Distributed Aggregation over Homomorphically Encrypted Data under Switching Networks

Abstract

This paper presents a distributed algorithm for additive aggregation over encrypted data. Our algorithm allows networked systems to compute the sum of all individual data without revealing each node's private contribution. To conceal the private data, the local computations on each node and the communications with neighbor nodes are conducted by utilizing the additively homomorphic cryptosystem. As a consequence, the leader node which has the secret key of the cryptosystem recovers the correct outcome after decryption in a finite time. Moreover, the leader cannot learn the private data of the other nodes. Furthermore, the proposed algorithm can operate on encrypted data for an infinite time horizon even when the network varies intermittently with time.