Search

We present a Software Defined Radio (SDR) implementation of a Ground Penetrating Radar which uses much larger bandwidth than the SDR instantaneous RF bandwidth in order to increase radar resolution. It is based on the Stepped Frequency principle, where the baseband signal is an Orthogonal Frequency Division Multiplexing (OFDM) signal, transmitted in different subbands by modulating the RF stepped frequencies. Using OFDM in each subband provides faster signal transmission and simpler detection. The receive signal-to-noise ratio (SNR) is improved by repeating a number of times the transmit baseband signal before modulating each RF frequency. The random phase discontinuities between the adjacent subbands that appear when the RF frequency is changed, are removed by postprocessing the received signal. Singular value decomposition is used to remove the effects of the direct and the ground reflected waves. We perform field experiments with anti-tank mines buried in the ground. The resulting B-scans provide excellent detection and localization of the buried objects.

We present an integration of elevation correction algorithm with Singular Value Decomposition (SVD) based surface reflected wave removal algorithm for ground penetrating radar (GPR) systems positioned above ground with elevation variation. The GPR system and the experimental environment are simulated using the gprMax open-source software. The simulated scenario can be used for GPR mounted on Unmanned Aerial Vehicle (UAV), e.g. drone. The results show excellent reduction of both the effect of terrain with elevation variation and the effect of the surface reflected wave, increasing the probability of object detection, especially when the targets are low-reflective objects. The improved detection capability is tested using Faster-RCNN network, which shows the effectiveness of the method.

We investigate an implementation of a compressive sampling (CS) stepped frequency ground penetrating radar. Due to the small number of targets, the B-scan is represented as a sparse image. Due to the nature of stepped frequency radar, smaller number of random frequencies can be used to obtain each A-scan (sparse delays). Also, the measurements obtained from different antenna positions can be reduced to a smaller number of random antenna positions. We also use the structure in the B-scan, i.e. the shape of the targets, which can be known, for instance, when detecting land mines. We demonstrate our method using radar data available from the Web from the land mine targets buried in the ground. We use group sparsity, i.e. we assume that the targets have some non-zero (and presumably known) dimension in the cross-range coordinate of the B-scan. For such targets, we also use the Markov chain model for the targets, where we simultaneously estimate the model parameters using the EMturboGAMP algorithm. Both approaches result in improved performance.

The main goal of this project is to demonstrate the advantages of sensor integration on a remotely controlled robotic platform for increasing operator safety and improving the classification of explosive targets. This is accomplished by combining the imaging provided by radars and an optoelectronic sensor, a time-of-flight (ToF) depth camera. An additional aim is to demonstrate the operability and practicality of the system in a field with landmine simulants having plastic cases.

As of 2017, there are an estimated 100 million abandoned land mines littered across 61 countries. Following the wars in Afghanistan, Libya, Syria, Yemen, and Ukraine, there has been a rise in casualties due to the triggering of previously-abandoned explosive devices. The above institutions combined specialties to develop a remotely-operable, multisensor, robotic device for the detection of land mines, UXO (1), and IEDs (2). The robotic detection device uses novel subsurface radar with imaging and target classification to differentiate between threatening landmines and innocuous clutter. The expected outcome of this research is the creation of a multi-sensor system on a semi-autonomous vehicle for detection and discrimination of explosive devices. This robotic platform will change the approach to detecting landmines in post-war zones, all the meanwhile lessening direct human-to-mine or animal-to-mine contact in detecting landmines. Architecture has a high potential to move the procedures of explosive detection toward completely remote and autonomous system.

