Advanced Intelligent Predictive Models for Urban Transportation

R. Sathiyaraj, A Bharathi, Balamurugan Balusamy

• Publisher: Chapman and Hall/CRC
• Year: 2022
• ISBN:
  9781000555981
• Edition: 1
• Topic: Computer Science
• Subtopic: Cloud Computing
• Index: Computer Science

The book emphasizes the predictive models of Big Data, Genetic Algorithm, and IoT with a case study. The book illustrates the predictive models with integrated fuel consumption models for smart and safe traveling. The text is a coordinated amalgamation of research contributions and industrial applications in the field of Intelligent Transportation Systems. The advanced predictive models and research results were achieved with the case studies, deployed in real transportation environments.

Features:

• 
  
  

• Provides a smart traffic congestion avoidance system with an integrated fuel consumption model.

• 
  
  

• Predicts traffic in short-term and regular. This is illustrated with a case study.

• 
  
  

• Efficient Traffic light controller and deviation system in accordance with the traffic scenario.

• 
  
  

• IoT based Intelligent Transport Systems in a Global perspective.

• 
  
  

• Intelligent Traffic Light Control System and Ambulance Control System.

• 
  
  

• Provides a predictive framework that can handle the traffic on abnormal days, such as weekends, festival holidays.

• 
  
  

• Bunch of solutions and ideas for smart traffic development in smart cities.

• 
  
  

• This book focuses on advanced predictive models along with offering an efficient solution for smart traffic management system.

• 
  
  

• This book will give a brief idea of the available algorithms/techniques of big data, IoT, and genetic algorithm and guides in developing a solution for smart city applications.

• 
  
  

• This book will be a complete framework for ITS domain with the advanced concepts of Big Data Analytics, Genetic Algorithm and IoT.

This book is primarily aimed at IT professionals. Undergraduates, graduates and researchers in the area of computer science and information technology will also find this book useful.

1 Overview

DOI: 10.1201/9781003217367-1

1.1 Introduction

With the rapid development of urban areas, the volume of vehicles has greatly increased, leading to issues such as collisions, traffic congestion, economic losses, environmental pollution, and excessive fuel waste. Among these issues, road traffic jams represent a significant problem related to the field of urban transportation. Intelligent transportation systems (ITS) is an interdisciplinary field that uses data analytics from different mathematical models, and is also seen as an important technology for alleviating congestion in urban traffic. Accurate traffic forecasting and traffic light regulation are important steps in the development of an ITS and are essential for transport system efficiency.

An efficient traffic management system is needed to forecast and control traffic flows in urban areas. Prediction of traffic helps to avoid traffic congestion before it develops. Typically, urban traffic forecasting uses historical and current traffic flow data to predict future road conditions (Niu et al. 2015). With the development of smart phone technology, sensors are widely used to analyze traffic conditions. For managing and forecasting traffic congestion, machine learning algorithms and Big Data analytics techniques are used. Big Data analytics plays an essential role in the intelligent traffic management system reach. Data analytics helps us to predict traffic congestion, and its occurrence can be avoided.

This book recommends a smart framework for the domain of transportation that performs traffic prediction with a fuel consumption model and analyzes traffic flow congestion using a genetic and regression model. Based on a multi-agent system, it will control traffic lights and deviate traffic routes.

1.2 Towards Intelligent Traffic Flow Prediction

Road traffic congestion is a persistent problem worldwide. With the huge growth of the population, the number of vehicles is increasing at a rapid rate. India is the second largest country in terms of growth in population and economy. Most cities in India are facing road congestion problems. There are practical difficulties in maintaining intelligent transport management systems (ITMS) in metropolitan cities in India. This is due to the slow growth in infrastructure compared to the rapid increase in the number of vehicles, as well as space and cost constraints.

Traffic flow information is needed to help travelers to make better travel decisions when it comes to congestion and to improve traffic operation efficiency. Predicting short-term traffic flow will be more helpful in managing freeway networks. This traffic flow prediction makes use of real-time data to predict traffic status in the subsequent 5–20 minutes. Every country in the world is striving to enhance their traffic management systems to make them more efficient. Researchers have used different methods to predict freeway traffic in urban areas.

