Adaptive Connection Admission Control Using Real-time Traffic Measurements in ATM Networks

Kohei SHIOMOTO  Shin-ichiro CHAKI  

IEICE TRANSACTIONS on Communications   Vol.E78-B    No.4    pp.458-464
Publication Date: 1995/04/25
Online ISSN: 
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Issue on Traffic and Quality Control for Communication Networks)
ATM,  CAC,  traffic measurements,  two state model,  bandwidth efficiency,  

Full Text: PDF>>
Buy this Article

Efficiency of network resource can be improved by statistical multiplexing in ATM networks. If cell traffic characteristics of each connection could be obtained beforehand, we could admit maximum connections while satisfying the QoS (Quality of Service) objective. Since such traffic characteristics as an average rate and a mean burst length are difficult to anticipate, only peak rate will be used for CAC (connection admission control). The peak rate assignment strategy will, however, lead to inefficient network utilization for bursty traffic. This paper proposes an adaptive admission control using real-time traffic measurements to overcome the above problem. This scheme is based on two-state cell stream model composed of overload and underload states. The two-state model simplifies the measuring algorithm, which is suited for online processing. Performance of this scheme is investigated through simulation study for multiplexing of on-off sources with a wide spectrum of traffic characteristics. Since the proposed control scheme exploits measurements of cell streams, it achieves nearly optimum bandwidth efficiency.