Portal de Periódicos da CAPES

Stochastic Channel Modeling for Railway Tunnel Scenarios at 25 GHz

2018; Electronics and Telecommunications Research Institute; Volume: 40; Issue: 1 Linguagem: Inglês

10.4218/etrij.2017-0190

2233-7326

Danping He, Bo Ai, Ke Guan, Zhangdui Zhong, Bing Hui, Junhyeong Kim, Heesang Chung, Kim Ilgyu,

Power Line Communications and Noise

ETRI JournalVolume 40, Issue 1 p. 39-50 ArticleFree Access Stochastic Channel Modeling for Railway Tunnel Scenarios at 25 GHz Danping He, Danping He orcid.org/0000-0002-0917-5013 Search for more papers by this authorBo Ai, Bo AiSearch for more papers by this authorKe Guan, Corresponding Author Ke Guan kguan@bjtu.edu.cn Search for more papers by this authorZhangdui Zhong, Zhangdui ZhongSearch for more papers by this authorBing Hui, Bing HuiSearch for more papers by this authorJunhyeong Kim, Junhyeong KimSearch for more papers by this authorHeesang Chung, Heesang ChungSearch for more papers by this authorIlgyu Kim, Ilgyu KimSearch for more papers by this author Danping He, Danping He orcid.org/0000-0002-0917-5013 Search for more papers by this authorBo Ai, Bo AiSearch for more papers by this authorKe Guan, Corresponding Author Ke Guan kguan@bjtu.edu.cn Search for more papers by this authorZhangdui Zhong, Zhangdui ZhongSearch for more papers by this authorBing Hui, Bing HuiSearch for more papers by this authorJunhyeong Kim, Junhyeong KimSearch for more papers by this authorHeesang Chung, Heesang ChungSearch for more papers by this authorIlgyu Kim, Ilgyu KimSearch for more papers by this author First published: 15 February 2018 https://doi.org/10.4218/etrij.2017-0190Citations: 15 Danping He (hedanping@bjtu.edu.cn), Bo Ai (boai@bjtu.edu.cn), Ke Guan (corresponding author, kguan@bjtu.edu.cn), and Zhangdui Zhong (zhdzhong@bjtu.edu.cn) are with the State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University and also with the Beijing Engineering Research Center of High-speed Railway Broadband Mobile Communications, China. Bing Hui (huibing@etri.re.kr), Heesang Chung (hschung@etri.re.kr), and Ilgyu Kim (igkim@etri.re.kr) are with the 5G Giga Service Research Laboratory, ETRI, Daejeon, Rep. of Korea. Junhyeong Kim (jhkim41jf@kaist.ac.kr) is with the 5G Giga Service Research Laboratory, ETRI, Daejeon, Rep. of Korea and the School of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Rep. of Korea. AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract More people prefer using rail traffic for travel or for commuting owing to its convenience and flexibility. The railway scenario has become an important communication scenario in the fifth generation era. The communication system should be designed to support high-data-rate demands with seamless connectivity at a high mobility. In this paper, the channel characteristics are studied and modeled for the railway tunnel scenario with straight and curved route shapes. On the basis of measurements using the "Mobile Hotspot Network" system, a three-dimensional ray tracer (RT) is calibrated and validated for the target scenarios. More channel characteristics are explored via RT simulations at 25.25 GHz with a 500-MHz bandwidth. The key channel parameters are extracted, provided, and incorporated into a 3rd-Generation-Partnership-Project-like stochastic channel generator. The necessary channel information can be practically realized, which can support the link-level and system-level design of the communication system in similar scenarios. 1 Introduction Currently, a growing number of people prefer to take rail traffic for traveling and commuting because of the comfortable experience and convenience that it provides. In order to meet the goals of efficiency, safety, and convenience, rail traffic is expected to be "smart" in the fifth generation (5G) era. The railway infrastructure, trains, travelers, and goods will be increasingly interconnected. Thus, rail-traffic communication has become an increasingly important topic, and seamless high-data-rate wireless connectivity is desired. Consequently, railway communications are required to evolve from handling only the critical signaling applications to support various high-data-rate applications 1, 2. In order to realize this vision, the millimeter wave (mmWave) band has become important to overcome spectrum scarcity, and many efforts have been carried out to develop novel technologies such as multiple-input–multiple-output (MIMO) beamforming 3-6, resource allocation, and multiple access. The 5GCHAMPION project 7, funded by the H2020 Europe–Korea collaborative program, aims to develop key enabling technologies for a proof-of-concept environment