Scott R. Cannon

Adding fault-tolerant transaction processing to LINDA

To simplify the difficult task of writing fault‐tolerant parallel software, we implemented extensions to the basic functionality of the LINDA or tuple‐space programming model. Our approach implements a mechanism of transaction processing to ensure that tuples are properly handled in the event of a node or communications failure. If a process retrieving a tuple fails to complete processing or a tuple posting or retrieval message is lost, the system is automatically rolled back to a previous stable state. Processing failures and lost messages are detected by time‐out alarms. Roll‐back is accomplished by reposting pertinent tuples. Intermediate tuples produced during partial processing are not committed or made available until a process completes. In the absence of faults, system overhead is low. The fault‐tolerance mechanism is implemented at the system level and requires little programmer effort or expertise. Two implementations of the model are discussed, one using a UNIX network of workstations and one using a Transputer network. Data measuring model overhead and some aspects of system performance in the presence of faults is presented for an example system.

Automatic Software Generation of ASIM Program Code from an xTEDS

The Satellite Data Model (SDM) developed at Utah State University (USU) is a plugand-play system for satellites. An xTEDS must be written for each sensor device that participates with the SDM. An interface module called an Appliqu e Sensor Interface Module (ASIM) must be developed to interface between the SDM and the sensor device. The program code that runs on an ASIM can be di cult to develop. This paper presents a technique for automatically generating ASIM program code by extrapolating the needed information from the sensor device’s xTEDS. This technique has been implemented in a tool called the ASIM Wizard. The ASIM Wizard reduces the time and e ort required to develop an ASIM.

An efficient fault-tolerant tuple space

A mechanism is presented for a fault-tolerant tuple space. By augmenting the MOM model of fault tolerance, a LINDA-style system is proposed that supports failure and recovery of worker processes and tuple-space in any state. System tuple space is replicated using local tuple-space segments. Each local tuple space replicates tuples and states produced or processed by local workers. All tuple requests are handled through local tuple-space agents prior to being forwarded to a system tuple-space manager in order to update tuple states. A MOM extension to LINDA operators allows local managers to know when tuple states can be committed. In the event of a local tuple-space failure, the local tuple-space segment is reconstructed from system tuples and states associated with the failed local node. The reconstructed local tuple space is inherited by another local tuple-space. For a system tuple-space failure, system tuple space is reconstructed by interrogating local tuple space managers. Tuple replicas and states are maintained with minimal system message overhead during nonfailure operation.

Usefulness of Doppler Catheters in Assessment of Coronary Artery Blood Flow

While most ultrasound transducers are large and used to obtain non-invasive data, Doppler transducers less than 1.0 mm in diameter can be manufactured and placed on intravascular catheters. Drs. Hartley and Cole1, 2 developed a 20 MHz pulsed-Doppler system for evaluating coronary artery blood flow. We have used the Hartley-Doppler-Sones system during routine coronary cineangiography in over 150 patients and found it useful in: characterizing phasic and mean left and right coronary artery blood velocity; monitoring changes in coronary artery blood velocity during transient events (i.e., arrhythmias, cough); monitoring changes in coronary artery blood velocity during diagnostic or therapeutic interventions.

The advent of the PnP Cube satellite

In terms of time and budget, integration is a significant time-consuming component of spacecraft development. While many useful COTS spacecraft components are available, interfacing and controlling these components in an integrated satellite system remains a complex engineering task. The Stanford/Cal Poly CubeSat and Poly-Picosatellite Orbital Dispenser (PPOD) standards have begun to standardize small satellite mechanical systems and revolutionize the way small satellites are deployed. NASA has recognized this as evident by their Educational Launch of Nanosatellites (ELaNa) program which recently selected 17 CubeSats for the ELaNa-4 launch in 2012 (including one high school). To capitalize on this momentum, the Air Force Research Lab (AFRL) has organized and supported a team of commercial and academic laboratories to develop and test an over-arching Space Plug-and-play Architecture (SPA) set of standards to support the rapid integration of independently developed satellite modular systems. SPA represents not only an electrical inter-connection and communication scheme, but a complete model for a self-organizing and self-configuring system to support the rapid assembly of mission-specific small satellites. Rather than forcing existing modules to be re-developed to a common messaging standard, SPA utilizes an XTEDS (eXtended Transducer Electronic Data Sheet) model. Each satellite module contains an electronic document describing its interface, capabilities, messages, data formats, etc. By reading a components XTEDS, other systems can quickly integrate and utilize a new module. While designed to initially take advantage of nanosatellites, everything developed can easily scale to larger spacecraft, UAVs or other aerospace and defense systems. This paper discusses our experience in developing the CubeSat Trailblazer, a 1U SPA-only spacecraft - launching in 2012 as a testbed for SPA technology. The mechanisms of self-organization for independent modules as a cooperating communications system are discussed. The simplifications associated with software development of a Command and Data Handler (CDH) is also presented.

