Baruch Awerbuch, Boaz-Patt Shamir and George Varghese, "Self-Stabilizing End to End Communication", In 25th Symposium on Theory of Computing , 1993  Home/Search   Document Details and Download   Summary   Related Articles   Check

....each computer may communicate directly with any other computer. This variant of searching with uncertainty, was introduced in [15] where the network topologies were the ring and the torus. Models with faulty information in the nodes have been considered before for the problem of routing (see [1, 3, 6, 8, 16]) However, in this problem it is assumed that the identity of the node that contains the information is known, and what is required is to reach this node following the best possible route. Search problems in graphs, where the identity of the node that contains the information sought is not known, ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese, Self-stabilizing end-to-end communication, Journal of High Speed Networks 5 (1996), 365--381.

....increased link traffic. Search problems that incorporate some degree of uncertainty or nondeterminism have been considered extensively in the past. One of the best known and most thoroughly studied is the routing problem for various computer network topologies in the presence of faulty nodes (see [2, 4, 8, 9, 14]) The routing problem differs from the problem we consider here in that in the former problem the identity of the target node is known in advance and what is required is to follow a shortest path to reach it, when some of the nodes or the interconnections may be faulty. On the other hand, graph ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese (1996) Self-stabilizing end-to-end communication. Journal of High Speed Networks, 5, 365--381.

....queries with a bounded probability of being incorrect. Additional investigations under the same model of searching with uncertainty were carried out for fully interconnected networks in [11] Models with faulty information in the nodes have been considered before for the problem of routing (see [1, 3, 7, 8, 15]) However, in this problem it is assumed that the identity of the node that contains the information is known, and what is required is to reach this node following the best possible route. Search problems in graphs, where the identity of the node that contains the information sought is not known, ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese, Self-stabilizing end-to-end communication, Journal of High Speed Networks 5 (1996), 365-381.

....Ainsi, le d ebit de transfert de donn ees d un protocole bas e sur l approche pr esent ee dans [Her92] se r eduit peu a peu a un let d eau . 4. 4 Protocole de transmission de donn ees via plusieurs chemins Des protocoles de communication point a point auto stabilisants sont pr esent es dans [APSV96, DW97] Le r eseau consiste en un ensemble de n uds routeurs connect es par les canaux FIFO de taille born e pouvant perdre des messages mais pas les dupliquer. Ces protocoles utilisent plusieurs chemins entre l emetteur et le destinataire; ainsi m eme si un chemin est hors service, la ....

....de n uds routeurs connect es par les canaux FIFO de taille born e pouvant perdre des messages mais pas les dupliquer. Ces protocoles utilisent plusieurs chemins entre l emetteur et le destinataire; ainsi m eme si un chemin est hors service, la d elivrance des paquets n est pas stopp ee. Dans [APSV96] il est suppos e que chaque canal de communication est un UDL. Un UDL est un canal de taille 1, ayant donc au plus un paquet en transit. Sachant qu un UDL peut etre r ealis e sur n importe quel canal de taille born e via une version autostabilisante du protocole du bit altern e [AB93] cette ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese. Self-stabilizing end-to-end communication. Journal of High Speed Networks, 5(4):365-381, 1996.

....which with high probability finds an item in O(d log n) steps. To the best of our knowledge these are the first results to be reported in 3 a model of searching as described above. We note that a significant amount of research has been done on routing in networks with random or malicious faults [1, 3, 6, 8, 14, 18] but this work does not appear to capture the search problem considered here. In terms of the restaurant finding problem, routing under faults corresponds to the problem where the exact address of the restaurant is known and one wishes avoid roadblocks or traffic jams. Other related lines of ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese, "Self-stabilizing endto -end communication," Journal of High Speed Networks, 5 (1996), 365381.

....increased link traffic. Search problems that incorporate some degree of uncertainty or nondeterminism have been considered extensively in the past. One of the best known and most thoroughly studied is the routing problem for various computer network topologies in the presence of faulty nodes (see [7, 8, 9, 10, 11]) The routing problem differs from the problem we consider here; in the former problem the identity of the target node is known in advance, and what is required is to follow a shortest path to reach it, when some of the nodes or the interconnections may be faulty. On the other hand, graph search ....

Awerbuch, B., Patt-Shamir, B. and Varghese, G. (1996) Self-stabilizing end-to-end communication. J. High Speed Networks, 5, 365--381.

....such links at most g times (g a constant) Once a direct link is established, the nodes may use it to exchange information for as many times as they wish. Many different protocols for end to end communication have been developed when links (and or intermediate nodes) may fail ( 1] 2] 3] 4] [5], 6] 8] 9] Also, previous work of the authors examined structural properties of unreliable networks ( 10] 11] 12] In this paper, we present a simple protocol which establishes (almost surely) a very long path in the network (involving a constant fraction, at least, of all the nodes) ....

Awerbuch, Patt-Shamir and Varghese. Self-Stabilizing End-to-End Communication. J. of High Speed Networks, vol. 5, no 4, 365--381, 1996.

....types occurring at any time and is insensitive to dynamic topology reconfiguration. This makes self stabilization an interesting solution for a large number of applications including distributed reset problems (for the most interesting solutions see [3,5] routing and communication protocols ([7,21, 29,32,44]) clock synchronization ( 19,22,40,52] graph theory ( 12,40,41,43,50,57] and others. Since the very first presentation of a self stabilizing algorithm, made by Dijkstra in 1974, progress made in this area has proved that it is one of most important and promising topics of fault tolerance. ....

B. Awerbuch, B. Patt-Shamir, G. Varghese, "Self Stabilizing End-to-End Communication", Journal of High Speed Networks, 1995.

