A QoE-aware dynamic bandwidth allocation algorithm based on game theory

Quality of Experience (QoE) is a wide concept including user perception, behavior and expectations as well as application and network performances. The Internet of the future should be able to increase the QoE offered to the users, also in relationship with their commercial profiles. This paper presents an innovative approach, based on game theory, which, according to the feedback of QoE, is able to dynamically assign the available bandwidth of a shared technology (e.g. a WLAN, a cell) to the running flows, in order to maximize the QoE of the users and to guarantee fairness. Moreover, flows could be prioritized, according to commercial profiles (e.g., flows for which users pay more must be served with higher quality). The proposed approach is independent of the way the feedback of QoE is computed (it could be given by a direct user quality expression, estimated from the QoS measurements, from the user's behavior, etc.): the only assumption is that the QoE is a non-decreasing continuous function of the bandwidth allocated to the application. Simulations, considering audio flows with different codecs for which the IQX hypothesis (i.e., exponential interdependency of QoE and QoS) holds, show high performances and fairness/prioritization properties.