SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Report presents theoretical study of dynamics of neural network organizing own response in both phase space and in position space. Postulates several mathematical models of dynamics including spatial derivatives representing local interconnections among neurons. Shows how neural responses propagate via these interconnections and how spatial pattern of neural responses formed in homogeneous biological neural network.
The Data Distribution Satellite (DDS), operating in conjunction with the planned space network, the National Research and Education Network and its commercial derivatives, would play a key role in networking the emerging supercomputing facilities, national archives, academic, industrial, and government institutions. Centrally located over the United States in geostationary orbit, DDS would carry sophisticated on-board switching and make use of advanced antennas to provide an array of special services. Institutions needing continuous high data rate service would be networked together by use of a microwave switching matrix and electronically steered hopping beams. Simultaneously, DDS would use other beams and on board processing to interconnect other institutions with lesser, low rate, intermittent needs. Dedicated links to White Sands and other facilities would enable direct access to space payloads and sensor data. Intersatellite links to a second generation ATDRS, called Advanced Space Data Acquisition and Communications System (ASDACS), would eliminate one satellite hop and enhance controllability of experimental payloads by reducing path delay. Similarly, direct access would be available to the supercomputing facilities and national data archives. Economies with DDS would be derived from its ability to switch high rate facilities amongst users needed. At the same time, having a CONUS view, DDS would interconnect with any institution regardless of how remote. Whether one needed high rate service or low rate service would be immaterial. With the capability to assign resources on demand, DDS will need only carry a portion of the resources needed if dedicated facilities were used. Efficiently switching resources to users as needed, DDS would become a very feasible spacecraft, even though it would tie together the space network, the terrestrial network, remote sites, 1000's of small users, and those few who need very large data links intermittently.
Two satellite architectures for delivering broadband integrated services digital network (B-ISDN) service are evaluated. The first is assumed integral to an existing terrestrial network, and provides complementary services such as interconnects to remote nodes as well as high-rate multicast and broadcast service. The interconnects are at a 155 Mbs rate and are shown as being met with a nonregenerative multibeam satellite having 10-1.5 degree spots. The second satellite architecture focuses on providing private B-ISDN networks as well as acting as a gateway to the public network. This is conceived as being provided by a regenerative multibeam satellite with on-board ATM (asynchronous transfer mode) processing payload. With up to 800 Mbs offered, higher satellite EIRP is required. This is accomplished with 12-0.4 degree hopping beams, covering a total of 110 dwell positions. It is estimated the space segment capital cost for architecture one would be about $190M whereas the second architecture would be about $250M. The net user cost is given for a variety of scenarios, but the cost for 155 Mbs services is shown to be about $15-22/minute for 25 percent system utilization.
Neural networks are a very useful tool for analyzing and modeling complex real world systems. Applying neural network simulations to real world problems generally involves large amounts of data and massive amounts of computation. To efficiently handle the computational requirements of large problems, we have implemented at Los Alamos a highly efficient neural network compiler for serial computers, vector computers, vector parallel computers, and fine grain SIMD computers such as the CM-2 connection machine. This paper describes the mapping used by the compiler to implement feed-forward backpropagation neural networks for a SIMD (Single Instruction Multiple Data) architecture parallel computer. Thinking Machines Corporation has benchmarked our code at 1.3 billion interconnects per second (approximately 3 gigaflops) on a 64,000 processor CM-2 connection machine (Singer 1990). This mapping is applicable to other SIMD computers and can be implemented on MIMD computers such as the CM-5 connection machine. Our mapping has virtually no communications overhead with the exception of the communications required for a global summation across the processors (which has a sub-linear runtime growth on the order of O(log(number of processors)). We can efficiently model very large neural networks which have many neurons and interconnects and our mapping can extend to arbitrarily large networks (within memory limitations) by merging the memory space of separate processors with fast adjacent processor interprocessor communications. This paper will consider the simulation of only feed forward neural network although this method is extendable to recurrent networks.
Our previous prototype, RAID-1, identified several bottlenecks in typical file server architectures. The most important bottleneck was the lack of a high-bandwidth path between disk, memory, and the network. Workstation servers, such as the Sun-4/280, have very slow access to peripherals on busses far from the CPU. For the RAID-2 system, we addressed this problem by designing a crossbar interconnect, Xbus board, that provides a 40MB/s path between disk, memory, and the network interfaces. However, this interconnect does not provide the system CPU with low latency access to control the various interfaces. To provide a high data rate to clients on the network, we were forced to carefully and efficiently design the network software. A block diagram of the system hardware architecture is given. In the following subsections, we describe pieces of the RAID-2 file server hardware that had a significant impact on the design of the network interface.
