Experimental Validation of a Zero-Trust-Based Layered Cybersecurity and Hardening Framework for Smart Metering and IoT Modem Infrastructures

Smart meters and Internet of Things (IoT) modems are now essential for remote monitoring and management in energy distribution systems, but they also widen the attack surface. Vulnerabilities span the hardware, firmware, communication, and data-management layers, and this study addresses them with a five-layer security and hardening framework based on the zero-trust model. Seven mechanisms are integrated within a single architecture: endpoint hardware protection, secure boot, mutual authentication, network micro-segmentation, operational command verification, artificial intelligence-based anomaly detection, and privacy-preserving data aggregation. The framework was evaluated in a laboratory testbed across eight scenarios: physical tampering, malicious firmware, unauthorized connection, network scanning, denial-of-service, replay attacks, low-rate stealthy network anomalies, and consumer profiling via data exposure. Cryptographic key zeroization had a mean latency of 14.2 ± 1.1 ms, unauthorized network traffic was blocked according to the configured security policies, and the SIEM alert associated with replay rejection was generated within 127.08 ms. The Long Short-Term Memory (LSTM)-based time-series model achieved 95.73% accuracy with an area under the curve (AUC) of 0.9696. For the privacy layer, anonymized load-profile data obtained from field-deployed smart meters operating within the Advanced Metering Infrastructure of Dicle Electricity Distribution Inc. were used. The MCFE hourly aggregate of the 11 complete meter profiles matched the plaintext reference with zero error in all 24 hourly slots. These laboratory results support the feasibility and effectiveness of the proposed layered design for smart-meter and IoT-modem infrastructures under the tested conditions.