Fault detection and diagnosis for solar-powered Wireless Mesh Networks using machine learning
| Autor: | Diego Passos, Joacir de Oliveira Silva, Ricardo C. Carrano, Débora C. Muchaluat-Saade, Vinicius C. Ferreira, Célio Albuquerque |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Wireless mesh network
Computer science business.industry 020206 networking & telecommunications 02 engineering and technology computer.software_genre Machine learning Fault detection and isolation Support vector machine Statistical classification Naive Bayes classifier Knowledge extraction Software deployment 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Data mining Artificial intelligence Decision table business computer |
| Zdroj: | IM |
| DOI: | 10.23919/inm.2017.7987312 |
| Popis: | The inherent complexity of Wireless Mesh Networks (WMNs) makes management and configuration tasks difficult, specially for fault detection and diagnosis. In addition, manual inspections are extremely costly and require a highly skilled workforce, thus becoming impractical as the problem scales. To address this issue, this paper proposes a solution that makes use of machine learning techniques for automated fault detection and diagnosis (FDD) on solar-powered Wireless Mesh Networks (WMNs). We have used the Knowledge Discovery in Databases (KDD) methodology and a pre-defined dictionary of failures based on our previous experience with the deployment of WMNs. Thereafter, the problem was solved as a pattern classification problem. Several classification algorithms were evaluated, such as Naive Bayes, Support Vector Machine (SVM), Decision Table, k-Nearest Neighbors (k-NN) and C4.5. The SVM presented the best results, achieving a 90.59% overall accuracy during training and over 85% in validation tests. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|