Showing papers presented at "Parallel and Distributed Processing Techniques and Applications in 2007"

Consistent Checkpoint Protocol for Wireless Ad-hoc Networks.

Abstract: Switch-controlled apparatus for actuating a power switch to enable application of power from an associated power supply to a heater filament of a glow plug in a diesel engine is disclosed. The filament is one which is heated to a predetermined design temperature when a specified voltage is applied thereto. The power source has a supply voltage greater than the specified voltage of the filament. A switch accessible to an operator of the engine is operable in a first position to prevent the application of the supply voltage to the apparatus, operable in a second position to apply the supply voltage through a first terminal of the apparatus, and operable in a third position to apply the supply voltage through the first terminal and through a second terminal of the apparatus and to a starter for the engine. The apparatus comprises means operable, after the switch has been turned from the first to the second position, to actuate the power switch for a preheat time period which varies as an inverse function of available supply voltage and equals the time required to raise the temperature of the filament from ambient temperature to the design temperature. The apparatus also includes means operable, after the preheat time period, to alternately actuate and then deactuate the power switch in a cyclic fashion to maintain the design temperature of the filament for a prestart time period. The apparatus also includes means operable, after turning the switch from the second to the third position, to continue the cyclical actuation and deactuation of the power switch for an afterglow time period predetermined by the amount of time required for smooth engine idling and to minimize engine noise and white smoke emission.

An Analytical Model for Computer Systems with Non-Exponential Service Times and Memory Thrashing Overhead.

Abstract: This disclosure relates to countermeasure systems, particularly to systems for providing a protective cover against homing and/or fire control devices operating upon infrared, sonar, or microwave reflected energy or for confusing search and tracking devices, and more particularly to decoy rounds and their method of fabrication. The decoy rounds include at least one load of energy generating material, means for propelling the load, a burst charge for shattering the load, the burst charge being constructed so as to cause a line type explosion, and means for igniting the burst charge.

Towards a new Demand-Driven Message-Oriented Middleware in GIPSY.

Abstract: A vehicle-suspension type of permanent magnet-levitation transportation system has an elevated track structure defining a bed with a ferromagnetic levitation track on the underside thereof and a spaced parallel control track composed of ceramic magnetic material having transverse polarity orientation also running on the underside of the bed. A vehicle for traveling along the track has a series of spaced cobalt-rare earth levitation magnets positioned under the ferromagnetic track and a series of spaced cobalt-rare earth control magnets positioned under the control track in magnetic repulsion mode with the ceramic magnets. In levitated position the attraction of the levitation magnets for the ferromagnetic track is balanced by the repulsion between the control magnets and the ceramic track. Hydraulic, mechanical and electromagnetic means for positioning the levitation and control magnets in proper vertical relationship to maintain levitation of the vehicle are described.

Python-based Distributed Programming with Trickle

Abstract: Trickle is a an extension to the Python program- ming language that provides explicit but simple mechanisms to write distributed scripts and programs. Trickle links to - gether remote Python interpreters running on heterogeneou s machines so that work can be deployed and results collected. A Trickle program interacts with remote interpreters by inj ect- ing functions or classes. Remote objects can be instantiate d and invoked synchronously or asynchronously. Also, the in- jected code need not reside on the remote interpreters; code is dynamically transferred as needed. Trickle leverages of f of Python's list comprehensions and generators to simplify parallel invocation, result gathering, and dynamic schedu l- ing. The Trickle run-time system uses a broadcast mechanism to find eligible Trickle virtual machines. Python programme rs can immediately use Trickle to dispatch work to idle machine s with minimal setup and easy to learn mechanisms. This paper describes the Trickle extension interface, its implementa tion, and presents an example application and its performance.

A Two-phase Practical Parallel Algorithm for Construction of Huffman Codes.

Abstract: The construction of optimal prefix codes plays a significant and influential role in applications concerning information processing and communication. For decades, different algorithms were proposed treating the issue of Huffman codes construction and various optimizations were introduced. In this paper we propose a detailed practical time-efficient parallel algorithm for generating Huffman codes on CREW PRAM model exploiting n processors, where n is equal to the number of symbols in alphabet. We first compute the codewords lengths for all symbols concurrently with an innovative direct parallelization of the Huffman tree construction algorithm, alleviating the complexity of dealing with the original tree-like data structure. Then Huffman codes corresponding to symbols are generated in parallel based on a recursive formula introduced in [5]. The proposed algorithm achieves an O(n) time in the worst case when one-sided Huffman tree is formed, which is rarely encountered in practice, and O(log((logn – 1)!)) time in the best case when Huffman tree is nearly balanced.

