Resource-Efficient Sensor Fusion at the Edge via System-wide Dynamic Gated Neural Networks

• Published in: IEEE transactions on mobile computing Vol. In Press; pp. 1 - 15
• Main Authors: Singhal, Chetna, Wu, Yashuo, Malandrino, Francesco, Contreras, Sharon G. L., Levorato, Marco, Chiasserini, Carla Fabiana
• Format: Magazine Article
• Language: English
• Published: IEEE 2025
• Subjects: Artificial neural networks; Computational modeling; Computer architecture; Dynamic deep neural networks; Energy efficiency; Inference in the mobile-edge continuum; Logic gates; Mobile nodes; Network support to machine learning; Optimization; Sensor fusion; Servers; Soft sensors; Telecommunication network management
• ISSN: 1536-1233; 1558-0660; 1558-0660
• DOI: 10.1109/TMC.2025.3586882

Next-generation mobile systems will support multiple AI-based applications, each leveraging heterogeneous sensors and data sources through deep neural network (DNN) architectures collaboratively executed within the network. In this context, to minimize the cost of the AI inference task subject to requirements on latency, quality, and - crucially - reliability of the inference process, it is vital to optimize (i) the set of sensors/data sources and (ii) the DNN architecture, (iii) the network nodes executing sections of the DNN, and (iv) the resources to use. To achieve these goals, we leverage dynamic gated neural networks with branches, and propose a novel algorithmic strategy called Quantile-constrained Inference (QIC), based upon quantile-Constrained policy optimization. QIC makes joint, high-quality, swift decisions on all the above aspects of the system, with the aim to minimize inference energy cost. We remark that this is the first contribution connecting gated dynamic DNNs with infrastructure-level decision making. We evaluate QIC using a dynamic gated DNN with stems and branches for optimal sensor fusion and inference, trained on the RADIATE dataset offering Radar, LiDAR, and Camera data, and real-world wireless measurements. Our results confirm that QIC closely matches the optimum and outperforms existing approaches in reducing energy consumption (compute, communication, and total) and application requirements failure by over 70%.