Search

The distributed constrained state estimation (DCSE) and distributed optimal power flow (DOPF) problems are some kinds of large-scale block-additive constrained optimization problem (BCOP) in their exact formulation. In this paper, we propose a general form methodology, an efficient distributed asynchronous dual-type method, to cope with the computational difficulty of these problems. Our method achieves some attractive features, including being computationally efficient and numerically stable. The computational formulae of our method are explicit, compact, and easy to be programmed. We have demonstrated the computational efficiency of our method, by comparing with the state-of-the-art algorithm through the implementation of two real different types PC-networks in asynchronous computing environment on the IEEE 118-bus and IEEE 244-bus systems. Index Terms--Asynchronous computing environment, distributed asynchronous dual-type method, distributed constrained state estimation, distributed optimal power flow (DOPF), PC-networks.

A compile-time scheduler that exploits both parallelism and pipelining is described and discussed.

A logical hypercube structure is used to obtain scalability and fault-tolerance in distributed systems.

Linda is a parallel and distributed programming model. An extension of the Linda model is proposed, providing a general solution to the problem of false matching.

A distributed Java Virtual Machine, called DISK (DIStributed Kaffe), is discussed. DISK is used with networks of heterogeneous workstations.

When a group {I, ..., [I.sub.n]} of individuals wishes to collaboratively construct a complex multimedia document, the first requirement is that they be able to manipulate media-objects created by one another. For instance, if individual [I.sub.j] wishes to access some media objects present at participant [I.sub.k]'s site, he must be able to: 1) retrieve this object from across the network, 2) ensure that the object is in a form that is compatible with the viewing/editing resources he has available at his node, and 3) ensure that the object has the desired quality (such as image size and resolution). Furthermore, he must be able to achieve these goals at the lowest possible cost. In this paper, we develop a theory of media objects, and present optimal algorithms for collaborative object sharing/synthesis of the sort envisaged above. We then extend the algorithms to incorporate quality constraints (such as image size) as well as distribution across multiple nodes. The theoretical model is validated by an experimental implementation that supports the theoretical results. Index Terms - Multimedia networks, active networks, multimedia object delivery, quality of service guarantees, multimedia collaboration, multimedia authoring, distributed systems.

The topological design of distributed packet switched networks consists of finding a topology that minimizes the communication costs by taking into account a certain number of constraints such as the delay and the reliability. This paper proposes a genetic algorithm (GA) for generating low-cost feasible computer network topologies subject to these constraints. The implementation of this algorithm has been subjected to extensive tests in order to measure the quality of solutions. Computational results confirm the efficiency of the GA to provide good solutions for medium-sized computer networks, in comparison with well-tried conventional methods. Index Terms - Computer networks, distributed systems, genetic algorithms, heuristic method, topological design.

A study was conducted on a distributed system problem involving the replicating and scheduling periodic tasks in a multiple processor system. Several modules were considered for replication regardless of whether a failure did actually occur or not to obtain a high probability os successful completion of the tasks in the system. The results showed an improvement over other previous related work.

An investigation of data distribution in distributed memory parallel computers reveals that systems benefit from data redistribution to perform a series of Do-loops only when the communication cost of executing the Do-loops is greater than a determined threshold value. Proposed dynamic programming algorithms can establish efficient data distribution schema for Do-loops. Experimental studies demonstrate the importance of establishing the necessity for data redistribution.

Researchers examined methods of minimizing subtask execution times and communication times in a heterogeneous computer network. The subtasks in question are the routines which enable an application program to run on the different machines in the system, and the duration of communication times between machines can have a significant effect on overall system performance. The researchers developed a mathematical framework to represent the processes involved, and derived an algorithm to give the best scheduling and data relocation times for any subtask/machine matching.

