Edge Computing and IOT Security in Smart City Infrastructure

Abstract

Smart city ecosystems integrate millions of Internet of Things (IoT) devices generating massive volumes of real-time data, creating unprecedented security challenges due to device heterogeneity, resource constraints, and expanded attack surfaces. This paper proposes EdgeSecure, a lightweight deep learning-based security framework deployed at edge computing nodes for real-time threat detection in smart city infrastructure. The framework employs a hybrid CNN-LSTM architecture optimized through knowledge distillation for edge deployment, achieving a 94.2% overall detection rate across six attack categories while maintaining sub-40ms inference latency. EdgeSecure incorporates a lightweight mutual authentication protocol based on elliptic curve cryptography (ECC) for securing device-to-edge communication. Evaluation on the Bot-IoT and NSL-KDD datasets demonstrates that the proposed framework outperforms existing cloud-based and fog computing approaches in both detection accuracy and response time, processing over 9,200 packets per second per edge node. The framework reduces detection latency by 84.5% compared to cloud-only architectures while maintaining comparable accuracy, establishing edge-native security as a viable paradigm for protecting smart city infrastructure.