An expandable stochastic digital architecture for recurrent (Hopfield like) neural networks is proposed. The main features and basic principles of stochastic processing are presented. The stochastic digital architecture is based on a chip with n full interconnected neurons with a pipeline, bit processing structure. For large applications, a flexible way to interconnect many such chips is provided.
Requirements are carefully described in descriptions of systems to be acquired but often there is no requirement to provide measurements and performance monitoring to ensure that requirements are met over the long term after acceptance. A set of measurements for various UNIX-based systems will be available at the 1992 Goddard Conference on Mass Storage Systems and Technologies. The authors invite others to contribute to the set of measurements. The framework for presenting the measurements of supercomputers, workstations, file servers, mass storage systems, and the networks that interconnect them are given. Production control and database systems are also included. Though other applications and third party software systems are not addressed, it is important to measure them as well. The capability to integrate measurements from all these components from different vendors, and from the third party software systems was recognized and there are efforts to standardize a framework to do this. The measurement activity falls into the domain of management standards. Standards work is ongoing for Open Systems Interconnection (OSI) systems management; AT&T, Digital, and Hewlett-Packard are developing management systems based on this architecture even though it is not finished. Another effort is in the UNIX International Performance Management Working Group. In addition, there are the Open Systems Foundation's Distributed Management Environment and the Object Management Group. A paper comparing the OSI systems management model and the Object Management Group model has been written. The IBM world has had a capability for measurement for various IBM systems since the 1970's and different vendors were able to develop tools for analyzing and viewing these measurements. Since IBM was the only vendor, the user groups were able to lobby IBM for the kinds of measurements needed. In the UNIX world of multiple vendors, a common set of measurements will not be as easy to get.
We present a modified version of the Toffoletto and Hill (1989) open magnetosphere model that incorporates a tail-like interconection field with a discontinuity 10 represent the slow-mode expansion fan that defines the high-latitude tail magnetopause. (The interconnection field is defined as the perturbation on an initially closed magnetosphere model to make it open.) The expansion fan controls the open field line region in the tail, and the intersection of the fan with the tail current sheet is, by design, the x line. The new interconnection field allows greater control of the tail field structure; in particular, it enables us to eliminate the nightside mapping singularity that occurs in previous models when the interplanetary magnetic field is nonsouthward. Also, in contrast to earlier models, the far tail x line extends farther downstream on the flanks than in the center of the tail, consistent with observations.
Proposed fast, massively parallel data processor contains 8x16 array of processing elements with efficient interconnection scheme and options for flexible local control. Processing elements communicate with each other on "X" interconnection grid with external memory via high-capacity input/output bus. This approach to conditional operation nearly doubles speed of various arithmetic operations.
The first section of the report describes the AbNET system, a hardware and software communications system designed for distribution automation (it can also find application in substation monitoring and control). The topology of the power system fixes the topology of the communications network, which can therefore be expected to include a larger number of branch points, tap points, and interconnections. These features make this communications network unlike any other. The network operating software has to solve the problem of communicating to all the nodes of a very complex network in as reliable a way as possible even if the network is damaged, and it has to do so with minimum transmission delays and at minimum cost. The design of the operating protocols is described within the framework of the seven-layer Open System Interconnection hierarchy of the International Standards Organization. Section 2 of the report describes the development and testing of a high voltage sensor based on an electro-optic polymer. The theory of operation is reviewed. Bulk fabrication of the polymer is discussed, as well as results of testing of the electro-optic coefficient of the material. Fabrication of a complete prototype sensor suitable for use in the range 1-20 kV is described. The electro-optic polymer is shown to be an important material for fiber optic sensing applications. Appendix A is theoretical support for this work. The third section of the report presents the application of an artificial neural network, Kohonen's self-organizing feature map, for the classification of power system states. This classifier maps vectors of an N-dimensional space to a 2-dimensional neural net in a nonlinear way preserving the topological order of the input vectors. These mappings are studied using a nonlinear power system model.
