Context-Aware Federated DRL Enabled Efficient Bandwidth Allocation for Multimodal Data in Industry 4.0

With technological breakthroughs in human-machine interaction, modern industrial systems are transitioning toward immersive, multisensory environments. This transformation drives the adoption of cross-modal collaborative communication, a key enabler of intelligent, adaptive connectivity in Industrial Internet of Things (IIoT) systems. Aligned with the vision of Industry 4.0, the fusion of heterogeneous sensing modalities such as visual, auditory, thermal, and haptic inputs demands the development of realtime, context-aware resource management strategies to handle dynamic bandwidth and communication needs effectively. To meet these requirements, this work proposes a Context-Aware Multimodal Efficient Resource Allocation (CAMERA) strategy, introducing an intelligent mechanism for optimizing bandwidth allocation across diverse IIoT scenarios. At its core lies a contextual decision engine responsible for continuously analyzing each data modality's reliability, historical performance, and environmental relevance. This engine prioritizes data streams based on modality-specific QoS requirements, application-criticality, and operational context. The resource allocation task is modeled as a Markov Decision Process (MDP), enabling the system to adapt to time-varying network states and uncertainties. The proposed strategy leverages a Context-Aware Federated Deep Reinforcement Learning (CoFe-DRL) paradigm to solve this. In this approach, distributed IIoT agents independently learn optimal allocation policies using local observations while periodically synchronizing model parameters with a cloud server. The proposed approach not only preserves data privacy but also scales efficiently across large networks. Experimental evaluation in a simulated Industry 4.0 environment demonstrates that the proposed model significantly outperforms conventional non-contextual models, yielding improvements in network utilization, delivery accuracy, and resilience to sensory inconsistencies.