Linear Programming Based Affinity Scheduling for Heterogeneous Computing Systems.

Abstract: Resource management systems (RMS) are an important component in heterogeneous computing (HC) systems. One of the jobs of an RMS is the mapping of arriving tasks onto the machines of the HC system. Many different mapping heuristics have been proposed in recent years. However, most of these heuristics suffer from several limitations. One of these limitations is the performance degradation that results from using outdated global information about the status of all machines in the HC system. This paper proposes a new heuristic which addresses this limitation by only requiring partial information in making the mapping decisions. Simulation results show that our heuristic performs very competitively while requiring dramatically less information.

Evaluating Dependable Distributed Storage Systems.

Abstract: A method and apparatus for applying paste to battery grids wherein the paste is circulated in a primary circuit incorporating an open reservoir wherein the paste is relieved of all pumping pressure and portion of the paste is diverted from the primary circuit into a secondary circuit for delivery to a nozzle which may be selectively opened and closed to deliver paste to the grid, the secondary circuit also incorporates the open reservoir.

A More Efficient Algorithm for Finding a Maximum Clique with an Improved Approximate Coloring.

Abstract: A combine harvester with a threshing and a separating mechanism for threshing and separating crop material and a cleaning device for cleaning threshed and separated grain. The cleaning device comprises a cleaning fan structure with a fan housing and a fan therein. The fan housing comprises a generally cylindrical body extending transversely of the cleaning device with a pair of opposite end walls and an air inlet provided in each end wall and a further air inlet and an air outlet provided in the cylindrical body and extending transversely of the cleaning device.

A Parallel Simulation Framework for Integrated Regional Ecosytem Modeling.

Abstract: This paper presents a general framework to utilize high performance computations in regional ecosystem simulation. First, a comprehensive modeling package is introduced to demonstrate the challenges encountered due to the multiple spatial and temporal scales which arise in regional ecosystem modeling. Second, a parallel simulation framework is presented to support multi-component ecosystem modeling on high performance computational platforms. Third, two ecological models are summarized and presented as an example of model integration in the simulation framework. This work presents the first component-based integrated regional ecosystem simulation for natural resource management on high performance computational platforms.

Communication with Super Flexible Messaging.

Abstract: Super Flexible Messaging (SFM) provides a powerful and elegant message passing abstraction for transferring arbitrary data between remote processes. SFM achieves the simplicity of stream-oriented mechanisms and the ability to transfer structured data as found in high-level remote object invocation systems. Leveraging off of dynamic typing and named arguments in functions, SFM offers a simple syntax for sending structured data as attribute-value pairs between processes. A programmer can easily concentrate on algorithmic development because the specification of data to be transferred is done exclusively using a send() invocation. The recv() call can selectively choose incoming messages by specifying concrete attribute-value pairs or by constraining matches to a subset of attribute values. Our current implementation is in Python, but the SFM concept can be ported to other dynamically typed languages with named argument syntax. This paper presents the SFM mechanism, its implementation, some SFM examples, and a performance evaluation.

Speculative Parallelization of Many-Particle Collision Simulations.

Abstract: A manually actuated popcorn machine which heats, pops kernels of corn, and delivers a specified amount of popped corn to a waiting container in a short time. The popcorn machine includes a recirculating hot air conduit with an air blower portion, heating chamber portion, popping chamber portion for the kernels, a separating chamber portion, and a system of control baffles throughout the chambers for controlling the flow of air and popcorn. The conduits include a continuous main chamber and a spur conduit that leads to a cup dispensing chamber for placing the popped corn in a cup for distribution. When actuated, the popcorn machine delivers a set amount of stored kernels to a screened grid in the popping chamber, the air from the blower moves past the heating element and through the screened grid to work around the kernels and to keep the kernels in motion on and above the screen grid and to move the popped corn away from the screened grid and into a cup. When the hot air and the kernels reach a proper temperature, the kernels are popped and the volume and density are radically changed. The air, whose currents are controlled by baffles, delivers the more buoyant popped corn up and into the spur conduit for delivery to a container while the hot air is recycled into the blower.

