Concealed regression for aggregation in low power wireless networks
Andrey Tolstikhin, Željko Žilić · 2017
This paper investigates data aggregation techniques for low power, distributed wireless networks. Nodes in such networks are constrained in terms of processing power, cost, networking performance, and available energy for the transmission of generated data. Additionally, the sensitive nature of data collected by the networks and physical vulnerability of nodes suggests that end-to-end encryption needs to be used to maintain security. The strict security constraints pose challenges in aggregation and routing algorithm design, due to the complexity-security trade off in distributed computing. Information must be shared between nodes to calculate meaningful results while preserving individual privacy. A platform of networking, aggregation and security algorithms is suggested that can increase node contribution up to a factor of 1.8 and decrease bandwidth up to a factor of 4.1 compared to previous methods.