Abstract :
In this paper, we propose a protocol to evaluate conditional sum on encrypted data. Here, conditional sum is a sort of computation which takes pair of data (ai, di) ∈ A × D (i = 1, 2, ...) and a condition à ⊆ A as input and output summation of dis such that corresponding ai satisfies ai ∈ à (i.e., Σai∈à di.) Such a computation is used in statistical analysis of large scale data. For example, it is used to analyze a total medical cost of employee on their 50s, 40s and 30s. We construct protocols based on additive homomorphic encryption and 2-DNF cryptosystems, respectively, where 2-DNF cryptosystem is a homomorphic cryptosystem which allows arbitrary number of addition and single multiplication on encrypted data. The size of ciphertext for (ai, di) of the proposed schemes protocols on additive homomorphic encryption and 2-DNF encryption are O(n) and O(√n), respectively, where n denotes the cardinality of the set A.
Keywords :
cryptographic protocols; statistical analysis; 2-DNF cryptosystems; additive homomorphic encryption; ciphertext; conditional sum evaluation; employee; encrypted data; homomorphic cryptosystem; large scale data; medical cost; protocols; statistical analysis; Additives; Data privacy; Encryption; Protocols; Public key; Registers;