Roy Dong

Compressive Phase Retrieval From Squared Output Measurements Via Semidefinite Programming

Given a linear system in a real or complex domain, linear regression aims to recover the model parameters from a set of observations. Recent studies in compressive sensing have successfully shown that under certain conditions, a linear program, namely, l1-minimization, guarantees recovery of sparse parameter signals even when the system is underdetermined. In this paper, we consider a more challenging problem: when the phase of the output measurements from a linear system is omitted. Using a lifting technique, we show that even though the phase information is missing, the sparse signal can be recovered exactly by solving a semidefinite program when the sampling rate is sufficiently high. This is an interesting finding since the exact solutions to both sparse signal recovery and phase retrieval are combinatorial. The results extend the type of applications that compressive sensing can be applied to those where only output magnitudes can be observed. We demonstrate the accuracy of the algorithms through extensive simulation and a practical experiment.

A game theory model for electricity theft detection and privacy-aware control in AMI systems

We introduce a model for the operational costs of an electric distribution utility. The model focuses on two of the new services that are enabled by the Advanced Metering Infrastructure (AMI): (1) the fine-grained anomaly detection that is possible thanks to the frequent smart meter sampling rates (e.g., 15 minute sampling intervals of some smart meter deployments versus monthly-readings from old meters), and (2) the ability to shape the load thanks to advanced demand-response mechanisms that leverage AMI networks, such as direct-load control. We then study two security problems in this context. (1) In the first part of the paper we formulate the problem of electricity theft detection (one of the use-cases of anomaly detection) as a game between the electric utility and the electricity thief. The goal of the electricity thief is to steal a predefined amount of electricity while minimizing the likelihood of being detected, while the electric utility wants to maximize the probability of detection and the degree of operational cost it will incur for managing this anomaly detection mechanism. (2) In the second part of the paper we formulate the problem of privacy-preserving demand response as a control theory problem, and show how to select the maximum sampling interval for smart meters in order to protect the privacy of consumers while maintaining the desired load shaping properties of demand-response programs.

Fundamental limits of nonintrusive load monitoring

Provided an arbitrary nonintrusive load monitoring (NILM) algorithm, we seek bounds on the probability of distinguishing between scenarios, given an aggregate power consumption signal. We introduce a framework for studying a general NILM algorithm, and analyze the theory in the general case. Then, we specialize to the case where the error is Gaussian. In both cases, we are able to derive upper bounds on the probability of distinguishing scenarios. Finally, we apply the results to real data to derive bounds on the probability of distinguishing between scenarios as a function of the measurement noise, the sampling rate, and the device usage.

A dynamical systems approach to energy disaggregation

Energy disaggregation, also known as non-intrusive load monitoring (NILM), is the task of separating aggregate energy data for a whole building into the energy data for individual appliances. Studies have shown that simply providing disaggregated data to the consumer improves energy consumption behavior. However, placing individual sensors on every device in a home is not presently a practical solution. Disaggregation provides a feasible method for providing energy usage behavior data to the consumer which utilizes currently existing infrastructure. In this paper, we present a novel framework to perform the energy disaggregation task. We model each individual device as a single-input, single-output system, where the output is the power consumed by the device and the input is the device usage. In this framework, the task of disaggregation translates into finding inputs for each device that generates our observed power consumption. We describe an implementation of this framework, and show its results on simulated data as well as data from a small-scale experiment.

Incentive Design and Utility Learning via Energy Disaggregation

The utility company has many motivations for modifying energy consumption patterns of consumers such as revenue decoupling and demand response programs. We model the utility company--consumer interaction as a principal--agent problem. We present an iterative algorithm for designing incentives while estimating the consumers utility function. Incentives are designed using the aggregated as well as the disaggregated (device level) consumption data. We simulate the iterative control (incentive design) and estimation (utility learning and disaggregation) process for examples including the design of incentives based on the aggregate consumption data as well as the disaggregated consumption data.

Energy disaggregation via adaptive filtering

The energy disaggregation problem is recovering device level power consumption signals from the aggregate power consumption signal for a building. We show in this paper how the disaggregation problem can be reformulated as an adaptive filtering problem. This gives both a novel disaggregation algorithm and a better theoretical understanding for disaggregation. In particular, we show how the disaggregation problem can be solved online using a filter bank and discuss its optimality.

Privacy-enhanced architecture for occupancy-based HVAC Control

Large-scale sensing and actuation infrastructures have allowed buildings to achieve significant energy savings; at the same time, these technologies introduce significant privacy risks that must be addressed. In this paper, we present a framework for modeling the trade-off between improved control performance and increased privacy risks due to occupancy sensing. More specifically, we consider occupancy-based HVAC control as the control objective and the location traces of individual occupants as the private variables. Previous studies have shown that individual location information can be inferred from occupancy measurements. To ensure privacy, we design an architecture that distorts the occupancy data in order to hide individual occupant location information while maintaining HVAC performance. Using mutual information between the individuals location trace and the reported occupancy measurement as a privacy metric, we are able to optimally design a scheme to minimize privacy risk subject to a control performance guarantee. We evaluate our framework using real-world occupancy data: first, we verify that our privacy metric accurately assesses the adversarys ability to infer private variables from the distorted sensor measurements; then, we show that control performance is maintained through simulations of building operations using these distorted occupancy readings.

Nonlinear basis pursuit

In compressive sensing, the basis pursuit algorithm aims to find the sparsest solution to an underdetermined linear equation system. In this paper, we generalize basis pursuit to finding the sparsest solution to higher order nonlinear systems of equations, called nonlinear basis pursuit. In contrast to the existing nonlinear compressive sensing methods, the new algorithm is based on convex relaxation and is not a greedy method. The novel algorithm enables the compressive sensing approach to be used for a broader range of applications where there are nonlinear relationships between the measurements and the unknowns.

Modeling and open-loop control of IPMC actuators under changing ambient temperature

Because of the cost and complexity associated with sensory feedback, open-loop control of ionic polymer–metal composite (IPMC) actuators is of interest in many biomedical and robotic applications. However, the performance of an open-loop controller is sensitive to the change in IPMC dynamics, which is influenced heavily by ambient environmental conditions including the temperature. In this paper we propose a novel approach to the modeling and open-loop control of temperature-dependent IPMC actuation dynamics. An IPMC actuator is modeled empirically with a transfer function, the zeros and poles of which are functions of the temperature. With auxiliary temperature measurement, open-loop control is realized by inverting the model at the current ambient temperature. We use a stable but noncausal algorithm to deal with non-minimum-phase zeros in the system that would prevent directly inverting the dynamics. Experimental results are presented to show the effectiveness of the proposed approach in open-loop tracking control of IPMC actuators.

Delay and sampling independence of a consensus algorithm and its application to smart grid privacy

The consensus algorithm can represent many problems in cooperative behavior, and has been widely used in engineering and social sciences. In this work, we prove that the consensus model where the information that each agent receives from its neighbors has time-varying asynchronous delays and sampling, converges to an agreement independent of these communication constraints. This property is useful in the context of “data minimization,” which is one of the principles for privacy. As a practical example, we show how the independence of sampling rate can be used for microgrids with a consensus-based secondary control scheme where participants have incentives to share their states to ensure frequency synchronization, while at the same time minimizing the amount of data shared to preserve their privacy. We then propose two data sharing algorithms: 1) periodic sampling, and 2) discretionary sampling, and study their privacy as well as their performance. We show that even when a discretionary sampling scheme “lies” to their neighbors in order to preserve their privacy, the consensus algorithm performs almost as well as with periodic sampling.