We consider the problem of estimation of sparse time dispersive Single Input Multiple Output (SIMO) channels, using a single transmit and multiple receive antennas in pilot aided OFDM systems. The channels we consider are with a continuous time delays and sparse, and we assume a common support of the channel coefficients of the SIMO channels associated with different antennas, resulting from the same scatterer. To exploit these properties, we propose a new channel estimation algorithm based on the atomic norm minimization for the Multiple Measurement Vector (MMV) model. A joint estimation of the delays corresponding to the same scatterer is obtained using the combination of the atomic norm regularized minimization for the MMV model and the MUSIC method. Then, based on the availability of the channel correlation information, the path gains are estimated using the LS or the MMSE method. Additionally, we derive a theoretical estimate of the channel estimate Mean Square Error for the asymptotically increasing number of receive antennas. To evaluate the proposed algorithm, we compare its performance with other state of the art algorithms.

A solution for automatic detection and classification of buried objects by implementing Faster Region Convolutional Neural Network (Faster R-CNN) with Ground Penetrating Radar (GPR) system is presented. Specifically, Faster R-CNN Inception-v2 was chosen, as a compromise between computational load and accuracy, compared with other Faster R-CNNs. Although the solution is general, in the sense that it can be retrained for arbitrary number of classes, here we focus on the discrimination between anti-tank (AT) mines signatures and standard hyperbolic signatures obtained from other objects, including anti-personnel (AP) mines. The image dataset used for training and testing the R-CNN network consists of GPR B-scans obtained both by gprMax based simulations and from real measured GPR data. The method performance is evaluated using Confusion matrices and ROC curves. Post processing approach based on object size and depth below ground surface enables discrimination of AP mines.

Regenerating codes based on the approach of interference alignment for wireless interference channel achieve the cut-set bound for distributed storage systems. These codes provide data reliability, and perform efficient exact node repair when some node fails. Interference alignment as a concept is especially important to improve the repair efficiency of a failed node in a minimum storage regenerating (MSR) code. In addition it can improve the stored data security in presence of passive intruders. In this paper we construct a new code resilient against a threat model where a passive eavesdropper can access the data stored on a subset of nodes and the downloaded data during the repair process of a subset of failed nodes. We achieve an optimal secrecy capacity for the new explicit construction of MSR interference alignment code. Hence, we show that the eavesdropper obtains zero information from the original message stored across the distributed storage, and that we achieve a perfect secrecy., Comment: 20 pages, 3 figures

We analyze the asymptotic capacity of an OFDM system with pilot aided channel estimation over a Bernoulli-Gaussian sparse channel, with Lasso compressed sensing (CS) used for channel estimation. In the analysis, we utilize the CS asymptotic performance similarity when using Haar and DFT measurement matrices. We evaluate the mean square estimation error of an augmented Lasso estimator using the replica method results and then use it to obtain an asymptotic capacity lower bound for the OFDM system. For the considered system with equi-power pilot symbols, we optimize the average fraction of pilot subcarriers used for channel estimation and the pilot to data power ratio, given an average symbol power per subcarrier, and evaluate the asymptotic capacity bound increase due to CS.

We implement a software defined stepped frequency radar using a USRP. Stepped frequency radar suffers from large data collection time due to large number of pulses with different frequencies sent. Our approach uses a baseband stepped frequency signal, which is upconverted to RF, and, then, frequency hopping the RF frequency. In this way, a low cost hardware with moderate instantaneous RF bandwidth can be used. The total time for hopping all the frequencies, i.e. the data collection time, is also reduced. Resolution can be improved many-fold, depending on the USRP total frequency range.

We propose a new direction of arrival estimation method for wideband uncorrelated signals. The wideband signals are first decomposed into narrowband signals. A group sparse Lasso formulation is proposed that jointly fits the powers of the signals using overcomplete dictionaries of the directions of arrival, into the estimated covariance matrices for all narrowband signals. Then, we propose a new formulation that determines the regularization parameters and becomes the group Lasso formulation. Additionally, we propose a modified algorithm for direction of arrival estimation with lower complexity that uses conventional based methods in the preprocessing stage to reduce the number of variables in the optimization task. We compare the performance of the proposed method to the conventional methods for a circular antenna array.

