M. Gouda. Elements of security: Closure, convergence, and protection. Information Processing Letter, 77:109--114, 2001.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and study the number of steps it takes for all processors to be in unison (as if they started the computation in the same time) from the time some processor wakes up. They have shown that a synchronous network of N processors can reach unison within 2N synchronous rounds. Gouda and Herman [4] present a solution to the unison problem which is also stabilizing, i.e. the system is guaranteed to reach unison starting from any state. Attiya, Snir and Warmuth [1] consider the processor synchronization problem on anonymous, synchronous rings in order to reduce input collection and ....

Gouda, M. -- Herman, T.: Stabilising Unison, Information Processing Letters 35 (1990), 4, pp. 171--175.

.... the faulty processors can influence the clocks of the non faulty ones by speeding them up, ii) reaccession of repaired processors is not possible unless more than half of the processors are non faulty [12] Self stabilizing algorithms for the clock synchronization problem have also been proposed [11, 6, 1]. An algorithm is called self stabilizing if it can tolerate transient faults in the sense that, after a transient fault leaves the system in an arbitrary state, if no further fault occurs for a sufficiently long period of time then the system converges into a consistent global state and can solve ....

M.G. Gouda and T. Herman. Stabilizing Unison. Information Processing Letters 35, 1990, pp. 171--175.

.... faulty processors can influence the clocks of the non faulty ones by speeding them up, and that re accession of repaired processors is not possible unless more than half of the processors are non faulty [4] Self stabilizing algorithms for the clock synchronization problem have also been proposed [5,6]. An algorithm is called self stabilizing if it can tolerate transient faults in the sense that, after a 1 2 Parallel Processing Letters transient fault leaves the system in an arbitrary state, if no further fault occurs for a sufficiently long period of time, then the system converges into a ....

M.G. Gouda and T. Herman, Stabilizing Unison, Information Processing Letters 35 (1990) 171--175.

....of the same message in outgoing links. An application of this abstraction to clock synchronization in an asynchronous network can be found in [3] and the computational wave concept is widely used in synchronous systems [4] as well as in the design of internal clock synchronization protocols [5]. We distinguish two kinds of waves in our protocol: a request wave, and a synchronization wave: the former carries the remote clock reading request, the latter returns the reply, that is used to synchronize the local clock. One of the interesting features of the algorithm is selfstabilization. ....

Mohamed G. Gouda and Ted Herman. Stabilizing unison. Information Processing Letters, (35):171--175, 1990.

....situation (all clients synchronized with one multi cast) whatever is the initial state. As a consequence, perturbations are eventually compensated without undertaking exceptional actions. There is little or no relation with other clock synchronization algorithms that feature self stabilization, as [4]: they are meant to provide an internal synchronization, while we suggest an external clock synchronization, driven by a time server. In this sense our solution is less distributed. 2 Implementation of a virtual clock The overall synchronization algorithm consists in the generation of an ....

Mohamed G. Gouda and Ted Herman. Stabilizing unison. Information Processing Letters, (35):171--175, 1990.

....and the child have counter c 0 6= c. But in that case we can construct a bad global state in which each child of the root has a different counter value but each pair of neighbors appears to be in a good state locally. Thus the protocol of Figure 5 is not locally checkable. Independently, Gouda [Gou94] used the concept of observers to unify tree and ring stabilizing systems. The paper, however, uses a different proof from ours. Applications of Counter Flushing on Trees: Propagation of Information with feedback is a specific example of a centralized total algorithm [Tel89] A centralized ....

MG Gouda. Stabilizing observers. Information Processing Letters, 52:99--103, 1994.

.... the faulty processors can influence the clocks of the non faulty ones by speeding them up, ii) re accession of repaired processors is not possible unless more than half of the processors are non faulty [7] Self stabilizing algorithms for the clock synchronization problem have also been proposed [1, 2, 6]. An algorithm is called self stabilizing if it can tolerate transient faults in the sense that, after a transient fault leaves the system in an arbitrary state, if no further fault occurs for a sufficiently long period of time then the system A preliminary version of this work appeared in the ....

M.G. Gouda and T. Herman. Stabilizing Unison. Information Processing Letters 35, 1990, pp. 171--175.

.... the desired robustness of self stabilization, fast stabilization of output variables has been recently demonstrated in a number of algorithms [11, 9, 5] and some general methods to achieve time adaptivity [14, 15, 8] or local self stabilization [10, 1] Self stabilizing phase clocks are given in [12, 2, 6]. None of these constructions guarantee fast stabilization for cases of limited transient faults, and all appear to require lengthy stabilization time (proportional to the diameter of the communication graph) in some cases where only a single process variable is corrupted by a transient fault. ....

