D. Dolev, J. Halpern, and H. R. Strong, " On the Possibility and Impossibility of Achieving Clock Synchronization", J. of Computer and Systems Science, Vol. 32:2, pp. 230-250, 1986.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....assumption like at most f nodes may appear Byzantine also implies at most f pairs of perceptions may appear Byzantine , for any pair of non faulty receiving nodes. A perception based fault model thus preserves the corresponding global one, which means that classic impossibility results like [4, 3] remain valid. Node 1 (send) recv) Node s . faulty node non faulty q ) p ) Figure 3. Effect of a link fault between sender s and receiver p upon the perception vectors of all receivers r. Only V p is faulty here. 3 Formal Framework In order to formalize our ....

D. Dolev, J. Y. Halpern, and H. R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Sciences, 32:230--250, 1986.

....like at most f nodes may appear Byzantine also implies at most f pairs of perceptions may appear Byzantine , for any pair of non faulty receiving nodes. A perception based fault model thus preserves the corresponding global one, which reveals that classic impossibility results like [4, 3] remain valid. Node 1 (send) recv) Node s . faulty node non faulty q ) p ) Figure 3. Effect of a link fault between sender s and receiver p upon the perception vectors of all receivers r. Only V p is faulty here. Some recent papers [21, 28] reveal that other ....

....assumption like at most f nodes may appear Byzantine also implies at most f pairs of perceptions may appear Byzantine , for any pair of non faulty receiving nodes. A perception based fault model thus preserves the corresponding global one, which means that classic impossibility results like [4, 3] remain valid. 3 Formal Framework In order to formalize our perception based fault model, we assume that all nodes s, 1 s n, are somehow provided with a (virtual) event V ] that occurs at some specified in any suitable global time scale. Besides its pure occurrence, V ] may or may ....

D. Dolev, J. Y. Halpern, and H. R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Sciences, 32:230--250, 1986.

....of failures and clock drift was published in [7] where denotes the maximum number of processes, and C=E:G CFIRJ are the maximum and the minimum communication delays, respectively. An upper and lower bound for the maximum deviation achievable by any synchronization algorithms was derived in [3]. The authors show that the temporal imprecision between the two points in real time when the clocks of two correct processes show the same value, is at least S ; TU V W and it is at most C=E7G=HYC I2J . The effect of the drift of correct hardware clocks was neglected. The bounds derived ....

....authors show that the temporal imprecision between the two points in real time when the clocks of two correct processes show the same value, is at least S ; TU V W and it is at most C=E7G=HYC I2J . The effect of the drift of correct hardware clocks was neglected. The bounds derived in [3] are not directly applicable to convergence function based algorithms. Indeed, convergence function based algorithms restrict the form of communication that processes can use to synchronize their clocks in the sense that processes do only communicate to read each others clock. In particular, they ....

[Article contains additional citation context not shown here]

D. Dolev, J. Y. Halpern, and R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Science, 32(2):230--250, 1986.

....the local reading error is negligible, i.e. when process is correct at time , then 3 H F ( 3 : 2.2 Failure Hypotheses Each correct process has by definition a correct hardware clock. The total number of processes participating in the clock synchronization must be at least 3 1 [2], where denotes the maximum number of processes that can be faulty. For simplicity, we assume that no failed process recovers. We assume that processes and clocks fail in arbitrary ways. The reading error for approximating a clock of a process which has suffered an arbitrary failure is not ....

D. Dolev, J. Y. Halpern, and R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Science, 32(2):230--250, 1986.

....clocks achievable by an external internal clock synchronization algorithm in degraded mode. In systems without access to reference time and without message authentication, at least 1 time servers are needed to guarantee internal clock synchronization despite arbitrary time server failures [6]. We show below that 2 1 reference time servers are needed and sufficient to guarantee external synchronization of non reference time servers in case that up to reference time servers can suffer arbitrary failures. Since external clock synchronization implies internal clock synchronization, we ....

