Next generation mobile networks are expected to provide service to intelligent terminals with different categories. To fully optimize wireless networks, we argue that applications demands should be taken into account given the current channel and network conditions. Knowledge has to be shared between layers to obtain the highest possible adaptivity. In this context, we propose and evaluate an adaptive proportional fair algorithm with user multiplexing. Performance results show that the proposed algorithm is effective in reducing sojourn delays and in enhancing fairness among terminal users. The study investigates as well the impact of terminal categories with different levels of performance on our scheduling algorithm