URBAN DATA LAB

Prediction for Intelligent Transportation Systems

In recent years we have seen an increase in attention to intelligent transportation systems (ITS) around the globe. It is the technological pillar supporting the operations of sustainable smart cities and improving their accessibility and resilience. In this regard, 5G technology is on the way to revolutionize how we exploit different data sources in our daily activities. It aims to support ubiquitous network connections with billions of internet-of-things (IoT) devices, and improved user experience with extremely low latency.

In particular, by exploiting the recent advances in data science and leading-edge communications technologies which include 5G networks, wireless sensor networks (WSN), IoT, and intelligent control systems, an ITS can be developed. This can help in generating traffic prediction models that enable researchers and decision-makers to propose solutions to enhance the sustainability of transportation networks.

ITS can solve real-life problems

Globally, an increasing number of cities are interested in designing ITS because they can solve various municipal issues such as:

  1. ITS are in line with the United Nations Sustainable Development Goals of having sustainable transportation systems that have a central role in supporting sustainable development, enhancing economic growth, and improving infrastructure, environment, and human settlements.
  2. ITS can improve transportation service efficiency, safety, and drivers’ experience, reduce traffic congestion and accident rates, and help in the planning of roads and new cities.
  3. ITS play a significant role in the achievement of the Paris Agreement of controlling global greenhouse gas (GHG), given the fact that close to a quarter of the global GHG emissions come from vehicles and other transportation means.

 

Data sensing in ITS

The availability of data has increased dramatically because of the current revolutions in data sensing technologies with unprecedented levels of accuracy, reliability, and heterogeneity.  ITS data sources can be classified in the following categories:

  1. Roadside sensors: Deploying sensors on road sides is the most commonly used data source in ITS (i.e. Radar, LIDAR, computer-vision). However, the accuracy of the acquired data depends on the position of roadside sensors, their range and the geographical service area. It also may be affected by weather conditions such as snow, rain and fog.
  2. In-vehicle sensors: Recent commercial vehicles usually include position detection sensors that record the location and monitor other vehicle’s metrics at all times.
  3. Cooperative sensors: Examples include WSN, social media and crowd-sourced sensing.
  4. Static data sources: We can understand traffic flow in a certain area from public transit timetable and locations of bus stations. Data provided from these sources has a fixed format and static structure that makes it easier to be merged into the data modeling process.
  5. External data sources: There are many data sources that are not related to transportation system directly, but have a great impact on it such as weather conditions, social and sport events, holidays, etc.

 

The role of 5G in ITS

One of challenges that face the implementation of ITS is the need of reliable, high-speed, low latency and efficient data transmission technologies. By exploiting the advantages of 5G networks with up to 10 Gbps data rates, latency of less than 1 msec, and the ability to serve an increasing number of connected devices, the implementation of real-time data-driven ITS is practical nowadays. Massive flow of information about the transportation network can be realized for sharing data among vehicles, drivers, and data sensing devices through 5G networks by supporting the following ‘modes’ of connectivity:

  1. Vehicle-to-Network (V2N): Connects vehicles to the mobile network to support services like streaming media for entertainment and connectivity for dynamic route management, etc.
  2. Vehicle-to-Vehicle (V2V): Connects vehicles with each other for early warnings (e.g. an upcoming emergency) that increases the range of data sensing over a certain road.
  3. Vehicle-to-Infrastructure (V2I): Directly connects vehicles to roadside infrastructure like traffic lights which in turn can be connected to the wider mobile network.
  4. Vehicle-to-Person (V2P): Directly connects vehicles to pedestrians equipped with compatible mobile devices to issue alerts about potential dangers nearby.

In addition, 5G introduces new technologies (e.g. network visualization, cloud computing, new signal processing techniques) that can meet the needs of recent developments in ITS. This will help ITS keep up with the current revolution in data sensing, networking, processing, and controlling.

 

Traffic Prediction Models

Traffic prediction models are commonly used in ITS. We can exploit the huge amount of acquired data in designing efficient traffic prediction models. They can be classified into categories based on the used techniques as follows:

  1. Statistical Methods: The operation of most ITS was exclusively based on traditional statistics, Kalman filters, econometric methods, time-series models, auto-regressive models, and Bayesian regression.
  2. Nonlinear Methods: By exploiting the nonlinear characteristics of traffic flow time-series, nonlinear theories can be used to generate traffic flow prediction models such as mutation theory, wavelet theory, and chaos theory.
  3. Simulation based Methods: The transportation network, road infrastructure (i.e. road capacity), and other important metrics of vehicles (e.g., speed, classification, occupation) are described using algorithms and other relevant models such as M3 distribution model.
  4. Artificial Intelligence Methods: The fast expansion in machine learning and all other techniques in the Artificial Intelligence taxonomy provides new data processing methods that can predict the traffic flow more accurately and quickly, such as deep learning and reinforcement learning neural networks.
  5. Combined methods: In order to exploit the full advantages of different prediction methods, two or more methods can be used at the same time to generate a traffic flow prediction model.

 

Thus, with the help of emerging 5G and data science technologies, the efficiency of ITS systems can be improved, which in turn enhances the sustainability of transportation systems besides reducing traffic congestions, accident rates, and air pollution.

About the Authors
Dr. Anas Chaaban (Assistant Professor at the Okanagan School of Engineering, University of British Columbia) and team members Zhenyu (Charlus) Zhang and Mahmoud Hasabelnaby, are investigating traffic prediction methods for use in ITS, in addition to possible integration of such prediction methods in traffic-management and air-quality monitoring frameworks.

Scroll to top