Showing 1–8 of 8 results for author: Behnia, F

Deep Generative Models for Vehicle Speed Trajectories

Authors: Farnaz Behnia, Dominik Karbowski, Vadim Sokolov

Abstract: Generating realistic vehicle speed trajectories is a crucial component in evaluating vehicle fuel economy and in predictive control of self-driving cars. Traditional generative models rely on Markov chain methods and can produce accurate synthetic trajectories but are subject to the curse of dimensionality. They do not allow to include conditional input variables into the generation process. In th… ▽ More Generating realistic vehicle speed trajectories is a crucial component in evaluating vehicle fuel economy and in predictive control of self-driving cars. Traditional generative models rely on Markov chain methods and can produce accurate synthetic trajectories but are subject to the curse of dimensionality. They do not allow to include conditional input variables into the generation process. In this paper, we show how extensions to deep generative models allow accurate and scalable generation. Proposed architectures involve recurrent and feed-forward layers and are trained using adversarial techniques. Our models are shown to perform well on generating vehicle trajectories using a model trained on GPS data from Chicago metropolitan area. △ Less

Submitted 14 December, 2021; originally announced December 2021.

Three-Dimensional Localization of Active Aerial Targets Using a Single Terrestrial Receiver Site

Authors: Saber Kaviani, Fereidoon Behnia

Abstract: This paper proposes a method for the three-dimensional localization of an active aerial target by a single ground based sensor. The proposed method employs the time and frequency differences of arrival of the signal received directly from the aerial target and the signals received after being reflected from some large auxiliary terrestrial targets (pseudo-sensors) with known positions on the groun… ▽ More This paper proposes a method for the three-dimensional localization of an active aerial target by a single ground based sensor. The proposed method employs the time and frequency differences of arrival of the signal received directly from the aerial target and the signals received after being reflected from some large auxiliary terrestrial targets (pseudo-sensors) with known positions on the ground. Due to the terrestrial nature of the main and the pseudo sensors, it is impossible to solve for the target's altitude using traditional methods. The proposed method employs target motion analysis to obtain target position including its altitude with acceptable accuracy and low computational complexity. Presented simulations confirm acceptable accuracy of the proposed method in determining three dimensional position of the target despite limited number of the pseudo sensors and its low computational complexity. △ Less

Submitted 23 July, 2021; originally announced July 2021.

LASCA: Learning Assisted Side Channel Delay Analysis for Hardware Trojan Detection

Authors: Ashkan Vakil, Farnaz Behnia, Ali Mirzaeian, Houman Homayoun, Naghmeh Karimi, Avesta Sasan

Abstract: In this paper, we introduce a Learning Assisted Side Channel delay Analysis (LASCA) methodology for Hardware Trojan detection. Our proposed solution, unlike the prior art, does not require a Golden IC. Instead, it trains a Neural Network to act as a process tracking watchdog for correlating the static timing data (produced at design time) to the delay information obtained from clock frequency swee… ▽ More In this paper, we introduce a Learning Assisted Side Channel delay Analysis (LASCA) methodology for Hardware Trojan detection. Our proposed solution, unlike the prior art, does not require a Golden IC. Instead, it trains a Neural Network to act as a process tracking watchdog for correlating the static timing data (produced at design time) to the delay information obtained from clock frequency swee** (at test time) for the purpose of Trojan detection. Using the LASCA flow, we detect close to 90% of Hardware Trojans in the simulated scenarios. △ Less

Submitted 17 January, 2020; originally announced January 2020.

Code-Bridged Classifier (CBC): A Low or Negative Overhead Defense for Making a CNN Classifier Robust Against Adversarial Attacks

Authors: Farnaz Behnia, Ali Mirzaeian, Mohammad Sabokrou, Sai Manoj, Tinoosh Mohsenin, Khaled N. Khasawneh, Liang Zhao, Houman Homayoun, Avesta Sasan

Abstract: In this paper, we propose Code-Bridged Classifier (CBC), a framework for making a Convolutional Neural Network (CNNs) robust against adversarial attacks without increasing or even by decreasing the overall models' computational complexity. More specifically, we propose a stacked encoder-convolutional model, in which the input image is first encoded by the encoder module of a denoising auto-encoder… ▽ More In this paper, we propose Code-Bridged Classifier (CBC), a framework for making a Convolutional Neural Network (CNNs) robust against adversarial attacks without increasing or even by decreasing the overall models' computational complexity. More specifically, we propose a stacked encoder-convolutional model, in which the input image is first encoded by the encoder module of a denoising auto-encoder, and then the resulting latent representation (without being decoded) is fed to a reduced complexity CNN for image classification. We illustrate that this network not only is more robust to adversarial examples but also has a significantly lower computational complexity when compared to the prior art defenses. △ Less

Submitted 16 January, 2020; originally announced January 2020.

