Premium
Limiting distribution of the collision resolution interval
Author(s) -
Feldman P.,
Rachev S. T.,
Rüschendorf L.
Publication year - 1997
Publication title -
statistica neerlandica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.52
H-Index - 39
eISSN - 1467-9574
pISSN - 0039-0402
DOI - 10.1111/1467-9574.00034
Subject(s) - collision , interval (graph theory) , limiting , mathematics , normality , limit (mathematics) , resolution (logic) , asymptotic distribution , stability (learning theory) , statistical physics , statistics , mathematical analysis , combinatorics , physics , computer science , mechanical engineering , computer security , artificial intelligence , estimator , machine learning , engineering
[Note 1. Research supported by NATO, DFG and NSF grants. ...]We use the theory of probability metrics to study the asymptotic normality of the collison resolution intervals in the CTM multi‐access protocol under general conditions on the number of retransmitted messages and of new arrivals during the collision slots. Our main result establishes stability of the central limit theorem for the CTM algorithm. We provide extensive simulation results investigating the extent to which the mean of the collision resolution interval eventually becomes unstable for increasing values of n , the number of users who initially collide. The normal fit is numerically investigated and is shown to be quite satisfactory and stable with respect to moderate perturbations and n ≥50.