We propose an improved capacity lower bound for OFDM system with compressed sensing channel estimation for Bernoulli-Gaussian channel. We improve the known capacity lower bound which is based on Lasso compressed sensing channel estimation, by replacing the Lasso based estimate with an MMSE estimate, known to be optimal in the MMSE sense and achievable with practical algorithms for a broad range of system parameters setup. Additionally, for the system with equi-powered pilot subcarriers we optimize the capacity lower bound by finding the optimal average fraction of pilot subcarriers used for channel estimation and optimal pilot to data power ratio given the average symbol power per subcarrier, and propose an optimization procedure with polynomial complexity.

We analyze the use of discrete nonseparable shearlet transform for sparse representation of the MRI image. This transform is a non-tight frame, but has excellent localization properties in the spatial domain. Additionally, we exploit the frequency representation of the canonical dual filters of this transform and combine it with the measurement process, carried out in the Fourier domain, to implement the split Bregman algorithm for reconstruction. We compare the performance of the proposed approach to approaches using split Bregman algorithm with curvelets, wavelets and total variation.

We perform theoretical analysis of compressed sensing with Orthogonal Matching Pursuit (OMP) recovery. In the analysis we utilize an OMP simplification where the estimation step in each iteration is assumed to be perfect. We consider the case when the noise is AWGN, the signal is sparse with equal energy nonzero components and the measurement matrix has i.i.d. Gaussian entries. We estimate the probability of perfect support reconstruction. We then extend the analysis to the Multiple Measurement Vector case. The analysis sheds new light on the compressed sensing recovery and gives a close estimate of the OMP reconstruction performance.

We consider the problem of estimation of sparse time dispersive channels in pilot aided OFDM systems on Single Input Multiple Output (SIMO) channels, i.e. with a single transmit and multiple receive antennas. In such systems the channels are inherently continuous-time and sparse, and there is a common support of the channel coefficients of channels associated with different antennas, resulting from the same scatterer. To exploit these properties, we propose a new channel estimation algorithm that combines the atomic norm minimization of the Multiple Measurement Vector (MMV) model, the MUSIC and the least squares (LS) methods. The atomic norm minimization of the MMV model allows to exploit the common support assumption and the continuous-time nature of the channels, MUSIC allows for simple joint estimation of the delays corresponding to the same scatterer, and LS allows for estimation of the path gains. To evaluate the proposed algorithm, we compare its performance with the case when the common support assumption is not used.

Secure transmission in Multiple Input Multiple Output Multiple Eavesdropper (MIMOME) wiretap channels under quasi static Rayleigh fading is considered. Closed-form expression for the secrecy capacity of this type of channel is known only for some special cases. Therefore, we investigate the performance of two different suboptimal transmission methods under the assumption that the transmitter has full Channel State Information (CSI). In the first method the transmitter performs beamforming based on the Generalized Singular Value Decomposition (GSVD). In the second method the transmitter sends information along directions where there is gain to the legitimate receiver and artificial noise in all other directions. We determine the SNR regions where each of the methods outperforms the other.

As semiconductor technology continues to evolve, numerical modeling of semiconductor devices becomes an indispensible tool for the prediction of device characteristics. The simple drift-diffusion model is still widely used, especially in the study of subthreshold behavior in MOSFETs. The numerical solution of these two equations offers difficulties in small devices and special methods are required for the case when dealing with 3D problems that demand large CPU times. In this work we investigate the convergence properties of the Bi-CGSTAB method. We find that this method shows superior convergence properties when compared to more commonly used ILU and SIP methods.

