The Internet of Things (IoT) is a system of interrelated smart devices that share sensing data over networks. Heterogeneous data are collected, shared, and analyzed through the interconnection of massive physical devices. The analytical interaction between a series of IoT devices enables smarter decisions regarding the development of autonomous services. However, the scalability is essential to support the ever-growing number of connected computing devices. As the current IoT networks are centralized, the adoption of distributed peer-to-peer (P2P) networking could be a key solution to build successful decentralized blockchain-based IoT platforms. Therefore, this study attempts to characterize IoT networks in order to compare network performance under a variety of network conditions. In this paper, decentralized P2P computing is compared with the most commonly used centralized protocols, such as publish/subscribe and request/reply, within constrained IoT environments. First, cost modeling of network protocols is proposed to analyze the basic network performance. Moreover, the paper discusses a comparative analysis with push, pull, and P2P-based IoT data transmission approaches. The main purpose of the experiment is to identify the key factors that influence various IoT networks by comparing latency and availability generated by the request message. The results of the performance comparison show that the decentralized P2P-based approach has more key advantages than centralized networking in various performance metrics
This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science and ICT) (No. NRF-2018R1D1A1B07040573 and No. NRF-2019R1F1A1059036).
