近年來，行動通訊之資訊傳輸量快速地提升，為滿足用戶之需求，大、小細胞並存之異質網路在LTE系統中被廣泛討論。異質網路系統可減低發射功率、提高頻譜效率，進而提升總體資訊傳送量以達到5G網路之需求。然而，由於小細胞密度增加，其伴隨之行動用戶頻繁且不必要之細胞間交接(handover)程序，也引發系統複雜性及整體服務品質降低之疑慮。為避免行動用戶在大、小細胞間頻繁交接及乒乓效應(ping-pong effect)，一般演算法直接以用戶之移動速率作為執行交接程序之判斷條件，但無法有效避免乒乓效應的發生，也容易錯失分擔大細胞負載的機會。因此，本論文除了採用模糊Q學習(fuzzy Q-learning)，以達到細胞間之負載平衡外，亦針對乒乓效應以及大、小細胞間負載分擔的問題，提出使用鏈路預測演算法(link prediction algorithm, LPA)以估測用戶將於目標基地台服務範圍中所逗留的時間，並將之作為執行交接程序的判斷準則，以降低乒乓效應發生頻率。系統模擬結果顯示本論文提出之方法能有效降低不必要之交接程序，且不會伴隨嚴重的通話中斷率(call drop rate, CDR)。

In recent years, to satisfy the rapidly increased users’ demand of mobile data communications, heterogeneous network (HetNet), which is composed of macrocells and small cells (picocells, femtocells), has been widely discussed in LTE systems. HetNet system is capable of reducing transmission energy, improving spectrum efficiency, and increasing data throughput. However, as the increase of small cell density, users’ unnecessary handovers and ping-pong effects among cells cause higher complexity and lower quality of service. To avoid users’ improper handovers, most of the conventional handover algorithms adopted directly the user’s speed as the handover decision criterion. In practice, these schemes cannot reduce ping-pong effects and, moreover, will miss the opportunity to re-allocate macrocells’ loading. Hence, this project adopts not only the fuzzy Q-learning algorithm to achieve load balancing among cells, but also a link prediction algorithm (LPA) to estimate user’s time-of-stay (ToS) in target cell which is used as one of handover decision criteria. Simulation results show that the proposed handover algorithm based on ToS estimation can effectively reduce ping-pong effects and will not result in severe call drop rate (CDR).

[1] 3GPP TR 36.839, “Mobility enhancements in heterogeneous networks,” 2012.
[2] 3GPP TS 36.331, “Radio Resource Control (RRC);Protocol specification,” 2012.
[3] 3GPP TS 36.133, “Requirements for support of radio resource management,” 2012.
[4] Zhang, Haijun, et al. "Signalling cost evaluation of handover management schemes in LTE-advanced femtocell." Vehicular Technology Conference (VTC Spring), 2011 IEEE 73rd. IEEE, 2011.
[5] Lee, Choong-Hee, and Jae-Hyun Kim. "Time-of-Stay Estimation-Based Cell Selection Scheme in Multitier Heterogeneous Mobile Networks." IEEE Communications Letters 19.9 (2015): 1596-1599.
[6] Yang-Bang Li, “A Fuzzy Q-learning Approach for Load Balancing in LTE Systems,” National Cheng Kung University Thesis for Master of Science, July, 2015.
[7] Qin, Liang, and Thomas Kunz. “Increasing Packet Delivery Ratio in DSR by Link Prediction.” System Sciences, 2003. Proceedings of the 36th Annual Hawaii International Conference on. IEEE, 2003.
[8] 3GPP TR 36.814, “Further advancements for E-UTRA physical layer aspects” 2010.
[9] 3GPP TR 36.942, "Radio Frequency (RF) system scenarios," 2012.
[10] S. S. Mwanje and A. Mitschele-Thiel, "A Q-Learning Strategy for LTE Mobility Load Balancing," IEEE 24th International Symposium on Personal, Indoor and Mobile Radio Communications: Mobile and Wireless Networks, pp. 2154 - 2158, 2013.
[11] P. Munoz, R. Barco, I. de la Bandera, M. Toril and S. Luna-Ramírez, "Optimization of a Fuzzy Logic Controller for Handover-based Load Balancing," IEEE 73rd Vehicular Technology Conference (VTC Spring), pp. 1-5, 2011.
[12] Goldsmith, Andrea. Wireless communications. Cambridge university press, 2005.