E. Chang. Echo Algorithms: Depth Parallel Operations on General Graphs. IEEE Transactions on Software Engineering, pages 391--401, 1982.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Algorithm Algorithm GME 1 The first solution works in phases. Phases are initiated by root in response to requests made by the processes to use sessions. We use the well known Propagation of Information (PI) and Propagation of Information with Feedback (PIF) schemes to implement the phases[3, 14]. Assume that process p makes the first request for a session X . p records Session X as the requested session and sends a request message for X to its parent. The message is then forwarded to root following the processes in the path from p to root. Upon receipt of the request message, root ....

E. Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391--401, 1982.

.... A wave algorithm is a distributed algorithm where the participation of all processes in the network is required before a particular event, called a decision, is taken [Tel88] Wave algorithms include the token circulation (TC) and the propagation of information with feedback (PIF) Lan77, Cha82, Seg83] TC and PIF algorithms are also referred in the literature to as token traversal[Lan77] and echo algorithms [Cha82] respectively. The purpose of the TC scheme is to implement a token circulation scheme where the token, initiated by a process, in the remainder referred to as Root, is ....

.... a particular event, called a decision, is taken [Tel88] Wave algorithms include the token circulation (TC) and the propagation of information with feedback (PIF) Lan77, Cha82, Seg83] TC and PIF algorithms are also referred in the literature to as token traversal[Lan77] and echo algorithms [Cha82] respectively. The purpose of the TC scheme is to implement a token circulation scheme where the token, initiated by a process, in the remainder referred to as Root, is passed through every process in the network and return it to Root. The PIF scheme can be informally described as follows: Root ....

EJH Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391-401, 1982.

....node, or as part of a composed, more complex pattern, as discussed in the following section. 3.1.2 Constrained RemoteEcho Pattern The constrained remoteEcho pattern is a composition of the scout and the echo pattern. The echo pattern is presented and analyzed in [9] and is based on work in [3]. As this pattern is much more complex than the scout pattern, we abstain from describing its FSM in detail. We rather explain its mode of operation. The original echo navigation pattern comprises two phases of operation. The first or expansion phase is triggered by the network management ....

E. J. H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Transactions on Software Engineering, 8(4):391--401, July 1982.

....and yet valuable assessment of code mobility [12] the authors analyze and compare issues and concepts that belong to different abstraction levels. Similarly, in a recent work about autonomous objects [13] mechanisms like REV [14] and RPC [15] are compared to the Echo distributed algorithms [16], to applications like intelligent e mail and Web browsers, and to paradigms for structuring distributed applications, like mobile agents. We argue that these different concepts and notions cannot be compared directly. It is as inappropriate and misleading as trying to compare the emacs editor, ....

# E.J.H. Chang, "Echo Algorithms: Depth Parallel Operations on General Graphs," IEEE Trans. Software Eng., July 1982.

.... the election prob lem can obviously be solved by first determining the smallest value and then electing the processor which has this value; furthermore, once a leader is elected, a spanning tree can be easily constructed using a distributed (single initiator) graph traversal algorithm (e.g. see [7]) with an additional O(e) messages. Assuming distinct values, various 6The authors do not guarantee the outcome of the experiment if objects other than apples are used. upper and lower bounds for all three problems have been established depending on the actual topology of the network and on the ....

E.J. Chang, Echo algorithms: depth parallel operations on general graphs, IEEE Trans. Softw. Eng. SE-8, 1982, 391-400.

.... of total algorithms was introduced in [31] Such algorithms can be implemented in various ways; examples include a control message circulating on a (virtual) ring connecting all processes (see also Raynal and H elary [25] parallel distributed graph traversal schemes such as the echo algorithm [8], and (virtual) broadcast schemes on a spanning tree. Unfortunately, however, the values of the state indicators collected in that way do not allow the conclusion that the basic computation has terminated. Because of possible reactivations of processes behind the back of the wave, the ....

Chang, E. J.-H. Echo algorithms: Depth parallel operations on general graphs. IEEE Trans. Softw. Eng. SE--8 (1982), 391--401.

....In an early and yet valuable assessment of code mobility [46] the authors analyze and compare issues that belong to different abstraction levels. Analogously, in a recent work about autonomous objects [14] mechanisms like REV [100] and RPC [15] are compared to the Echo distributed algorithms [26], to applications like intelligent e mail and Web browsers, and to paradigms for structuring distributed applications, like mobile agents. These are different concepts and notions that cannot be compared directly. It is almost like comparing the emacs editor with the fork UNIX system call and ....

Chang, E. Echo Algorithms: Depth Parallel Operations on General Graphs. IEEE Trans. on Software Engineering (July 1982).

