Search

A mobile host has small memory, relatively slow processor, low power batteries, and communicate over low bandwidth wireless communication links. Existing mutual exclusion algorithms for distributed systems are not enough for mobile systems because of several limitations. In this study, a mutual exclusion algorithm that is more suitable for mobile computer systems is developed. The algorithm tends to minimize the number of messages needed to be transmitted in the system, by reducing the number of sites involved in the mutual exclusion decision, and reducing the amount of storage needed at different sites of the system. Key words: Distributed systems, synchronization, mutual exclusion, mobile computing, INTRODUCTION A mobile computing system is a distributed system consisting of a number of mobiles and fixed processing units. A distributed application consists of a collection of processes executing on [...]

This paper proposes an enhancement for the ADRW algorithm to achieve dynamic fragmentation and object allocation in distributed databases. The algorithm adapts to the changing patterns of object requests with the objective to dynamically adjust the allocation schemes of objects in order to minimize the total servicing cost of all requests. Objects may be replicated or fragmented depending on patterns of reads and writes. Qualitative analysis was used to characterize the performance of the enhanced algorithm. Keywords: Distributed Database, ADRW algorithm, Fragmentation, E-ADRW, INTRODUCTION Distributed systems are an important development in computing technology which is concerned with the delivery of constantly expanding data to points of query. Collections of data in the forms [...]

Grid computing focuses on large-scale resource sharing. Using a general reliability model for grid computing to relax some impractical assumptions, a heuristic algorithm is presented to evaluate grid program/service reliability. The heuristic algorithm is based on two heuristic criteria that determine the significance of an entity and prune those insignificant ones. Through algorithm analysis, the heuristic algorithm is shown to have a linear complexity. This is much better than the previous algorithms, which are of exponential complexity. Another advantage of the heuristic algorithm is that the running time is controllable by adjusting the parameter of significant level (SL) and significant rate. A regression method is proposed to adjust the SL and predict the running time. Two examples are given. Index Terms--Bayesian analysis, distributed computing, graph theory, grid program/service reliability, heuristic algorithm.

Abstract: The rapidly increasing complexity of computing systems is driving the movement towards autonomic systems that are capable of managing themselves without the need for human intervention. Without autonomic technologies, [...]

Pervasive computing provides an attractive vision for the future of computing. Computational power will be available everywhere. Mobile and stationary devices will dynamically connect and coordinate to seamlessly help people in accomplishing their tasks. For this vision to become a reality, developers must build applications that constantly adapt to a highly dynamic computing environment. To make the developers' task feasible, we present a system architecture for pervasive computing, called one world. Our architecture provides an integrated and comprehensive framework for building pervasive applications. It includes services, such as discovery and migration, that help to build applications and directly simplify the task of coping with constant change. We describe our architecture and its programming model and reflect on our own and others' experiences with using it. Categories and Subject Descriptors: D.4.7 [Operating Systems]: Organization and Design-Distributed systems; D.4.1 [Operating Systems]: Process Management; D.4.2 [Operating Systems]: Storage Management; D.4.4 [Operating Systems]: Communications Management-Network communication; D.2.11 [Software Engineering]: Software Architectures--Domain-specific architectures, patterns General Terms: Design, Performance, Reliability Additional Key Words and Phrases: Asynchronous events, checkpointing, discovery, logic/operation pattern, migration, one.world, pervasive computing, structured I/O, tuples, ubiquitous computing

