Security Aspects in the Advanced Metering Infrastructure

The traditional electrical grid is transitioning into the smart grid. New equipment is being installed to simplify the process of monitoring and managing the grid, making the system more transparent to use but also introducing new security problems. Smart meters are replacing the traditional electrical utility meters, offering new functionalities such as remote reading of the consumption indexes, different time of use tariffs, automatic error reporting, and the possibility for the electricity providers to remotely turn off and on the electricity service at one location. This research thesis studies this last feature through two scenarios where we emphasize the effects of an attack exploiting the remote turn off feature, both on a theoretical level and through a simulation. In the first scenario, the frequency property of the grid is the target in an attempt to cause a widespread blackout. In the second scenario, the voltage is driven out of bounds by the adversary, causing physical damage to the electrical appliances of the affected customers. Data provided by the smart meters can be used to develop fraud and attack detection and mitigation tools. Obtaining real data can sometimes be cumbersome, due to privacy concerns. We propose an anonymization technique for sensitive data, based on a cryptographic procedure; this provides consistent results even if it is used over different traces. An implementation of this technique is also provided. In the process of developing fraud and attack detection and mitigation techniques, the case of off-line centralized data is covered, for both individual smart meters and clusters of smart meters.