....stabilization time (0 rounds) and in the state each processor is required to have. Keywords: Distributed systems, fault tolerance, PIF, self stabilization, snap stabilization, wave algorithms. 1 Introduction The wave scheme is a fundamental and widely used approach in distributed computing [Cha82, Seg83] The concept of wave can be used to solve various versions of the following two problems: the token circulation (TC) problem (also, called the Token Traversal problem) and the propagation of information with feedback (PIF) problem. The solution to these basic problems (TC and PIF) can ....

EJH Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391-401, 1982.

....algorithm can be easily used to implement any self stabilizing system which requires a (self stabilizing) wave protocol running on an arbitrary network. Keywords: Fault tolerance, propagation of information with feedback, reset, self stabilization, snapshot, wave algorithms. 1 Introduction Chang [11] and Segall [19] defined the concept of Propagation of Information with Feedback (PIF) also called wave propagation) A node, called a root, initiates the first phase (propagation or broadcast) of the wave. Every node, upon receiving the first broadcast message, chooses the sender of this message ....

E. Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391--401, 1982.

.... et e largement etudi ees et utilis ees en algorithmique distribu ee, notamment en ce qui concerne la d etection de leur terminaison, et ce, sous diff erentes appellations. Dans [3] les auteurs proposent une technique g en erale de d etection de terminaison pour ces diffusing computations. Dans [2], l id ee d un retour vers l instigatrice (unique) est propos ee pour s assurer de la terminaison de la vague, et la technique est appel ee echo algorithms. Dans [6] l auteur etudie les protocoles PI (Propagation of Information) demi vague) et PIF (Propagation of Information with Feedback) ....

E. J. H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE 8(4):391--401, July 1982.

....intermediate node waits for a message from all its children. Next it sends a message to its own father. The root of the diffusion tree (i.e. the entity that began the PI) knows that the protocol is ended when it has received a message from all its neighbors. This protocole has also been studied in (Chang, 1982) under the name of echo algorithms. In (H elary et al. 1987) an improvement in term of number of messages of these protocols is proposed. In (Raynal and H elary, 1988) they are used to compute a global function, on several topologies and with several communication hypotheses. In (Tel, 1987) ....

Chang, E. J. H. (1982). Echo algorithms: Depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE 8(4):391--401.

....with a new proof. When the binary operator is not idempotent, the global computation needs a leaves to root rooted spanning tree, as explain above for the plus Gammadirect Gammaassociation. This technique is similar to Propagation Information with Feedback [Segall, 1983] and to echo algorithms [Chang, 1982], which are powerful ways to build distributed algorithms . Other comparisons could be done with parallel programming paradigms. Indeed, since a direct Gamma association with an s operator gives the same result for each entity of a strongly connected component, this primitive can be compared ....

Chang, E. J. H. (1982). Echo algorithms: Depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE 8(4):391--401.

....Data and Structures 2 the most general technique to design a system to tolerate arbitrary transient faults. A self stabilizing system, regardless of the initial states of the processors and initial messages in the links, is guaranteed to converge to the intended behavior in finite time. Chang [12] and Segall [17] defined the concept of Propagation of Information with Feedback (PIF) also called wave propagation) The PIF scheme was also independently introduced in [15] The PIF scheme can be informally described as follows: A node initiates a wave, called the propagation wave. Every node, ....

....set X . Definition 1 (Snap stabilization) The protocol P is snap stabilizing for the specification SP P on E if and only if the following condition holds: 8a 2 C : 8e 2 E a : e SP P . 3 Propagation of Information With Feedback (PIF) Scheme Let us quickly review the well known PIF scheme [12, 17] on tree structured networks. The PIF scheme is the repetition of a PIF cycle. The PIF cycle can be informally defined as follows: Starting from an initial configuration where no message has yet been broadcast (Figure 1 (i) the root initiates the broadcast phase and its descendants (except the ....

CHANG, E. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering SE-8 (1982), 391--401.

....condition for a deadlocked task is: a blocked task T is deadlocked if T is in a knot or T can only reach deadlocked tasks. There are similarities between the detection of an OR model deadlock and the detection of the termination of a group of cooperating tasks in a distributed computation[1, 2, 16, 19, 18, 23, 24, 46]. In a distributed system tasks cooperate with each other in a computation by means of message exchange. A distributed computation is said to be globally terminated if it reaches a final state which, in turn, relies on its member tasks reaching their final states and being ready to terminate. The ....

....in CSP environments, and proposed a detection algorithm for them. Dijkstra and Scholten[19] introduced the notion of diffusing computation (See Section 5.2.3) and suggested an algorithm to detect the termination of an arbitrary diffusing computation in any network environment. Cohen and Lehmann[16] extended Francez s CSP termination model and solution to distributed systems where new cooperating tasks are created and terminated dynamically. Misra and Chandy[46] discussed how a termination computation algorithm can detect deadlock for the OR model. And then in [47] Misra and Chandy presented ....

