Systems and Means of Informatics

2024, Volume 34, Issue 4, pp 31-47

OPTIMIZATION OF THE THRESHOLD PARAMETER OF A RED-LIKE QUEUE MANAGEMENT ALGORITHM FORA G/M/1 QUEUE

  • Ya. M. Agalarov

Abstract

The article considers the problem of calculating the optimal threshold value of a RED-like algorithm by a threshold value for a G/M/1 queue with incomplete queue renovation and probabilistic clients reset. The RED-like algorithm uses a single-threshold mechanism for probabilistic reset of clients from the queue, according to which at each moment of release of the service device, some clients are removed from the queue with a given probability. Customers queue in the order of arrival and those whose queue number exceeds the specified threshold at the time of release of the service device (they are in the “congestion zone”) are reset with a given probability. It is assumed that the “congestion zone” has a limited number of seats and if there are no empty seats in the “congestion zone” at the time of arrival of the customer, the customer is reset. The objective function is a weighted sum of the average customer delay time, the average number of customers dropped from the queue per unit of time, the average number of customers rejected at the entrance per unit of time, the average downtime of the device, and customer service payment. The mathematical problem of optimizing the objective function by a threshold value at a fixed size of the “overload zone” is formulated. The proofs of some relations between the characteristics of the queuing system and the unimodality of the objective function in terms of the threshold value are presented. A simple algorithm for the guaranteed solution of the formulated problem is proposed.

[+] References (22)

[+] About this article