Search

Space and Earth observation programs demand stringent guarantees ensuring smooth and reliable operations of space vehicles and satellites. Due to unforeseen circumstances and naturally occurring faults, it is desired that a fault-diagnosis system be capable of detecting, isolating, identifying, or classifying faults in the system. Unfortunately, none of the existing fault-diagnosis methodologies alone can meet all the requirements of an ideal fault-diagnosis system due to the variety of fault types, their severity, and handling mechanisms. However, it is possible to overcome these shortcomings through the integration of different existing fault-diagnosis methodologies. In this paper, a novel learning-based, diagnostic-tree approach is proposed which complements and strengthens existing efficient fault detection mechanisms with an additional ability to classify different types of faults to effectively determine potential fault causes in a subsystem of a satellite. This extra capability serves as a semiautomatic diagnostic decision support aid to expert human operators at ground stations and enables them to determine fault causes and to take quick and efficient recovery/reconfiguration actions. The developed diagnosis/analysis procedure exploits a qualitative technique denoted as diagnostic tree (DX-tree) analysis as a diagnostic tool for fault cause analysis in the attitude control subsystem (ACS) of a satellite. DX-trees constructed by our proposed machine-learning-based automatic tree synthesis algorithm are demonstrated to be able to determine both known and unforeseen combinations of events leading to different fault scenarios generated through synthetic attitude control subsystem data of a satellite. Though the immediate application of our proposed approach would be at ground stations, the proposed technique has potential for being integrated with causal model-based diagnosis and recovery techniques for future autonomous space vehicle missions.

The problem of segmentation of multispectral satellite images is addressed. An integration of rough-set-theoretic knowledge extraction, the Expectation Maximization (EM) algorithm, and minimal spanning tree (MST) clustering is described. EM provides the statistical model of the data and handles the associated measurement and representation uncertainties. Rough-set theory helps in faster convergence and in avoiding the local minima problem, thereby enhancing the performance of EM. For rough-set-theoretic rule generation, each band is discretized using fuzzy-correlation-based gray-level thresholding. MST enables determination of nonconvex clusters. Since this is applied on Gaussians, determined by grannies, rather than on the original data points, time required is very low. These features are demonstrated on two IRS-1A four-band images. Comparison with related methods is made in terms of computation time and a cluster quality measure. Index Terms--Clustering, granular computing, minimal spanning tree, mixture modeling, rough knowledge encoding.

To provide plasma-physics scientists with worldwide access to historical and current measurement data from the Textor-94 Tokamak experiment, we developed a new architecture: the 'WebUmbrella.' It uses HTTP to transfer data in (X)HTML, plain text, or other MIME-type documents. We provide a single point of access to our servers via HTTP redirection. Server implementations for several data stores are already present, and the first user tests are positive. In particular, we present one server implementation that provides access to already existing common object request broker architecture (CORBA) servers. It uses the Apache Web server and the Tomcat Servlets engine, and consists of Java Servlets that convert HTTP requests into CORBA requests for the existing interface. These developments are the aftermath of an earlier experience with CORBA interface definition language (IDL) interfaces between user clients and several data servers, which provide access to the different data stores. The interfaces were developed in a telematics application program project, 'Dynacore.' After the evaluation of this demonstrator, we designed a new system to improve the service. Because of the positive experience, we inherited the data representation, the data-browsing features, and the single access point for clients to several data stores. We tried to eliminate the obligatory use of CORBA, as its added value did not weigh up against the problems we faced using it: CORBA is difficult to understand for scientists inexperienced in object-oriented design and difficult to access with simple analysis tools like IDL or MATLAB. We also did not find a suitable CORBA implementation for Virtual Address eXtension/Virtual Memory System (VAX/VMS) with C/Fortran bindings. Finally, WebUmbrella's transport protocol, contrary to CORBA's, does not suffer from firewall and access problems. Index Terms--Data communication, distributed database systems, internetworking, tree data structures.

