0018-9545 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TVT.2017.2711582, IEEE Transactions on Vehicular Technology IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. XX, NO. XX, XXX 2017 1 Optimization of handover parameters for LTE/LTE-A in-building systems Diego Castro-Hernandez, Raman Paranjape, Member, IEEE Abstract—The optimization of handover parameters for in- building systems is investigated in this paper. We proposed a novel methodology that provides in-building base stations with the flexibility to customize handover (HO) parameters to specific radio frequency conditions at the cell-edge for different loading scenarios. We propose the use of machine learning and data mining techniques to allow the base stations to autonomously learn and identify characteristic patterns in the received signal strength values (reported by users during the handover process), and apply optimal HO parameters for each case. Our optimiza- tion strategy jointly considers the radio frequency conditions at the cell-edge and the load levels of the base stations, to determine optimal handover parameters that maximize the quality of service and guarantee the continuity of service at the cell-edge. We evaluated our methodology with experimental data collected from two fully operational LTE in-building systems deployed in a university campus. Our results show that with our methodology the spectral efficiency at the cell-edge can be greatly improved. Downlink data rate gains at the cell-edge reached a value close to 150% for a certain loading scenario compared to the traditional approach of selecting a unique set of handover parameters for the entire in-building system. Index Terms—Mobility, clustering, in-building, handover, self- optimizing, HetNets. I. I NTRODUCTION &MOTIVATION H ETEROGENEOUS networks (HetNets) are becoming an attractive option for network operators to increase the capacity of their cellular networks as well as improve coverage in strategic locations, e.g. high traffic areas. In HetNets, the capacity of the network is increased by deploying low power small cells (e.g. microcell, picocells, femtocells), within the coverage of high power tower-mounted macrocells. The low power consumption and the small size factor are aspects that make HetNets a convenient and low cost solution, in particular to provide service in indoor environments as “in building systems”. Nowadays, in-building systems are being largely deployed in different venues like: shopping centers, airports, stadi- ums and university campuses. However, the optimization of handover parameters in in-building systems is a challenging task for network operators, specially due to the nature of HetNets that combine low power and high power base stations. Typically, operators define the handover parameters with the Copyright (c) 2017 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending a request to pubs-permissions@ieee.org. The authors are with the Department of Electronic Systems Engineer- ing, University of Regina, SK, Canada. (e-mail: castrohd@uregina.ca, Ra- man.Paranjape@uregina.ca) Manuscript received XXX, XX, 2016; revised XXX, XX, 2016. objective of guarantying the continuity of service at the cell edge. However, with the implementation of new technologies like VoLTE, the quality of service provided to users as they move between coverage areas becomes also a relevant aspect. Two main factors make the optimization of handover pa- rameters a complex task for in-buildings systems: irregular cell-edge conditions and dynamic variation of the load. In the first place, the irregularity of the radio frequency (RF) conditions at the cell-edge of in-building systems is due to the uneven levels of interference caused by the outdoor macrocell [1]. In certain situations, it may be preferred to execute a handover as early as possible due to rapidly degrading radio frequency conditions as users approach the cell-edge of the in-building system. In other situations, it may be preferred to delay the execution of the handover to avoid unnecessary execution of handovers. Currently, most network operators define a unique set of handover parameters for the entire in-building system. However, such unique set of parameters could be too aggressive in some cases or too conservative in others. Additionally, the optimization of handover parameters becomes more complicated if we also consider the second factor: the load. Commonly, due to their large footprint, macrocells tend to handle a large number of users, in some cases even reaching congestion. Therefore, in such scenario, it would be advantageous for the base station of the in-building system to delay the execution of the handover, in order to keep the quality of service provided to cell-edge users at an acceptable level before handing them over to the macrocell. Hence, the conditions of the cell-edge (e.g. signal strength, interference level) and the loading conditions of the cells determine the proper set of handover parameters for optimal operation. In this paper, we propose a methodology to optimize the handover parameters of in-building systems. Such methodol- ogy considers the two factors described before jointly (i.e. RF conditions at the cell-edge and load level of the cells) to determine the optimal values of the handover parameters. Our approach comes to answer the question of how late or how early a handover can be executed in order to maximize the quality of service at the cell-edge while minimizing handover failures and reducing the triggering of unnecessary handovers. We propose the use of machine learning and data mining techniques to allow the in-building system to autonomously learn and identify characteristic patterns in the signal strength received from users as they approach the cell-edge, and apply optimal handover (HO) parameters for each case. We evaluated the performance of our approach with experimental data collected from fully operational LTE in-building systems.