@InProceedings{Kalutharage:2026:TradingApplianceLiveness,
    author = "Kalutharage, Chathuranga Sampath and Fowley, Brendon and Brady, Cason and Bradbury, Matthew",
    booktitle = "{Computer Security. ESORICS 2026 International Workshops}",
    title = "{Trading Appliance Liveness for Home Electricity Consumption Privacy}",
    year = "2026",
    address = "Rome, Italy",
    month = "14--18 September",
    publisher = "Springer Nature Switzerland",
    abstract = "Homes have been equipped with smart meters in recent years to automate the monitoring of energy usage to facilitate better scheduling of electricity generation and offer better billing of customers. This has increased the time resolution of energy consumption, which leads to privacy threats such as occupancy monitoring, pattern of live analysis, and device identification. Much work has been performed on using energy storage and renewable generation to obfuscate observations made via a smart meter. In this paper, we consider how appliance liveness can be traded-off to reduce behavioural distinguishability. We identify that liveness can be traded-off by undertaking an analysis of the REFIT smart home dataset and transforming the energy consumption of each day using a mixed integer programming model to and investigate all 512 combinations of liveness constraints of 9 appliances. The Pareto frontier is found for each subset of combinations where a single appliance must be live in order to demonstrate the trade-offs that occupants would need to make in order to reduce behavioural distinguishability. We identify that while it is possible to trade-off liveness for privacy, appliances which have a high usage in a day can limit the ability to reduce behavioural distinguishability. We also find a strong positive correlation between energy consumption and behavioural distinguishability, where reducing behavioural distinguishability leads to a reduction in energy consumption. This analysis helps identify which appliance categories have the greatest influence on behavioural distinguishability and highlights the trade-offs between appliance liveness and observability in smart meter data.",
    file = ":MIST2026.pdf:PDF"
}