Two Graph Algorithms On an Associative Computing Model.

Abstract: The MASC (for Multiple Associative Computing) model is a SIMD model enhanced with associative properties and multiple synchronous instruction streams (IS). A number of algorithms have been developed for this model and some have been implemented using its associative programming language. In this paper, we present two graph algorithms for the restricted MASC model with one IS (called the ASC model). These are an ASC version of Kruskal algorithm for the minimum spanning tree (MST) problem and an ASC version of Dijsktra algorithm for the single source shortest path (SSSP) problem. Both of our algorithms have worst case time of O(n) which provide speedups of at least n over their sequential versions. Also, our second algorithm is cost optimal. These algorithms have advantages of being easy to program, highly scalable and having small overhead costs. They provide a better understanding about the power of the ASC/MASC model in terms of effectiveness of algorithms it supports.

A Personal Web Service Based on a User-Friendly Personal Identification Method in a Ubiquitous Environment.

Abstract: An obstetric device for assisting in the birth of animals, has an end frame that abuts the pelvis of the dam and defines an aperture through which the offspring is delivered, and a longitudinal frame member that extends perpendicularly away from the end frame member and is connected thereto by a U-shaped portion that lies in a plane perpendicular to the end frame. The legs of the U-shaped portion are secured to opposite side portions of the end frame at positions on the side portions below their midpoints but above their lower ends so that the greater part, but not all, of the delivery aperture is positioned above the longitudinal frame member. At the other end of the longitudinal frame member, a T-shaped portion provides axles for winding drums that are freely rotatable in either direction thereon, independently of each other. A delivery rope is wound up on each drum, for attachment to the forelegs of the offspring. Handles are provided on the winding drums to rotate them manually so as to tension the delivery ropes.

Suppressing Independent Loops in Packing/Unpacking Loop Nest to Reduce Message Size for Message-passing Code.

Abstract: Suppressing Independent Loops in Packing/Unpacking Loop Nests to Reduce Message Size for Message-Passing Code. Martin II, Phillip, 2010. Capstone Paper, University of North Carolina Wilmington. Two optimization techniques are explored for a parallelizing compiler that automatically generates parallel code for distributed-memory systems. Two problems arise from message-passing code generated by the compiler: 1) messages contain redundant data, and 2) the same data is sometimes transmitted to different processors, yet the messages are repacked for each processor. In this paper, the technique that eliminates the first problem, redundant data, is found to yield benefits. The steps necessary to implement the technique into the parallelizing compiler are provided in detail. The second optimization technique yields negligible benefits and is not implemented. Testing of the implementation is described and the results of the testing are given to show the benefits achieved by the first optimization technique.

Markov Reliability Modeling of Star Networks.

Abstract: A spark-arresting module for disposition in a duct for hot stack gases comprising a rigid frame structured with spaced, parallel bars alternating with spaced, parallel slots, a shaft at the upstream side of the frame and parallel thereto, a plurality of blades provided with diametrically-located vanes fixed to the shaft for rotation therewith, said vanes being movable through the slots in shearing relation to the bars by rotation of the shaft and being so shaped that some portion of each blade is at all times situated between the bars and a motor for effecting rotation of the shaft.

Moving Threads: A Non-Conventional Approach for Mapping Computation to MP-SOC.

Abstract: A multiple speed ratio manual power transmission shifting mechanism and interlock for an automotive vehicle comprising primary torque delivery gears and a cluster gear assembly having gear elements engageable with each of the main torque delivery gears; two double-acting and one single-acting synchronizer clutch mechanisms for selectively actuating the five primary torque delivery gears a single shift rail and reverse relay assembly for actuating the synchronizer clutch mechanism, a reverse idler shiftable to a driving position or to an inoperative position by the action of a first selector pin mounted on the shift rail that selectively engages either of the double acting synchronizer clutch mechanisms or the reverse idler relay, and an interlock cooperating with the selector pin to block axial movement of the unselecting actuator mechanisms. A second selector pin is mounted on the shift rail for engaging and axially moving the single acting synchronizer clutch mechanism into engagement with the fifth forward speed gear. An interlock sleeve cooperates with the second selector pin to block axial movement of the single acting synchronizer clutch mechanism when the shift rail is in a position to engage the other gear actuating mechanisms.

Resource Allocation in a Cluster Based Imaging System.

