Efficient Conditional Handover Algorithm in 5G with Blockages using Recurrent Neural Network

In the Third Generation Partnership Project (3GPP) specification [1], the conditional hand over procedure is defined to reduce the hand over failures in 5G networks by early preparation of the radio resource for user equipment (UE). In this paper, in order to minimize the resource reservation time while maintaining the radio link failures (RLF) rate of a UE, we make the first attempt to trigger the conditional handover of the UE only when the UE would experience an RLF in the near future. Our idea is to adjust the handover margin of triggering the conditional hand over of a UE based on the prediction on whether the UE would experience an RLF in the near future. To achieve an accurate prediction in the urban scenario with blockages, we reduce the conditional handover problem to a classification problem and solve the classification problem using a recurrent neural network (RNN). Simulations show that the proposed algorithm has extraordinary performance in terms of the RLF rate and resource reservation time, as compared with the state-of-the-art methods.