We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.DS

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Data Structures and Algorithms

Title: A Fast and Accurate Failure Frequency Approximation for $k$-Terminal Reliability Systems

Abstract: This paper considers the problem of approximating the failure frequency of large-scale composite $k$-terminal reliability systems. In such systems, the nodes ($k$ of which are terminals) are connected through components which are subject to random failure and repair processes. At any time, a system failure occurs if the surviving system fails to connect all the k terminals together. We assume that each component's up-times and down-times follow statistically independent stationary random processes, and these processes are statistically independent across the components. In this setting, the exact computation of failure frequency is known to be computationally intractable (NP-hard). In this work, we present an algorithm to approximate the failure frequency for any given multiplicative error factor that runs in polynomial time in the number of (minimal) cutsets. Moreover, for the special case of all-terminal reliability systems, i.e., where all nodes are terminals, we propose an algorithm for approximating the failure frequency within an arbitrary multiplicative error that runs in polynomial time in the number of nodes (which can be much smaller than the number of cutsets). In addition, our simulation results confirm that the proposed method is much faster and more accurate than the Monte Carlo simulation technique for approximating the failure frequency.
Comments: 17 pages, 3 figures, 5 tables
Subjects: Data Structures and Algorithms (cs.DS); Networking and Internet Architecture (cs.NI)
Cite as: arXiv:1712.09666 [cs.DS]
  (or arXiv:1712.09666v1 [cs.DS] for this version)

Submission history

From: Anoosheh Heidarzadeh [view email]
[v1] Wed, 27 Dec 2017 20:05:35 GMT (6302kb,D)

Link back to: arXiv, form interface, contact.