....seventh phase will be a local reset and hence L u;v will hold after 7 phases. Since data packets are only sent in between phases of checking correction, this justifies our claim that only a constant number of bad packets are sent per link. The formal code for our stabilizing Slide is given in [APV93]. The code for the basic Slide protocol is almost identical to the code described in [AGR92] except for the use of the sending discipline described earlier to deal with the bounded capacity links. The code for the stabilizing reset and snapshot procedures follows the informal description given ....

....receiver must also flow control the network by sending a free signal to the network when RMQ is empty. The Slide code at the receiver node also has to be modified to only send packets when the receiver message queue is thought to be free. More details of these two technicalities can be found in [APV93]. 5.3 Improved Efficiency for Large Messages The simplest message delivery protocol is to use the majority encoder we described earlier Since each packet within a message can travel n hops, the resulting communication complexity is O(Cn) O(n 2 m) and the normal and stabilization time ....

[Article contains additional citation context not shown here]

Baruch Awerbuch, Boaz Patt-Shamir, and George Varghese. Self-stabilizing end-to-end communication. Technical Memo to appear, MIT, Lab. for Computer Science, October 1993.

....total number of bits contained in the packet. A natural question to ask is the following: how large must packet headers be for reliable communication This problem is addressed in [AF99] part of a large body of research on the end to end communication problem [AAF 94] AAG 97] AG88] AMS89] [APV96], DW97] KOR95] LLT98] F98] The end to end communication problem is to send information from one designated processor (the sender S) to another designated processor (the receiver R) over an unreliable communication network. This is a fundamental problem in distributed computing, since (a) ....

B. Awerbuch, B. Patt-Shamir, and G. Varghese, Self-stabilizing End-to-End Communication, Journal of High Speed Networks, 5(4), (1996), 365-381.

....total number of bits contained in the packet. A natural question to ask is the following: how large must packet headers be for reliable communication This problem is addressed in [AF99] part of a large body of research on the end to end communication problem [AAF 94] AAG 97] AG88] AMS89] [APV96], DW97] KOR95] LLT98] The end to end communication problem is to send information from one designated processor (the sender S) to another designated processor (the receiver R) over an unreliable communication network. This is a fundamental problem in distributed computing, since (a) ....

Awerbuch, Patt-Shamir, and Varghese, Self-stabilizing End-to-End Communication,

....With this assumption, we can define the following metrics for a stabilizing end to end protocol N for graph G. Intuitively, Stabilization time is the worst case time it takes N to stabilize; Message delivery time is the worst case time to deliver a message. Formal definitions can be found in [APV95]. Space is the the maximal amount of space required by a node program. 3 Reducing a Fail Stop Network to a C Channel using Slide The (non stabilizing) end to end protocol described in [AGR92] has two parts: first part is the Slide protocol that is used to reduce a fail stop network to a non FIFO ....

....responses at u if they match the number of the last request sent. Using this we can show [Var93] that within 4 invocations of the snapshot protocol that the snapshot reset protocol will begin correctly even after starting in an arbitrary state. The formal code for our stabilizing Slide is given in [APV95]. We do not provide the details here for lack of space and because our changes to the Slide protocol are quite simple. Instead we concentrate next on providing details of the second part of our solution. 4 Stabilizing Bounded Channel Protocol We turn to the second part of our solution, that of ....

[Article contains additional citation context not shown here]

Baruch Awerbuch, Boaz Patt-Shamir, and George Varghese. Self-stabilizing end-to-end communication. Technical Memo to appear, MIT, Lab. for Computer Science, October 1995.

....of our local correction theorem. Our local correction method and theorem has been used to construct stabilizing solutions to a number of non trivial applications including a network reset protocol[APV91, Var93] a synchronizer [AV91] and a stabilizing solution to the end to end problem [APV96] We refer the reader to those papers for these applications. In this section, we give a complete example of the Local Correction theorem to a simple application:token passing. Token protocols that stabilize in time proportional to the height of the tree have been described before by [DolevIM90] ....

....protocol with snapshot and reset actions. We gave an example application of stabilizing a mutual exclusion protocol. Other, more non trivial applications include stabilizing a network reset protocol[APV91, Var93] a synchronizer [AV91] and a stabilizing solution to the end to end problem [APV96] Protocols that are locally checkable but work on a tree topology can be stabilized in time proportional to the height of the tree. Thus we can remove the need for local correctability if the underlying topology is a tree. Another way to remove the need for local correctability (without ....

Baruch Awerbuch, Boaz Patt-Shamir, and George Varghese. Self-stabilizing end-to-end communication. Journal of High Speed Networks, 5(4), October 1996.

No context found.

Baruch Awerbuch, Boaz-Patt Shamir and George Varghese, "Self-Stabilizing End to End Communication", In 25th Symposium on Theory of Computing , 1993

No context found.

Awerbuch, Patt-Shamir, and Varghese, Selfstabilizing End-to-End Communication, Journal of High Speed Networks, vol. 5, no. 4, 1996, pages 365--381.

No context found.

Awerbuch, Patt-Shamir, and Varghese, Selfstabilizing End-to-End Communication, Journal of High Speed Networks, vol. 5, no. 4, 1996, pages 365--381.

No context found.

B Awerbuch, B Patt-Shamir, and G Varghese. Self-stabilizing end-to-end communication. Journal of High Speed Networks, 5(4):365--381, 1996.

No context found.

B. Awerbuch, B. Patt-Shamir, and G. Varghese, Self-stabilizing End-to-End Communication, Journal of High Speed Networks, vol. 5, no. 4, 1996, pages 365--381.

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