Blockage Prediction and Fast Handover of Base Station for Millimeter Wave Communications

We propose a blockage prediction and fast base station (BS) handover (BP-FBSH) scheme based on the reference signal received power (RSRP) of the mobile terminal (MT) and the indices of the BS transmit beams for millimeter wave communications. Using a specific beam tracking method called neighborhood beam search, the BS transmits multiple neighborhood beams to the MT and collects the RSRPs of these beams from the MT. Then, the BP-FBSH scheme uses the beam-associated information sequences composed of the RSRPs and the indices of the BS transmit beams to train a long short-term memory (LSTM)-based blockage prediction neural network (BPNN). If the BPNN predicts the MT is to be blocked, the scheme triggers a handover of this MT to an adjacent BS as well as determining an initial access (IA) beam for it by an LSTM-based BS handover neural network. Simulation results based on Wireless Insite software show that the scheme can achieve high success rate for both the blockage prediction and the IA beam prediction.