From Connected Devices to Trusted Intelligence: A Survey of Secure, Energy-Efficient and AI-Enabled Internet of Things Systems

DOI

https://www.doi.org/10.59256/ijrtmr.20260605047

Published In

Volume 6, Issue 5 (2026)

Published Online

September-October 2026

Pages

430-440

Article Type

Review Article

Check for updates

The Internet of Things (IoT) has grown from simple remote-control links into large networks of sensors, gateways and analytics engines that run homes, hospitals, factories and power grids. Each stage of this growth has added a new constraint. Early systems had to connect devices reliably over whatever network was available. Later systems had to stretch battery life in wireless sensor networks and protect data on devices too small for standard cryptography. Current systems must also learn from large, noisy data streams, often at the network edge. Research on these concerns is scattered across communication, security, optimization and application studies, and few surveys follow how solutions at one level shape the problems at the next. This survey reviews IoT research published mainly between 2019 and 2026, together with earlier foundational work, and organizes it around four research questions on connectivity, energy efficiency, security and intelligent analytics. It proposes a five-branch taxonomy and a layered reference architecture that place each reviewed study. The review finds that application-level gateways and 5G service classes have largely solved basic connectivity, while energy-aware clustering and routing remain the main levers for extending sensor network lifetime. Lightweight blockchain consensus can raise spoofing detection to 97.6% and cut packet loss from 18.2% to 5.1% at an energy cost of about 0.002 J per transmission. AI-driven nonlinear optimization lowers data mining error to 0.9%, but doubles computation time compared with linear baselines. Across studies, security and intelligence are rarely evaluated together, and energy and latency overheads are reported unevenly. Future work should develop joint security-analytics designs, common benchmarks with overhead reporting, privacy-preserving learning across sites and lightweight cryptography that follows the new NIST standard.