Particle-filter-enabled real-time sensor fault detection without a model of faults

Abstract

Abstract-We are experiencing an explosion in the amount of sensors measuring our activities and the world around us. These sensors are spread throughout the built environment and can help us perform state estimation and control of related systems, but they are often built and/or maintained by third parties or system users. As a result, by outsourcing system measurement to third parties, the controller must accept their measurements without being able to directly verify the sensors' correct operation. Instead, detection and rejection of measurements from faulty sensors must be done with the raw data only. Towards this goal, we present a method of detecting possibly faulty behavior of sensors. The method does not require that the control designer have any model of faulty sensor behavior. As we discuss, it turns out that the widelyused particle filter state estimation algorithm provides the ingredients necessary for a hypothesis test against all ranges of correct operating behavior, obviating the need for a fault model to compare measurements. We demonstrate the applicability of our method by showing its ability to reject faulty measurements and accuracy in state estimation of a nonlinear vehicle traffic model, without information of generated faulty measurements' characteristics. In our test, we correctly identify nearly 90% of measurements as faulty or non-faulty without having any fault model. This leads to only a 3% increase in state estimation error over a theoretical 100%-accurate fault detector.

A Framework for Robust Assimilation of Potentially Malign Third-Party Data, and its Statistical Meaning

“…However, for our current purposes this is not easily implementable because in our filtering context, we are trying to make a decision as to whether to accept or reject a measurement as we receive it. Taking a "soft" view, and considering a range of state space values based on our current range of belief of whether we should accept or reject H 0 , while potentially giving us a view of a broader range of possibilities (and separate measurement hypotheses) that we could revisit in light of future data, leads to a blow-up when the number of separate sensors increase [29]. Instead, for expedience, in what follows we adopt the Neyman-Pearson and null hypothesis significance test and select a hard significance level α, and reject or accept the measurement based on whether our estimated p-value ( 17) is larger or smaller.…”

“…In this paper, we will not exhaustively define all PDFs of interest in the interest of readability. See [29] for a more lengthy discussion of a precursor to the Fisher-type hypothesis-testing particle filter discussed in the present work.…”

“…The work described in this paper was originally inspired by technical problems we encountered in our prior work in this area. In [28], [29], we report on our efforts to use anonymized third-party data from connected vehicles to estimate the state of traffic on a freeway. That is, the assimilation of records consisting of times, positions, and speeds from transponders near the freeway, but without certainty of the correctness of the data.…”

Robust Data Fusion of UAV Navigation Measurements with Application to the Landing System

“…The federation scheme is very often used in practice, but it does not take into account the possible correlation between parameter under estimation, and for correction of long-term U-type errors it needs to develop a special methodology. It is interesting to use the particle filter to combat anomalies like in [13]. The application of this algorithm in redundant systems also requires additional research.…”

“…For the mathematical description of such errors the heavy-tailed distributions are used instead of the Gaussian distribution [31]. The filtering problem for heavy-tailed distributions can be solved approximately using the so-called particle filters, which are based on the Monte Carlo method [13,32]. However, the complexity of such algorithms increases exponentially with the increase of the state space dimension, which requires large computational resources.…”