Evaluating QoE in Cognitive Radio Networks for Improved Network and User Performance

The Cognitive Radio (CR) technique is widely recognized as a promising solution to the spectrum scarcity problem. Previously, extensive research focused on resource allocation (RA) algorithms, targeting the optimization of the MAC layer network performance; these papers assume an unrealistic queue model where the backlog is unchanging over time. The packet level performance of any Secondary Network (SN) is challenging to evaluate due to the highly variable resource availability. However, some level of guaranteed performance is vital to the success of CR technique, because user Quality of Experience (QoE) depends heavily on it. We develop a packet level network performance evaluation platform for CR to evaluate the effect of key factors controlling the QoE. We use QoE as a unified evaluation metric combining both loss and delay. This allows us to evaluate the effect of performance for users with different service requirements, and allows us to expose the limitation of existing RA schemes. We further demonstrate that QoE informed RA schemes can significantly compensate poor quality performance with limited cost.