Paul D. Fiore

Efficient Approximate Wordlength Optimization

In this paper, the problem of finding good wordlength combinations for fixed-point digital signal processing flowgraphs is addressed. By formulating and solving an approximate optimization problem, an estimated Pareto-optimal curve for attainable cost/quality combinations is rapidly calculated. This curve and the associated wordlength combinations are useful in several situations and can serve as starting points for real design searches. Examples that illustrate these concepts are given.

Design for HMM-based SAR ATR

This paper describes progress on the Automatic Target Recognition (ATR) system for Synthetic Aperture Radar (SAR) imagery. The system is based upon a feature extraction, data ordering, and statistical modeling paradigm. Feature extraction is performed by applying image segmentation to convert the SAR imagery into one of four pixel classes. A description of a real-time image segmentation design is given. The segmented imagery is re-ordered from a 2D spatial representation to a sequential representation through the use of multiple Radon Transforms. Finally, the re-ordered data is classified by target type by applying Hidden Markov Model decoding techniques. Performance results on the MSTAR public targets database is provided.

Constrained maximum likelihood solution of linear equations

The total least squares (TLS) is used to solve a set of inconsistent linear equations Ax/spl ap/y when there are errors not only in the observations y but in the modeling matrix A as well. The TLS seeks the least squares perturbation of both y and A that leads to a consistent set of equations. When y and A have a defined structure, we usually want the perturbations to also have this structure. Unfortunately, standard TLS does not generally preserve the perturbation structure, so other methods are required. We examine this problem using a probabilistic framework and derive an approach to determining the most probable set of perturbations, given an a priori perturbation probability density function. While our approach is applicable to both Gaussian and non-Gaussian distributions, we show in the uncorrelated Gaussian case that our method is equivalent to several existing methods. Our approach is therefore more general and can be applied to a wider variety of signal processing problems.

Image registration using both distance and angle information

Image registration is an important task in matching two or more images with unknown rotation, displacement, and scaling. A class of registration algorithms uses the location of feature points in the images, rather than pixel intensity values to perform the registration. An O(N/sup 3/) algorithm of this type that uses both relative distance and angle information is examined. The algorithm allows for missing and spurious feature points, and can provide scale mismatch information. It requires less computation than a previous method.

Low-Complexity Implementation of a Polyphase Filter Bank

This paper describes the tradeoffs and design method for a polyphase filter bank implemented in field programmable gate array technology. Consideration is given to the minimization of required hardware resources by careful choice of the FFT approach, multiplier implementation, and wordlength optimization. These choices are made subject to maintaining specified output performance. Finally, the sigma?delta technique is examined to further reduce the computational complexity.

Full-duplex communication via adaptive nulling

Simultaneous communication in the same frequency band is gaining interest as an approach to increasing spectral efficiency. This functionality is usually prevented by the close proximity of the transmit and receive antennas, resulting in interference at, or saturation of, the receiver electronics. Furthermore, it is of interest to minimize complexity of mobile units; a worst case is to assume that the mobile units are unable to cooperate in the channel estimation and beamforming processes. In this paper, we develop an uncooperative wide-band MIMO adaptive channel estimation and nulling approach for simultaneous transmit and receive on the same channel. Low-power training waveforms are used to estimate the time-varying MIMO interference channel. Transmit beamforming weights that span the null space of the interference channel are then calculated via an algebraic approach. Results are reported, which show null depths at the receive antennas of greater than 40dB in simulation and 30dB in a slow-update wide-band experiment.

Efficient Wordlength Reduction Techniques for DSP Applications

Intermediate results in digital signal processing (DSP) hardware frequently must be truncated or rounded to maintain reasonable wordlengths. Noise and bias are introduced into the signal due to these operations. For the addition operation, we investigate two methods which have reduced variance and bias and yet maintain the computational simplicity of truncation. Essentially, each method drives the least significant carry input of the adder string with a very simple boolean function. We demonstrate the utility of these approaches by calculating bias and variance reductions, and by using the methods in several simple but important DSP examples.

Systolic array for acceleration of template-based ATR

The performance of Hausdorff distance-based clutter/target discrimination is characterized. This allows selection of detection thresholds and pre-processing parameters to achieve the desired detection performance. An efficient systolic array implementation of Hausdorff distance-based target/clutter discrimination is presented.

Parallel multiplication using fast sorting networks

A recent paper describes the use of Svobodas binary counter in the construction of fast parallel multipliers. The resulting approach was shown to be faster than the conventional Dadda multiplier when the wordlength N was small. Unfortunately, the growth in the number of gates of that method was O(N/sup 3/) and the speed was O(N). In this paper, Batchers bitonic sorting network and other efficient networks replace the Svoboda counter. The asymptotic growth rate in gates of these new methods is O(N/sup 2/ log/sup 2/ N), and the speed is O(log/sup 2/ N).

Scale-Recursive Lattice-Based Multiple-Access Symbol Constellations

This paper describes scale-recursive symbol constellations for the synchronous multiple-access (MA) channel, which can be seen as a generalization of Voronoi coding. The constellations are related by a fixed scaling and rotation, and are designed so that in the noise-free case, the combined signal always falls onto an appropriate lattice. This allows high throughput and a linear complexity decoding algorithm. Novel constellations of up to 24 dimensions are given that show improved performance over conventional rectangular lattices.