Fractal-Based N-Environment Multiagent IoT System Reliability

• Published in: IEEE internet of things journal 2024-08, Vol.11 (16), p.27463-27479
• Main Authors: Eroshkin, Ivan, Vojtech, Lukas, Neruda, Marek
• Format: Article
• Language: English
• Subjects: Computation; Customer services; Estimation; Fractal; Fractals; Internet of Things; Internet of Things (IoT); Maintenance; mathematical model; Mathematical models; Multiagent systems; multiagent systems (MASs); multienvironment (ME); Real time; Real-time systems; Reliability; Reliability analysis; reliability estimation; Reliability theory; System reliability; unified reliability
• ISSN: 2327-4662; 2327-4662
• DOI: 10.1109/JIOT.2024.3398406

The tremendous growth of Internet of Things (IoT) systems and the popularity of distributed systems, which were brought back into fashion by containerization, create new challenges for system maintenance. Each IoT-specific service is characterized by a specific domain, or environment, making the maintenance of distributed IoT systems with dozens of agents a challenging task. This article proposes a fractal-based mathematical model, called the fractal multiagent IoT system (FMAIS), for scalable multiagent systems (e.g., cloud-native services, wide-scale IoT services). A reliability estimation model called the multienvironment enhanced real-time customer-oriented reliability estimation (ME-ERT-CORE) model is proposed as a method for estimating the reliability of a multiagent IoT system, where each agent can be specified by its own domain-specific environment (e.g., healthcare, logistics, smart factories, etc.). An estimation of the time complexity of the FMAIS is performed, and a model called ME-ERT-CORE for estimating its reliability is formulated. The results show that the FMAIS can be used for real-time simulations of system behavior and ME-ERT-CORE can be used for the real-time estimation of its reliability, monitoring purposes, and, for instance, Quality-of-Service (QoS) descriptions based on service level agreement (SLA) requirements. The conducted measurements show that the reliability of the FMAIS with four layers and 96 applications, where each application deploys 96 instances, can be computed in 8.022 ms. With the proposed optimization method, the reliability for the same FMAIS can be computed in 0.219 ms, which is 36 times faster than the ME-ERT-CORE reliability evaluation based on the definition. An FMAIS with the aforementioned parameters can be simulated in 84.300 ms.