Sojourn time estimation-based small cell selection in Ultra-Dense Networks

摘要

Ultra-Dense Networks (UDNs) is a clear trend for future 5G while dense deployment of small cells (SCs) would result in unnecessary handovers (UHOs) due to a shorter sojourn time in the target SC and thus user experience is degraded. The main reason for this problem is that existing small cell selection (SCS) schemes ignore the importance of sojourn time. Therefore, in this paper, a SCS scheme based on sojourn time estimation is proposed. The Cumulative Distribution Function of the sojourn time depending on the user velocity and SC intensity is presented firstly. Furthermore, the instantaneous sojourn time is calculated by estimating the chord length of user trajectory in the coverage of candidate SC, in which user direction is obtained by applying a direction prediction scheme meanwhile the coverage of SC is determined by Poisson-Voronoi Tessellations (PVTs). In addition, the optimal sojourn time threshold is determined for the purpose of lower handover failures and those candidate SCs whose sojourn time are shorter than the predefined threshold would be skipped. Simulation results show that compared with a typical parameters setting (TTT=320ms, Offset=3dB) in LtE, the proposed scheme can reduce the Ping Pong (PP) rate and UHO rate by up to 50% and 60%, respectively. Moreover, the handover failures can be reduced by up to 30%.