QAAs: QoS provisioned artificial intelligence framework for AP selection in next-generation wireless networks

Abstract

Emerging trend of ubiquitous data access is driving the demand for effective wireless communication connectivity. In essence to this, wireless local area network (WLAN) technology seems to be a reliable and cost effective access for the next-generation wireless ecosystem. But the pivotal challenge for WLAN in the next generation wireless networks is to cater the legions of heterogeneous services with characteristic sets of quality of service requirements. However, the strategies present in the existing literature are not accoutered for the application-agnostic association and are incompetent in handling the enormous WLAN state space. Realising the pitfalls of the existing strategies, a novel software-defined networking enabled artificial intelligence framework has been proposed. The proposed framework implements a novel invalid action reduction scheme and double deep reinforcement learning to guarantee the flow based association in a multi-service WLAN environment. Moreover, it allows the multi-parametric optimisation of the association decision and faster convergence to the stable solution. The analytical results validated through the extensive simulations revealed that the proposed scheme achieves high performance gain in terms of convergence, stability and network utility as compared to the other solutions in the literature.