Radio Resource Management for 5G MU MIMO over vector computing
5G MU MIMO RRM is a family of algorithms that solves a high complexity resource allocation problem including user grouping, MCS selection, and frequency resources assignment, and this in online manner, with a new decision done every TTI, that is, every 0.5ms. In the current state, with 64 antennas of one gNB and dozens of UE with up to 8 antenna each, and with up to 16 layers in every group, it represents a complexity challenge even for traditional approaches targeting at eMBB traffic and maximal spectral efficiency. But if we consider URLLC scenario with strict latency QoS and besides, other dimensions such as gNB owning 128 or 256 antenna and serving up to 32 layers, and Multi-Carrier operations at one gNB, then the intractable nature of the NP-hard problem to be solved becomes a more and more severe limitation. In similar cases, e.g., in Big Data, the technology had to climb new computational resources, such as NVidia GPU cards. In our talk we will consider RRM from this perspective and try to answer the very first questions and mostly invite the audience to look towards these perspectives.