A system for coupling an orbital replacement unit (ORU) to a space station structure via the actions of a robot and/or astronaut is described. This system provides mechanical and electrical connections both between the ORU and the space station structure and between the ORU and the ORU and the robot/astronaut hand tool. Alignment and timing features ensure safe, sure handling and precision coupling. This includes a first female type spline connector selectively located on the space station structure, a male type spline connector positioned on the orbital replacement unit so as to mate with and connect to the first female type spline connector, and a second female type spline connector located on the orbital replacement unit. A compliant drive rod interconnects the second female type spline connector and the male type spline connector. A robotic special end effector is used for mating with and driving the second female type spline connector. Also included are alignment tabs exteriorally located on the orbital replacement unit for berthing with the space station structure. The first and second female type spline connectors each include a threaded bolt member having a captured nut member located thereon which can translate up and down the bolt but are constrained from rotation thereabout, the nut member having a mounting surface with at least one first type electrical connector located on the mounting surface for translating with the nut member. At least one complementary second type electrical connector on the orbital replacement unit mates with at least one first type electrical connector on the mounting surface of the nut member. When the driver on the robotic end effector mates with the second female type spline connector and rotates, the male type spline connector and the first female type spline connector lock together, the driver and the second female type spline connector lock together, and the nut members translate up the threaded bolt members carrying the first type electrical connector up to the complementary second type connector for interconnection therewith.
Processors are becoming faster and multiprocessor memory interconnection systems are not keeping up. Therefore, it is necessary to have threads and the memory they access as near one another as possible. Typically, this involves putting memory or caches with the processors, which gives rise to the problem of coherence: if one processor writes an address, any other processor reading that address must see the new value. This coherence can be maintained by the hardware or with software intervention. Systems of both types have been built in the past; the hardware-based systems tended to outperform the software ones. However, the ratio of processor to interconnect speed is now so high that the extra overhead of the software systems may no longer be significant. This issue is explored both by implementing a software maintained system and by introducing and using the technique of offline optimal analysis of memory reference traces. It finds that in properly built systems, software maintained coherence can perform comparably to or even better than hardware maintained coherence. The architectural features necessary for efficient software coherence to be profitable include a small page size, a fast trap mechanism, and the ability to execute instructions while remote memory references are outstanding.
This paper describes the packet switched Instrumenters Communication System (ICS) that was developed for the Command Management Facility at GSFC to support the Gamma Ray Observatory (GRO) spacecraft. The GRO ICS serves as a vital science data acquisition link to the GRO scientists to initiate commands for their spacecraft instruments. The system is ready to send and receive messages at any time, 24 hours a day and seven days a week. The system is based on X.25 and the International Standard Organization's (ISO) 7-layer Open Systems Interconnection (OSI) protocol model and has client and server components. The components of the GRO ICS are discussed along with how the Communications Subsystem for Interconnection (CSFI) and Network Control Program Packet Switching Interface (NPSI) software are used in the system.
In this paper, the system architecture for an advanced Canadian ka-band geostationary mobile satellite system is described, utilizing hopping spot beams to support a 256 kbps wideband service for both N-ISDN and packet-switched interconnectivity to small briefcase-size portable and mobile terminals. An assessment is given of the technical feasibility of the satellite payload and terminal design in the post year 2000 timeframe. The satellite payload includes regeneration and on-board switching to permit single hop interconnectivity between mobile terminals. The mobile terminal requires antenna tracking and platform stabilization to ensure acquisition of the satellite signal. The potential user applications targeted for this wideband service includes: home-office, multimedia, desk-top (PC) videoconferencing, digital audio broadcasting, single and multi-user personal communications.
A mobile, self-powered, self-contained, and remote-controlled robot is presented. The robot is capable of safely operating in a combustible atmosphere and providing information about the atmosphere to the operator. The robot includes non-sparking and non-arcing electro-mechanical and electronic components designed to prevent the robot from igniting the combustible atmosphere. The robot also includes positively pressurized enclosures that house the electromechanical and electronic components of the robot and prevent intrusion of the combustible atmosphere into the enclosures. The enclosures are interconnected such that a pressurized gas injected into any one of the enclosures is routed to all the other enclosures through the interconnections. It is preferred that one or more sealed internal channels through structures intervening between the enclosures be employed. Pressure transducers for detecting if the pressure within the enclosures falls below a predetermined level are included. The robot also has a sensing device for determining the types of combustible substances in the surrounding atmosphere, as well as the concentrations of each type of substance relative to a pre-determined lower explosive limit (LEL). In addition, the sensing device can determine the percent level of oxygen present in the surrounding atmosphere.
