Graph Neural Networks Based Anomalous RSSI Detection

TL;DR

The paper addresses anomaly detection in wireless IoT links by enabling per-time-step localization within RSSI time series. It introduces a time-series to graph transformation using Markov Transition Field and a Graph Attention Network (GAT) to classify each node, producing per-point anomaly probabilities. The approach achieves competitive accuracy with state-of-the-art image-based methods while using approximately 171× fewer trainable parameters, and it provides localization and duration information for detected anomalies. Validated on the Rutgers WiFi dataset with synthetic anomalies, the method offers a scalable, efficient solution for real-time wireless link monitoring with actionable timing details.

Abstract

In today's world, modern infrastructures are being equipped with information and communication technologies to create large IoT networks. It is essential to monitor these networks to ensure smooth operations by detecting and correcting link failures or abnormal network behaviour proactively, which can otherwise cause interruptions in business operations. This paper presents a novel method for detecting anomalies in wireless links using graph neural networks. The proposed approach involves converting time series data into graphs and training a new graph neural network architecture based on graph attention networks that successfully detects anomalies at the level of individual measurements of the time series data. The model provides competitive results compared to the state of the art while being computationally more efficient with ~171 times fewer trainable parameters.

Figures (3)

• Figure 1: Time-series representation of the links and of the four types of anomalies considered in cerar2020anomaly.
• Figure 2: A simple example of TS to graph transformation. Non-anomalous nodes are coloured in green, while the anomalous are coloured red.
• Figure 3: Proposed GNN architecture.