to be showcased at the 2018 Winter Olympics in PyeongChang, Korea. One of the key applications is to provide a high-mobility broadband connection via a 5G mmWave high-capacity backhaul in the range of 24 GHz to 28 GHz. The Electronics and Telecommunication Research Institute (ETRI), as a member of the 5GCHAMPION project, prototyped a new wireless communication system named "Mobile Hotspot Network (MHN)," which works in the range of 24 GHz to 30 GHz with 125-MHz, 250-MHz, 500-MHz, and 1-GHz bandwidths. It targets the support of services with a data rate of gigabits per second with a speed over 400 km/h 8, and several trials have been carried out in the Seoul Subway. Standardization organizations are becoming active in the promotion of the priority in standardizing mmWave high-speed railway (HSR) communication technologies as well. In IEEE 802.15, the "High Rate Rail Communications" interest group was founded in 2014 to invite proposals and studies on broadband HSR communications. In 2016, the Macro + Relay deployment for the HSR scenario was agreed to be included in the 3rd Generation Partnership Project (3GPP) evaluation 9-11. The working frequency is around 30 GHz, and the bandwidth is above 100 MHz. Each baseband unit is attached to three remote radio heads (RRHs). RRHs are uniformly deployed along the two rail tracks. The suggested distance from the RRH to the rail track is 5 m. MIMO systems with a unidirectional beam or bidirectional beam are recommended to compensate the high attenuation of mmWave band propagation. A correct understanding of the mmWave propagation channel characteristics in railway scenarios is mandatory to effectively support the design and evaluation of communication systems 12. However, the mmWave band channel has been explored mainly for urban indoor and outdoor scenarios 13-16. According to 17, the railway environments are divided into 19 scenarios on the basis of a review of four different HSR lines and eight HSR stations. Five applications are defined in 1, 2 from the viewpoint of propagation and wireless channels, including the train-to-infrastructure, intercar, intracar, inside-the-station, and infrastructure-to-infrastructure communications. The route shapes are also very important to the mobility and communication design. Straight and curved route shapes are typical in railways. However, most of the current studies focus on the straight route owing to a lack of measurements of the curved route. The works on existing standardized channel modeling rarely provide dedicated parameters for railway scenarios 18, 19. In this study, the channel characteristics of tunnel scenarios are studied and modeled. Measurements are conducted in the 5G mmWave band in Seoul Subway Line 8 for both straight and curved routes. The 3D environments of the measurement campaign are modeled, and a ray tracer (RT) is calibrated and leveraged with the measurements to explore more channel characteristics. The key time-variant channel parameters and their correlations are modeled. The stochastic channels are realized on the basis of a 3GPP-like channel generator. The validated results indicated that the 3GPP-like framework is suitable for describing high-mobility scenarios. With the provided parameters, researchers and engineers can practically realize 3GPP-like channels to evaluate the designed communication technology in similar scenarios. The remainder of this paper is organized as follows: The measurements and RT calibration are introduced in Section II. The channel parameters are analyzed and modeled in Section III. The conclusions are drawn in Section IV. 2 Measurement Campaign and Ray-Tracer Calibration 2.1 Measurement Campaign An MHN Radio Unit (mRU), which is also called a transmitter (Tx) in this work, is installed on the side wall of the tunnel (see Fig. 1(a)), and the MHN Terminal Equipment (mTE), which is also called a receiver (Rx), is installed at the middle of the front window in the cab. The heights of the mRUs and mTE are 3.2 m and 3.0 m, respectively. The shortest two-dimensional distance between an mRU and the mTE is 2.8 m. The antenna used for both the mRUs and mTE is an 8 × 8 patch array antenna, as shown in Fig. 1(b). The measurements are conducted by installing the MHN test bed in Seoul Subway Line 8 from Jamsil station to a place after Songpa station, as shown in Fig. 1(c). A curved route is connected to a straight route in the measurement campaign. The radius of curvature of the tunnel is around 500 m. The lengths of the curved route and the straight route are 600 m and around 1,700 m, respectively. The half-power beam width (HPBW) is 8°, and the antenna gain is 22 