AFIT neural network research

A brief summary of research done at the Air Force Institute of Technology (AFIT) in the area of neural networks is provided. It has been shown that backpropagation, used for feedforward artificial neural networks, is just a degenerate version of an extended Kalman filter, and that networks can do about as well as the optimum statistical classification technique. A method of finding the importance of features for use by a neural network classifier has been determined. Techniques for using neural networks for image segmentation have been developed. In optical pattern recognition, techniques that allow the processing of real FLIR (forward-looking infrared) images with existing binary spatial light modulators have been devised. An optical direction of arrival detector applicable to laser illumination direction determination has been designed and tested; the design is similar to a fly's eye. Coated mirrors for the optical confocal Fabry-Perot interferometer have been designed, specified, fabricated, and installed. Significant progress has been made in the use of neural networks for processing multiple-feature sets for speech recognition     

A wavelet multiresolution analysis for spatio-temporal signals

The wavelet filters of the conventional 3D multiresolution analysis possess homogeneous spatial and temporal frequency characteristics which limits one's ability to match filter frequency characteristics to signal frequency behavior. Also, the conventional 3D multiresolution analysis employs an oct-tree decomposition structure which restricts the analysis of signal details to identical resolutions in space and time. This paper presents a 3D wavelet multiresolution analysis constructed from nonhomogeneous spatial and temporal filters, and an orthogonal sub-band coding scheme that decouples the spatial and temporal decomposition processes     

Design and implementation of near-field, wideband synthetic aperture beamformers

A coarray-based near-field, wideband synthetic aperture beamformer using stepped-frequency signal synthesis and post-data acquisition processing is presented. While coarray techniques offer significant reduction in the number of array elements for a given angular resolution, the hybrid subarray-stepped frequency realization of wideband systems simplifies implementations and offers flexibility in beamforming. Proof of concept is provided using real data collected in an anechoic chamber for several pulse shapes and array weightings.     

Synthetic-aperture radar processing using fast factorized back-projection

Exact synthetic aperture radar (SAR) inversion for a linear aperture may be obtained using fast transform techniques. Alternatively, back-projection integration in time domain can also be used. This technique has the benefit of handling a general aperture geometry. In the past, however, back-projection has seldom been used due to heavy computational burden. We show that the back-projection integral can be recursively partitioned and an effective algorithm constructed based on aperture factorization. By representing images in local polar coordinates it is shown that the number of operations is drastically reduced and can be made to approach that of fast transform algorithms. The algorithm is applied to data from the airborne ultra-wideband CARABAS SAR and shown to give a reduction in processing time of two to three orders of magnitude.     

RPC-MAG The Fluxgate Magnetometer in the ROSETTA Plasma Consortium

The fluxgate magnetometer experiment onboard the ROSETTA spacecraft aims to measure the magnetic field in the interaction region of the solar wind plasma with comet 67P/Churyumov-Gerasimenko. It consists of a system of two ultra light (about 28 g each ) triaxial fluxgate magnetometer sensors, mounted on the 1.5 m long spacecraft boom. The measurement range of each sensor is ±16384 nT with quantization steps of 31 pT. The magnetometer sensors are operated with a time resolution of up to 0.05 s, corresponding to a bandwidth of 0–10 Hz. This performance of the RPC-MAG sensors allows detailed analyses of magnetic field variations in the cometary environment. RPC-MAG furthermore is designed to study possible remnant magnetic fields of the nucleus, measurements which will be done in close cooperation with the ROSETTA lander magnetometer experiment ROMAP.     

A method for constructing an airfoil using a specified chord velocity diagram

A problem of determining a shape of the airfoil being streamlined by a potential incompressible inviscid flow is solved by the successive approximation method using a specified chord velocity diagram. It is shown that a closed airfoil that possesses a specified chord velocity diagram can be constructed with a sufficient accuracy; if the chord diagram is unsuccessfully specified, the closed airfoil may prove to be not univalent, that is, physically unrealizable.     

Two Methods of Spectral Estimation for SARSAT Signals

Search and rescue satellite aided tracking (SARSAT) depends on the processing of emergency locator transmitter (ELT) signals which are received by a satellite in low polar orbit. Since the signal from a distressed vehicle is normally immersed in a background of other ELT signals (false alarms), interference, and noise, different methods of spectral estimation can provide advantages in estimating carrier frequency. A comparison between average spectrum and minimum spectrum for several real signals is provided here.     

Subspace array processing for the suppression of FM jamming in GPS receivers

The mitigation of FM interference in GPS receivers is considered. In difference to commonly assumed wideband and narrowband interferers, the FM interferers are wideband, but instantaneously narrowband, and as such, have clear time-frequency (TF) signatures that are distinct from the GPS coarse acquisition (C/A) spread spectrum code. In the proposed technique, the estimate of the FM interference instantaneous frequency (IF) and the interference spatial signature are used to construct the spatiotemporal interference subspace. The IF estimates can be provided using existing effective linear or bilinear TF methods. The undesired signal arrival is suppressed by projecting the input data on the interference orthogonal subspace. With a multisensor receiver, the distinctions in both the spatial and TF signatures of signal arrivals allow effective interference suppressions. The deterministic nature of the signal model is considered and the known underlying structure of the GPS C/A code is utilized. We derive the receiver signal-to-interference-plus-noise ratio (SINR) under exact and perturbed IF values. The effect of IF estimation errors on both pseudorange measurements and navigation data recovery is analyzed. Simulation results comparing the receiver performances under IF errors in single and multiantenna GPS receivers are provided.     

Microdosimetric measurements of heavy ion tracks.

The biological effectiveness of radiations depends on the spatial pattern of ionizations and excitations produced by the charged particle tracks involved. Ionizations produced by both the primary ion and by energetic delta rays may contribute to the production of biologically relevant damage and to the concentration of damage which may effect the probability of repair. Although average energy concentration (dose) can be calculated using homogeneous track models, the energy is actually concentrated in small volumes containing segments of the ion and delta ray tracks. These local concentrations are studied experimentally using low pressure proportional counters, and theoretically, using Monte Carlo methods. Small volumes near an ion track may be traversed by a delta ray. If they are, the energy deposited will be similar to that produced by a single electron track in a low-energy x-ray irradiation. The probability of a delta ray interaction occurring decreases with the square of the radial distance from the track. The average energy deposited is the product of this probability and the energy deposited in an interaction. Average energy deposited calculated from measured interaction probability is in good agreement with the results of homogeneous track models.     

Optimal number of phased array faces and signal processors for horizon surveillance

The problem of the optimal number of phased array faces for performing 360/spl deg/ horizon surveillance is considered. Assuming the detection performance is the same in all beam positions and the total number of T/R modules is constant, it is shown that the optimal number of array faces is three. This is true whether the arrays are operating simultaneously or sequentially. A parametric analysis is performed between the number of array faces operating simultaneously and the associated cost of simultaneous operation in terms of the size of the array.