Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Powiadomienia systemowe

Sesja wygasła!
Sesja wygasła!

Znaleziono wyników: 1

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: hidden Markov models (HMM's)

Sortuj według:

Ogranicz wyniki do:

Analytical investigation of congestion -avoidance strategies in closed-type queuing models of computer networks with priority scheduling

Oniszczuk W.

TASK Quarterly : scientific bulletin of Academic Computer Centre in Gdansk

2007

Vol. 11, No 3

237-252

A new approach is presented to modeling intelligent admission control and congestion avoiding mechanism, without rejecting new requests, embedded into a priority closed computer network. Most Call Admission Control (CAC) algorithms treat every request uniformly and hence optimize network performance by maximizing the number of admitted and served requests. In practice, requests have various levels of importance to the network, for example priority classes. Here, the investigated closed network with priority scheduling has been reduced to two service centers, which allows for decomposition of a larger network into a chain of individual queues, where each queue can be studied in isolation. A new algorithm (approach) of this special type of closed priority queuing systems is presented, including a node consisting of several priority sources generating tasks, designated as an Infinite Server (IS), and a service centre with a single service line. This model type is frequently described as a finite source, pre-emptive-resume priority queue (with general distribution of service time). The pre-emptive service discipline allows a task of lower priority to be returned to the head of a queue when a new task of higher priority arrives. A mathematical model of provisioning and admission control mechanism is also described. The idea behind this mechanism has been derived from the Hidden Markov Model (HMM) theory. It is crucial in the CAC process that the network manager obtains correct information about the traffic characteristics declared by the user. Otherwise, the quality of service (QoS) may be dramatically reduced by accepting tasks based on erroneous traffic descriptors. Numerical results illustrate the strategy's effectiveness in avoiding congestion problems.