由於現今嵌入式系統及無線感測技術被廣泛的應用在各式各樣的領域當中，因此對於CPS (cyber-physical systems)的需求也逐漸被重視，CPS為一個結合感測器與演算法的控制系統，廣泛被應用在無論是在交通、航空、化學、醫療領域的數據偵測與資料監控上。
而有鑑於CPS系統需要負荷龐大的歷史資料量，此篇論文提出一個新穎的多版本控管索引架構，針對新形態非揮發性記憶體資料存取較為快速的特性來做設計。而先前的論文所提出的Multiversion B-tree (MVBT) [1] 雖已成功建立一個針對嵌入式多版本資料庫系統的索引結構，且MVBT架構維持容易，然而MVBT由於當資料在索引目錄中做更動時，常會導致節點的分裂(split)或結合(merge)動作，以至於索引目錄需要占用龐大的記憶體空間，而對於非揮發性記憶體來說，空間是一個需要被節省的有限資源，因此，之後有論文提出針對空間利用率較高的多版本控管索引架構SEMI (Space-Efficient Multiversion Index, SEMI) [2]，而SEMI利用PCM byte-addressable特性，使資料範圍查詢(key-range query)的效率提高。儘管SEMI改善了原本MVBT空間利用率不佳的問題，卻因為Binary Search Tree的結構在資料量增大時樹高難以控管而導致效能下降，因此本篇論文提出一個對於空間和時間皆能改善其存取效能的多版本控管索引架構，且降低冗餘資料量。

In this paper, we present a novel index structure called the hybrid index structure, which is suitable for indexing the embedded multiversion databases using NVM storage as a medium. Embedded database systems are popularly applyed in various kinds of systems, e.g., cyber-physical systems (CPS). For the requirement of accessing the historical data, a multiversion index is introduced to maintain multiple versions of data items in an effective way. However, the structrural maintainance problem is a complex issue for the designer both because the space and time utilization.
　　In previous work, the multiversion B-tree (MVBT) [1] has already provided an efficient indexing structure for indexing multiversion data items in an embedded multiversion databases and the structure of MVBT is not that difficult to maintain. Besides, it also bring out a good response of version-range query. However, the MVBT indexing structure may lead to a large waste of space due to the frequent version/key split. As the consequence, the space-efficient multiversion index (SEMI) [2] have proposed to solve this problem. Although SEMI has a brilliant space-efficiency on multiversion indexing structure, but the structure of it is oppositely hard to maintain. Furthermore, it may occur thrashing when the key is getting larger. Therefore, in our approach, we proposed an hybrid index structure for both of the efficiency of space and time complexity, by cut down the amount of redundant data.

[1] Becker, S. Gschwind, T. Ohler, B. Seeger, and P. Widmayer. 1996. An asymp-totically optimal multiversion B-tree. VLDB Journal 5, 4 (Dec. 1996).
[2] Y.-H. Kuan, Y.-H. Chang, P.-C. Huang, and K.-Y Lam. 2014. Space-efficient multiversion index scheme for PCM-based embedded database systems. Design Automation Conference (DAC), (2014).
[3] J. R. Driscoll, N. Sarnak, D. D. Sleator, and R. E. Tarjan. 1989. Making data structures persistent.ElsevierJournal of Computer and System Sciences26, 1 (1989), 86–124.
[4] J. Wang, K.-Y. Lam, Y.-H. Chang, J.-W. Hsieh, and P.-C. Huang. 2014. Block-based multi-version B+-tree forflash-based embedded database systems.IEEE Transactions on Computers64, 4 (April 2015), 925–940.
[5] Chen, G. Xing, X.Wang, and X. Fu. 2011c. Fidelity-aware utilization control for cyber-physical surveillance systems. In Proceedings of the IEEE Real-Time Systems Symposium. 117–126.
[6] H.-S. P. Wong, S. Raoux, S. B. Kim, J. Liang, J. P. Reifenberg, B. Rajendran, M. Asheghi, and K. E. Goodson. Phase change memory. Proceedings of the IEEE 98, 12 (2010), 2201–2227.