2004
DOI: 10.1016/j.tcs.2004.07.029
|View full text |Cite
|
Sign up to set email alerts
|

Least-recently-used caching with dependent requests

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

6
116
0

Year Published

2005
2005
2017
2017

Publication Types

Select...
4
3
1

Relationship

1
7

Authors

Journals

citations
Cited by 68 publications
(122 citation statements)
references
References 19 publications
6
116
0
Order By: Relevance
“…The theorem presented in this subsection complements the results of Jelenkovic and Radovanovic who showed in [13,12] that for dependent request processes, the hit probability is asymptotically, for large cache sizes, the same as in the corresponding LRU system with i.i.d. requests.…”
Section: Lru Insensitivitysupporting
confidence: 80%
“…The theorem presented in this subsection complements the results of Jelenkovic and Radovanovic who showed in [13,12] that for dependent request processes, the hit probability is asymptotically, for large cache sizes, the same as in the corresponding LRU system with i.i.d. requests.…”
Section: Lru Insensitivitysupporting
confidence: 80%
“…Our method of proof uses probabilistic and sample path arguments introduced in [12] for the case of the ordinary LRU (PAC(β, 1)) algorithm. The starting point of our analysis is given by the representation formula in (7) from Section 2.…”
Section: Resultsmentioning
confidence: 99%
“…Note that the performance of the PAC algorithm drastically improves even for the small values of parameter k and, thus, it is nearly optimal; (ii) For the ordinary MTF (k = 1) searching, the convergence of C (N) /N in distribution as N → ∞ and the Laplace transform of the limiting function are obtained in Lemma 4.5 of [6]. The result in its presented form, for the ordinary LRU, was derived in [11]; (iii) It is possible to relax the assumption q i = h N /i α , 1 ≤ i ≤ N, e.g., by assuming that for any > 0, there exists i 0 , such that for all i 0 ≤ i ≤ N, inequality…”
Section: Theorem 2 Assume That Q Imentioning
confidence: 98%
“…Though this model does not take into consideration the correlation between accesses, studies [10] show that the behavior modeled assuming independent reference gives results that are very close to those obtained using models that do indeed incorporate such correlation. We will also show this with empirical results which reconfirm that IRM is indeed able to accurately predict SPM and cache performance for data-structures such as trees where the access-pattern is clearly data-dependent and correlated.…”
Section: Model Descriptionmentioning
confidence: 72%