[Paper Review] Transport-domain applications of widely used data sources in the smart transportation: A survey

Sina Dabiri, Kevin Heaslip|arXiv (Cornell University)|Mar 29, 2018

Traffic Prediction and Management TechniquesEngineering18 citations

TL;DR

This survey systematically reviews transport-domain applications of six widely used data sources—traffic sensors, video processors, GPS/probe vehicles, mobile networks, social media, smart cards, and environmental data—in smart transportation systems. It classifies existing applications, proposes data fusion architectures to enhance mobility understanding, and identifies key challenges and future research directions for intelligent, cost-effective traffic management.

ABSTRACT

The rapid growth of population and the permanent increase in the number of vehicles engender several issues in transportation systems, which in turn call for an intelligent and cost-effective approach to resolve the problems in an efficient manner. Smart transportation is a framework that leverages the power of Information and Communication Technology for acquisition, management, and mining of traffic-related data sources, which, in this study, are categorized into: 1) traffic flow sensors, 2) video image processors, 3) probe people and vehicles based on Global Positioning Systems (GPS), mobile phone cellular networks, and Bluetooth, 4) location-based social networks, 5) transit data with the focus on smart cards, and 6) environmental data. For each data source, first, the operational mechanism of the technology for capturing the data is succinctly demonstrated. Secondly, as the most salient feature of this study, the transport-domain applications of each data source that have been conducted by the previous studies are reviewed and classified into the main groups. Thirdly, a number of possible future research directions are provided for all types of data sources. Moreover, in order to alleviate the shortcomings pertaining to each single data source and acquire a better understanding of mobility behavior in transportation systems, the data fusion architectures are introduced to fuse the knowledge learned from a set of heterogeneous but complementary data sources. Finally, we briefly mention the current challenges and their corresponding solutions in the smart transportation.

Motivation & Objective

• To identify and categorize the transport-domain applications of widely used data sources in smart transportation systems.
• To analyze the operational mechanisms and limitations of each data source, including traffic sensors, GPS probes, mobile networks, and social media.
• To propose data fusion architectures that combine heterogeneous data sources to improve mobility behavior modeling and transportation decision-making.
• To highlight current challenges in data quality, privacy, and integration, and suggest future research directions for scalable and intelligent transportation solutions.
• To provide a comprehensive, evidence-based review of existing studies to guide researchers and practitioners in leveraging data for smarter mobility.

Proposed method

• Categorizes six major data sources: traffic flow sensors, video image processors, GPS/mobile probe data, location-based social networks, transit smart card data, and environmental sensors.
• Reviews and classifies prior studies based on application domains such as traffic state estimation, incident detection, route planning, and transit optimization.
• Proposes data fusion frameworks that integrate complementary data sources to overcome individual data limitations and improve accuracy in mobility modeling.
• Analyzes technical and privacy challenges in data collection and integration, including data sparsity, latency, and anonymization.
• Uses a systematic literature review approach to synthesize findings from 52 pages of research, supported by 10 figures and 153 references.
• Identifies future research directions, such as real-time data processing, AI-driven analytics, and privacy-preserving data sharing mechanisms.

Experimental results

Research questions

• RQ1 What are the primary transport-domain applications of widely used data sources in smart transportation systems?
• RQ2 How do different data sources (e.g., GPS, mobile networks, smart cards) contribute uniquely to understanding urban mobility?
• RQ3 What are the key limitations of individual data sources in capturing comprehensive traffic behavior?
• RQ4 How can data fusion architectures effectively combine heterogeneous data sources to improve transportation analytics?
• RQ5 What are the major challenges in deploying data-driven smart transportation systems, and what future research is needed to address them?

Key findings

• GPS-based probe data and mobile network signals are widely used for real-time traffic speed and volume estimation, with accuracy improving when fused with other data sources.
• Smart card data from transit systems enables detailed analysis of public transport demand patterns and transit network efficiency.
• Video image processors and traffic sensors provide high-resolution spatial-temporal data but are limited by high deployment and maintenance costs.
• Location-based social networks offer insights into human mobility and event-driven traffic changes, though with lower spatial and temporal precision.
• Data fusion architectures significantly enhance the reliability and completeness of mobility modeling by mitigating individual data source biases and gaps.
• Despite advances, challenges remain in data privacy, real-time processing, and interoperability across heterogeneous data systems, requiring further research in secure and scalable integration frameworks.