RUDN mathematicians have built a model of a queue of requests with different priorities. Unlike analogues, it is more fair to low-priority tasks. The calculations will be useful for calculating the optimal parameters of real systems, for example, in telecommunications. The results are published in the Journal of Ambient Intelligence and humanized Computing.

Newswise — In medicine, telecommunications, banking, and many other areas, it can be a scenario with one performer and several simultaneous requests of different importance. For example, one ambulance and several patients waiting for it with different urgent levels. Or a 5G mobile hotspot and multiple requests from different devices. Or a queue at a bank that has regular customers and loyalty program customers. For mathematics, all these examples are identical. It does not matter what or who exactly plays the role of the server, and what is the client or the request. Such a situation is described in queue theory by probabilistic equations. Solving them helps finding the optimal parameters for such a prioritized system. A RUDN University mathematician with colleagues from Belarus and Korea built a model for a queue with priority customers. It helps optimizing the system and make it more fair to customers who are not included in the priority group.

“Queuing theory is a recognized mathematical tool for optimization in many real systems. One of its important sections is the theory of priority systems. In such systems, clients are classified into different categories based on urgency and importance. Service is provided according to a priority scheme. For example, hospitals use different priority schemes, that is, they sort incoming patients according to the severity of an injury or disease,” said Alexander Dudin, Doctor of Physical and Mathematical Sciences, Head of the RUDN Research Center for Applied Probabilistic Analysis.

Mathematicians have considered a model in which requests are divided into two groups - the first has priority over the second. There are several servers in the system. Some of them are reserved for clients of the first group, some for clients of the second group, and the rest are considered common. Clients of both groups can use shared servers only under certain conditions set in the system.

Unlike analogues, the new model does not assume full prioritization of requests from the first group. If a request from the first group is received, and all servers are busy, the service of the request from the second group will not be interrupted. That is, the priority task may be in the queue.

“We have researched a new priority scheme that is friendlier than most other schemes. For example, it is possible that when a priority client arrives, service to a non-priority client will not be immediately terminated. Reservation of servers for exclusive service of both groups is allowed. Our results can be used to solve optimization problems,” said Alexander Dudin, Doctor of Physical and Mathematical Sciences, Head of the Research Center for Applied Probabilistic Analysis, RUDN University.

 

Journal Link: Journal of Ambient Intelligence and Humanized Computing (2022)