Forensic characterization of RF circuits

Given the wide use of Radio Frequency (RF) devices for applications ranging from data networks to wireless sensors, it is of interest to identify the types of RF devices that are located in an environment. For example, one could be on campus at a university and would want to prevent their RF device from interfering with other devices. There are currently two techniques that are used to identify RF devices in the environment, which are known as passive and active fingerprinting. These fingerprinting techniques require that the RF device comply to some transmission standard (IEEE 802.11) and can therefore only be used to identify a subset of RF devices. We propose an approach that can theoretically be used to identify any RF device. The proposed approach is used to determine sources of nonlinearities and identify filters that are present in the RF device. In our approach, a probe signal is transmitted to the RF device. The RF device retransmits or reradiates a return signal in response to the probe signal. The properties of the return signal are measured and these properties can be shown to contain distinct information that is inherent to sources of the nonlinearities and the filters of the RF device. This information constitutes features that can be used to identify the RF device. In an effort to identify sources of nonlinearities, we will analyze the output signal obtained from a RF circuit model when the input is either a two-tone probe signal or a windowed linear chirp probe signal. It will be shown that the power spectral density of the output signal contains distinct features caused by nonlinearities in the RF circuit. These features are used to identify the nonlinearity and consequently the RF device. In addition, we will analyze the output signal from the bandpass filter in the RF circuit when the input is either a windowed linear chirp probe signal or a linear chirp probe signal. In this dissertation, we will use simulations to verify our methods. For each simulation, several RF circuit models are considered, where each circuit model contains a unique nonlinearity and bandpass filter. The features are classified using statistical pattern recognition techniques. Some of our results indicate that our simulation approach needs further investigation to validate it against actual measurement data