Over the past years, mobile agent technology has attracted considerable attention, and a significant body of literature has been published. To further develop mobile agent technology, reliability mechanisms such as fault tolerance and transaction support are required. This article aims at structuring the field of fault-tolerant and transactional mobile agent execution and thus at guiding the reader to understand the basic strengths and weaknesses of existing approaches. It starts with a discussion on providing fault tolerance in a system in which processes simply fail. For this purpose, we first identify two basic requirements for fault-tolerant mobile agent execution: (1) non-blocking (i.e., a single failure does not prevent progress of the mobile agent execution) and (2) exactly-once (i.e., multiple executions of the agent are prevented). This leads us to introduce the notion of a local transaction as the basic building block for fault-tolerant mobile agent execution and to classify existing approaches according to when and by whom the local transactions are committed. In a second part, we show that transactional mobile agent execution additionally ensures execution atomicity and present a survey of existing approaches. In the last part of the survey, we extend the notion of fault tolerance to arbitrary Byzantine failures and security-related issues of the mobile agent execution. Categories and Subject Descriptors: A. 1 [General Literature]: Introductory and Survey; D. 1.3 [Programming Techniques]: Concurrent Programming--Distributed programming; D.2.11 [Software Engineering]: Software Architectures--Patterns (e.g., client/server, pipeline, blackboard); D.4.5 [Operating Systems]: Reliability-Checkpoint/restart; fault-tolerance; D.4.6 [Operating Systems]: Security and Protection--Cryptographic controls; invasive software General Terms: Reliability, Algorithms, Design, Theory Additional Key Words and Phrases: ACID, agreement problem, asynchronous system, Byzantine failures, commit, crash failures, fault tolerance, malicious places, mobile agents, replication, security, transaction

Internet users increasingly rely on publicly available data for everything from software installation to investment decisions. Unfortunately, the vast majority of public content on the Internet comes with no integrity or authenticity guarantees. This paper presents the self-certifying read-only file system, a content distribution system providing secure, scalable access to public, read-only data. The read-only file system makes the security of published content independent from that of the distribution infrastructure. In a secure area (perhaps off-line), a publisher creates a digitally signed database out of a file system's contents. The publisher then replicates the database on untrusted content-distribution servers, allowing for high availability. The read-only file system avoids performing any cryptographic operations on servers and keeps the overhead of cryptography low on clients, allowing servers to scale to a large number of clients. Measurements of an implementation show that an individual server running on a 550-Mhz Pentium III with FreeBSD can support 1,012 connections per second and 300 concurrent clients compiling a large software package. Categories and Subject Descriptors: D.4.3 [Operating Systems]: File System Management--Distributed file systems; D.4.6 [Operating Systems]: Security and Protection--Authentication; D.4.7 [Operating Systems]: Organization and Design--Distributed systems General Terms: Design, Management, Performance, Security Additional Key Words and Phrases: File systems, read-only, security

This paper provides a comprehensive study and comparison of two state-of-the-art direct solvers for large sparse sets of linear equations on large-scale distributed-memory computers. One is a multifrontal solver called MUMPS, the other is a supernodal solver called SuperLU. We describe the main algorithmic features of the two solvers and compare their performance characteristics with respect to uniprocessor speed, interprocessor communication, and memory requirements. For both solvers, preorderings for numerical stability and sparsity play an important role in achieving high parallel efficiency. We analyse the results with various ordering algorithms. Our performance analysis is based on data obtained from runs on a 512-processor Cray T3E using a set of matrices from real applications. We also use regular 3D grid problems to study the scalability of the two solvers. Categories and Subject Descriptors: G.1.3 [Numerical Analysis]: Numerical Linear Algebra--Sparse, structured, and very large systems (direct and iterative methods); G.4 [Mathematics of Computing]: Mathematical Software--Parallel and vector implementations General Terms: Algorithms, Performance Additional Key Words and Phrases: Sparse direct solvers, parallelism, distributed-memory computers, multifrontal and supernodal factorizations

Communication-induced checkpointing protocols can be used in distributed computing to achieve a property called Rollback Dependency Trackability. This attribute holds that all rollback dependencies between local checkpoints are capable of being monitored on line by using a transitive dependency vector. It is, however, important to study the issue of designing efficient RDT protocols and a few techniques for analyzing existing RDT protocols and comparing their performances are discussed.

