(B∥C∥A), (C∥A∥B) or (C∥B∥A) as a query to the remote server. Then the server tests the Bloom filter against the trapdoor and retrieves the associated matched document to DU without needing for the posting list intersection protocol.
Our scheme consists of six algorithms KeyGenerator, BuildIndex, DocEncrypt, Trap- doorGen, SearchIndex and DocDecrypt which are scattered between two phases, Sender
Phase and Retrieval Phase.
3.2.3.1
Sender phase
This phase includes three algorithms as detailed below:
I. Key generator: the data owner DO initiates the scheme by using KeyGenerator(µ) …show more content…
This algorithm takes the security parameter µ as input to obtain the pub- lic parameters CP = {DUpub,DOpub,V,r,g, ˆ e,Hbloom,H1,H2,H3} and the private keys PR = {DUpr,DOpr}.
II. Index construction: for each document Di ∈ D, DO dedicates a secure index
IDXDi, which is stored at the service provider that will help DO to perform a keyword search by calling BuildIndex(D,WDi,CP,DOpr). Each document Di comprising of an unique identifier IDi ∈ {0,1}n. Firstly, to protect the docu- ment Identifiers IDi, DO encrypts this IDiwith El Gamal encryption technique, such technique assurances that if the same document identifier is encrypted multiple times, it will create different ciphertexts but all decrypted to the same value. Then
DO creates one Bloom filter BF for each document, this filter consists of an array of x-bits, and uses q independent hash functions h1,...,hq. The filter allows the data owner to perform keyword searches efficiently, but could result in some false positive retrievals. A classical Bloom Filter may reveal information about the …show more content…
After that the array bits at the positions h1(Encperj),...,hq(Encperj) are set to 1. Finally, DO stores the encrypted ID EncIDiand associated bloom filter BFIDiin IDXDi.
In Bloom Filter, the number of 1s is reliant on the number of BF entries, in this case, the number of different permutations. As a consequence, the scheme reveals the number of permutations in each document. To avoid this problem, padding number of dummy keywords may be used to make sure that the number of 1s in the
Bloom Filter is nearly the same for various documents. Padding process is costly compared to the scheme without it because the higher rate of false-positive.
III Document collection encryption: to protect data privacy and undesired accesses, the document collection should be encrypted before outsourcing them onto re- mote servers which are not within their trusted domains. To do so, DO calls
DocEncrypt(D,CP,DOpr) algorithm to encrypt each file Di∈ D using El Gamal encryption technique[28].
The final step in the sender phase algorithm is sending the IDX and encrypted documents set EncDoc to remote