Private Weighted Sum Aggregation for Distributed Control Systems

Abstract

Data aggregation in distributed networks is a critical element in Internet of Things applications ranging from smart grids and robot swarms to medical monitoring over multiple devices and data centers. This paper addresses the problem of private weighted sum aggregation, i.e., how to ensure that an untrusted aggregator is able to compute only the weighted sum of the users’ private local data, with proprietary weights. We propose a scheme that achieves the confidentiality of both the users’ local data and the weights, as long as there are at least two participants that do not collude with the rest. The solution involves two layers of encryption based on the Learning With Errors problem. We discuss how to achieve efficient multi-dimensional data aggregation by using plaintext packing in the homomorphic crypto-system used, such that the communication between the users and the aggregator is minimized.