....are themselves internally synchronized. Achieving this internal synchronization of the servers in has to be provided by another algorithm, i.e. an internal clock synchronization algorithm. However, the clocks in cannot be internally synchronized because of the 2 1 servers can fail [6]. When we replace the servers in by reference time servers, these servers receive external time signals which allow them to synchronize their clocks externally and therefore also internally. The 2 1 lower bound result for external clock synchronization depends on the assumption that ....

[Article contains additional citation context not shown here]

D. Dolev, J. Y. Halpern, and R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Science, 32(2):230--250, 1986.

....our approach can also be used to custom design fault tolerant clock boards in case the reliability of available COTS boards is not sufficient for the target application. We show that 2F 1hardware clocks (and not 3F 1like one might expect from the lower bound for internal clock synchronization [3]) are sufficient to mask F clock failures. We cautioned that one should place only a limited trust on GPS signals since they are suseptible to delay and jamming attacks. Authenticated time servers with a request reply pattern are more trust worthy than a broadcast time service (like GPS) since one ....

D. Dolev, J. Y. Halpern, and R. Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Science, 32(2):230--250, 1986.

....the membership after it has updated its member set and switched to mode up. 4. 3 Internal Time Base A major implementation problem with the above specification (D,A,S) is that even in synchronous systems one cannot enforce that the processes change their member sets at the same point in real time [7]. The best one can guarantee is that two processes change their member sets within, say, i time units of each other (see Figure 10) Thus, we relax the requirement (A) by replacing real time by clock time . As in [3] we will assume the existence of an internal time base provided by an internal ....

DOLEV, D., HALPERN, J. Y., AND STRONG, R. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Science 32, 2 (1986), 230--250.

....for crash and omission failures. We show that to achieve optimal accuracy, fewer than half the clocks in the system can be faulty. With arbitrary failures and in the absence of authentication, synchronization can be achieved only if fewer than a third of the clocks in the system are faulty [3]. For all the models of failure that we consider, our algorithms are optimal with respect to the number of faulty clocks they can tolerate. The solution presented in this paper is simple and efficient, and can be easily implemented [ 1 ] Its message complexity is similar to those previously ....

.... dmn(l o) From Lemma 7 we see that P a dmi, 1 ) Therefore the smallest possible Dm that this algorithm can achieve is given by Dmax dmi, l )3 tdr. Dolev et al. have shown that the real time between when clocks read the same value cannot be guaranteed to be better than ta, 2 [3]. Hence, when optimal accuracy is required, Dmax must be at least (I p) ta, 2. LEMMA 8. The algorithm in Figure 1 achieves agreement. PROOF. If assumptions S1 and S2 hold for some k 1, then Lemma 6 states that the agreement condition is satisfied for k. We now show, by induction on k, that ....

[Article contains additional citation context not shown here]

DOLEV, D., HALPERN, J. Y., AND STRONG, R. On the possibility and impossibility of achieving clock synchronization. In Proceedings of the 16th Annual ACM STOC (Washington D.C., Apr.). ACM, New York, 1984, pp. 504-511. (Also to appear in J. Comput. Syst. Sci.)

....modeling decisions and a stimulus for the development of algorithm verification techniques. Similar results should be possible for real time systems. Some examples of complexity results that have already been obtained for real time systems are the many results on clock synchronization, including [8, 11, 17, 20, 32] (see [31] for a survey) In this paper, we embark on a study of complexity results for real time systems. We begin this study by considering timing based variations of certain problems that have previously been studied in asynchronous concurrent systems. In particular, we study a variant of the ....

DOLEV, D., HALPERN, J., AND STRONG, H.R. (1986), On the possibility and impossibility of achieving clock synchronization, d. Cornput. System Sci. 32, 230-250.

....are permitted to take arbitrary steps, faulty clocks would not increase their power to affect the behavior of nonfaulty processes. 2) There are at most f faulty processes, for a fixed constant f, and the total number of processes in the system, n, is at least 3f 1. Dolev, Halpern and Strong [DHS] show that it is impossible without authentication to synchronize clocks unless more than 2 3 of the processes are nonfaulty. 3) The message delay for every message is in the range [ 8, for some nonnegative constants and 8 with . 4) A START message arrives at each process p at ....

