Global State-Dependent QoE based Routing Hai Anh TRAN, Abdelhamid MELLOUK, Said HOCEINI, Brice AUGUSTIN University of Paris-Est Creteil Val de Marne (UPEC) Image, Signal and Intelligent Systems Lab-LiSSi Lab Transport Infrastructure and Network Control Group - TINC 122 rue Paul Armangot, 94400 Vitry sur Seine, France {hai-anh.tran, mellouk, hoceini, brice.augustin}@u-pec.fr Abstract—For years, wireless network systems have been trying to satisfy end-users and support high quality multimedia applications such as Mobile TV, VoIP, etc. Combining wireless networks with multimedia content distribution needs efficient routing protocols. We develop in this paper a new routing protocol, namely DOQAR (Dynamic Optimized QoE Adaptive Routing), to improve the user perception and optimize the usage of network resources. In our end-to-end model, smartphone users connect to content servers in a wired network across a wireless access network. In order to evaluate the QoE, we use a Multi-Layer Perception-based (MLP) method. Experimental results show a significant performance against other traditional routing protocols. Index Terms—Quality of Experience (QoE), Quality of Ser- vice (QoS), network services, wireless network, routing system, Reinforcement Learning, autonomous system. I. I NTRODUCTION Nowadays, the Next Generation Networks (NGN) trend is the Fixed Mobile Internet Convergence (FMIC) in deploying Wireless Broadband Access (WBA) technologies and migrat- ing the traditional telecom networks to the Internet Protocol (IP) technology. While NGN network experts are going to employ a common network layer protocol in core networks to accomplish the current network services, the access net- works will use various technologies, such as WLAN, WPAN, Ethernet cable, DSL, 2G/3G, LTE, WiMAX, UWB, optical fiber, etc. to meet the diversified requirements of end-users [1]. Using a network environment with multiple operators and multiple networks, end-users expect to use a heterogeneous wired and wireless high-bandwidth ubiquitous network access and diversified services. In fact, accessing the network from mobile devices (laptops, smartphones, or mobile phones) has now many choices for connectivity with the advent of Fourth Generation (4G) com- munication networks that is considered as a solution of an all- IP network layer. The notion of converged network is based on a model that combines a common network core having all network functionalities and different access networks. This combination develops a single network with various access technologies. With this increased choice, network service providers have to consider technical factors (i.e traditional QoS parameters) that influence the usefulness of the service. While the known QoS concept designates a set of technical criteria to ensure network service, Quality of Experience Fig. 1: End-to-end network system (QoE) is a notion that represents the overall level of end-user satisfaction of a service. QoE is expressed by human feelings like “good”, “excellent”, “bad”, etc. A QoE-aware network system is a promising new notion in which service providers are aware of user perception and can consequently adapt to the dynamic environment to obtain acceptable and predictable QoE. In addition, the QoE impacts on the setting of internal parameters of the network. Our work takes into account the end-to-end (e2e) QoE model (Fig. 1). The goal is to maintain the e2e quality between terminal users and servers through the network system. An e2e QoE system includes 3 components on which QoE has impact: • The user terminals and Content servers: in a network system, the terminals represent end-user devices such as laptops, smartphones, etc. The QoE measurement can be realized in this part to give feedbacks into the network. • Access network: end-users have to make a decision to choose which access network they use to connect to the system. Nowadays access network selection based on QoE is a new trend in NGN. • Transport network: The core network may not be wired but may well be wireless by using “mobile routers”. In order to improve the network quality, QoE is considered as an efficient criteria to design routing methods in core networks. According to wireless AP selection, various traditional approaches are based on connection quality between the AP and the end-user, but do not take into account the flow in the core network. Our approach is more general: the customer