具社區階層整合雲端智慧家庭之隱私保護資料分析

Privacy Preserving Data Analytics in an Integrated Cloud-Based Smart Home Management System with Community Hierarchy

李英宗(Ying-Tsung Lee)

指導教授 ： 曹承礎

國立臺灣大學/管理學院/資訊管理學研究所/博士(2017年)

https://doi.org/10.6342/NTU201701963

摘要

本研究透過觀察現象，發現產學研究單位樂觀看待智慧家庭，但在市場面上有家戶滲透率偏低的問題。探究可能發生的原因，在於除了系統面上要達到安全、有用與易用以外，還需要處理使用者關心的隱私與信任議題。釐清使用者在想要系統智慧化自動串連元件背後，在隱私保護上也意圖要能夠自動化匿名處理各階段資料，才能達到真正信任。因此，抽象化這個信任議題來看，可參考信任第三方模型，在限制目標是維持最大化外部查詢匿名集、確保最小化內部受駭單點損失的想法前提下，延伸調整既有單點模式，提出解決方案。經過方案比較，提出符合目標的階層式架構，聚焦在具社區階層的智慧家庭解決方案。在效用評估的相對價值上，本研究除了在隱私保護面提供了端到端的機制、同時在資料分析面也兼顧了自主控制的彈性，可供使用者體驗選擇轉換隱私公開等級以取得分析應用上的平衡。最後，本架構框架在延伸應用上，可思考延伸導入社區端於實體地理階層中，已具備階層式架構的環境，套用隱私保護機制強化居家端匿名，增加使用者參與計畫意願，擴大計畫的精細度與規模。

總結本研究的貢獻重點如下：
一、在智慧家庭領域透過文獻探討，可以發現使用者採用面向所關心的從基本的安全、有用、易用、延伸到處理隱私與信任，才能夠結合資料分析輔助達到好用。同時在技術發展趨勢上，包含了從裝置整合、多重介面到雲端分析，同樣的帶出全面性資料隱私保護的需求。
二、對應的隱私保護-技術、流程與架構，抽象化來看就是信任第三方模型，可分拆成接收、處理、分享等模組。社區階層化的智慧家庭，符合限制目標:『維持最大化外部查詢匿名集』與『確保最小化內部受駭單點損失』，是適合的研究應用場域。
三、本研究在智慧社區實際建置應用成果，重點特色是使用者識別內建匿名、家戶端可自主控制、透過社區端進行實體隔離，最後在雲端可具有分析應用彈性。
四、可實現的分析內容包含雲端整體的統計排名、社區區域性的關聯比較以及深入各時間、空間屬性篩選的儀表板探索式資料分析，達到圖表展示效果。
五、具體來說，本研究的貢獻價值包含:
1.學術上進行信任第三方模型延伸，經過評估比較，可套用到社區階層架構，具備實務性管理應用價值。
2.智慧家庭的日常操作時間統計，可供匯流調研機構參考，與傳統居家電視、電腦、手機與平板的模式交叉比對。
3.本框架可泛用延伸導入社區端於實體地理階層，套用隱私保護機制，強化居家端匿名，擴大精細度與規模。
4.功能性的統計排名，結合各時間空間屬性篩選的儀錶板分析成果，可協助業者優化使用者介面，針對用戶提供更貼心的功能建議。
5.透過區域性，對各行政區與社區進行關聯比較，輔助政府跨區管理，可挖掘低度使用區域，並可依使用量適度分配管理資源。
6.最後，藉由本階層信任模型，在各社區均具備管理主機，可善用運算資源，符合新興從雲端移轉至在地邊緣運算之趨勢，除了本研究所提之隱私保護議題外，尚有節省骨幹頻寬、強化可靠度以及在地快速回應等效益。

關鍵字

智慧家庭 ； 社區階層 ； 雲端服務 ； 隱私保護 ； 資料分析

並列摘要

This research observed a phenomenon in which smart home systems fail commercially despite optimistic expectations from both industrial and academic research institutions. This phenomenon may be attributed to privacy and trust concerns, in addition to the safety concerns, perceived usefulness, and perceived ease of use of smart home systems. Smart home systems must not only fulfill users’ desires to automatically connect various components, but must also be capable of automatically handling hierarchical data anonymously for privacy protection, thus earning users’ trust. The trusted third party model (TTP) can be utilized to address the trust concerns with abstraction. To maximize an anonymous dataset for external search and minimize internal hacked loss at a single point, the existing single-point framework must be modified and a suitable hierarchical framework must be proposed as a solution. When applied to practice, this implies smart home systems that integrate community hierarchy and cloud data analysis for privacy protection. Performance assessments and comparisons emphasize the relative value of smart home systems. The proposed community hierarchy application provides an end-to-end mechanism for privacy protection while also accounting for the flexibility of autonomous control for data analysis. The proposed model enables users to experience the system and designate the privacy level accordingly. Finally, the proposed framework can be regarded as a template hierarchical framework similar to the smart districts project in the European Union. The proposed framework can be introduced to communities according to actual geographical features, and the strengthened privacy protection mechanisms can be implemented to ensure the anonymity of home units, thereby incentivizing user participation and increasing the accuracy and scope of the project.

The key contributions of this study are summarized as follows:
The review of the literature on smart home systems revealed that user concerns (comprising safety, perceived usefulness, perceived ease of use, privacy protection, and trust) must all be addressed to achieve satisfactory actual system use supplemented by data analyses. Meanwhile, various technological trends, including device integration, multiple interfaces, and cloud analysis, have prompted the demand for comprehensive data privacy protection.
The privacy protection technique, procedure, and framework correspond to those of the TTP model when viewed from an abstraction perspective, and can be decomposed to modules for receiving, handling, and sharing. Smart home systems with a community hierarchy are aligned with the limitations of maximizing an anonymous dataset for external search and minimizing internal hacked loss at a single point, and are a suitable field for developing applications.
The achieved practical highlights of the smart home system proposed in this study are the built-in anonymity of user recognition, autonomous control from home units, ability to perform physical isolation through community ends, and a flexible cloud service for analytical applications.
The achievable analytical applications include comprehensive rankings from the cloud service, regional comparison among communities, and exploratory data analysis featuring a dashboard that visualizes temporal and spatial attributes.

In essence, the contributions of this study are as follows:
1) Academically, this study extended the TTP model. The assessment and comparison can be applied to community hierarchical frameworks and have practical implications.
2) The daily data of smart home system usage can serve as a reference for digital convergence research institutions to compare with daily usage of home TVs, personal computers, mobile phones, and tablet devices.
3) The proposed framework can be introduced to community levels according to actual geographical features, and the strengthened privacy protection can be applied to ensure the anonymity of home units, thereby incentivizing user participation and increasing the accuracy and scope of the project.
4) The dashboard analysis that visualizes rankings of various temporal and spatial attributes enables smart home system operators to optimize user interface and provide user-friendly recommendations.
5) The regional comparison among administrative regions and communities facilitates cross-district management for governments, identifies underutilized areas, and organizes resource allocation according to system usage.
6) Finally, the hierarchical trust model features a data management host in each community to utilize computational resources, aligning the model with the emerging trend of changing cloud computing to edge computing. In addition to addressing the privacy protection concerns examined in this study, the proposed model offers benefits such as saving backbone bandwidth, enhancing reliability, and enabling community responses to be quickly dispatched.

並列關鍵字

Smart Home ； Community Hierarchy ； Cloud Services ； Privacy Preserving ； Data Analytics

參考文獻

[104] S. A. Osia, A. S. Shamsabadi, A. Taheri, H. R. Rabiee, N. D. Lane, and H. Haddadi, “A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics,” arXiv.org. [Online]. Available: https://arxiv.org/abs/1703.02952.

[2] G. Ramamoorthy, “IoT Drives Innovation in the Semiconductor Industry,” Gartner Webinar., Jul.2015.

[3] S. Kong, Y. Kim, R. Ko, and S. K. Joo, "Home appliance load disaggregation using cepstrum-smoothing-based method," IEEE Trans. Consumer Electron., vol. 61, no. 1, pp. 24-30, Feb. 2015.

[4] J. Han, C. S. Choi, W. K. Park, I. Lee, and S. H. Kim, "Smart home energy management system including renewable energy based on ZigBee and PLC," IEEE Trans. Consumer Electron., vol. 60, no. 2, pp. 198-202, May 2014.

[5] J. Han, C. S. Choi, W. K. Park, I. Lee, and S. H. Kim, "PLC-based photovoltaic system management for smart home energy management system," IEEE Trans. Consumer Electron., vol. 60, no. 2, pp. 184-189, May 2014.