SPC-Cloud: 應用智慧型手機探偵車於道路品質及交通資訊收集與分析的雲端系統

Abstract

智慧型手機所具有的多樣化的感測功能與通訊介面，是實現探偵車概念的最佳平台， 可作為感測資料的收集與通訊設備，用來開發道路資訊系統（Road Information System, RIS）與交通資訊系統（Traffic Information System, TIS）。為有效收集、處理與分析大 量的感測資料，我們將利用雲端技術來進行大規模的資料收集、處理與分析。 第一年，我們的目標在於建立通訊與資料收集的雲端系統雛型及發展行動感測的核心 技術。相關工作重點包含：了解各類感測元件的誤差來源以發展校正技術與智慧型手 機與車輛動態結合的探偵車系統座標轉換問題、學習智慧型手機感測元件進階使用技 術；開發互動式TMC 雲端系統及學習Hadoop 雲端資料分散處理技術。 第二年，我們的目標是建立靜態RIS 雲端系統。相關工作重點包含：事件基礎的震動 波模型尋找演算法、路況判別演算法、整合現有線上交通資料庫系統、處理具時空特 性的感測資料（如：通訊資料減量、資料庫設計）及靜態道路及交通資訊雲端系統。 第三年，我們的目標是建立動態的RIS 及TIS 雲端系統。相關工作重點包含：具時空 特性的雲端資料彙整及探勘技術（如：異常路面資料的分群技術、適性及適時的動態 交通資訊彙整）、Map-Reducer 資料處理技術、以時空索引的資訊查詢技術、路面品質 與其他即時路況查詢及行車時間預估。

Smartphones equipped with various sensing abilities and communication interface is the best platform to fulfill the concept of probe car. The smartphone-embedded probe car systems can be used for sensing data collection and communication, and developing Road Information System (RIS) and Traffic Information System (TIS). We apply cloud techniques to efficiently process, analyze and collect lots of sensing data. To achieve the above goals, we need to deal with problems introduced from smartphones and cloud techniques, including sensing problems, e.g., sensor error handles, dynamical smartphone installation and feature extraction; In addition, we need to design user interface for data query and event report; As for cloud techniques, we need to design databases to efficiently process temporal and spatial related data, and develop cloud data processing techniques. In the first year, we will focus on developing prototype system for communication and data collection, and core techniques for mobile sensing. Tasks include studying sources of sensor errors, develop calibration techniques and study advanced sensor utilization techniques, dynamic tri-axis rotation technique for dynamic smartphone installation, developing an interactive cloud-based Traffic Message Channel (TMC) system and studying Hadoop distributed cloud systems. In the second year, we will focus on developing static cloud-based RIS. Tasks include, develop event-based shockwave model discovery algorithms, developing road and traffic condition recognition algorithms, integrating current online databases, handling temporal-spatial related data, e.g., reduction of amount of communication data and database design, and developing static cloud-based traffic and road information systems. In the third year, we will focus on dynamic cloud-based RIS and TIS. Tasks include cloud-based data fusion and mining techniques for temporal-spatial related data, e.g., clustering algorithms for abnormal road conditions and adaptive dynamic traffic data fusion, map-reducer techniques, query techniques for temporal-spatial related and traffic time estimation.