[Article contains additional citation context not shown here]

E. J. H. Chang, "Echo Algorithms: Depth Parallel Operations on General Graphs," IEEE Transactions on Software Engineering, Vol. SE-8, No. 4, pp. 391--401, July 1982. 71

....early and yet valuable assessment of code mobility [12] the authors analyze and compare issues and concepts that belong to different abstraction levels. Similarly, in a recentwork about autonomous objects [13] mechanisms like REV [14] and RPC [15] are compared to the Echo distributed algorithms [16], to applications like intelligent e mail and Web browsers, and to paradigms for structuring distributed applications, like mobile agents. We argue that these different concepts and notions cannot be compared directly. It is as inappropriate and misleading as trying to compare the emacs editor, ....

E.J.H. Chang, "Echo Algorithms: Depth Parallel Operations on General Graphs," IEEE Trans. on Software Engineering,July 1982.

....maximum compression rate is achieved in sparse matrices networks with many nodes of degree 2. 4. 2 Step 2: Determine the Fundamental Cycle Matrix Spanning Tree T : For determining the fundamental cycle matrix C f ,wefirst compute a spanning tree T over the graph G with a distributedecho algorithm [2,3,6].Echo algorithms havetwo phases: a forward phase, where the network is flooded with explorer messages, and an echo phase, where the receipt of explorers is acknowledged by returning echo messages. The echo algorithm is initiated by sending explorer messages from an arbitrary 1 reference node to ....

E.J.H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Trans. Softw. Eng. SE-8,4(July 1982), 391--401.

....with bounded (O(n) storage, our protocol improves upon the known, classical broadcast protocols, Echo, PIF and Intelligent Flood, used in many actual networks. Details follow. Furthermore, these classical protocols work only for static asynchronous networks. The Echo and PIF protocols [Cha82, DS80, Seg83] have lower throughput. Namely, the maximal time required between broadcasting two messages using Echo is proportional to the diameter of the network, compared to a constant in our protocol It is simple to improve the throughput of PIF using a window, however this seems to increase ....

....hand, the protocol should have high throughput (rate) To achieve this, the source does not wait for network wide progress, before enabling a new accept event. Instead, the source waits only for progress of its immediate neighbors. This should be contrasted with the Echo and PIF protocols [DS80, Cha82, MRR80, Seg83] that wait before accepting a new message until an entire spanning tree converges. broad:prot update:January 3, 1994 L a T E X:August 26, 1996 24 On the other hand, the protocol should not cause high congestion on links. High congestion may be caused, for example, by the ....

Ernest J. H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Trans. Software Eng., 8(4):391--401, July 1982.

....thus detecting a phantom deadlock. The algorithm is designed for the single resource model. It makes the additional assumption that messages arrive in the order in which they were sent. 4.5. 2 Diffusing Computation In algorithms in this class a transaction starts a diffusing computation [DS80,Cha82] when it has to wait for a lock. A deadlock is indicated, if the computation terminates. A node of a directed graph starts a diffusing computation by sending messages to its successors. Upon receiving such a message a node can send messages to its successors, and so on. In order to determine ....

....computation by sending messages to its successors. Upon receiving such a message a node can send messages to its successors, and so on. In order to determine when a diffusing computation terminates, nodes can also receive signals from their successors and send signals to their predecessor [DS80, Cha82] Based on the signals it receives, the initiator can decide whether it is deadlocked or not. Using diffusing computation is an overkill for the detection of single resource and AND model deadlocks. In these models a deadlock is indicated by a probe that returns to its initiator, which induces ....

E. J. H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Trans. Software Eng., 8(4):391--401, 82.

No context found.

E. Chang. Echo Algorithms: Depth Parallel Operations on General Graphs. IEEE Transactions on Software Engineering, pages 391--401, 1982.

No context found.

E. Chang, Echo algorithm: depth parallel operations on general graphs, IEEE Transactions on Software Engineering 8 (4) (1982) 391--401.

No context found.

E. Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391401, 1982.

No context found.

E. Chang. Echo Algorithms: Depth Parallel Operations on General Graphs. IEEE Transactions on Software Engineering, pages 391--401, 1982.

No context found.

EJH Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391-401, 1982.

No context found.

EJH Chang. Echo algorithms: depth parallel operations on general graphs. IEEE Transactions on Software Engineering, SE-8:391-401, 1982.

No context found.

E.J.H. Chang. Echo algorithms: Depth parallel operations on general graphs. IEEE Trans. Softw. Eng. SE-8,4#July 1982#, 391#401.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC