Venue | Category |
---|---|
TDSC'16 | Secure Deduplication |
Enhanced Secure Thresholded Data Deduplication Scheme for Cloud Storage1. SummaryMotivation of this paperEnhanced threshold-based deduplicated encryptionImplementation and Evaluation2. Strength (Contributions of the paper)3. Weakness (Limitations of the paper)4. Future Works
Outsourced data may require different levels of protection, depending on how "popular" the datum is.
this differentiation based on popularity also can allevuates the user's need to mannually sort the data as common: deduplicated potentially sensitive (non-deduplicate)
Duplicate occurs only when the file becomes popular (i.e., shared by many users)
strictly confidential files that must never be deduplicated.
two backup types:
- uploaded by many users: benefit strongly from deduplication
- uploaded by one or very few users only: require confidentiality
- identity provider: users are identified when a user first joins the system.
- index repository service: generate the file identity.
- semantic security for unpopular data
- conventional convergent security for popular data prevent the storage provider which is honest-but-curious.
This paper just considers a single threshold .
- pirate bay dataset: each torrent represents a file (file popularity)
- Ubuntu popularity contest
- the value of (the tunable parameter of the scheme)
- the popularity distribution of files in the dataset
- Client computation
- Server computation
- Communication
- Dropbox interact
trade-off between security and storage efficiency
how to improve it?
Pirate Bay dataset