Mobile Phone Based Context-Aware Traffic State Estimation Tran Minh Quang Graduate School of Engineering and Science Shibaura Institute of Technology A thesis submitted for the degree of Doctor of Engineering (Dr.) 2012 September 2 Acknowledgements I feel tremendously lucky to have the opportunity to work with Prof. Eiji Kamioka on the ideas in this dissertation. Kamioka instilled in me a love for researching on mobile multimedia communications and ubiquitous computing, agreed to take me on as a graduate student, and encouraged me to immerse myself in something I had a passion for. He is always beside me, understands my difficulties and works side by side with me in finding the right direction.
He incubated me in research method, scientific writing skill, etc. He even took me in his car to collect real-time traffic data for evaluations. I have never met a professor more generous with his time and experience. I am grateful to Prof.
Shuji Kubota at Shibaura Institute of Tech- nology (SIT), Prof. Shigeki Yamada at National Institute of Infor- matics (NII), Prof. Hiroaki Morino and Prof. Takumi Miyoshi at SIT for serving on my oral committee.
Kubota also serves as a co-supervisor who has inspired me to new ideas proposed in this dissertation. Yamada has advised me precious experiences on research approaches and gave several useful comments on the effec- tiveness of the proposed approaches when I visited and presented at NII. Morino has provided me both ideas and fruitful comments on my path of doing this research. His vigorous discussions always inspire me to something new.
I would like to thank Prof. Takashi Watanabe at SIT, Mr. Hiroyuki Ishizaki at International Center for Social Entrepreneurship (ICSE), Mr. Takashi Yoshida at PLANS, Ltd., who gave me constructive comments related to the feasibility of the mobile phone based traffic state model proposed in this thesis.
I wish to thanks all staff and faculties at SIT for their kindly taking care of me for not only on study but also on daily life related issues when I study in Japan. I still remember Ms. Reiko Kageyama on the first day when she picked me up at Narita Airport. Yukiko Sonoi has spent a lot of time to help me find suitable apartment.
Midori Yabe has supported me on any daily difficulty, from teaching me Japanese, interpreting documents issued by the city hall, to ex- plaining the worries coming from the Great Tohoku Earthquake to me. I am grateful to Ms. Yukari Moriuchi, Ms., at the Graduation School Section who always kindly support and ad- vise the procedures related to my research duty at SIT. I would like to thank Prof.
Ayao Tsuge, Prof. Masato Murakami, Prof. Chiaki Nakayama for their great Human Resource Fostering Program which positively influences the way I do this research. I also wish to thank SIT’s Hybrid Twining Program (HBT), Japanese Government (Mon- bukagakusho: MEXT) scholarship, and Dept.
of Computer Science and Engineering (HCM Uni. Vietnam) for their supports when I study in Japan for my Ph. I would like to acknowledge the fine work of the other individuals who have contributed to this research. Pawel Gora (from Univ.
of Warsaw, Poland) provided traffic simulation software, namely the traffic simulation framework (TSF), to generate traffic synthetic data. He also generously donated his time and expertise in customizing the TSF for special requirements on data generation in this research. Issei Yajima at SIT donated his time taking me on his motorbike to collect surrounding noise data on streets which is used for evaluating the proposed vehicle classification model. I would like to thank all my wonderful friends and colleagues at Mobile Multimedia Communications Laboratory.
Hiroki Murata, Mr. Keizaburo Nishina, Mr. Kaoru Koshimizu, Mr. Shigeki Nakamura, Mr.
Tomoki Takemura, Mr. Yuji Yoshimura have spent a lot of time instructing me on the campus life. We have also had enjoyable time in Okinawa when participating in a scientific conference there. I would especially like to thank my collaborators, Mr.
Muhammad Ariff Buharudin and Mrs. Effiyana Binti Ghazali Nurzal. Finally, I would like to thank my family, my parents, my wife, and my little daughter for their love and support. My parents give me the life and have provided me with steady guidance, knowledge, and encour- agement.
