Smart-home surveillance systems increasingly rely on heterogeneous IoT data streams, requiring efficient fusion, scalability, and robustness under noisy sensing conditions. This paper proposes a Quantum-Inspired Deep Neuro-Fusion Architecture (QDNFA) for anomaly detection in edge–cloud IoT environments. The framework integrates modular encoders, temporal alignment, and a quantum-inspired optimisation mechanism to support multi-modal data processing while maintaining real-time performance. Experimental evaluation is conducted on the CASAS Smart Home dataset to validate sensor-centric anomaly detection, scalability across multiple devices, and edge–cloud inference efficiency. While the architecture is designed to support audio and video modalities, the present study focuses on low-dimensional sensor data, and large-scale benchmarking on audio–visual surveillance datasets is identified as future work. Results demonstrate improved detection accuracy and reduced latency compared to baseline methods in sensor-driven smart-home scenarios.
M et al. (Sun,) studied this question.