Abstract: Recently there has been an increased demand for imaging systems in support of high-speed digital printing. The required increase in performance in support of such systems can be accomplished through an effective parallel execution of image processing applications in a distributed cluster computing environment. We present a mathematical model of such a cluster based raster imaging system. The output of the system must be presented to a raster based display at regular intervals, effectively establishing a hard deadline for the production of each output image. Failure to complete a rasterization task before its deadline will result in an interruption of service that is unacceptable. The goal of this research was to design a resource allocation heuristic capable of completing each rasterization task before its deadline, thus, preventing any service interruptions. This new heuristic is evaluated within a simulation of the studied raster imaging system. We clearly demonstrate the effectiveness of the heuristic by comparing its results with the results of two resource allocation heuristics commonly used in this type of system.

Unified Data Model (UDM): A Library for Parallel I/O and Data Management.

Abstract: Funct onal ty for Unstructured Meshes and Var ables One of the important and powerful functions in the UDM library is the management of unstructured meshes and variables. The library supports a broad range of unstructured meshes, which include meshes with fixed shapes, arbitrary polygons, and arbitrary polyhedrons. A mesh element may be a zone, or face, or edge; i.e., zone-mesh, face-mesh, and edge-mesh. An edgemesh may be 1-D, 2-D, or 3-D; and a face-mesh may be 2-D or 3-D. Mesh elements of a zone-mesh may be made directly from nodes, the elements may be made from edges, and the elements may also be made from faces, and the faces are made from either edges or nodes. The UDM library also supports ghost mesh elements, boundary faces, boundary edges, boundary nodes, slip faces, slip edges, slip nodes, and other options. The variables the library supports include node-, or edge-, or face-, or zone-variables, and the variables may be scalars, vectors, and tensors.

Self-Timed Stream Processor for Surrounding Computing Environment.

Abstract: An apparatus for the investigation and/or control of tachycardia comprising an implantable pacemaker having a short (typically 180 ms or lower) refractory period and providing in use, output pulses in response to external stimuli applied to the skin of the patient, and an external stimulus generator comprising means for generating the external stimuli. In the control of tachycardia, the implanted unit provides output pulses in response to natural heart beats. These output pulses are detected by the external stimulus generator through electrodes thereon held in contact with the skin. The external generator determines the pulse rate of the detected pulses and, if above a preset level, generates the external stimuli of an appropriate characteristic in time and number to cause the implanted unit to issue output pulses to arrest the tachycardia. The apparatus may also be used to induce a tachycardia, enabling non-invasive electrophysiological studies to be made.

A proposal on Ad Hoc and Ubiquitous Communication Environment supported by Data-Driven Networking Processor.

Abstract: This paper proposes a new environment for ad hoc and ubiquitous communicationt that is supported by data-driven networking processor. Te environment is based on mobile ad hoc network (MANET) which is without any centered servers. Over the MANET, there is a ubiquitous environment that realizes communication among people wherever they are. Underlying platform need satisfy requirements given by upper layer function. To implement this environment, this paper studies technical requirements to each layer and it proposes three basic ideas. One is information discovery mechanism as the first step of achieving ubiquitousness. We propose an autonomous contents discovery methods based on social communities. The second idea is the reliable information transmission over the ad hoc network based on the discovered information in above layer by combining Multiple Description Coding with multiple source transport. The other is the data-driven platform to support communication environment, which achieves efficient real-time multi-processing capability without any runtime overhead.

Improving Dynamic Load Balancing Under CORBA with a Genetic Strategy in a Neural System of Off-line Signature Verification.

Abstract: A flat plate solar boiler, having one face which is exposed to solar radiation to heat a liquid within the boiler, also has its opposing face, or the dark side of the boiler, indirectly heated by solar radiation thereby to minimize heat losses from the boiler and to apply further input heat to the boiler. The dark side or second face of the solar boiler is indirectly heated through a metallic plate which is in conductive engagement with the second face of the boiler, extends beyond at least one of the side edges of the boiler, and has a radiation absorptive material on an extension portion of the plate facing the sun. In an alternative embodiment, the second face of the boiler is heated by use of a second solar boiler which is disposed in laterally spaced relation to the first boiler to provide a heated liquid which is circulated past the second face of the first boiler.

Brokering Algorithms for Composing Low Cost Distributed Storage Resources.