My wife, Nguyen Thi Hang, has made countless sacrifices to be always with me giving me her great support and encouragement. My little daughter, Tran Thi Minh Tam (Chip-chan) always gives me happy laughs which ultimately helps me to forget the work whenever it was at a standstill. My parents, my wife and my little daughter are levers for me to accomplish the thesis. This dissertation is dedicated to them.
Abstract Traffic congestion causes several social issues such as economic loss, air and noise pollution, quality of life degradation, and so forth. Traffic state estimation (TES) is one of the most important fields in intelli- gent transportation systems (ITS) aiming at alleviating traffic conges- tion. Conventional TES systems which utilize road-side fixed sensors such as loop detectors, RFID readers, video cameras, etc., confront essential issues of coverage limitation, real time effect and invest- ment/maintenance cost, since it is impractical to install a huge num- ber of sensors at every street. In the recent years, with the advances of mobile phone technologies, mobile devices such as mobile phones, PDA, etc., are utilized as traffic probes to overcome the aforemen- tioned road-side fixed sensor systems’ disadvantages.
However, mobile phone based traffic state estimation (M-TES) approach introduces emerging challenges which relate to (1) Processing lack informative data reported by “on-the-shelf” mobile phones; (2) Managing limited resources such as low computational capacity and narrow bandwidth; and (3) Solving difficulties rooted from low and uncertain penetration rate. This dissertation proposes a novel mobile phone based context- aware traffic state estimation (MC-TES) framework which consists of notable solutions for difficulties mentioned about. Among achieved results, most important ones include: (1) Inventing a notable traffic state quantification (TSQ) model by which traffic state is granularly quantified; (2) Proposing a novel data collection policy based on the “3R” philosophy (the Right data is collected at the Right time by the Right devices) to control data transmission load and min- imize data redundancy; (3) Designing a unique vehicle classification method, using only the data reported by mobile phones, to improve the effectiveness of the proposed TSQ model, especially when the model is applied in developing countries where several vehicle types participate in road networks; (4) Investigating thoroughly the effect of low penetration rate on traffic state estimation to figure out the so-called “acceptable” penetration rate for a predefined “expected” accuracy; (5) Inventing two novel velocity-density inference circuits (VDICs), namely the adaptive and the adaptive feedback VIDCs, to improve the traffic state estimation effectiveness even if the pene- tration rate is low; (6) Proposing a notable genetic algorithm (GA) based velocity-density estimation (GA-VDEM) model to optimize the previously proposed VDICs; (7) Introducing a practical artificial neu- ral network (ANN) based prediction model to assure the traffic state estimation accuracy even when the penetration rate becomes unac- ceptably low, namely just several percent or even zero; (8) Discussing and proposing suitable context-aware approaches to the difficulties of “less informative” data remaining in each issue mentioned above. Each of the proposed solutions mentioned above was evaluated thor- oughly using both the real field experimental data and numerous sim- ulated data.
Evaluation results reveal the effectiveness, the robustness as well as the scalability of the proposed solutions. For example, the proposed TSQ accurately provides quantitative traffic state of even irregular traffic flows where velocity and density do not relate to each other in a common way. The GA-VDEM provides traffic state esti- mation as high accuracy as around 90% when the penetration rate is low but still relevant, namely larger than 18%. The ANN based prediction approach ensures the accuracy as around 73% even if the penetration rate is unacceptably low, namely just several percent or even zero.
Consequently, this thesis opens a new research direction on the mobile probe based traffic state estimation, accelerating the M-TES, or more general the mobile phone based ITS (M-ITS), into realization. Contents Contents vi List of Figures ix 1 Introduction 1 1.2 Existing Technologies and Challenges .3 Mobile Phone Based Context-Aware Traffic State Estimation (MC- TES) .4 Contributions and Thesis Organization. 14 2 Mobile Phone Based Traffic State Quantification 19 2.2 Traffic State Indicators .3 Traffic State Quantification. 31 3 Controlling Communication Cost and Ensuring Estimation Ac- curacy 34 3.2 The “3R” Data Collection Policy .4 Traffic State Quantification with Different Participating Vehicle Modes .1 Effecticveness of the “3R” Data Collection Policy .2 Effectiveness of the Vehicle Classification Method.
