Runtime Adaptive LTE MAC Schedulers with Stochastic Guarantees

We describe a process for the automated design and refinement of a class of scheduling algorithms, such as those used in the MAC layer of state-of-the-art LTE cellular systems. We provide a simulation-based setup that produces a trained knowledge base that drives a simple scheduling algorithm within a range of operating conditions. The advantage of the resulting setup is that the scheduling algorithm can be trained to adapt to a variety of disparate runtime constraints not normally considered as part of the algorithm design phase. [placeholder]