Intelligent Road Traffic Status Detection System Through Cellular Networks Handover Information
M G Demissie, G H Correia, C L Bento, Coimbra University, PT
This study explores the use of cellular networks handover information in order to complement the effort of traffic flow rate estimation on the arterial roads, through the use of multinomial logistic regression and artificial neural networks.
The growing development of various Intelligent Transportation Systems (ITS) schemes needs comprehensive and high quality road traffic information. Successful deployment of ITS schemes requires installation of large data collection points and deployment of substantial amount of probes in to the traffic stream. However, the cost prohibitive nature of primary road traffic data collection methods, either human observation or different forms of remote sensing does not allow the coverage of the entire transportation network in a city.
Therefore, they are only providing information from the limited part of the road network. Thus, traffic management sectors would be forced to rely on an incomplete picture of the traffic stream in the city. In this study, an innovative method of road traffic status detection system is developed. We explore the use of cellular networks handover information in order to complement the effort of traffic flow rate estimation in the road traffic stream. To test this method, handover counts were obtained from the cellular towers in the vicinity of arterial roads on five case-study places in Lisbon, Portugal. Traffic counts were also obtained from the same roads and their hourly traffic classified into three categories: high, medium, and low traffic levels. An initial correlation analysis proved the existence of a strong relationship between handover and traffic volume, and then half of the data collected was used to build a multinomial logistic regression model (MNL) and to train an artificial neural network (ANN) that relates these traffic levels and the handover counts. The other half of the data was used to validate both models. The estimation results from MNL and ANN are competitive with good classification accuracy of 72 percent and 81 percent, respectively. The estimation outputs give indicative values about the intensity of the traffic that can be used as a baseline to trigger different road traffic management schemes to regulate the operation of traffic flow and maximize throughput. The results from this study have various practical applications for the active traffic management systems in the field of advanced traveler information systems, advanced traffic management systems, and to evaluate system performance for modeling and planning purposes. We conclude that this study suggests the interest of using cell phone handover information in estimating the road traffic status.
Keywords: traffic estimation, cellular handover, multinomial logistic regression, artificial neural network
Association for European Transport