Abstract
This paper provides an accurate target cell's RSRP (received signal received power) prediction technique for cellular handovers, ensuring robust connectivity for autonomous vehicles (AVs). We propose an extreme gradient boosting (XGBoost)-based mechanism to predict channel state information (CSI) in advance prior to a cell handover request due to lower RSRP. Our test results indicate that for speeds ranging from 0 to 120 km/h, the proposed prediction technique improves the handover success rate (HSR) by up to 4%. In particular, the average achieved success rate with the proposed algorithm is 97% compared to the conventional algorithm providing only 93% success rate. The proposed solution can work for any frequency pair and wireless technology.