Machine Learning approaches for predicting Throughput of Very High and EXtreme High Throughput WLANs in dense deployments

Wireless local area networks standards like IEEE 802.11ax and higher are being exploited thoroughly across the globe due to many encouraging factors. Dynamic Channel Bonding is a technique used extensively to increase the throughput of these networks often. Komondor is a Linux-based simulator that provides accurate estimates of throughput of WLANs based on the comprehensive inputs of the deployment detail. In complex and dense deployments, in the overlapping basic service sets (OBSS), the actual throughput of BSS varies quite a bit from its individual deployment. Machine learning techniques offer immense potential and flexibility to explore a variety of their models to ascertain the suitability for a given problem in hand of estimating the throughput under dense deployment of BSSs. The paper dwells on the ML techniques with the experimentation results and suggests the possibility of improving the results through fine-tuning the model parameters and so on.