Luis Tirado

Consensus-based imaging using ADMM for a Compressive Reflector Antenna

This paper describes a novel norm-one-regularized, consensus-based imaging algorithm, based on the Alternating Direction Method of Multipliers (ADMM), that can be used by a high-sensing-capacity Compressive Reflector Antenna (CRA). The proposed method outperforms current state of the art iterative reconstruction algorithms in terms of computational cost; and it ultimately enables the use of a CRA in quasi-real-time, compressive sensing imaging applications.

Inverse Fast Multipole Method for Monostatic Imaging Applications

A 3-D imaging technique for monostatic radar cross section measurement is presented. The method is based on an existing formulation conceived for bistatic geometries which takes into account the scattered field polarization and is accelerated using the fast multipole method. The synthetic aperture radar (SAR) image processing includes acquired field, which is of interest for polarimetric SAR imaging. In addition, the proposed technique overcomes the limitation of fast-Fourier-transform-based inverse techniques on canonical domains (planar, cylindrical, and spherical). Validation with simulations and measurements is presented. Calculation time is also drastically reduced using graphics processing unit implementation.

Norm-1 Regularized Consensus-based ADMM for Imaging with a Compressive Antenna

This letter presents a novel norm-1-regularized, consensus-based imaging algorithm, based on the alternating direction method of multipliers (ADMM). This algorithm is capable of imaging metallic targets by using a limited amount of data. The distributed capabilities of the algorithm enable a fast imaging convergence. Recently, a compressive reflector antenna (CRA) has been proposed as a way to provide high sensing capacity with a minimum cost and complexity in the hardware architecture. The ADMM algorithm applied to the imaging capabilities of the CRA outperforms current state-of-the-art iterative reconstruction algorithms, such as Nesterov-based methods, in terms of computational cost, enabling the use of the CRA in quasi-real-time, compressive sensing imaging applications.

Ray Tracing for Simulation of Millimeter-Wave Whole Body Imaging Systems

A ray tracing algorithm for modeling millimeter waves in a whole body imaging system is presented. Ray tracing is a well-known method for approximating high-frequency wave behavior and is well suited for implementation on graphics processing units (GPUs), presenting computational speed advantages over conventional full-wave modeling techniques. This method leverages the NVIDIA OptiX engine to ensure computational efficiency. Numerical results in this work are compared with conventional two-dimensional method of moments solutions to assess accuracy and computational times are compared with a three-dimensional GPU implementation of the modified equivalent current approximation.

A GPU implementation of the inverse fast multipole method for multi-bistatic imaging applications

This paper presents a parallel implementation of the Inverse Fast Multipole Method (IFMM) for multi-bistatic imaging configurations. NVIDIAs Compute Unified Device Architecture (CUDA) is used to parallelize and accelerate the imaging algorithm in a Graphics Processing Unit (GPU). The algorithm is validated with experimental data, collected by a Frequency-Modulated Continuous Wave (FMCW) radar system operating in the 70-77 GHz frequency band. The proposed GPU-based IFMM algorithm accelerated the single-core CPU version by a factor of 46.

Ray tracing simulation tool for portal-based millimeter-wave security systems using the NVIDIA® OptiX™ ray tracing engine

Summary form only given. Person-borne weapons and explosives present major security threats at civil infrastructures such as airports. Millimeter-wave whole body imaging systems are used to identify anomalous objects hidden underneath peoples clothing. Improvements to these systems can be investigated using electromagnetic computational models. Such forward models allow prediction of scattering from objects and give insight into the most effective sensor configurations and imaging techniques. Conventional full-wave simulation methods such as Finite Difference Frequency Domain and the Method of Moments provide extremely accurate solutions but are usually slow, ranging from minutes to hours in 3D depending on the domain size and frequency range. Less accurate but faster methods include physical optics and the Modified Equivalent Current Approximation (MECA). Simulating a portal-based scanning system involves evaluating the scattered field from the human body at multiple frequencies and at multiple receiver points, which is computationally expensive when using these methods. This work presents a fast forward ray tracing simulation tool. Rays are launched in the direction of a simulated tessellated human body and traced until reaching simulated system receivers. The scattered field at these receivers is computed by using the pathlength phase of each ray and adding ray fields within finite size receiver elements. Although the algorithm presented in this work excludes the modeling of phenomena such as refraction and diffraction, the path distance information encoded in reflected rays gives acceptable accuracy in scattered field amplitude and phase data and for images processed with SAR methods. The figure shows reflectivity magnitude of 2D SAR of a human body, using field data generated from ray tracing (left) and a 2D Method of Moments (right). The green contour is the true transverse body profile with torso and arms.

Coherent image formation and calibration for multi-bistatic radar configurations

This paper presents a new calibration algorithm for coherent image formation in a multi-bistatic radar imaging system. The algorithm uses two spheres as calibration targets. Preliminary results show that positional errors larger than one wavelength can be corrected, thus improving the performance of current-state-of-the art algorithms like the Iterative Field Matrix (IFM) method [4].