IDOS: Lightweight Nonparametric Outlier Detection Algorithm for Resource-Constrained Applications

Outlier detection is critical for numerous real-world applications. However, traditional methods, ranging from rule-based approaches to deep-learning models, often require extensive computational resources and intricate parameter tuning, limiting their usability in resource-constrained environments such as wearable devices and IoT systems. This paper proposes a lightweight, nonparametric outlier detection algorithm IDOS (Interpolated Density for Outlier Score) that balances efficiency and adaptability. The approach eliminates the need for parameter tuning, reduces computational over-head, and supports flexible deployment across diverse scenarios. Through a case study on health monitoring, the algorithm demonstrates competitive performance and low resource consumption compared to existing methods. These results highlight its potential for real-time anomaly detection in constrained environments.