Anonymous networks have processors that do not have identity numbers. Distributed algorithms are used to solve various problems about networks, which include leader election, edge detection, spanning tree construction and topology recognition. These problems may or may not be solved, depending on network attributes such as no information, upper bound on network size, network size and network topology. A graph property designated as network symmetricity is introduced to describe the solvable classes of networks according to graph theory.

Anonymous networks have processors that do not have identity numbers. Network symmetricity is a graph property which is important in analyzing anonymous networks and solving various representative problems using distributed computing. The determination of graph symmetricity entails complex computation, and depending on the class of the network, this complexity may vary from P-time to NP-complete or co-NP-complete.

A new method for the automatic parallelization of sequential programs is developed for application to a distributed memory message passing type multiprocessor. This new method incorporates all the problems involved in automatic parallelization. The method is developed based on the various techniques used for adapting shared memory multiprocessors for specific applications. The method separates the loop nests of the multiprocessors into different groups for easier classification.

We analyze a distributed asynchronous algorithm, proposed by Tsitsiklis and Bertsekas, for optimal routing in a virtual-circuit data network. We show that, under a strict convexity assumption on the link delay functions, the sequence of routings generated by the algorithm converges in the space of path flows and the convergence rate is linear. Our analysis is based on estimating the distance from a routing to the set of optimal routings and, for the synchronous case, it gives an explicit estimate of the convergence ratio in terms of the network parameters.

A novel algorithm in O(mu) time for the module allocation problem is asymptotically optimal and quicker than general algorithm of Bokhari. The problem is executed under conditions of uniform communication costs and the existence of undirected communication graph as a tree. The solution differs from the generalized algorithm of Fernandez-Baca which has the communication graph as a partial K-tree.

Communication problems hindering efficient information routing in distributed sensor networks are examined in terms of the delay constraint and the reliability constraint. It is shown that the maximum end-to-end lag is a function of the network's diameter for which the SDIA and MDIA diameter finding algorithms are introduced. A discussion on link/node failures and other network topological design standards concerning reliability and delay is also given.

A discussion is presented of the difficulties encountered in the topological design of a network with a two-level hierarchical structure and a full-meshed embedded backbone network. The problem is treated as a quadratic 0-1 programming model and as a linear one as well. The proposed efficient solution assumes that the latter is a version of the uncapacitated facility problem. The solution can also be used to design large data communication and transportation systems.

An important measure of the performance of an interconnection network is its capacity. The capacity of an [Mathematical Expression Omitted] binomial network is discussed as a measure of the average number of inputs that can be connected to the outputs. It is shown that for input subset of size less than or equal to d + 2, all inputs can be connected to outputs. For input subset of size greater than d + 2, the network is considered in terms of its prime decomposition class (pdc). We present an empirical formula for the capacity of [Mathematical Expression Omitted] networks based on the number of configurations and capacity contribution of different factors in the pdc. It is shown that an [Mathematical Expression Omitted] binomial network can provide capacity close to that of the complete crossbar switch while avoiding the latter's expensive hardware cost of contact switches. Next, we suggest a method to maximize the capacity of binomial network when selecting inputs. We illustrate how concepts of constant weight codes and residue classes can be applied to construct a type of [Mathematical Expression Omitted] network called Optimally Designed Binomial (ODB) Network. We present an algorithm which selects inputs corresponding to different residue classes and uses the left cyclic shift feature to optimize the capacity of the network.

Factors related to solving the mutual exclusion problem in distributed processing are examined. These include coteries which are groups of node subsets which in turn are termed quorums. The join algorithm, which constructs a composite coterie from nonempty coterie inputs, is presented. It composes both dominated and nondominated coteries without a priori information about the composite coterie's quorum. The generalization of the binary tree protocol via the join algorithm is also discussed.

A programming and communications system which simulates shared memory on a heterogeneous computer network to accommodate synchronous and asynchronous distributed control systems is proposed. The Hetero Helix automatically implements the communication system for the network which produces an increase in reliability and maintainability. The system's compiler bases this automatic generation on an analysis of the data structures in the shared memory.