The APi (which is basically what you ask) design should address the following scheduling challenges:

• Ultra-densification: As the number of connected devices increases, the number of users per cell increases dramatically. This ultra-densification will require more efficient resource management and scheduling algorithms.
• Massive multiple-input and multiple-output (MIMO): In 5G systems, where the base station (BS) will be equipped with a large number of antennas. Massive MIMO will provide significant gains in spectral and energy efficiency, as well as improved spatial multiplexing capabilities. However, it will introduce additional challenges to the scheduling process.
• User mobility: As user mobility increases, the channel characteristics will change more rapidly. In order to handle the rapidly changing environment, the scheduler must adapt to these changes in real time.
• Heterogeneous networks (HetNets): HetNets are a major component of 5G systems, where small cells are overlaid within the macrocell. The scheduling process will need to account for the different channel characteristics and traffic loads in each cell.
• QoS requirements: 5G systems will support a wide range of applications, with different QoS requirements. The scheduler must be able to support these.

You can pass Queue state information (QSI) and channel state information (CSI) to the ML program. The ML algorithm would output back the matrix of RB allocations to UEs.

Thanks for your answers. Please advise on these two follow on questions (a) What would be an appropriate ML algorithm to try out? (b) Should we write this code in C or is it better to write it in Python?

Where can I find some summary information on how NetSim's MAC scheduler works? Without having to read the detailed documentation.