traffic flow data is very dynamic and highly dependent on the area, day of the week, time of day, local events occurring, etc, making accurate predictions of traffic flow difficult to produce. By applying algorithms such as intelligent swarm-based neural networks, deep learning such as deep belief networks, stacked autoencoders, and long short term memory, accurate predictive models are produced. These models may result in fewer collisions, more optimal traffic flows, and information necessary for determining future roadway infrastructure. This survey summarizes five methods of traffic prediction, critiques these methods, and offers some open questions for future research.
- Gianforte School of Computing