Intelligent In‐Content‐Aware Elastic Resource Allocation Framework Based on Dynamic Network Slicing for End‐to‐End Tactile IoT Applications in 6G Environment

ABSTRACT The tactile internet of things (TIoT) demands ultralow latency, high reliability, and dynamic resource allocation for real‐time human–machine interactions. Traditional network slicing methods (e.g., peak or average‐based provisioning) fail to adapt to fluctuating haptic signals, causing inefficiencies like overprovisioning or underprovisioning. This paper proposes an intelligent in‐content‐aware elastic resource allocation framework for 6G‐enabled TIoT, combining federated learning (FL) and dynamic network slicing to optimize resource allocation. FL enables distributed prediction of resource demands at tactile devices, preserving privacy and reducing bandwidth overhead. The framework dynamically adjusts network slices using an AI‐driven close‐loop feedback mechanism, improving elasticity. Building on prior FL‐based architecture, this study further refines dynamic allocation for fluctuating tactile signals. Experiments show that the framework outperforms static methods, reducing packet loss by 17%, improving resource efficiency by 30%, and enhancing fairness under varying workloads. Key innovations include modeling TIoT signals with generalized Pareto–Poisson models, an end‐to‐end semantic orchestration workflow, and empirical validation of resilience to asynchronism and service level agreement (SLA) violations. The results highlight its potential to boost quality of service (QoS) and cost‐efficiency in next‐gen tactile applications.