Comments: 6 pages, Accepted and to appear in ISQED 2020

Robust Wiener filter based time gating method for detection of shallow buried objects

Authors: Ali Gharamohammadi, Fereidoon Behnia, Arash Shokouhmand

Abstract: In detecting shallow buried underground objects, reflected power from ground, i.e. ground surface clutter makes the task extremely difficult. In order to remove ground clutter, conventional methods in the literature are not as much effective as we need for objects buried detection in shallow depths. In this paper, a robust method, based on Time gating and wiener filtering, is proposed, which is ve… ▽ More In detecting shallow buried underground objects, reflected power from ground, i.e. ground surface clutter makes the task extremely difficult. In order to remove ground clutter, conventional methods in the literature are not as much effective as we need for objects buried detection in shallow depths. In this paper, a robust method, based on Time gating and wiener filtering, is proposed, which is very precise and effective in Ultra wideband (UWB) imaging. The problem with time gating method solely is that the timing window length for unknown target depths cannot be determined beforehand with sufficient accuracy. Imprecise window length selection removes parts of target signals along with the clutter and increases missed detection probability. This paper proposes an algorithm to circumvent this problem by first using a wiener filter for cancellation of ground clutter to a reasonable extent and pre detection of target positions by average similarity function (ASF). The time gating method is then used in the second step using the information provided from the first step for window length selection. The combination of the two steps provides better detection of shallow buried objects with less missed detection of targets. △ Less

Submitted 17 October, 2019; originally announced October 2019.

Machine learning based identification of buried objects using sparse whitened NMF

Authors: Ali Gharamohammadi, Fereidoon Behnia, Arash Shokouhmand

Abstract: In this paper, a whitening-based algorithm has been applied to sparse non-negative matrix factorization (NMF) as a preprocessing practice enhancing the identification of buried object, significantly. In fact, without utilizing a suitable whitening algorithm, the input signals are very similar to one another. Therefore, conventional approaches are not able to identify and properly discriminate desi… ▽ More In this paper, a whitening-based algorithm has been applied to sparse non-negative matrix factorization (NMF) as a preprocessing practice enhancing the identification of buried object, significantly. In fact, without utilizing a suitable whitening algorithm, the input signals are very similar to one another. Therefore, conventional approaches are not able to identify and properly discriminate desired buried objects. A suitable whitening algorithm makes the covariance matrix diagonal, such that the processed signals highly resembles the unprocessed versions. Such a similarity would be helpful to identify new received signal. In this study, we have employed the zero-phase component analysis (ZCA) as our whitening method. Hence, although signals of different targets are very similar in many cases, the mentioned advantages, achieved by whitening method, make identification of targets much easier. △ Less

Submitted 16 October, 2019; originally announced October 2019.

Optimum window length of Savitzky-Golay filters with arbitrary order

Authors: Mohammad Sadeghi, Fereidoon Behnia

Abstract: One of the widely used denoising methods in different domains is the Savitzky-Golay (SG) filter. The SG filter has two design parameters: window length and the filter order. As the length of the window increases, the estimation variance decreases, but the bias error increases at the same time. Mean square error (MSE) measure includes both bias and variance criteria. In this paper, we obtain the op… ▽ More One of the widely used denoising methods in different domains is the Savitzky-Golay (SG) filter. The SG filter has two design parameters: window length and the filter order. As the length of the window increases, the estimation variance decreases, but the bias error increases at the same time. Mean square error (MSE) measure includes both bias and variance criteria. In this paper, we obtain the optimal window length of an SG filter with arbitrary order which minimizes the MSE. To achieve the optimal window length, we propose an algorithm whose performance is better than the existing methods. In this paper, we follow the viewpoint proposed by Persson and Strang and design the filter on the basis of Chebyshev orthogonal polynomials △ Less

Submitted 30 August, 2018; originally announced August 2018.

NLOS Mitigation Using Sparsity Feature And Iterative Methods

Authors: Abbas Abolfathi, Fereidoon Behnia, Farokh Marvasti

Abstract: Well-known methods are employed to localize mobile station (MS) using line of sight (LOS) measurements. These methods may result in large error if they are fed with non LOS (NLOS) measurements. Our proposed algorithm, referred to as Sparse Recovery of NLOS using IMAT (SRNI), considers NLOS as unknown variables and solves the resultant underdetermined system emphasizing on its sparsity feature base… ▽ More Well-known methods are employed to localize mobile station (MS) using line of sight (LOS) measurements. These methods may result in large error if they are fed with non LOS (NLOS) measurements. Our proposed algorithm, referred to as Sparse Recovery of NLOS using IMAT (SRNI), considers NLOS as unknown variables and solves the resultant underdetermined system emphasizing on its sparsity feature based on IMAT methods. Simulations are conducted to investigate the performance of SRNI in comparison of other conventional algorithms. Results demonstrate that SRNI is fast enough to deal with large combination of BSs and also accurate in lower number of BSs △ Less

Submitted 19 March, 2018; originally announced March 2018.

Comments: 6 pages, 7 figures