Busy-wait techniques are heavily used for mutual exclusion and barrier synchronization in shared-memory parallel programs. Unfortunately, typical implementations of busy-waiting tend to produce large amounts of memory and interconnect contention, introducing performance bottlenecks that become markedly more pronounced as applications scale. We argue that this problem is not fundamental, and that one can in fact construct busy-wait synchronization algorithms that induce no memory or interconnect contention. The key to these algorithms is for every processor to spin on separate locally-accessible ag variables, and for some other processor to terminate the spin with a single remote write operation at an appropriate time. Flag variables may be locally-accessible as a result of coherent caching, or by virtue of allocation in the local portion of physically distributed shared memory. We present a new scalable algorithm for spin locks that generates O(1) remote references per lock acquisition, independent of the number of processors attempting to acquire the lock. Our algorithm provides reasonable latency in the absence of contention, requires only a constant amount of space per lock, and requires no hardware support other than

A fault-tolerant distributed mutual exclusion algorithm which adjusts to node mobility is presented, along with proof of correctness and simulation results. The algorithm requires nodes to communicate with only their current neighbors, making it well-suited to the ad hoc environment. Experimental results indicate that adaptation to mobility can improve performance over that of similar non-adaptive algorithms when nodes are mobile.

this paper, we have used the terms mobile and portable computing interchangeably. However, strictly speaking, we view portable computing as a more restrictive form of mobile computing: though a portable computer may connect to the network from different locations, it essentially connects to the network at any given access point through an individual wired connection and therefore, cannot simultaneously move and maintain its connection to the network. The notion of a "cell" and the associated wireless communication with a broadcast capability, is absent from portable computing.

bound D, the bounded-diameter minimum spanning tree problem seeks a spanning tree on G of lowest weight in which no path between two vertices contains more than D edges. This problem is NP-hard for 4 1, where n is the number of vertices in G. An existing greedy heuristic for the problem, called OTTC, is based on Prim's algorithm. OTTC usually yields poor results on instances in which the triangle inequality approximately holds; it always uses the lowest-weight edges that it can, but such edges do not in general connect the interior nodes of low-weight boundeddiameter trees. A new randomized greedy heuristic builds a bounded-diameter spanning tree from its center vertex or vertices. It chooses each next vertex at random but attaches the vertex with the lowest-weight eligible edge. This algorithm is faster than OTTC and yields substantially better solutions on Euclidean instances. An evolutionary algorithm encodes spanning trees as lists of their edges, augmented with their center vertices. It applies operators that maintain the diameter bound and always generate valid o#spring trees. These operators are e#cient, so the algorithm scales well to larger problem instances. On 25 Euclidean instances of up to 1 000 vertices, the EA improved substantially on solutions found by the randomized greedy heuristic.

This report presents a token-based K-mutual exclusion algorithm. The algorithm uses K tokens and a dynamic forest structure for each token. This structure is used to forward token requests. The algorithm is designed to minimize the number of messages and also the delay in entering the critical section, at low as well as high loads.

Distributed synchronization is an essential component of parallel and distributed computing. Several fast and low-overhead distributed synchronization algorithms have been proposed. Each of these algorithms required O(log n) messages per critical section entry and O(log n) bits of storage per processor. Asymptotic performance estimates do not proclaim any of the algorithms to be a winner, but no performance comparison of the algorithms has yet been performed. In this paper, we make a comparative performance study of four distributed semaphore algorithms. Each of these algorithms represents a different approach to maintaining distributed information. Since the algorithms we study are the basis for distributed synchronization, distributed virtual memory, coherent caches, and distributed object systems, our results have implications about the best methods for their implementation. We find that the distributed synchronization algorithm of Chang, Singhal, and Liu has the overall best perfor...

Distributed multipoint applications for group interaction across wide-area networks, such as for simulation and telecollaboration, are becoming increasingly popular. While reliable multicasting has made significant advances in recent years, effective mechanisms to synchronize and coordinate work within large multicast groups and across long distances are still lacking. Synchronous sharing of resources, whose operational semantics prohibits parallel usage, typically creates race conditions among users, which can be resolved through an access discipline called floor control. Existing solutions on floor control, implemented either at the session or application layer, are mostly proprietary, limited in scope and not scalable. Furthermore, no performance comparison of floor control protocols has been attempted to date. We present a novel taxonomy and comparative performance analysis of known classes of floor control protocols, ranging from socially mediated control to protocols operating on ring and tree topologies. We find that aggregation and selective transmission of control information in a tree structure is the most promising solution with regard to scalability, efficacy, and robustness. The principal operation of such a tree protocol is outlined, which dynamically organizes participants in a multi-level control tree and aggregates resource sharing directives on the paths between interacting stations.

The bounded diameter minimum spanning tree problem is an NP-hard combinatorial optimiza-tion problem with applications in various fields like communication network design. We propose a general variable neighborhood search approach for it, utilizing four different types of neighbor-hoods. They were designed in a way enabling an efficient incremental evaluation and search for the best neighboring solution. An experimental comparison on instances with complete graphs with up to 1000 nodes indicates that this approach consistently outperforms the so far leading evolutionary algorithms with respect to solution quality and computation time.

INTRODUCTION. Over the last decade distributed computing systems have attracted a great deal of attention. This is due, in part, to the technological advances in the design of sophisticated software and communication interfaces, the availability of low-cost processors and the rapid decline in hardware costs. The motivations for building distributed computing systems are many. Resource sharing, parallel processing, system availability and communication are four major reasons. By distributing a computation among various sites, processes are allowed to run concurrently and to share resources, but still work independently of each other. Many distributed computations involving the sharing of resources among various processes require that a resource be allocated to a single process at a time. Therefore, mutual exclusion is a fundamental problem in any distributed computing system. This problem must be solved to synchronize the access to shared resources in order to maintain their consist

Abstract not found