Translated title of the contribution: I-Tree: A Spatial Time-series Indexing Mechanism for Supporting Integrated Retrieval of Sensing Data

木實 新一, 石塚 宏紀, 岩井 将行, 宮崎 純, 瀬崎 薫, 戸辺 義人

Research output: Contribution to journalArticle


物理空間の各所に設置されたセンサのデータを統合的に検索し,適切なタイミングでユーザに情報やサービスを提供するためには,空間時系列データを高速に処理する必要がある.本論文では,大量の空間時系列データの高速な検索を可能にする空間時系列索引機構 I-Tree の構造と検索アルゴリズムについて述べ,人工的に生成した大量の人数計測データと市街地に設置した微気象センサのデータを用いた評価実験の結果を示す.I-Tree を用いれば,空間時系列問合せに対してユークリッド距離の近い系列を,従来手法よりも安定して高速に検索することができる.Fast and efficient retrieval of spatial time series data is indispensable for applications that must provide users with timely and relevant information and services using complex sensor data in the real world. This paper introduces an indexing mechanism called I-Tree, which allows for fast retrieval of a massive amount of spatial time-series data, and describes its structure and the search algorithms. We also discuss the results of our performance evaluation using a large amount of artificially-generated spatial time-series data as well as temperature data from micro-weather sensors deployed in a city. I-Tree allows for fast, Euclidean distance-based similarity search of spatial time-series under various conditions.
Original languageJapanese
Pages (from-to)26-39
Number of pages14
Issue number1
Publication statusPublished - Mar 25 2011

Cite this