Rimedo Labs RF Channel Reconfiguration rApp

The RF Channel Reconfiguration rApp significantly reduces RAN energy consumption by dynamically adapting the active antenna array geometry (e.g., scaling from 64T64R to 32T32R or 16T16R and back) depending on the traffic load. It employs a Deep Q-Learning (DQL) agent to predict the impact of reconfiguration on user throughput, ensuring savings do not violate SLAs. This granular control offers higher performance than simple cell sleep modes because the cell remains active and available to users, operating merely in a more power-efficient state. Supports sustainability goals and OpEx reduction for 5G M-MIMO deployments.

Unlike binary "Cell Off" approaches, RCR offers a "Low Power Mode" middle ground. It keeps the cell operational for signalling and basic coverage but drastically reduces power draw (up to a 24.8% of the total power). This is ideal for medium-load scenarios where traffic is too high for a complete shutdown but too low to justify powering 64 RF chains. Moreover, the transition time for RF channel reconfiguration is much shorter compared to the cell on/off switching, taking advantage of short low-load periods, e.g., of a few minutes.

The RF Channel Reconfiguration rApp can cooperate with other energy-saving algorithms operating at different time-scales, e.g., cell on/off switching. While cell on/off switching provides energy saving in a long-time scale, RF Channel Reconfiguration rApp complements it during medium-load, or short off-peak periods.