Date of Publication

6-2020

Document Type

Master's Thesis

Degree Name

Master of Science in Manufacturing Engineering

Subject Categories

Digital Communications and Networking | Manufacturing

College

Gokongwei College of Engineering

Department/Unit

Manufacturing Engineering and Management

Thesis Adviser

Renann G. Baldovino

Abstract/Summary

In this study, the development of an optimized decentralized Internet of Things (IoT) architecture for Long Range Wireless Area Network (LoRaWAN) with IBM NODE-RED is presented. An architecture, based on the guidelines provided by The Things Network, was deployed for the collection of reference data. A new optimized architecture, equipped with Bit Shifting and a Support Vector Machine (SVM) Classifier as payload filter, was also deployed. The new architecture is designed for a more efficient payload preparation process, ideally using less data than the reference architecture. Moreover, the SVM classifier filters unhealthy payloads out, triggering a downlink to request a replacement payload. To avoid errors and an infinite loop of request, the proponent has included a counter. Lastly, the new architecture is mapped in IBM Node-RED for ease of use and data visualization purposes.

Results have shown that Manual

Bit Shifting has reduced payload size by as much as 15.55% and airtime by as much as 11.04%. This increases the capacity of the system to send 40 more messages per day. In addition, the SVM classifier has improved the credibility of the system to collect data, as tested repeatedly. Overall, the study has optimized the entire process, as shown in the comparison of RSSI and SNR results between the two architectures. Hence, it can be concluded that through the improvement made in various parts of the architecture, the new architecture has been proven to be faster, more efficient, and more reliable in collecting usable data, collectively optimizing it compared to the reference architecture by The Things Network.

Abstract Format

html

Language

English

Format

Electronic

Keywords

Internet of things; Support vector machines; Wireless sensor networks

Upload Full Text

wf_yes

Embargo Period

4-11-2022

Download

Share

COinS