Privacy-Aware Cost-Effective Scheduling Considering Non-Schedulable Appliances in Smart Home

Boyang Li, Jie Wu, Yiyu Shi

TL;DR

This paper introduces a runtime scheduling framework that balances electricity cost savings with consumer privacy in smart homes. Unlike previous models that only consider flexible appliances (like dishwashers), this approach accounts for "non-schedulable" devices (like TVs or ovens) and uses a rechargeable battery to mask power signatures, ensuring user habits remain private even under worst-case scenarios.

What Problem Does This Paper Address?

Smart meters are essential for modern energy efficiency, but they come with a significant "privacy tax." By recording fine-grained power usage, these meters inadvertently leak data about a resident's daily habits—when they cook, when they sleep, or even what TV shows they watch. Attackers can use this "load signature" to cause physical or financial harm.

While researchers have previously proposed using rechargeable batteries to "level" the load and hide these signatures, most models assume all appliances can be scheduled. In reality, many household activities are spontaneous (non-schedulable). These unpredictable power spikes can break privacy protections. The authors tackle the challenge of maintaining privacy and minimizing costs in a home where some energy use is completely unpredictable.

Methods

The authors propose a framework called PACES (Privacy-Aware Cost-Effective Scheduling). The core of their approach involves three main components:

  1. Mathematical Modeling: They categorize appliances into two groups: schedulable (can be shifted to off-peak hours) and non-schedulable (must run immediately). They also model a rechargeable battery as a buffer.
  2. Dynamic Programming (Algorithm 1): To handle the complexity of scheduling over a 12- or 24-hour horizon, they use a backward recursive approach to build a "schedule table." This table tells the system the optimal action (charge/discharge battery or start an appliance) for any given state.
  3. Iterative Alternative Algorithm (Algorithm 2): This is the "secret sauce." Since non-schedulable appliances introduce uncertainty, the algorithm iteratively identifies the "worst-case privacy scenario"—the time window where spontaneous appliance use would most likely reveal a user's patterns. It then optimizes the schedule of other devices to ensure that even if the worst happens, the total power draw stays within a predefined "privacy bound" ($\lambda$).

Datasets

The researchers evaluated their framework using the ECO data set, which contains real-world household electricity consumption data from Switzerland collected over eight months.

  • Appliances used: Schedulable (clothes dryer, washing machine, dishwasher) and Non-schedulable (PC, stereo, TV, laptop).
  • Pricing: They used publicly available day-ahead electricity prices from the Ameren Corporation to simulate realistic cost-saving incentives.

Key Insights and Ideas

  • The Battery as a Mask: In this model, the battery isn't just for "buying low and selling high." It acts as a privacy shield. By charging when usage is low and discharging during peaks, it flattens the home's "power profile," making it impossible for an observer to tell which specific appliances are running.
  • Worst-Case Robustness: Instead of trying to predict exactly when a user will turn on the TV, the algorithm prepares for the worst possible time they could do so. By scheduling flexible loads around these potential "worst-case" windows, the system provides a privacy guarantee.
  • The Privacy-Cost Trade-off: There is a direct relationship between the "privacy bound" ($\lambda$) and the electricity bill. A tighter privacy bound (smaller $\lambda$) makes the load curve smoother but limits the system's ability to shift all usage to the cheapest hours, slightly increasing the cost.

Key Results

  • Privacy Protection: Experimental results showed that the scheduled power consumption curve successfully stayed within the upper and lower privacy boundaries, effectively masking the "spiky" signatures of individual appliances.
  • Cost Efficiency: Despite the privacy constraints, the system was able to lower total billing costs by shifting schedulable loads to lower-priced time slots.
  • Battery Sensitivity: The study found that the system is highly sensitive to battery capacity. As battery size increases, the billing price drops rapidly at first before converging, indicating there is a "sweet spot" for battery sizing in smart homes (around 20-30 kWh in their specific simulation).

Limitations and Future Work

The authors acknowledge that their current approach can be computationally expensive. As the number of appliances in a home increases, the "state space" for the dynamic programming algorithm grows exponentially, making it time-consuming to build the schedule table.

Future research directions include:

  • Optimizing the algorithm for speed to handle homes with dozens of smart devices.
  • Exploring more complex battery models (e.g., accounting for battery degradation over time).
  • Integrating renewable energy sources (like solar panels) into the scheduling framework.

Summary updated

Abstract

A Smart home provides integrating and electronic information services to help residential users manage their energy usage and bill cost, but also exposes users to significant privacy risks due to fine-grained information collected by smart meters. Taking account of users' privacy concerns, this paper focuses on cost-effective runtime scheduling designed for schedulable and non-schedulable appliances. To alleviate the influence of operation uncertainties introduced by non-schedulable appliances, we formulate the problem by minimizing the expected sum of electricity cost under the worst privacy situation. Inventing the iterative alternative algorithm, we effectively schedule the appliances and rechargeable battery in a cost-effective way while satisfying users' privacy requirement. Experimental evaluation based on real-world data demonstrates the effectiveness of the proposed algorithm.
BibTeX
@inproceedings{li2019privacy,
  author = {Boyang Li and Jie Wu and Yiyu Shi},
  title = {{Privacy-Aware Cost-Effective Scheduling Considering Non-Schedulable Appliances in Smart Home}},
  booktitle = {ICESS},
  year = {2019},
  doi = {10.1109/icess.2019.8782440},
}

Loading related content...

Figure