Repudiative information retrieval

On preserving privacy in content-oriented networks

“…For content-analysis attacks, we can't use existing approaches to provide users plausible deniability, such as repudiable information retrieval [3] or off-the-record messaging [6], since they require cooperation from particular storage nodes (that could be shut down if known by the government, and the government knows everything users know). Approaches such as deniable encryption [8] require that content publishers share secret keying information with end users.…”

“…Assuming a fast hash function implementation on a single CPU pre-computing all the possible chunk names is likely to take the adversary less than 10 seconds. 3 10 seconds of pre-computation time is not a limiting factor even if the cover blocks are set to change periodically. Increasing (n + m) by decreasing the size of the data blocks or increasing the number of cover files as well as increasing k results in a likely decrease in how fast the adversary can pre-compute a table of chunk names.…”

Achieving Efficient Query Privacy for Location Based Services

“…Asonov et al [3,4] propose a relaxation of the strong privacy requirement of PIR to reduce preprocessing complexity; their PIR requires the support of a secure coprocessor (SC). This relaxation intends to replace the strong privacy requirement of no information about the user query being revealed to a weaker privacy notion of not much information about the user query being revealed.…”

“…PIR with relaxed privacy is called Repudiative Information Retrieval (RIR). RIR introduced in [4] requires a PIR with SC support; however, the tradeoff solution presented in our work has no such requirements. The server running the PIR protocols does not need to have a SC.…”

“…A collection of POIs is saved in a database such that each row corresponds to one POI. 4. Each cell of the grid is mapped to a portion of the database, i.e., a particular set of database rows (each containing a POI).…”