IoT-Based Adaptive Load Shedding for Smart Grid Stability: A Real-Time Edge-Computing Framework for Residential Demand Management

Abderrahmen Sellami, Kamel Cherif

Power grids in developing regions are particularly vulnerable to demand-side imbalances, resulting in prolonged and unplanned outages. This paper presents an IoT-based adaptive load-shedding framework designed to stabilize residential electricity consumption in real time. The system integrates ESP32 microcontrollers, PZEM004T energy metering modules, SIM800L GSM transceivers, multi-channel relays, and a Firebase-backed mobile application. Communication relies on the ESP-NOW protocol for sub-50ms local latency, supplemented by GSM/GPRS for remote alerting. A deterministic, priority-based decision engine classifies appliances into four categories (critical, essential, distinguishable, optional) to minimize user disruption. Experimental validation on a prototype testbed demonstrated a 32.4% peak demand reduction, average response latency below 50ms, and 98.7% system uptime. The proposed approach outperforms centralized SCADA and cloud-based demandresponse systems in latency and resilience while maintaining lower deployment costs.