RL-PMAgg: Robust aggregation for PM2.5 using deep RL-based trust management system

Amir Rezapour*, Wen Guey Tzeng

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Air pollution has become a major environmental issue in large cities. Air pollutants, especially fine particulate matter (PM2.5) has raised various concerns on human health. As a result, several low-cost PM2.5 monitoring systems have been deployed worldwide. However, an accurate air pollution monitoring system profoundly relies on data quality. In this paper, we propose RL-PMAgg for robustly computing PM2.5 pollution rates in existence of faulty sensors. Our method consists of three modules. The outlier detector gives quality assessments to the measurements. We use an RL-based trust management system to create a profile for each sensor and track its behavior in the long run. Then, an aggregated PM2.5 rate is computed by using a set of honest sensors along with their trust levels and measurements. We evaluate RL-PMAgg on both simulated and real-world datasets. We compare the proposed method with relevant works. Experimental results show that RL-PMAgg resists the majority of attacks as compared with other works.

Original languageEnglish
Article number100347
JournalInternet of Things (Netherlands)
Volume13
DOIs
StatePublished - Mar 2021

Keywords

  • Anomaly detection
  • Internet of Things
  • PM2.5
  • Reinforcement learning
  • Smart city
  • Trust management system

Fingerprint

Dive into the research topics of 'RL-PMAgg: Robust aggregation for PM2.5 using deep RL-based trust management system'. Together they form a unique fingerprint.

Cite this