Engineering topics
Publications and source records attributed to Yeh, Pen-Shu.
Report presents analysis of performances of conceptual Rice universal noiseless coders designed to provide efficient compression of data over wide range of source-data entropies. Includes predictive preprocessor that maps source data into sequence of nonnegative integers and variable-length-coding processor, which adapts to varying entropy of source data by selecting whichever one of number of optional codes yields shortest codeword.
Report discusses some algorithms for fast, high-performance universal noiseless coder adaptively generating nearly optimal variable-length codes. Coders based on principles discussed in reports described in "Some Practical Universal Noiseless Coding Techniques" (NPO-18833) and "Optimality of Variable-Length Codes" (NPO-18834). Coding algorithms of present report independently implemented in custom-made very-large-scale integrated-circuit coding modules.
Improved method of encoding digital data provides for efficient lossless compression of partially or even mostly redundant data from low-information-content source. Method of coding implemented in relatively simple, high-speed arithmetic and logic circuits. Also increases coding efficiency beyond that of established Huffman coding method in that average number of bits per code symbol can be less than 1, which is the lower bound for Huffman code.
Lossless data compression was studied for many NASA missions. The Rice algorithm was demonstrated to provide better performance than other available techniques on most scientific data. A top-level description of the Rice algorithm is first given, along with some new capabilities implemented in both software and hardware forms. Systems issues important for onboard implementation, including sensor calibration, error propagation, and data packetization, are addressed. The latter part of the guide provides twelve case study examples drawn from a broad spectrum of science instruments.
Data compression has been proposed for several flight missions as a means of either reducing on board mass data storage, increasing science data return through a bandwidth constrained channel, reducing TDRSS access time, or easing ground archival mass storage requirement. Several issues arise with the implementation of this technology. These include the requirement of a clean channel, onboard smoothing buffer, onboard processing hardware and on the algorithm itself, the adaptability to scene changes and maybe even versatility to the various mission types. This paper gives an overview of an ongoing effort being performed at Goddard Space Flight Center for implementing a lossless data compression scheme for space flight. We will provide analysis results on several data systems issues, the performance of the selected lossless compression scheme, the status of the hardware processor and current development plan.
This paper first provides an overview of an adaptive, lossless, data compression algorithm originally devised by Rice in the early '70s. It then reports the development of a VLSI encoder/decoder chip set developed which implements this algorithm. A recent effort in making a space qualified version of the encoder is described along with several enhancements to the algorithm. The performance of the enhanced algorithm is compared with those from other currently available lossless compression techniques on multiple sets of test data. The results favor our implemented technique in many applications.
A VLSI architecture for an adaptive data compression encoder capable of sustaining fixed or variable bit-rate output has been developed. There are three modes of operation: lossless with variable bit-rate, lossy with fixed bit-rate and lossy with variable bit-rate. For lossless encoding, the implementation is identical to the USES chip designed for Landsat 7. Obtaining a fixed bit-rate is achieved with a lossy DPCM algorithm using adaptive, nonuniform scalar quantization. In lossy mode, variable bit-rate coding uses the lossless sections of the encoder for post-DPCM entropy coding. The encoder shows excellent compression performance in comparison to other current data compression techniques. No external tables or memory are required for operation.
Rice developed a universal noiseless coding structure that provides efficient performance over an extremely broad range of source entropy. This is accomplished by adaptively selecting the best of several easily implemented variable length coding algorithms. Variations of such noiseless coders have been used in many NASA applications. Custom VLSI coder and decoder modules capable of processing over 50 million samples per second have been fabricated and tested. In this study, the first of the code options used in this module development is shown to be equivalent to a class of Huffman code under the Humblet condition, for source symbol sets having a Laplacian distribution. Except for the default option, other options are shown to be equivalent to the Huffman codes of a modified Laplacian symbol set, at specified symbol entropy values. Simulation results are obtained on actual aerial imagery over a wide entropy range, and they confirm the optimality of the scheme. Comparison with other known techniques are performed on several widely used images and the results further validate the coder's optimality.
This paper provides the basic algorithmic definitions and performance characterizations for a high-performance adaptive noiseless (lossless) 'coding module' which is currently under separate developments as single-chip microelectronic circuits at two NASA centers. Laboratory tests of one of these implementations recently demonstrated coding rates of up to 900 Mbits/s. Operation of a companion 'decoding module' can operate at up to half the coder's rate. The functionality provided by these modules should be applicable to most of NASA's science data. The hardware modules incorporate a powerful adaptive noiseless coder for 'standard form' data sources (i.e., sources whose symbols can be represented by uncorrelated nonnegative integers where the smaller integers are more likely than the larger ones). Performance close to data entries can be expected over a 'dynamic range' of from 1.5 to 12-15 bits/sample (depending on the implementation). This is accomplished by adaptively choosing the best of many Huffman equivalent codes to use on each block of 1-16 samples. Because of the extreme simplicity of these codes no table lookups are actually required in an implementation, thus leading to the expected very high data rate capabilities already noted.
The algorithmic definitions and performance characterizations are presented for a high performance adaptive coding module. Operation of at least one of these (single chip) implementations is expected to exceed 500 Mbits/s under laboratory conditions. Operation of a companion decoding module should operate at up to half the coder's rate. The module incorporates a powerful noiseless coder for Standard Form Data Sources (i.e., sources whose symbols can be represented by uncorrelated non-negative integers where the smaller integers are more likely than the larger ones). Performance close to data entropies can be expected over a Dynamic Range of from 1.5 to 12 to 14 bits/sample (depending on the implementation).
A universal noiseless coding structure was developed that provides efficient performance over an extremely broad range of source entropy. This is accomplished by adaptively selecting the best of several easily implemented variable length coding algorithms. Custom VLSI coder and decoder modules capable of processing over 20 million samples per second are currently under development. The first of the code options used in this module development is shown to be equivalent to a class of Huffman code under the Humblet condition, other options are shown to be equivalent to the Huffman codes of a modified Laplacian symbol set, at specified symbol entropy values. Simulation results are obtained on actual aerial imagery, and they confirm the optimality of the scheme. On sources having Gaussian or Poisson distributions, coder performance is also projected through analysis and simulation.
Addressed in this paper is the dependence of telecommunication channel, forward error correcting coding and source data compression coding on integrated circuit technology. Emphasis is placed on real time high speed Reed Solomon (RS) decoding using full custom VLSI technology. Performance curves of NASA's standard channel coder and a proposed standard lossless data compression coder are presented.
The process of finding the coordinate transformation between a robot and an external sensor system has been addressed. This calibration is equivalent to solving a nonlinear optimization problem for the parameters that characterize the transformation. A two-step procedure is herein proposed for solving the problem. The first step involves finding a nominal solution that is a good approximation of the final solution. A varational problem is then generated to replace the original problem in the next step. With the assumption that the variational parameters are small compared to unity, the problem that can be more readily solved with relatively small computation effort.
The data compression performance of the Hartley transform on a Markov-1 signal is theoretically compared to that of the Fourier transform. Covariance distribution and residue correlation measurements have been computed for the Hartley, Fourier, and cosine transforms. The Hartley transform achieves better coding performance than the Fourier transform, but is inferior to the cosine transform.