A stable distributed tuple space

An approach is presented which extends the MOM fault-tolerant implementation of the Linda model of parallel programming. The original MOM system provided persistence of tuples and tuple states across both tuple-space and worker node halt failures. Unfortunately, the requirement that system tuple space reside in a central location restricted the scalability of the MOM model. In this work, an approach is presented for distributed system tuple space and tuple states using a hashing function on tuple labels. This approach compares favourably with other tuple-space distribution methods in terms of message costs during non-fault operation, and allows preservation of the fault-tolerant mechanisms of the MOM model.

An equal compression-ratio comparison of beat-to-beat and SAPA2 compression techniques

To adequately compare beat-to-beat substraction (BBS) and scan-along polynomial approximation Hash 2 (SAPA2) electrocardiogram (ECG) direct compression methods, the authors applied both techniques to fifty randomly selected ECG segments from the MIT compression test database. The parameters of both techniques were adjusted to produce equal compression ratios close to 6.5 for each segment. After compression, SAPA2 segments were reconstructed using a cubic spline, and BBS segments with linear interpolation. Unlabeled plots of the original segment and both reconstructed segments in random order were graded for fidelity and clinical accuracy by a cardiologist blind to the nature and strategy of the two compression techniques. BBS was graded higher than SAPA2 in 64% of segments, SAPA2 better than BBS in 14%. Mean absolute errors and percent root-mean-square differences were nearly equal.<<ETX>>

Comparison of Continuous and Pulsed Doppler in Diagnosis of Mitral and Aortic Stenosis in Adults

The basic concepts necessary to utilize Doppler echocardio-graphic techniques to diagnose and quantitate valvular heart disease are simple but must be thoroughly understood. An understanding of the flow abnormalities induced by valve stenosis, the Dopplercardiographic manifestations of the abnormalities, and the ability of various types of instrumentation to detect these abnormalities are important. Similarly, an organized approach to acquisition of diagnostic information and extraction of appropriate parameters is essential if accurate diagnosis is to be accomplished.

Cryocooler vibration control with an inaccurate transfer function

Cryocooler vibrational stability is an important issue in IR sensing. Unfortunately, several common approaches to this problem require that the cryocooler transfer function be accurately measured. We present a digital solution using a simple iterative algorithm for an axially aligned dual-piston cooler. We derived a formula for this algorithm to predict the vibration force attenuation Ak equals [1 - (mu) (cos(phi) + j sin(phi) )/1 + (epsilon) ]k where k is the algorithm iteration number, (mu) is an algorithm parameter, (phi) is the maximum absolute error in the measured transfer function phase, and (epsilon) is the relative error in measured transfer function gain. If (mu) is chosen so that 0 less than (mu) less than 2 (1 + (epsilon) ) cos(phi) , the algorithm will provide an exponential vibration attenuation. As long as (phi) less than (pi) /2, it is possible to find a value for (mu) so the algorithm will converge. To demonstrate this property of the algorithm, we constructed a cryocooler vibration model using two large axially-mounted audio speakers mounted on a rigid structure with numerous vibration modes. Speakers were driven using 2 D/A channels and vibration forces were measured using an A/D and an accelerometer mounted on the structure. After accurately measuring the vibration response transfer function of the model, we corrupted phase angles between -(pi) /2 and (pi) /2. In each corrupted transfer function case, the control algorithm quickly converged to greater than 25 db below uncompensated vibration power and within 5 db of the static model vibration floor. Tests were then conducted on a Hughes 65K SSC cryocooler. The algorithm was able to significantly reduce vibrations and remain stable under a variety of changing operation parameters and cooling loads.

Approaches to teaching parallel processing on the undergraduate level (abstract)

The ACM/IEEE-CS Joint Curriculum Task Force in its Computing Curricula 1991 listed several knowledge units related to parallel processing, which they consider should be “common requirements for all undergraduate programs in the field of computing.” (p. 4) But incorporating use of parallel processors poses unique problems for the professor. How can one provide the appropriate equipment--simulation, transputers, UNIX distributed workstations, or perhaps a computer available through Internet? What are the advantages and tradeoffs of these systems? What are the support requirements? Should the professor expose the students to several architectures or delve deeply into one? Which programming language(s) should be taught? What textbooks are appropriate for the undergraduate? What kinds of projects are reasonable? Exposure can be part of several traditional courses, such as Operating Systems, Programming Languages, and Simulation; or there can be a separate course or enough courses to create an emphasis in parallel computing. Which approach is best for your school? How can one obtain funding to develop a course in the area? What short courses, workshops, and conferences are available to help the professor learn more about parallel computing? Professors, who have been deeply involved in parallel processing on the undergraduate level, will discuss these and related issues during this panel.