58 4 Adaptive Approaches to Low Penetration Rate Issues 60 4.2 Influence of Penetration Rate on Traffic State Estimation Accuracy 63 4.3 Velocity-Density Inference Circuits (VDICs) .1 Conventional Velocity-Density Inference Models .2 The Adaptive Velocity-Density Inference Circuit .3 The Adaptive Feedback Velocity-Density Inference Circuit 75 4.2 Effect of Penetration Rate on the Estimation Accuracy .3 Effectiveness of the Velocity-Density Inference Circuits. 85 5 Synergistic Approaches to Optimize the Velocity-Density Infer- ence Model 87 5.2 The Modified VDEM .3 Intelligent Context-Ware Approach to Velocity-Density Estimation 90 5.1 Overview of the Genetic Algorithm .2 GA-based Velocity-Density Inference Mechanism .3 GA-Based Velocity-Density Estimation Model (GA-VDEM) 95 5.4 Traffic State Prediction Under Unacceptably Low Penetration Rate 98 5.1 The Proposed ANN-based Prediction Model .2 Narrowing the Computational Space .1 Effectiveness of the GA-VDEM .1 The Detailed Optimization Capacity .2 The Overall Effectiveness of the Proposed GA- VDEM .2 Effectiveness of the ANN-based Prediction Model .3 Effect of Related Road Segments on Prediction Accuracy .1 Traffic State Quantification Model .2 Solutions for Low and Uncertain Penetration Rate Issues .3 Implementation Difficulties in MC-TES .5 Potential Business Model. 120 7 Conclusion and Future Work 123 7. 125 8 Summary 128 Glossary 131 References 132 viii List of Figures 1.1 The architecture of a context-aware system .2 The overall architecture of the MC-TES .3 Organization of the thesis .1 Road segmentation: Principle and practice .2 Capacity, C k,i = Qi0 + Qk,i , of a road segment i with 2 lanes .3 The traffic-state quadrant space .4 Better qualifying traffic states based on the Goodness metric .5 Quantifying traffic states based on the Goodness metric .6 A visual view of identifying traffic state level based on the Good- ness metric .1 A vehicle reports its information based on its velocity change rate 37 3.2 The boundary of a walker mobile phone at point P0 .3 Data uploading decision in a walker mobile phone .4 Vehicle classification model .5 Vehicle classification at the server side based on GPS data .6 Vehicle classification based on acceleration patterns .7 An ANN model for vehicle classification using relative acceleration and surrounding noise patterns .8 Two connected biological neurons (Source: “Neural Network De- sign”, T.
Martin, at el.9 Modeling an artificial neuron .11 A Feed-forward, multilayer neural network. 48 ix LIST OF FIGURES 3.12 The velocity change calculating and data upload timing .13 AAThe error rate of pedestrian estimation .14 Data transmission load under the “3R” data collection policy .15 Vehicle classification result using an ANN with the combination of acceleration and surrounding noise patterns as contextual input data 58 4.1 Vehicles that report data are denoted as the car-shape ones.2 Effect of penetration rate on velocity and density estimations .3 The adaptive Velocity-Density inference circuit .4 Relation between the meanspeed capacity and the velocity estima- tion error (both are estimated from sensed data) .5 The adaptive Feedback Velocity-Density inference circuit .6 The road segmentation in TSF .7 The relation between the penetration rate and the everage velocity and density estimation errors .8 Comparing the effect of penetration rate on velocity estimation error with the Herring work .9 Effectiveness of the velocity-density inference circuits in estimating the average velocity .10 Effectiveness of the velocity-density inference circuits in estimating the density .1 The flowchart of a GA’s process .2 The pseudo code of the Genetic Algorithm .