Abstract: In this paper we investigate the composition of cheap network storage resources to meet specific availability and capacity requirements. We show that the problem of finding the optimal composition for availability and price requirements can be reduced to the knapsack problem, and propose three techniques for efficiently finding approximate solutions. These include a dynamic programming algorithm, a heuristic and an genetic algorithm. The algorithms can be implemented on a broker that intermediates between buyers and sellers of storage resources. Finally, we show that a broker in an open storage market, using the combination of the three algorithms can more frequently meet user requests and lower the cost of requests that are met compared to a broker that simply matches single resources to re-

A Reduct Solving Parallel Algorithm Based on Relational Extension Matrix.

Abstract: In this paper, on the basis of studying the limitations of the basic rough set model, we present Tolerance Information Systems, which is based on a family set of tolerance relations between objects when given a set of tolerance relations. The model inherits most of the characteristics of the basic model of rough set; and they also have a better effect of approximation classification. Based on this model, we propose two algorithms that will give us one near-optimal attributes reduct in memory and process efficient way, the first one is a single processor algorithm which uses concepts from relation and extension matrices. Based on the first algorithm, its parallel processor version is proposed, the parallel version is far more efficient.

Aspects for forming structured Bluetooth based Scatternets

Abstract: Currently many approaches for wireless network management are in progress Especially Bluetooth based networks require an elaborate management of restrictions the Bluetooth standard implies In this paper we will point out the main requirements to be considered before forming such networks We will also describe the most common Bluetooth based network topologies and discuss their advantages and disadvantages In addition to this, we will give an overview of a selected set of Bluetooth network forming algorithms and examine these under consideration of said aspects and topology decisions Furthermore, we will compare these algorithms to the requirements of the BlueSpot system, developed by us

Dynamic decomposition algorithm merging control flow analysis.

Abstract: A torque converter for inclusion between a driving shaft and a driven shaft. Preferably there is more than one torque generator and a gear train in parallel between the input and output shafts. The torque generators and any attached driving gears rotate about shafts whose axes are preferably symmetrically displaced about, and are stationary relative to, the axis of the input shaft. The above mentioned torque generators each preferably have a pair of radially opposed eccentrically mounted weights thereon. Preferably there are at least two pairs of weights on each torque generating train, together with means for varying the phasing between the pairs of weights. Means is also provided for accommodating slight misalignments between the shafts of the torque generating trains. Each of the torque generating trains has attached thereto at least one unidirectional clutch which allows the shafts to turn in only one direction. Means are also provided for delivering torque from the torque generating trains to the output shaft and for reversing the direction of the output shaft torque.

The Gossiping on OTIS-Hypercube Optoelectronic Parallel Computer.

Abstract: A watering cup especially useful for watering chickens having a first contact actuated valve for admitting water which first valve bears against a second interior valve and holds the second valve open. The first valve can be removed for cleaning and inspection without the necessity of shutting down the water pressure since removal of the first valve closes the second valve and prevents leakage.

An Input Queueing Implementation for Low-Latency Speculative Optical Switches.

Abstract: An automotive automatic power transmission system having a torque converter contained in a converter housing and a planetary gear system powered by the torque converter. The planetary gear system is contained in a gear housing detachably secured to the converter housing. The planetary gear system has an output shaft carrying an output gear meshing with an idler gear which is in turn in meshed relationship with a driven gear of a final drive unit. The aforesaid gears are of the helical type and are rotatable about parallel axes, with a resulting axial thrust on the output shaft in a direction towards the planetary gear system during rotation of the output gear in the forward direction. The output shaft is journalled between first and second thrust bearings, with the first thrust bearing being carried on a hollow member detachably fixed to the converter housing and protruding axially therein to support the torque converter stator on a one-way clutch. The second thrust bearing is radially and axially supported exclusively by the converter housing and is provided with a loading capacity greater than that of said first bearing.

Atmospheric Effects of IEEE802.11g over HTTP.

Abstract: Atmospheric effects on radio frequency transmissions have been studied extensively since the advent of radio usage in military applications, yet the effects on shorter range transmissions have been less well covered. This study attempts to find correlations between atmospheric conditions and the transmission efficiency of IEEE 802.11g, using HTTP as the high level transport mechanism. This study examines the effects of atmospherics on application layer protocol performance. Prior work on this topic concentrates on the effect of weather conditions on signal strength. However, signal strength itself may not be the best predictor of application layer performance.