Decentralized Q-learning for uplink power control

Fractional power control (FPC) is the simplified version of open loop power control (OLPC) in long term evolution (LTE) that relies on downlink path loss information from base station (BS). This allows user equipment (UE) to decide which power to use for uplink transmission. However, asymmetric behavior of uplink and downlink transmission in crowded network might cause unfair transmit power estimation. This motivates our investigation of implementing uplink path loss and q-learning algorithm to enable UE to decide appropriate transmit power on its own. In this study we apply the concept of FPC into q-learning, enabling UE to find suitable transmit power with respect to uplink path loss. 3GPP uplink path loss model is exploited in our study. We compare outputs between our proposed method and FPC. . From simulation, we find out that DQL performs better as compared to fractional power control in terms of signal-to-interference-noise-ratio (SINR) with average increase factor of 3.5.