Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Privacy-friendly smart metering

Many smart metering proposals threaten users' privacy by disclosing fine-grained consumption data to utilities. We have designed protocols that allow for precise billing of consumption while not revealing any consumption information to third parties. We also have developped protocols that allow for privacy-friendly real-time aggregation of smart-meter readings.

Information computed on the basis of fine-grained smart-meter readings has multiple uses within the energy industry, including billing, providing energy advice, settlement, forecasting, demand response, and fraud detection. Microsoft Research has developed technologies that allow for these computations to be executed without the need for customers to disclose raw meter readings. In brief, smart-meters transmit encrypted certified meter readings, that are processed by any customer device (smart phone, web browser, home gateway, personal computer) to compute the information required, and further provide them to authorised parties. These privacy-friendly computations can include time-of-use bills, settlement values, fraud detection flags, or usage profiles. Cryptographic mechanisms protect the privacy of the data and the correctness of the computations even when performed on customer devices.

Energy industry processes, such as settlement, monitoring, financial forecasting, transmission network development or demand response, require real-time aggregates of readings across populations of meters. Microsoft Research has developed privacy technologies that allow the direct aggregation of encrypted meter readings. The sum of readings, as well as their mean and variance, can be computed in real-time, without revealing individual meter readings.

Our protocols are generic enough to be used in other settings such as pay-as-you-drive car insurance, electronic traffic pricing and on-line services billing.

For more information contact George Danezis (gdane@microsoft.com), Markulf Kohlweiss (markulf@microsoft.com), Cedric Fournet (fournet@microsoft.com)