Opportunistic Scheduling for Users with Heterogeneous Minimum Rate QoS Requirements

Future wireless applications, such as online gaming, live video streaming, augmented/virtual reality need high fidelity data transfer with reliable control information exchange to provide an immersive user experience. Developing wireless scheduling algorithms that can guarantee a minimum data rate for users over a few hundred milliseconds with possibly heterogeneous requirements and/or channel quality is crucial. We propose a class of measurement based wireless schedulers that are able to trade off between opportunistically scheduling users with good channels and prioritizing users with service deficits. The proposed algorithms continuously track the QoS provided to each user and switch between conservative or greedy resource allocation depending on user channel strength, QoS deadline, and past channel realizations. We also formulate an oracle-aided linear optimization of the minimum rate QoS constrained scheduling problem to derive a spectral efficiency bound. Through extensive simulations, we show that our proposed algorithms provide a much higher user rate than popular scheduling algorithms such as QoS-PF, EXP-QoS, and weighted MaxQuantile with better reliability when the network is critically loaded. We observe at least a two-fold increase in the mean rate for each user in the network when compared to the EXP-QoS algorithm.