Widespread use of wireless networks nowadays raises many problems for service providers in managing their resources. These problems are caused mainly by restricted bandwidth and variable radio condition in this type of network. Moreover, with the emergence of multimedia traffic and its requirements in terms of quality, admission control is hence an inevitable choice to optimize network resources while maintaining high service quality at users. In this paper, we propose an admission control mechanism based on Quality of experience (QoE) perceived by users. The human QoE is obtained by a tool called Pseudo Subjective Quality Assessment (PSQA), which is based on statistic learning using Random Neural Network (RNN). Instead of relying on technical parameters such as bandwidth, loss, or latency, which do not correlate well with human perception, our scheme is based on Mean Opinion Score (MOS) but without interaction from real humans. The simulation results demonstrate the better performance of our proposition compared to the loss-based approach regarding user satisfaction evaluated by achieved QoE at user and bandwidth utilization of the network evaluated by goodput.