Document Type



Doctor of Philosophy


Electrical Engineering

First Adviser

Rick S. Blum


The presented research investigates topics relating to sensor systems focusing mainly on estimation. The first topic studies Byzantine attacks on sensor systems estimating the value of an unknown deterministic parameter based on quantized observations. The presented work initially describes aspects of the optimal processing under a practical family of attacks where the sensors employ bad data detectors to check if the observed sensor data fits the observation models assumed by the estimation algorithm under no attack. Next, the performance of any estimation approach employed by the sensor system under any general attack is described for cases where any number of observations, sensors, and quantization levels could be employed. The second topic studies sensor networks focused on estimating ocean waveforms. Our work is the first to derive the Cramer-Rao bound (CRB) for the short-term forecasting of ocean waves. The CRB is a lower bound on the mean square estimation error for an unbiased estimator. The obtained results are general in the sense that they apply to a number of types of wave sensors. A low-complexity estimation method is presented along with numerical results demonstrating its accuracy and runtime performance. We also describe a method that relies on the CRB to calculate the expected loss in absorbed power, under optimal control, by a single or multiple WEC devices due to errors in the estimation of short-term future waveforms experienced by the devices. The third topic focuses on developing a novel model accounting for the dependence of power grids on communication networks for their safe and economic operation. Our model is formulated as a two-settlement stochastic optimal power flow (OPF) problem where we account for errors in forecasting while also considering random failures in the communication network. Our work jointly optimizes the topology of the communication network and the control actions in anticipation of the different random failures and errors that a given power grid may face. We present results that identify optimal topologies for a communication network supporting the operation of an IEEE standard 9-bus system under different conditions and we discuss some properties of the optimal solutions. Finally, the fourth topic studies topology estimation in power distribution networks. Accurate topology estimates are crucial for maintaining situational awareness, properly dispatching distributed energy sources, detecting cyber attacks, and many other key tasks. The presented work takes advantage of the ever increasing adoption of novel metering and sensing devices providing data from network locations that were traditionally unmonitored by grid operators. We present a topology estimation scheme for radial distribution networks relying on power flow measurements and nodal load forecasts. We also describe a sensor placement method that enables the presented scheme to identify all the detectable faults in the network. The performance of our detection scheme is then demonstrated through several numerical results on the IEEE 123-bus test feeder.