The fast recursive shortest spanning tree (FRSST) algorithm achieves image segmentation and edge detection in a running time of O(n) for an image of n pixels. The region-based mapping procedure and non-requirement of a sorting process makes FRSST algorithm faster than conventional algorithms. The memory requirements also decrease by 20% or more, specifically for large size images. FRSST utilizes a spanning subtree to extend the spanning tree and makes use of updating neighboring link weights.

An approach to organizing storage in database systems is presented that, under a wide range of conditions, saves both storage space and processing time. Text values in a database are replaced by short, fixed-length, rank-preserving numeric tokens. The actual values are stored in separate, nonredundant storage. Database operations that depend only on the relative magnitude of data values can be performed directly on the tokens. Tokenization is shown to improve database performance most in situations where there are a lot of ad hoc queries and a low volume of database insertions relative to other operations. Index Terms - Abstract data types, database management, design, file organization, performance, tokenization.

Mesh-connected trees (MCT) are a class of parallel architectures that execute grid algorithms as efficiently as the grid, and the mesh of tree (MOT) algorithms as efficiently as the MOT. The MCT matches the running time of well known hypercube algorithms. A simple variant of the MCT architecture admits both the MOT and the torus as its subgraphs. The MCT network has large enough band width, to be considered as a low cost alternative for general purpose parallel computers.

A new algorithm RELSYN-OPT synthesizes real-time, reliable, user-specific multiprocessors via a heterogeneous library of N processors and L link types. RELSYN-OPT applies to dynamic and static task-allocation for ultra-reliable distributed processing environments. The worst case computational complexity for a series-parallel graph, with M subtasks and nested-depth d, is O(M.(L+N).N(super d)). For tree-structured task graphs, RELSYN-OPT runs in O(M.(L+N)) with asymptotic optimization. A multiprocessor architecture for a real-time robot arm controller is synthesized.

The pipelined execution of multijoin queries in a multiprocessor-based database system is explored in this paper. Using hash-based joins, multiple joins can be pipelined so that the early results from a join, before the whole join is completed, are sent to the next join for processing. The execution of a query is usually denoted by a query execution tree. To improve the execution of pipelined hash joins, an innovative approach on query execution tree selection is proposed to exploit segmented right-deep trees, which are bushy trees of right-deep subtrees. We first derive an analytical model for the execution of a pipeline segment, and then, in light of the model, develop heuristic schemes to determine the query execution plan based on a segmented right-deep tree so that the query can be efficiently executed. As shown by our simulation, the proposed approach, without incurring additional overhead on plan execution, possesses more flexibility in query plan generation, and can lead to query plans of better performance than those achievable by the previous schemes using right-deep trees. Index Terms - Pipelining, parallel query processing, bushy trees, right-deep trees, hash joins.

A new method has been developed for achieving multicast communication in parallel processing systems. The proposed protocol uses propagation trees to ensure total ordering of message delivery and to guarantee that all recipients will receive the messages concurrently in the same relative order. It also reduces the time needed to multicast the messages to all the destinations.

Texture analysis is performed efficiently by tree-structured wavelet transform which provides natural and effective wavelet transforms for analyzing textures with prominent middle frequency channels. The traditional pyramid structured wavelet transform and Gabor filters having constant Q-property are primarily suitable for images with low frequency energies. Wavelet transform yields a group of space localization bases which can reconstruct signals from the wavelet coefficients.

Electronic experts have developed a new method for signal compression by decomposing wavelet packets on the model of an adaptive tree structure. Decomposition of a direct signal takes place by using wavelet packets as the basis of a given wavelet kernel to discover the best wavelet packet model. A layout that includes both rate and distortion is worked on to prune the tree, which denotes the entire collection of permissible wavelet packet basis, to arrive at the most suitable basis subtree which would reduce the coding bit rate for better target quality.