학술논문

Interference resilience of Thread: A practical performance evaluation
Document Type
Conference
Source
2021 IEEE 18th Annual Consumer Communications & Networking Conference (CCNC) Consumer Communications & Networking Conference (CCNC), 2021 IEEE 18th Annual. :1-4 Jan, 2021
Subject
Communication, Networking and Broadcast Technologies
Computing and Processing
Power, Energy and Industry Applications
Signal Processing and Analysis
Semiconductor device measurement
Wireless sensor networks
Protocols
Instruction sets
Buildings
Interference
Loss measurement
IIoT
OpenThread
WSN
IEEE 802.15.4
Language
ISSN
2331-9860
Abstract
Currently, most IoT services are hosted centralized in a cloud. However, they rely on data gathered by numerous distributed sensors, which require to be connected to these clouds. At the moment this is realized with non-IP based proprietary wireless sensor networks that use inflexible networking concepts and require to be carefully configured. The Thread protocol, with its implementation OpenThread, is a new wireless mesh approach to solve this problem. OpenThread is maintained by the Thread group, which is an organization supported by almost 300 companies, including Google, Amazon, and Apple. Its design is based on an IPv6, for seamless integration in existing IP networks. This work, after a short a practical overview of the Thread protocol in general, investigates the network performance in different interference scenarios based on results from a hardware testbed, set up in an office building. This testbed is based on the Nordic Semiconductor nRF52840 System on a Chip (SoC), running OpenThread from the official github repository. Besides a thorough examination for round-trip-time (RTT) and losses, a stability test was performed over five weeks, to see the impact of interference during business hours. We measured RTTs in the range of 10 to 100 ms and losses up to 40 % depending on interfering wireless services. Despite its obvious advantages, from our experimental results, it is evident how some practical limitations make Open Thread not always a good choice.