DMCA
Endpoint admission control with delay variation measurements for QoS (2002)
| Venue: | in IP networks,” ACM Computer Communication Review |
| Citations: | 17 - 0 self |
BibTeX
@ARTICLE{Bianchi02endpointadmission,
author = {G. Bianchi and Universitá Di Palermo and F. Borgonovo and A. Capone and L. Fratta and C. Petrioli},
title = {Endpoint admission control with delay variation measurements for QoS},
journal = {in IP networks,” ACM Computer Communication Review},
year = {2002},
pages = {61--69}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In this paper we describe a novel Endpoint Admission Control scheme (EAC) for IP telephony. EAC mechanisms are driven by independent measurements taken by the edge nodes on a flow of packets injected in the network to probe the source to destination path. Our scheme is characterized by two fundamental features. First, it does not rely on any additional procedure in internal network routers other than the capability to apply different service priorities to probing and data packets. Second, the connection admission decision is based on the analysis of the probing flow delay variation statistics. Simulation results, which focus on a IP telephony scenario, show that, despite the lack of core routers cooperation, toll-quality performance figures (99th delay percentiles not greater than few ms per router) can be obtained even in severe overload conditions. Finally, a comparison with an EAC scheme driven by probe losses only, shows that the use of delay variation statistics as endpoint decision criterion is a key factor for EAC effectiveness.
Keyphrases
endpoint admission control delay variation measurement internal network router different service priority endpoint decision criterion novel endpoint admission control scheme flow delay variation statistic delay variation statistic probe loss core router cooperation toll-quality performance figure connection admission decision eac effectiveness data packet fundamental feature eac mechanism independent measurement eac scheme ip telephony key factor delay percentile severe overload condition simulation result ip telephony scenario edge node additional procedure