This invention provides a new hierarchical approach for supervised neural learning of time dependent trajectories. The modular hierarchical methodology leads to architectures which are more structured than fully interconnected networks. The networks utilize a general feedforward flow of information and sparse recurrent connections to achieve dynamic effects. The advantages include the sparsity of units and connections, the modular organization. A further advantage is that the learning is much more circumscribed learning than in fully interconnected systems. The present invention is embodied by a neural network including a plurality of neural modules each having a pre-established performance capability wherein each neural module has an output outputting present results of the performance capability and an input for changing the present results of the performance capabilitiy. For pattern recognition applications, the performance capability may be an oscillation capability producing a repeating wave pattern as the present results. In the preferred embodiment, each of the plurality of neural modules includes a pre-established capability portion and a performance adjustment portion connected to control the pre-established capability portion.
The integration of high quality, single crystal thin film gallium arsenide (GaAs) and indium phosphide (InP) based photonic and electronic materials and devices with host microstructures fabricated from materials such as silicon (Si), glass, and polymers will enable the fabrication of the next generation of micro-opto-mechanical systems (MOMS) and optoelectronic integrated circuits. Thin film semiconductor devices deposited onto arbitrary host substrates and structures create hybrid (more than one material) near-monolithic integrated systems which can be interconnected electrically using standard inexpensive microfabrication techniques such as vacuum metallization and photolithography. These integrated systems take advantage of the optical and electronic properties of compound semiconductor devices while still using host substrate materials such as silicon, polysilicon, glass and polymers in the microstructures. This type of materials optimization for specific tasks creates higher performance systems than those systems which must use trade-offs in device performance to integrate all of the function in a single material system. The low weight of these thin film devices also makes them attractive for integration with micromechanical devices which may have difficulty supporting and translating the full weight of a standard device. These thin film devices and integrated systems will be attractive for applications, however, only when the development of low cost, high yield fabrication and integration techniques makes their use economically feasible. In this paper, we discuss methods for alignment, selective deposition, and interconnection of thin film epitaxial GaAs and InP based devices onto host substrates and host microstructures.
Advanced radar and communication systems rely heavily on state-of-the-art microelectronics. Systems such as the phased-array radar require many transmit/receive (T/R) modules which are made up of many millimeter wave - microwave integrated circuits (MMIC's). The heart of a MMIC chip is the Gallium Arsenide (GaAs) field-effect transistor (FET). The transistor gate length is the critical feature that determines the operating frequency of the radar system. A smaller gate length will typically result in a higher frequency. In order to make a phased array radar system economically feasible, manufacturers must be capable of producing very large quantities of small-gate-length MMIC chips at a relatively low cost per chip. This requires the processing of a large number of wafers with a large number of chips per wafer, minimum processing time, and a very high chip yield. One of the bottlenecks in the fabrication of MIMIC chips is the transistor gate definition. The definition of sub-half-micron gates for GaAs-based field-effect transistors is generally performed by direct-write electron beam lithography (EBL). Because of the throughput limitations of EBL, the gate-layer fabrication is conventionally divided into two lithographic processes where EBL is used to generate the gate fingers and optical lithography is used to generate the large-area gate pads and interconnects. As a result, two complete sequences of resist application, exposure, development, metallization and lift-off are required for the entire gate structure. We have baselined a hybrid process, referred to as EBOL (electron beam/optical lithography), in which a single application of a multi-level resist is used for both exposures. The entire gate structure, (gate fingers, interconnects and pads), is then formed with a single metallization and lift-off process. The EBOL process thus retains the advantages of the high-resolution E-beam lithography and the high throughput of optical lithography while essentially eliminating an entire lithography/metallization/lift-off process sequence. This technique has been proven to be reliable for both trapezoidal and mushroom gates and has been successfully applied to metal-semiconductor and high-electron-mobility field-effect transistor (MESFET and HEMT) wafers containing devices with gate lengths down to 0.10 micron and 75 x 75 micron gate pads. The yields and throughput of these wafers have been very high with no loss in device performance. We will discuss the entire EBOL process technology including the multilayer resist structure, exposure conditions, process sensitivities, metal edge definition, device results, comparison to the standard gate-layer process, and its suitability for manufacturing.