The wide-sense stationary-uncorrelated scattering (WSSUS) channel model is a commonly employed model for the multipath channel experienced in mobile communications. The second order statistics of these channels are described by the delay cross-power density /spl phi//sub h/(/spl tau/;/spl Delta/t) or by its /spl Delta/t-Fourier transform, the scattering function S/sub h/(/spl tau/;/spl lambda/). This paper presents an analysis of the delay cross-power density and scattering functions for mobile communications channels. We assume an arbitrary spatially uncorrelated scattering (US) field with arbitrary propagation-loss factors. Our first result is a general integral expression for /spl phi//sub h/(/spl tau/;/spl Delta/t) that holds with both transmitter and receiver being mobile. We then derive more detailed results for the case of a stationary base station. We derive an infinite Bessel series for /spl phi//sub h/(/spl tau/;/spl Delta/t) and a closed-form expression for S/sub h/(/spl tau/; /spl lambda/). These results generalize the well-known classical approximation for the time-correlation function /spl phi/~/sub h/(/spl Delta/t)=/sup def//spl int//spl phi//sub h/(/spl tau/;/spl Delta/t)d/spl tau//spl ap/ J/sub 0/(2/spl pi//spl lambda//sub m//spl Delta/t), which corresponds to the zeroth term of our Bessel series.

We propose a new direction of arrival estimation method for wideband uncorrelated signals. The wideband signals are first decomposed into narrowband signals. A group sparse Lasso formulation is proposed that jointly fits the powers of the signals using overcomplete dictionaries of the directions of arrival, into the estimated covariance matrices for all narrowband signals. Then, we analyze the problem formulation and propose a new formulation that determines the regularization parameters and becomes the group Lasso formulation. We compare the performance of the proposed method to the conventional methods.

We compare in a unified way the outage probability performance of the three cooperative strategies: decode-and-forward (DF), compress-and-forward (CF) and amplify-and-forward (AF). The strategies are treated in the slow Rayleigh fading relay channel under different channel conditions for both full-duplex and half-duplex modes of operation. Perfect channel state information (CSI) is assumed to be available at the transmitter and the relay. The systematic comparison reveals the cases where the use of certain cooperative strategy is beneficial.

We propose a joint source channel coding framework for real time transmission of H.264/AVC single layer video signals over wireless MIMO channels. In the proposed framework the different layers of the communication system exchange information with central controller which decides on the parameters that should be used within different layers to optimize the quality of the received video. The MIMO channel is characterized by the probability density function of its mutual information and the performance of the H.264/AVC encoder is modeled by a rate distortion model that takes into account the channel errors. The framework allows for the video encoder adaptation, given the current video frame, to minimize the total distortion taking into account the available system resources and channel statistics. We show the effectiveness of the proposed framework in video transmission over MIMO Rayleigh fading channels, when channel distribution information is available at the transmitter.

We propose a new indirect method for direction of arrival estimation of a bandlimited communication signal based on compressed sensing. The analyzed signal in our system is sparse in the frequency domain. Also, the exact carrier frequency of the communication signal is unknown but the frequency range where it belongs is known and assumed to be narrowband compared to the carrier frequency. The signal is converted to baseband using the center frequency of the observed frequency range and then sampled using random sampling in time and circularly from different antennas. The samples are compensated according to an assumed direction of arrival and then reconstructed using reconstruction algorithms from compressed sensing framework. Based on the concentration of the signal energy in the frequency domain a decision on the correctness of the assumed direction is made. A grid of assumed directions of arrival over the entire direction of arrival space is applied in order to find the best candidate. Simulations are carried out to evaluate the performance of the method.

This paper investigates the ergodic rates of three cooperative strategies for the fast Rayleigh fading relay channel, namely, decode-andforward (DF), compress-and-forward (CF) and amplify-and-forward (AF). The transmitter-relay, relay-receiver and transmitter-receiver links experience independent fast Rayleigh fading. Our analysis considers both fullduplex and half-duplex modes of operation of the relay channel, assuming that perfect transmit channel distribution information (CDI) is available at the transmitter and the relay. The ergodic rates are compared under different channel conditions to the corresponding upper capacity bounds of the relay channel and to the rate of direct transmission in order to investigate when it is beneficial to use a certain cooperative strategy. Cooperative transmission makes the ergodic rates approach or even become identical to the upper capacity bounds if some conditions are satisfied. The results show that DF and CF outperform AF as they are more complex strategies, while AF itself achieves higher rates than direct transmission in most channel scenarios, thus trading performance for implementation simplicity., ACM Computing Classification System (1998): E.4, C.2.1.