Dynamic Class of Service mapping for Quality of Experience control in future networks

Abstract:
The present work introduces a novel approach to cope with some key limitations of the present communication networks. In particular, the need for effectively managing heterogeneous resources over heterogeneous networks while guaranteeing personalized Quality of Experience (QoE) requirements to the applications, claims for a full cognitive approach which is realized through the introduction of an appropriate Future Internet architecture. The paper, taking this architecture in mind, introduces the innovative concept of dynamic association between applications and Classes of Services, performed by means of proper Reinforcement Learning algorithms. The resulting procedure guarantees QoE personalization, requires low processing capabilities and entails limited signalling overhead.