0018-9545 (c) 2018 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.2018.2795650, IEEE Transactions on Vehicular Technology JOURNAL OF L A T E X CLASS FILES, VOL. 11, NO. 4, DECEMBER 2012 1 Over-the-Air Tests of High-Speed Moving LTE Users in a Reverberation Chamber Davide Micheli, Massimo Barazzetta, Riccardo Diamanti, Pietro Obino, Renzo Lattanzi, Luca Bastianelli, Valter Mariani Primiani, Senior Member, IEEE, and Franco Moglie, Senior Member, IEEE Abstract—High-speed trains are nowadays a reality in most countries. Internet services and connectivity are required by the passengers on board using smartphones and laptops. Neverthe- less, the higher the speed, the bigger the issues that operators have to face to implement mobility and to reduce the effect of fast fading signal and Doppler shift, which may cause throughput reduction and problems to access the network. In the past, operators tried to optimize the cellular access to UMTS cells covering railways. Now that 4G networks have become mature and deployed, a similar approach needs to be optimized and implemented to LTE cells covering the same targets. We present the results of a testing campaign replicating real “high speed train” propagation scenarios, like a train running in open space or in tunnels with the help of a reverberation chamber. The chamber was partially filled by absorbing panels and equipped with rotating stirrers in order to approach the multipath prop- agation and to replicate the fading conditions that is typical of smartphones in a train coach. The effects of Doppler and its fast variations are presented, together with an analysis of the efficiency of different transmission solutions for tunnels (single input single output – SISO – cells propagating inside them). The testing sessions were performed under a collaboration program between Telecom Italia (in the paper referred as “operator”), Nokia Networks and Universit` a Politecnica delle Marche. After the experiments in our laboratory, some features were introduced in the operator’s live network, covering the high-speed railway between the two Italian cities of Piacenza and Bologna. Index Terms—4G, Doppler effect, high speed train, live radio access network, LTE, MIMO, PRACH, radio base station, ran- dom access procedure, reverberation chambers, SISO, through- put. I. I NTRODUCTION L ONG-term evolution (LTE) networks are nowadays im- plemented in many countries. With respect to third gener- ation systems, LTE is able to grant higher bandwidth, serving higher number of users and providing higher throughput [1]. Laboratory OTA testing of LTE communication systems can be carried out by different methods. The two-stage method is able to model the real multipath channel, it is based on anechoic chamber measurements of the antenna radiation pattern [2]. D. Micheli, P. Obino and R. Lattanzi are with TIM S.p.A., Via Parco de’ Medici 69, 00148 Rome, Italy e-mail: davide.micheli@telecomitalia.it; pietro.obino@telecomitalia.it; renzo.lattanzi@telecomitalia.it M. Barazzetta is with Nokia Networks Italia, Energy Park 14, 20871 Vimercate, Italy e-mail: massimo.barazzetta@nsn.com R. Diamanti is with TIM S.p.A., Via Guido Miglioli 11, 60131 Ancona, Italy e-mail: riccardo.diamanti@telecomitalia.it L. Bastianelli, V. Mariani Primiani and F. Moglie are with Dipartimento di Ingegneria dell’Informazione, Universit` a Politecnica delle Marche, via Brecce Bianche 12, 60131 Ancona, Italy e-mail: l.bastianelli@pm.univpm.it; v.mariani@univpm.it; f.moglie@univpm.it Manuscript received Xxxxx XX, XXXX; revised Xxxxxxxx XX, XXXX. The anechoic chamber multiprobe method (ACMP) simulates mathematical multipath environment [3]. It is conceptually simple and reproduces an over the air test (OTA) even if it suffers the highest cost and needs extensive system calibration effort. The dimension of the obtained test area depends on the number of employed probes located to synthesize the radio channel [4]. For that reason, the ACMP was also simulated by a conductive setup [5]. The reverberation chamber method is based on the generation of a random multipath environment whose characteristic can be tuned from a Rayleigh distribution to a Rice distribution [6] according to the environment under replication [7]. A valuable comparison among all methods can be found in [8], together with the description of the radiated version of the wireless cable method. A reverberation chamber (RC) provides a lower cost test facility very close to some real life environments and this is the technique adopted in the present paper. While LTE access allows very high peak throughput (currently up to 1 Gbps in the downlink direction with the aid of carrier aggregation [9]), the speed of fast- moving mobiles may have negative impacts on performance and call-setup success rate, which reflects the user’s perception for network availability. Access to cells could be problematic for a fast moving user equipment (UE), and random access procedure (RAP) may fail. Specifications say that access to physical random access channel (PRACH) must occur via cyclic shifts of few root sequences [10], and that signatures generated from the cyclic shifts of a common root sequence are orthogonal, while signatures generated from cyclic shifts of different root sequences are not orthogonal. Depending on the required cell range, cells may have few root sequences with a lot of cyclic shifts (little cell range, optimal orthogonality) or many root sequences with few cyclic shifts (high cell range, sub-optimal orthogonality). Cellular network operators some- times use non-orthogonal sequences in order to expand the cell range and provide wide coverage. Moreover, orthogonality could be further compromised due to the high speed of the UE, if no precautions are taken. In the present paper, a 3rd Generation Partnership Project (3GPP) defined solution to maintain the signature orthog- onality (called PRACH usage of restricted set) has been implemented and tested. Another aspect highlighted by this work is that the pure Doppler effect itself weakly affects LTE performance, even if the downlink suffers from high-speed conditions more than the uplink. The performance starts to degrade heavily when propagation inside tunnels is reproduced in an RC by using two simultaneous transmissions of the same signal, one affected by positive and the other affected by