An Offline Edge AI System for Sustainable Resource Management in Low-Connectivity Communities

Abstract

Intelligent systems with sustainable resource management in low connectivity areas require intelligent systems that
operate independent of cloud infrastructure. In this paper, we introduce SAHAYAK, which is an Edge AI system that
senses, detects anomalies, and makes decisions all on low-cost edge devices. SAHAYAK uses rule-based inference
combined with machine learning for detecting both sudden and gradual anomalies of resources along with maintaining
computational efficiency. The experiments conducted using synthesized datasets show a 15-25% decrease in resource
wastage, enhanced anomaly detection precision, and a response latency of less than 200ms.