QoS investigation of proportional fair scheduling in LTE networks

In this paper, the quality of service in a Long Term Evolution (LTE) network with respect to the employed scheduling strategy and the cell-load is investigated. Considering a traffic-mix with services that have different requirements, the scheduler needs to be aware of the type of data to be scheduled and of the state of each user, e.g., the availability of user-data for transmission or how long a packet of a user has been in the queue already. Classical schedulers are not designed for such scenarios. To overcome this situation, we propose two new scheduling strategies based on proportional fair scheduling since the metric for proportional fair scheduling is already well suited for the kind of traffic-mix that we assume in our network. We compare the performance of the classical scheduling strategies to the new approaches in terms of average-delay but also throughput by applying them in traffic-situations, where various services are used under different load-situations.