The Time Warp optimistic technique is one method of ensuring correct results in parallel and distributed simulations but its efficiency depends on the efficacy of the technique used for state saving and operation. A hybrid method for checkpointing using both periodical and incremental state saving techniques for large Time Warp distributed simulations is presented. It is also analytically compared with the periodical state saving and incremental state saving methods.

The linear properties of quasi-periodic complex-coupled distributed-feedback (DFB) structures, with coupling constants have longtudinal variations along the propapgation axis, have been investigated. The results show that the coupling gradient can provide a shift of the apparent average absorption or of detuning from Bragg resonance.

A tree network consisting of communicating processors is considered. The objective is to minimize the processing time by distributing the processing load to other nodes. The effect of the order of load distribution on the processing time is addressed. An algorithm which optimally determines the order of load distribution is developed. It is shown that the order depends only on the channel capacity between nodes but not on the computing capability of each node.

Chair and Varshney have derived an optimal rule for fusing decisions based on the Bayesian criterion. To implement the rule, the probability of detection [P.sub.D] and the probability of false alarm [P.sub.F] for each detector must be known, but this information is not always available in practice. An adaptive fusion model which estimates the [P.sub.D] and [P.sub.F] adaptively by a simple counting process is presented. Since reference signals are not given, the decision of a local detector is arbitrated by the fused decision of all the other local detectors. Furthermore, the fused results of the other local decisions are classified as 'reliable' and 'unreliable.' Only reliable decisions are used to develop the rule. Analysis on classifying the fused decisions in term of reducing the estimation error is given, and simulation results which conform to our analysis are presented.

The Active Nodal Task Seeking (ANTS) approach to the design of multicomputer systems is named for its basic component: an active nodal task-seeker (ANT). In this system, there is no load balancing or load sharing, instead, each ANT computing node is actively finding out how it can contribute to the execution of the needed tasks. A run-time partition is established such that some of the ANT computing nodes are under exhaustive diagnosis at any given time. An ANTS multicomputer system can achieve a mean time to failure of more than 20 years with just 8 computing nodes and 3 buses, while the minimum requirements are 3 computing nodes and 1 bus, and with a worst case computing node failure rate of 5 x [10.sup.-4]/h. This work has been motivated by the need to develop high-performance multicomputer systems for radar, active and passive sonar, and electronic warfare that can provide ultradependable performance for more than 20 years without field repairs. We argue that high performance is also an attribute of an ANTS computing system, because the overhead of dynamic task scheduling is reduced and because efficient use is made of the available processing resources.

The paper introduces and studies an idea of distributed modeling treating it as a new paradigm of fuzzy system modeling and analysis. This form of modeling is oriented towards developing individual (local) fuzzy models for specific modeling landmarks (expressed as fuzzy sets) and determining the essential logical relationships between these local models. The models themselves are implemented in the form of logic processors being regarded as specialized fuzzy neural networks. The interaction between the processors is developed either in an inhibitory or excitatory way. In more descriptive way, the distributed model can be sought as a collection of fuzzy finite state machines with their individual local first or higher order memories. It is also clarified how the concept of distributed modeling narrows down a gap between purely numerical (quantitative) models and the qualitative ones originated within the realm of Artificial Intelligence. The overall architecture of distributed modeling is discussed along with the detailed learning schemes. The results of extensive simulation experiments are provided as well.

A dynamical model for the distribution of tasks on a system of heterogenous processors is introduced. The task distribution procedure is based on the perceived benefits a task provider sees in using a particular processor for the execution of his requirements. The decision is usually based on a limited amount of information and is therefore a statistical process. The degree of certainty can be regulated by the so-called gain factor and ranges from a total lack of knowledge to a complete knowledge which renders the distribution dynamics a deterministic one. An entropy function is introduced for each task. The values of the task entropies at any moment in time are a measure for the distribution of tasks on the available processors. A second entropy function, the processor entropy, is introduced. The values of the processor entropy give useful information on the real utilisation of the processor system. Some concepts from general equilibrium theory are introduced and their relevance for achieving optimal processor utilisation discussed.