Abstract:Road traﬃc data are conventionally obtained by ﬁxed-location detectors e.g., induction loop coil sensors and closed-circuit television systems. However, these conventional on-road sensors for traﬃc data collection cost highly both in installation and maintenance, and they are only available in a limited part of the road network. Thanks to the current ubiquity of cellular communication infrastructure, the mobile phone location data, obtained from the cell towers, are an alternative to collect real-time on-road traﬃc data with low costs for traﬃc management operations. In this paper, the classiﬁcation of vehicle types is considered from the mobile phone location data that can be made available at the cell towers whose locations are known. Considering data that can be mixed from mobile phones carried by passengers in the diﬀerent modes of urban transportation, this work deals with the motorcycle, car, and Skytrain via the Simulation of Urban Mobility (SUMO) with the Sublane model. The random forest, k-nearest neighbor, and support vector machine algorithms are applied for the vehicle type classiﬁcation. From the random forest and k-nearest neighbor algorithms, the classiﬁcation accuracies are greater than 85% and 80%, respectively, with the cell tower inter-spacing less than 500 meters and greater than 80% for the average classiﬁcation accuracy for both random forest and k-nearest neighbor algorithms. On the other hand, the support vector machine algorithm’s classiﬁcation accuracy is greater than 85% only in the case of 100 meters of cell tower inter-spacing. Since the existing cell tower inter-spacing is typically in the order of 500 meters for the current 4G mobile phone technology, and as lows as 100 meters for the upcoming 5G technology, the SUMO-based ﬁnding in this paper suggests a clear potential of using this ubiquitously available mobile phone data for the vehicle type classiﬁcation and other subsequent travel information extraction analysis in the future.

The Diversity and Scale Matter: Ubiquitous Transportation Mode Detection Using Single Cell Tower Information

A Road Congestion Detection System Using Undedicated Mobile Phones

Detecting traffic congestions using cell phone accelerometers.

Using Smart Phone Sensors to Detect Transportation Modes

Measuring Cell-Id Trajectory Similarity for Mobile Phone Route Classification

A Deep Learning Model for Transportation Mode Detection Based on Smartphone Sensing Data

Travel Mode Detection Method Based on Big Smartphone Global Positioning System Tracking Data

A Transport Mode Detection Framework Based on Mobile Phone Signaling Data Combined with Bus GPS Data

CellTrans

A hybrid method for intercity transport mode identification based on mobility features and sequential relations mined from cellular signaling data

Travel Mode Identification for Non-Uniform Passive Mobile Phone Data

Intercity Traffic Travel Mode Identification Method Based on Mobile Signalling Data

Feature Pyramid biLSTM: Using Smartphone Sensors for Transportation Mode Detection

Travel Modes Recognition Method Based on Mobile Phone Signaling Data

Graph Supported Mode Detection within Mobile Phone Data Trajectories

Smartphone detector examination for transportation mode identification utilizing imbalanced maximizing-area under the curve proximal support vector machine

Towards predictive crowd based transport infrastructure monitoring system

Transportation Mode Detection Using Learning Methods and Self-Contained Sensors: Review

Transportation Mode Detection Combining CNN and Vision Transformer with Sensors Recalibration Using Smartphone Built-In Sensors

Computer Simulation Study of Vehicle Type Classiﬁcation Using Machine Learning Techniques with Mobile Phone Location Data

Inferring transportation mode from smartphone sensors: Evaluating the potential of Wi-Fi and Bluetooth