Prediction-Based Conditional Handover for 5G mm-Wave Networks: A Deep-Learning Approach
Changsung Lee, Hyoungjun Cho, Sooeun Song, Jong‐Moon Chung
Abstract
Conditional handover (CHO) is one of several promising mobility enhancements in 5G networks. By making preparation decisions earlier than in LTE HO, CHO can provide an improved HO success rate. This article, analyzes the strengths and weaknesses of CHO by comparing CHO to 5G baseline HO. Since millimeter-wave communications are vulnerable to blockages, sudden changes in signal reception power can mislead CHO into making undesired early preparations in 5G networks. To enhance the robustness of CHO, current studies propose using an increased number of preparations, resulting in considerable signaling overhead. This article offers a novel prediction-based CHO (PCHO) scheme that uses deep-learning technology to overcome the weaknesses of CHO and make more intelligent preparation decisions. Based on the changes in the signal patterns of the base stations, PCHO uses former blockage information to predict the best next base station to which to conduct HO. Performance evaluation demonstrates that PCHO can improve the early preparation success rate while reducing signaling overhead compared to current CHO schemes.