Article DOI: 10.1111/exsy.12120 An adaptive fuzzy handover triggering approach for Long-Term Evolution network Chiew Foong Kwong, 1 * Teong Chee Chuah, 2 Su Wei Tan 2 and Ayyoub Akbari-Moghanjoughi 1 (1) Faculty of Science, Technology, Engineering and Mathematics, INTI International University, Nilai, Malaysia E-mail: chiewfoong.kwong@newinti.edu.my (2) Faculty of Engineering, Multimedia University, Cyberjaya, Malaysia Abstract: To cope with the increasing demand for efficient data delivery, self-organizing networks have been introduced in the Long Term Evolution (LTE) system to provide autonomous and flexible mobility management. The existing handover triggering scheme for LTE is not flexible enough to incorporate new performance metrics, and it introduces handover latency. There are studies on non-conventional handoff algorithms for LTE applications, for instance, the fuzzy logic approach. However, the fuzzy logic approach needs regular manual tuning to constantly produce optimal output. In this paper, we address this issue by proposing an adaptive fuzzy logic-based handoff decision algorithm, which can cope with environmental changes and improve efficiency by reducing human intervention. Performance results show that the proposed algorithm can reduce unnecessary handovers by about 20% compared with the fuzzy logic and conventional LTE handover triggering scheme, leading to reduced packet loss rates. Keywords: expert system < system, fuzzy system < system, knowledge base < system, artificial intelligence 1. Introduction Recent studies have shown that the operation of a 3G network, including mobility management, accounts for ~20% of the total operational expenditure (OPEX) cost (Xenakis et al., 2014). This is expected to increase proportionally with the size and sophistication of the technology deployed especially with today’s highly complex and advanced cellular networks such as the Long Term Evolution (LTE). One of the major goals in LTE’s mobility management is to reduce both the ping-pong handover (PPHO) and the radio link failure (RLF) in order to ensure a manageable OPEX (Akyildiz et al., 2014). The PPHO is an undesirable phenomenon in LTE networks in which a repeated handover occurs between two enhanced Node-Bs (eNBs) because of rapid fluctuations of the received signal strengths from both eNBs at the cell edge. An RLF, on the other hand, occurs when user equipment (UE) is out of the coverage or experiences a severed link for a period (Akyildiz et al., 2014). The conventional LTE handover decision-making based on the power budget algorithm (PBGT) relies mainly on the reference signal received power (RSRP) (Akyildiz et al., 2014). Handoff triggering is normally regulated by two parameters, namely the time-to-trigger (TTT) and handover margin (HOM). A right balance between the TTT and the HOM is vital to ensure that handovers are not delayed too long while avoiding unnecessary PPHOs. However, optimisation of both parameters requires extensive measuring campaign and statistical analysis (Munoz et al., 2013). There are non-conventional methods to achieve both timely handovers and flexibility, such as the fuzzy logic-based handover algorithm (FLHA) (Zekri et al., 2012). The FLHA also includes the ability to incorporate additional performance metrics for greater choice to configure complex networks. Besides linguistic in nature, which allows for ease of maintenance, the FLHA can make decisions based on multivariate analysis in environments of uncertainty (Zekri et al., 2012). However, the FLHA is not the ubiquitous solution and does not fully mitigate the problems of the PBGT algorithm. It requires human experts and experience to formulate practical fuzzy rules and membership functions through a manual tuning process to obtain the desired output. In addition, the FLHA also does not entirely address the issue of timely triggering of handovers and the flexibility to adapt to rapid changes in the mobile radio channel. In this work, we develop an extended version of the conventional FLHA algorithm, herein known as the adaptive fuzzy logic handover algorithm (adaptive-FLHA) for LTE networks. Although the FLHA has been proposed to be used in LTE networks, for example, the work by Aziz et al. (2010), to our best knowledge, the adaptive-FLHA approach has not been studied for use in LTE systems. In contrast to the FLHA approach, the proposed adaptive- FLHA allows tuning of the fuzzy membership functions © 2015 Wiley Publishing Ltd 30 Expert Systems, February 2016, Vol. 33, No. 1