1.3 Broad Factors Impacting Traffic Flow

Traffic congestion on the road can be defined as the condition in which the number of vehicles in the lane is higher than the lane capacity. Traffic congestion may occur due to various reasons. The primary reason may vary depending on the location. It occurs when the demand exceeds road capacity. Reasons for traffic congestion include an increased number of vehicles in the lane, improper parking, road maintenance work, accidents, etc.

Traffic congestion has a wide range of consequences, including squandering time for users and causing delays in reaching their destinations, increased fuel consumption, pollution, and a higher risk of collisions, among others. This is a serious issue which needs to be dealt with. Solutions for the problem need to be developed. Some possible solutions are parking restrictions, changes in school timings to reduce rush hours, traffic counters, better traffic management, speed limit enforcement, lane splitting, provision of flyovers, construction of metro systems, public education programs, etc.

To reduce traffic congestion, a novel traffic flow management model is needed. In this book, the focus is to develop a model which can eliminate traffic congestion, thereby resulting in a uniform flow of traffic. The causes and effects of traffic congestion and the most appropriate solutions to the problem vary according to the location.

1.4 Prediction Techniques on Traffic Flow

With the advent of information and communication technologies, many traffic forecasting models have been developed to help traffic management and control. This section discusses work related to traffic analysis, prediction, congestion, and traffic light control.

To predict traffic flow, a novel approach based on long short-term memory (LSTM) (Ma et al. 2015) has been suggested. Lin et al. (2017) propose a novel fuzzy deep-learning approach called FDCN to predict the citywide flow of traffic. This method is based on the theory of fuzzy logic and the model of the deep residual network. Do et al. (2019) propose a deep learning–based traffic flow predictor with spatial and temporal attentions (STANN).

To establish the spatial dependencies between road segments and temporal dependencies between time steps, spatial and temporal attention is deployed (Hou et al. 2019) – an adaptive hybrid model to predict the short-term flow of traffic. To predict traffic flow, the linear autoregressive integrated moving average (ARIMA) and non-linear wavelet neural network (WNN) method was used. The outputs of the two individual models were then evaluated and combined by fuzzy logic, and the weighted result was regarded as the final predicted traffic volume of the hybrid model.

Atta et al. (2018) proposed an intelligent system which maintains the active timings of traffic signals by sensing traffic density and reducing overcrowding with IoT (Internet of Things) sensors, which offer powerful and advanced communication technology. A new model by Zaki et al. (2019), based on hidden Markov model and contrast, defines traffic states during peak hours in two-dimensional space (2D). This model uses mean speed and contrasts to capture the variability in traffic patterns.

1.5 Generic Traffic Flow Prediction Models and Measurements

Existing research on traffic flow forecasting primarily focused on reducing traffic congestion and providing uninterrupted traffic flow.A large number of research studies are focused on traffic forecasting, with the majority of models focusing on short-term traffic forecasting. Approaches are normally categorized as either parametric or nonparametric (Van Lint et al. 2002) (Vlahogianni & Karlaftis 2013).

These approaches mainly focus on the structure of the model and real-time traffic scenarios. To improve the accuracy of prediction, ARIMA (Williams & Hoel 2003) and STARIMA approaches (Pfeifer & Deutsch 1980) were proposed.

Several data mining and machine learning methods are employed in predicting traffic. These approaches are the classification trees model (CTM) (Kim et al. 2008) (Zhan et al. 2011), decision trees (DT), artificial neural networks (ANN) (Wei & Lee 2007) (Vlahogianni & Karlaftis 2013), genetic algorithms (GA) (Valenti et al. 2010), and support/relevance vector machines (SVM/RVM) (Luo et al. 2019).

Intelligent transportation systems (ITS) are indispensable in proposing solutions to these types of problems. However, the abrupt volume of vehicles involved in urban transportation necessitates the use of Big Data analytics to process information and utilize traffic systems in an optimized way (Vlahogianni et al.2008). Regression techniques are proposed to model vehicle arrivals based on real-time data (Lippi et al. 2013). Accurate traffic prediction (Liu et al. 2016) is proposed as a traffic estimation method which makes use of road network correlation and sparse traffic sampling.