D. Dolev, J. Halpern and R. Strong, On the possibility and impossibility of achieving clock synchronization, Proceedings of the Sixteenth Annual ACM Symposium on Theory of Computing (1984). "

....squad problem follows from a reduction to weak agreement in [4J. We provide a direct proof. For approximate agreement, the 3m l bound was noted, but not proved, in [7] while the 2m l connectivity requirement was previously unknown. For clock synchronization, the 3m l node bound was proved in [6], with a complicated proof. The authors of [6] also claimed that they knew how to prove the corresponding 2m l connectivity lower bound, but we presume that such a proof would also be complicated. We prove both the 3m l node and the 2m l connectivity bounds, for a much more general notion of clock ....

....weak agreement in [4J. We provide a direct proof. For approximate agreement, the 3m l bound was noted, but not proved, in [7] while the 2m l connectivity requirement was previously unknown. For clock synchronization, the 3m l node bound was proved in [6] with a complicated proof. The authors of [6] also claimed that they knew how to prove the corresponding 2m l connectivity lower bound, but we presume that such a proof would also be complicated. We prove both the 3m l node and the 2m l connectivity bounds, for a much more general notion of clock synchronization than in [6] These ....

[Article contains additional citation context not shown here]

Dolev D, Halpern J, Strong H (1984) On the possibility and impossibility of achieving clock synchronization. Proceedings of the 16th STOC, April 30-May 2, 1984, Washington, DC, pp 504-510

....permitted to take arbitrary steps, faulty clocks would not increase lheir power to affect the behavior of nonfaulty processes. 2) There are at most f faulty processes, for a fixed constant f, and the total number of processes in the system, n, is at least 3f 1. Dole, Halpern and Strong [DHS] show that it is impossible without authentication to synchronize clocks unless more than 2 3 of the processes are nonfaulty. 3) Tte message delay for every message is in the range [ r . for some normegahvo colstafts ald v. ilh , r. 4) A START message arrives at each pro( ess p ....

D. Dolev, J. Halpern and R. Strong, On the possibility and impossibility of achieving clock synchronization, Proceedings of the Sixteenth Annual ACM Symposium on Theory of Computing (1984).

.... to estimate the required amount of adjustment; ii) provides the abstraction of a synchronised clock whose reading is the sum of the physical clock reading and the adjustment that is stored in a memory location, as the processor s physical clock is read only and cannot be directly adjusted (see [Dolev84] for details) and (iii) phases in the adjustment gradually to avoid sudden jumps or rollbacks in the readings of the synchronised clock. Thus, keeping processors clocks in bounded synchronism involves estimating and phasing in adjustments periodically. The software component that provides the ....

D. Dolev et. al., "On the Possibility and Impossibility of Achieving Clock Synchronisation," Proc. 16th Annual ACM STOC, pp. 504-511, April 1984.

....provided n, the number of FCUs, satisfies n 3a 2s m. It is provably impossible (i.e. it can be proven that no algorithm can exist) to tolerate a arbitrary faults in clock synchronization with fewer than 3a 1 FCUs and 2a 1 disjoint communication paths (or a 1 disjoint broadcast channels) DHS86,FLM86] unless digital signatures are employed which is equivalent to reducing the severity of the arbitrary fault mode) Synchronization is approximate (i.e. the clocks of different FCUs need to be close together, not exactly the same) those problems that require exact agreement (e.g. group ....

Danny Dolev, Joseph Y. Halpern, and H. Raymond Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Sciences, 32(2):230--250, April 1986.

....pulse generators for each processors. It is easy to see that the less centralized the clock implementation is the more resilient to faults it is. 2. The Model 2 In the past clock synchronization solutions that can tolerate faults have been proposed for the case of arbitrary, or Byzantine, faults [19, 18, 20, 8, 21, 23]. In those model characteristics they proved that no algorithm can work unless more than one third of the processors are nonfaulty [8] In the case of authenticated Byzantine faults the things are not so bad; there exist algorithms that can tolerate any number of faulty processors [12] The ....

