Asymmetric homomorphisms for secure aggregation in heterogeneous scenarios

In multicast communication, a single source transmits the same content to a large amount of receivers. This kind of communication is usually represented following a tree model where the root of the tree is the multicast source and the leaves are the receivers. Scalability problems arise when the root needs to collect data (sensor information, metering data, etc.) from the leaves. This results in a many-to-one (leaf-to-root) communication. The matter is further complicated if there are security requirements on the leaf-to-root traffic. In this paper we present a method for secure and scalable many-to-one lossy transmission based on asymmetric homomorphisms which enables the root of the tree to compute any mathematical function (e.g. minimum, maximum, average, … ) on the data sent by the leaves. Our proposal preserves the confidentiality of those data. Authentication is guaranteed in the sense that only authorized nodes can participate in the protocol. Integrity against compromised leaves is also achieved. In the case of a compromised intermediate node which colludes with a compromised leave, they can only cause a limited deviation in the final aggregate value.

► We propose a scalable protocol that provides secure and lossy data aggregation.
► The base station can compute any mathematical function on the data sent by sensors.
► Asymmetric homomorphic encryption techniques are applied to provide integrity.
► The proposed protocol can be implemented in a wide range of environments.
► The message length grows logarithmically with the number of sensors.