Time-series is the major factor in almost any of the traffic flow prediction models. A local linear regression model was used in short-term traffic forecasting (Sun et al. 2004). In the same traffic flow forecasting, a Bayesian network approach was employed (Sun, et al. 2006) to predict short-term traffic flow, and a support vector–based online learning weighted support vector regression approach was proposed (Jeong et al. 2013). Many researchers have focused on providing an efficient solution for traffic management and control. The focus was turned on machine learning techniques, which provided a path for academics and industries to focus their attention on smart intelligent transport management systems (SITMS).

A prediction methodology based on deep learning was proposed (Lv et al. 2014) for analyzing the system. A stacked auto encoder model was employed in a greedy layer-wise fashion. An approach for forecasting travel speed for multi-step-ahead based on 2-min travel speed data using fuzzy neural networks was proposed (Tang et al. 2017).

Chiu and Chand (1993) have proposed an approach adapting the fixed type controller approach. These have been programmed to function at preset timings for each series. This solution was used for various types of traffic scenarios and especially for undisciplined traffic. Several research techniques have attempted to avoid using a fixed cycle controller system, but ITS still requires a cost-effective method for reducing traffic congestion. Addressing a single issue may not be an efficient solution; several parameters have to be considered to avoid traffic congestion in advance. Vehicle speed and length are considered the two most important factors in any traffic management system.

VANET is a technology that considers vehicles as nodes in a network and creates a mobile network for data exchange and communication, whereby each vehicle will participate in exchanging information related to traffic conditions. VANETs provide better techniques to collect real-time traffic related information in a cost effective manner and traffic information delivery (Hartenstein & Laberteaux 2009). V2V and V2R communications in VANETs help to gather traffic updates from vehicles and roadside units.

To gather real-time traffic related data, many techniques rely on loop and mobile detectors. Because cellular phones are not dedicated to traffic data swap, this type of detection is costly and results in huge amounts of traffic information. The cost of implementing the loop detector systems is very high (Wang et al. 2014). Delays in communication and the degradation of safety result in an incompetent system. Timing and security issues in ITS with VANETs are addressed (Zheng et al. 2017).

An efficient traffic light system is proposed by Shandiz et al. (2009) which uses a genetic algorithm to evaluate the stochastic data, thereby arriving at an optimized transport system. The data is compiled, each traffic signal is coded, and then the data is analysed collectively for each path to determine the ideal condition. The imitation is completed using the length of the route and average vehicle speed.

The research helps to determine the conditions in which the maximum number of vehicles can go through the traffic lights. The concurrent examination of a lot of objects with a number of statistical measures is provided through a multivariate study (Hair, William, Barry & Rolphe 2010). Using this method, a concurrent study is carried out on the above two variables. Different multivariate study techniques have been proposed that could be applied as per the requirement.

Petracca et al. (2013) propose a prototype for a vehicle which is capable of interacting with other roadside vehicles and also with an internal electronic device. The model also specifies the different mechanisms in the on-board element. This offers a variety of applications that could be used to develop well-organized operations.

This study (Hair, William, Barry & Rolphe 2010) (Hair et al. 1998) gives an overview of the various requirements for designing an efficient ITS system. Simulations are completed using MATLAB to establish the accuracy of the proposed scheme based on real-world circumstances. The observations show that the proposed environment helps in identifying vehicle positions in different environments.

The algorithm is examined in terms of running time and result accuracy. (De la Garza et al.2013) propose novel schemes which integ...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Contents
• Preface
• Authors
• 1 Overview
• 2 Related Works
• 3 Smart Traffic Prediction and Congestion Avoidance System (S-TPCA) Using Genetic Predictive Models for Urban Transportation
• 4 Short-Term Traffic Prediction Model (STTPM)
• 5 An Efficient Intelligent Traffic Light Control and Deviation System
• 6 IoT-Based Intelligent Transportation System (IoT-ITS)
• 7 Intelligent Traffic Light Control and Ambulance Control System
• 8 Conclusions and Future Research
• Bibliography
• Index