Tactile IoT-Driven Predictive Representation for Swift Content Delivery in Augmented and Virtual Reality

Tactile IoT envisions seamless communication across the network with ultra-low latency features, enabling real-time interactions among various users and heterogeneous devices, such as Augmented Reality (AR) and Virtual Reality (VR) Head-Mounted displays (HMDs). To achieve this vision, complementary technologies such as semantic source coding play a crucial role by efficiently representing and compressing information at its source, capturing meaningful patterns and context. This paper introduces a novel solution for AR/VR to achieve an unparalleled immersive experience. The scope ranges from the intricacies of precise human pose estimation to the intelligent optimization of content delivery tailored to those predictions. We propose a transformative temporal-spatial Markov Decision Process (MDP), which optimizes video frame delivery by aligning the Age of Information (AoI) with the user's Field of View (FoV). This AI/ML-augmented MDP algorithm facilitates responsive content adaptation, ensuring relevance and enhancing user engagement in AR/VR environments. Furthermore, integrating Long Short-Term Memory (LSTM) predictions with a Deep Q-Network (DQN) model allows for sophisticated content adjustment, adapting to user behavior and preferences. Our approach establishes a new paradigm in interactive AR/VR experiences, bridging the gap between user actions and virtual responses, thus mitigating latency issues and enriching the overall user experience. This cutting-edge combination of technologies sets a new benchmark in immersive AR/VR, highlighting our commitment to developing user-centric, efficient solutions. Simulation results compare the solution with other techniques and confirm the effectiveness of the proposed strategy.