| Future green WiFi networks will require Access Points (APs) to minimize their energy usage and maximize their spectral efficiency; equivalently, APs need to maximize their Energy Efficiency (EE). To this end, transmit power control and channel bonding are promising approaches to improve the EE of APs. However, these approaches may cause interference between neighboring APs that lead to low capacity or data rates. To this end, we first outline a Continuous Time Markov Chain (CMTC) to study an AP's channel bonding process given random traffic arrivals and interference. After that, we outline a Deep Reinforcement Learning (DRL) solution that controls when an AP uses a single or bonded channel and its transmit power. Simulation results show that an AP equipped with our DRL solution improves its EE by 40\% to 560\% as compared to competing solutions. |