Alexandria Engineering Journal (Oct 2022)
Formalization of the metric of parameters for quality evaluation of the subject-system interaction session in the 5G-IoT ecosystem
Abstract
The highlight of an article is a mathematical model of the process of subject-system information interaction in the IoT ecosystem. In contrast to the existing models, the studied process is described by a simplified queuing system, reduced to a discrete Markov chain. The controlled parameter is the total amount of occupied system resources for all active sessions subject system interaction. Based on the created model, the metric of parameters that fully characterize the quality of functioning of the studied IoT system is analytically described. The proposed metric includes such stochastic characteristic parameters as the average number of active sessions of subject-system interaction in the information environment of the IoT system, the average amount of occupied system resources, and the probability of losing incoming requests due to insufficient free system resources or lack of free communication channels. The dependence of the mentioned stochastic characteristic parameters on the type and settings of the distribution function of the released system resources volume is also analytically formalized. The proposed expressions also consider the workload level of the front-end interface of the IoT system and the desired amount of system resources specified in the input request. The experiments showed that with the increasing workload on the front-end interface of the studied IoT system, the values of all characteristic parameters of the metric increase. In particular, for the studied system with an increase in load in the range from 0.8 to 1.1: - the average number of active sessions of subject-system interaction increased linearly from 11 to 16; - the average amount of occupied system resources gradually increased nonlinearly from 72% to 93%; - the probability of losing an incoming request increased exponentially from 5% to 18%. Note that the increase in the value of the probability of loss of incoming requests is exponential. It is established that the reason for such growth is the growth of variance of the main characteristic parameter of incoming requests – the value of the desired amount of system resources. The geometric distribution-based system resource management scheme also found that it mainly satisfies incoming requests with a smaller value of the desired amount of system resources and, in general, is focused on supporting already active sessions of subject-system interaction. The proposed mathematical apparatus can be used at the design stage of client-oriented 4G-LTE, 5G, and, potentially, 6G ecosystems.