.... Model 2 In the past clock synchronization solutions that can tolerate faults have been proposed for the case of arbitrary, or Byzantine, faults [19, 18, 20, 8, 21, 23] In those model characteristics they proved that no algorithm can work unless more than one third of the processors are nonfaulty [8]. In the case of authenticated Byzantine faults the things are not so bad; there exist algorithms that can tolerate any number of faulty processors [12] The negative results in that model are that: i) the faulty processors can influence the clocks of the non faulty ones by speeding them up, ii) ....

D. Dolev, J.Y. Halpern and H.R. Strong. On the Possibility and Impossibility of Achieving Clock Synchronization. Journal of Computer Systems Science 32, 2, 1986, pp. 230--250.

....initial state q 0;i and a distinguished fail state. A configuration is a vector C = q 1 ; q n ) where q i is the local state of p i ; denote state i (C) q i . The initial configuration is the vector (q 0;1 ; q 0;n ) Processes communicate by sending messages 4 See [13, 22, 25, 27, 37, 39], for example. 5 These definitions could be expressed in terms of the general timed automaton model described in [1] and [29] however, we choose here to present the definitions directly, in order to avoid the intervening layer of definitions. 5 (taken from some alphabet M) to each other. A ....

Dolev, D., Halpern, J. Y., and Strong, H. R. On the possibility and impossibility of achieving clock synchronization. J. Comput. Syst. Sci. 32, 2 (1986), 230--250.

No context found.

D. Dolev, J. Halpern, and H. R. Strong, " On the Possibility and Impossibility of Achieving Clock Synchronization", J. of Computer and Systems Science, Vol. 32:2, pp. 230-250, 1986.

No context found.

D. Dolev, J. Halpern, and H. R. Strong, " On the Possibility and Impossibility of Achieving Clock Synchronization", J. of Computer and Systems Science, Vol. 32:2, pp. 230-250, 1986.

No context found.

D. Dolev, J. Halpern, and H. R. Strong, On the Possibility and Impossibility of Achieving Clock Synchronization, J. of Computer and Systems Science, Vol. 32:2, pp. 230-250, 1986.

No context found.

D. Dolev, J. Y. Halpern, and H. R. Strong. On the possibility and impossibility of achieving clock synchronization. J. Comput. Syst. Sci., 32(2):230--250, Apr. 1986.

No context found.

D. Dolev, J. Halpern, and H. R. Strong, On the Possibility and Impossibility of Achieving Clock Synchronization, J. of Computer and Systems Science, Vol. 32:2, pp. 230-250, 1986.

....squad synchronization seems to be a basic problem in distributed computing. Much of the theoretical work on fault tolerant distributed computing has concentrated on the agreement problem. More recently, attention has shifted in part to other basic problems such as clock syn chronization [DHS, HSSD, LM, LL] and distributed coin flipping [BD] Through this paper, we hope to add firing squad synchronization to the list of basic problems in distrib uted computing which are amenable to theoretical under standing. In Section 2 we give definitions. Section 3 contains re sults (DFS ....

Dolev, D., Haipern, J., and Strong, H. R., On the possibility and impossibility of achieving clock synchronization, Proc. 16th ACM Syrup. on Theory of Computing, 1984, pp. 504-512.

No context found.

D. Dolev, J.Y. Halpern, and R. Strong. On the possibility and impossibility of achieving clock synchronization. In Proceedings of the ACM Symposium on the Theory of Computing, pages 504--511, April 1984.

No context found.

Danny Dolev, Joseph Y. Halpern, and H. Raymond Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Sciences, 32(2), pages 230--250, April 1986.

No context found.

Danny Dolev, Joseph Y. Halpern, and H. Raymond Strong. On the possibility and impossibility of achieving clock synchronization. Journal of Computer and System Sciences, 32:230--250, 1986.

First 50 documents  Next 50

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