SecVKQ: Secure and verifiable kNN queries in sensor–cloud systems

Abstract

Sensor–cloud has gained increasingly popularity since it bridges the physical world and the cyber world through pervasive computation. This paper focuses on secure and verifiable k nearest neighbor (kNN) queries over large-scale outsourced datasets in sensor–cloud systems. Existing work in this line often incurred high computational and communication overheads, remaining far away from practical and scalable. To this end, we propose SecVKQ, a two-phase search framework, which mainly includes a preliminary screening phase and an exact search phase. SecVKQ purposely takes advantage of secure data separation and adaptive encryption strategy, embracing edge servers into the classic dual-cloud model, so as to optimize query performance. Under SecVKQ, we design a series of secure protocols and develop a succinct verification strategy to derive a unified solution. The experimental results demonstrate the effectiveness of SecVKQ. Compared to the state-of-the-art work, SecVKQ achieves a speed-up of two orders of magnitude in search latency, and a savings of 50% communication cost for verification.