RUDN Mathematicians Suggested a Way to Increase Internet Speed without Changing Tariffs

RUDN mathematicians calculated the characteristics of a queueing system with processor sharing on the example of the Internet. The model takes into account uneven request distribution as well as impatient user behavior (e.g. cases when a user may disconnect from the system if the connection speed is too low). The development will help mobile providers increase the speed of Internet access without changing their tariffs. The results of the study were published in the Operations Research Perspectives journal.

The speed of Internet connection available to mobile device users depends on the provider's strategy of request processing and the distribution of server resources. Mathematicians try to find ways of operation that take into account limited resource availability and the accidental nature of the request flow. To do so, they use various mathematical models such as the queueing theory based on mathematical statistics and the probability theory. RUDN mathematicians analyzed a system in which the processing of all incoming signals started at the same time and available resources were shared equally. They also developed a user behavior model taking into account the impatience of the users,

"We considered a queueing system in which all incoming requests are processed simultaneously (at the speed that is respectively reduced). An example of such a system may be a data downloading channel of a basic station belonging to a mobile network node," said Sergey Dudin, a candidate of physics and mathematics from the Institute of Applied Mathematics and Telecommunication, RUDN.

The model has 3 specific features. First, the number of requests that can come into the system simultaneously, is finite. Second, there are certain increases and decreases in the speed of request inflow which is typical for modern telecom networks. Finally, the processing time has phasic-type, not traditional exponential distribution.

RUDN mathematicians calculated a number of important system characteristics, including request quantity distribution and the distribution of the timing of a random request in the system. The model may be used, among other things, to design wireless communication networks. It will help find optimal limitations on the number of requests including given parameters of the flow, the server, and the patience level of network users.
 
"The advantage of our study lies in the applicability of the model to the analysis of practically important systems with up to 1,000 simultaneously processes requests. Other existing developments don't take into account the fact that users tend to disconnect when the Internet speed is low," commented Sergey Dudin.

The results of modeling may be used to design new wireless communication networks and to optimize existing ones. This will help reduce time and resources spent on the design process and improve service quality. For example, providers may develop more efficient tariff policies and increase the speed of Internet connection for their subscribers.

More information:
A.N. Dudin et al. Analysis of queueing model with processor sharing discipline and customers impatience, Operations Research Perspectives (2018). DOI: 10.1016/j.orp.2018.08.003

Provided by RUDN University