Section: New Results

Embedded Data Management

Participants : Nicolas Anciaux, Saliha Lallali, Philippe Pucheral, Iulian Sandu Popa [correspondent] .

Embedded keyword indexing: In this work, we revisit the traditional problem of information retrieval queries over large collections of files in an embedded context. A file can be any form of document, picture or data stream, associated with a set of terms. A query can be any form of keyword search using a ranking function (e.g., TF-IDF) identifying the top-k most relevant files. The proposed search engine can be used in sensors to search for relevant objects in their surroundings, in cameras to search pictures by using tags, in personal smart dongles to secure the querying of documents and files hosted in an untrusted Cloud, or in a personal cloud securely managed using a tamper resistant smart object. A search engine is usually based on a (large) inverted index and queries are traditionally evaluated by allocating one container in RAM per document to aggregate its score, making the RAM consumption linear with the size of the document corpus. To tackle this issue, we designed a new form of inverted index which can be accessed in a pure pipeline manner to evaluate search queries without materializing any intermediate result. Successive index partitions are written once in Flash and maintained in the background by timely triggering merge operations while files are inserted or deleted from the index. By combining this new index and the corresponding evaluation techniques, our embedded search engine is capable of reconciling high insert/delete/update rate and query scalability. We have demonstrated the search engine on a secure USB token in the context of a personal cloud, and have conducted in depth performance evaluations on a development board representative for different smart objects characteristics. The experimental results demonstrate the scalability of the approach and its superiority compared to state of the art methods. This work was published at VLDB’15 [21] and demonstrated at SIGMOD’15 [24] . It constitutes the main contribution of the PhD thesis of Saliha Lallali

Spatio-temporal indexing in Flash storage: The convergence of mobile computing, wireless communications and sensors has raised the development of many applications exploiting massive flows of spatio-temporal data such as in location-based services, participatory sensing, or traffic management [15] . Spatio-temporal data indexing is among the most active research topics in this area. Nevertheless, since a few years a new fundamental parameter has made its entry on the database scene: the NAND flash storage. The peculiar characteristics of flash memory require redesigning the existing data storage and indexing techniques that were devised for magnetic hard-disks. TRIFL, proposed in [16] is an efficient and generic TRajectory Index for FLash, designed around the key requirements of both trajectory indexing and flash storage. TRIFL is generic in the sense that it is efficient for both simple flash storage devices such as the SD cards and more powerful devices such as the solid state drives. In addition, TRIFL includes an online self tuning algorithm that allows adapting the index structure to the workload and the technical specifications of the flash storage device to maximize the index performance. Moreover, TRIFL achieves good performance with relatively low memory requirements, making it appropriate for many application scenarios. The experimental evaluation shows that TRIFL outperforms the representative indexing methods on flash disks but also on magnetic disks. This work [15] [16] is part of Dai Hai Ton That’s Ph.D. thesis, co-supervised by Iulian Sandu Popa.