Open AccessAn Exact Analytical Model for an IoT Network with MMPP Arrivals
Author(s) -
Osama$AAUP$Palestinian Salameh
Publication year - 2021
Publication title -
international journal of computer networks and communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.159
H-Index - 8
eISSN - 0975-2293
pISSN - 0974-9322
DOI - 10.5121/ijcnc.2021.13106
Subject(s) - computer science , network packet , markov chain , preemption , computer network , queue , scheduling (production processes) , markov process , real time computing , markov model , false alarm , alarm , mathematical optimization , mathematics , statistics , materials science , machine learning , composite material , operating system
Analytical modeling of the Internet of Things (IoT) networks is challenging. This is due to the presence of a large number of devices in these networks and the complexity of the priorities between different types of traffic. Taking these aspects into account, the objective of this paper is to analyze the performance of an IoT network where the IoT devices work independently of one another. To this end, we developed a novel multi-dimensional Continuous-Time Markov Chain (CTMC) model with threshold-based preemption. In this model, each IoT device is modeled as a Markov Modulated Poisson Process (MMPP) that can transmit regular and alarm packets. Alarm packets have higher priority over regular packets. To measure access to the channel between alarm and regular packets, we introduced a threshold parameter where the threshold is the number of packets in the alarm queue that indicates when preemption starts. The performance measures include blocking probability, the average delay of regular packets and alarm packets, discard rate , and success probability of regular packets. Comprehensive numerical analysis was conducted. Our results indicate that impact of the threshold on performance measures is higher on the boundary values of the threshold. The model was proven to be efficient in analyzing the performance of IoT networks on a wide range of parameter values. These results may be used in the future to develop and assess a protocol that utilizes a scheduling algorithm with a dynamic preemption threshold to optimize the performance of the IoT network.