Coordinates-Based Resource Allocation Through Supervised Machine Learning

Appropriate allocation of system resources is essential for meeting the increased user-traffic demands in next generation wireless technologies. Traditionally, relies on channel state information (CSI) users optimizing resource allocation, which becomes costly fast-varying conditions. In such cases, an estimate terminals’ position provides alternative to estimating condition. this work, we propose a coordinates-based scheme using supervised machine learning techniques, and investigate how efficiently performs comparison traditional approach under various propagation We consider simple setup as first step, where single transmitter serves mobile user. The performance results show that achieves very close CSI-based scheme, even when available user’s coordinates are erroneous. quite consistent, especially complex frameworks like random forest neural network used allocation. terms applicability, training time about 4 s required algorithm, appropriate predicted less than 90 $\mu \text{s}$ with learnt model size < 1 kB.