MG Gouda and T Herman. Stabilizing unison. Information Processing Letters, 35:171-175, 1990.

.... that a protocol P is secure against an adversary D, one needs to partition the reachable states of P into safe states and unsafe states, then identify the critical variables of P (those that need to be protected from the actions of D) and show that the following three conditions hold ( 1] and [7]) i. Closure: The set of safe states is closed under any execution of a P action, and the set of reachable states (i.e. the union of the safe state set and the unsafe state set) is closed under any execution of a P action or a D action. ii. Convergence: Starting from any unsafe state, any in ....

....V to V , then there is a safe state s such that the values of the critical variables in s equals to V, and execution of the same action starting at s changes the values of the critical variables of P from V to V . Note that this condition is a generalization of the corresponding condition in [7] which states that each execution of a P action starting at an unsafe state cannot change the values of the critical variables of P. Following this de nition, the security of the PING protocol can be established by identifying the safe, unsafe, and reachable states of the protocol, then ....

[Article contains additional citation context not shown here]

Gouda, M.G.: Elements of security: Closure, convergence, and protection. Information Processing Letters, Vol. 77, Nos. 2-4 (2001) 109-114

....and inspect the elapsed time since repair initiation. Measuring time in an asynchronous sense is the subject of many investigations, however few of these papers consider transient fault tolerance in an asynchronous system. Research on synchronizers, phase synchronizers, and integervalued clocks [2, 5, 8, 11, 13, 1] are the most closely related works to the repair timer of this paper. The only paper considering all cases of transient faults for a synchronizer in the asynchronous model is [3] but the algorithm presented there does not satisfy fault locality properties in all cases. The algorithm presented in ....

MG Gouda and T Herman. Stabilizing unison. Information Processing Letters, 35:171--175, 1990.

.... to the desired robustness of self stabilization, fast stabilization of output variables has been recently demonstrated in a number of algorithms [10, 8, 5] and some general methods to achieve time adaptive [13, 14, 7] or local self stabilization [9, 1] Self stabilizing phase clocks are given in [11, 2, 6]. None of these constructions guarantee fast stabilization for cases of limited transient faults, and all appear to require lengthy stabilization time (proportional to the diameter of the communication graph) in some cases where only a single process variable is corrupted by a transient fault. ....

MG Gouda and T Herman. Stabilizing unison. Information Processing Letters, 35:171--175, 1990.

....and Clocks The problems of clock synchronization and phase synchronization are natural to consider for research on self stabilization, since synchronization is a fundamental tool for system implementations. Self stabilizing synchronization has been the focus of much recent research [GH90, LZM90, ADG91, CFG92, AKM 93, DW93, CS94, PT94, DW95, HG95, LS95, ER95] 7 Fault Tolerance Dijkstra s initial work on self stabilization [Dij74] mentions no application to fault tolerance. Yet self stabilization is most often seen as a technique for fault tolerance in the area of distributed ....

....the natural interpretation of vertices as network nodes and edges as network links. Some of the research on self stabilizing communication protocols considers classic problems of window control, session control, message sequencing, connection setup and sequencing, and routing [BP89a, SG89, Spi90, AGH90, Her92b, AB93, Var93, Dol93, Dol94, AACD 94, DTOF94, APSV94, ADW95, CV96, DW97a, CW97, BK97] The paper [DIM97b] is a communication protocol for leader election; in [DPW96] mobile processes are considered, and [AACD 94] specializes to high speed networks, with self stabilizing protocols. ....

MG Gouda and T Herman. Stabilizing unison. Information Processing Letters, 35:171--175, 1990.

....state. Thus Q addresses one situation not specified in [6] if there are f n=2 faults, then Q converges to some state where all output bits are equal. Protocol Q need not be time adaptive, but is self stabilizing. The timer for the composite protocol consists of a self stabilizing phase clock [5]. The phaseclock is implemented by a integer variable clock i at processor i. For simplicity, let clock i be unbounded (it can be made finite in the obvious way) Let T be a constant, known 2 to all processors, satisfying T = O(diam) T max( 3 Delta diam ; stabilization time(Q) The ....

MG Gouda and T Herman. Stabilizing unison. Information Processing Letters, 35:171-- 175, 1990.

No context found.

M. Gouda. Elements of security: Closure, convergence, and protection. Information Processing Letter, 77:109--114, 2001.

No context found.

M. G. Gouda. Elements of security: Closure, convergence, and protection. Information Processing Letters, 77(2--4):109--114, 2001.

No context found.

MG Gouda. Stabilizing observers. Information Processing Letters, 52:99--103, 1994.

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