dBi with vertical–horizontal dual polarization, as shown in Fig. 2. The main lobes of both the mRUs and mTE are pointed at each other, and handover occurs when the train is near an mRU. Three mRUs (mRU1, mRU2, and mRU3) are installed in the curved route with distances less than 300 m to guarantee a stable communication link under possible non-line-of-sight conditions. Only two mRUs (mRU4 and mRU5) are installed on the straight route, and the largest tested link distance is 1,180 m at mRU5. Figure 1Open in figure viewerPowerPoint Measurement campaign: (a) MHN test bed mRU (Tx) installation details, (b) MHN test bed mTE (Rx) installation details, and (c) measurement along Seoul Subway Line 8. Figure 2Open in figure viewerPowerPoint Antenna pattern of mRU and mTE (V-polarization): HPBW = 8°, Gain = 22 dBi. 2.2 Environmental Modeling As shown in Fig. 3, the cross section of the tunnel is a rectangle with pylons in the middle that separate the two-way tracks. The width of the tunnel is 13.7 m, and the height is 7.1 m. Figure 4 shows the constructed straight and curved tunnel models for the RT simulation. The side wall and pylons are made of concrete, the tracks are made of metal, and the train is made of metal (train body) and glass (windows). The lengths of the two models are greater than 1,200 m, which are longer than the measured length. Figure 3Open in figure viewerPowerPoint Cross section of Seoul Subway Line 8 20. Figure 4Open in figure viewerPowerPoint (a) Straight and (b) curved tunnel scenarios. 2.3 Ray-Tracer Calibration The deployment and configuration of the mRUs and mTE in the RT are the same as those in the measurement campaign. A calibration algorithm based on a simulated annealing method 21, 22 is employed to reduce the error by seeking the appropriate material parameters. The calibrated dielectric parameters and scattering coefficients of the directive scattering model 23 of concrete, glass, and metal are provided in Table 1; note that the calibrated transmission loss of glass is 833.33 dB/m (the equivalent attenuation with a thickness of 6 mm is 5.0 dB). Figure 5 shows the progress of the calibration. As the number of iterations increases, the calibration error decreases and saturates after the 4th iteration. The received power of snapshot s is expressed as , and Fig. 6 compares the cumulative distribution functions (CDFs) of the error of Prx (mW) before and after RT calibration. The mean absolute errors of the calibrated RT are 4.2 × 10−5 mW (straight route) and 5.1 × 10−5 mW (curved route); the standard deviation (STD) of the absolute errors in milliwatts are 6.3 × 10−5 and 7.6 × 10−5, respectively. Thus, the calibrated RT results match the measurements well. Thereafter, intensive reliable RT simulations are conducted to practically explore more characteristics (azimuth/elevation angular spreads, omnidirectional channel characteristics) that could not be captured in the measurements. Figure 7 shows the transmitted and reflected rays traced for the curved tunnel. Table 1. Material parameters Material name ϵʹr ϵʺr A (dB/m) S α Concrete 5.310 0.307 Not involved 0.0011 40 Glass 6.270 0.149 833.33 6.27 × 10–4 91 Metal 1.000 107 INF 0 0 Figure 5Open in figure viewerPowerPoint Progress of the calibration results. Figure 6Open in figure viewerPowerPoint CDF of the absolute errors before and after calibration. Figure 7Open in figure viewerPowerPoint Rays traced for the straight and curved tunnels. 3 Simulation and Analysis of the Results The MHN system works from 24 GHz to 30 GHz and supports a maximum bandwidth of 1 GHz. Recently, the Korean government released a regulation that the total effective isotropic radiated power should be smaller than 36 dBm, and the allocated frequency is 25 GHz to 25.5 GHz, which is lower than that in the measurements. To re-evaluate the system performance, accurate channels are needed for straight and curved routes with new parameters. RT simulations are conducted at 25.25 GHz with omnidirectional antennas at the Tx and Rx. As most services have high requirements for the downlink in practice, the channel propagation and characteristics are analyzed by considering an mRU as the Tx and the mTE as the Rx in this work. The travel distance of the Rx is 1,000 m for the straight route and 350 m for the curved route. The distance between the Tx and the track center is 2.8 m. The snapshot sampling interval is 0.1 m. Thus, the total numbers of snapshots are 10,001 for the straight-route scenarios and 5,001 for the curved-route scenarios. Up to the 10th order of reflection, scattering and transmission are considered in the simulation. 3.1 Path-Loss Model Figure 8 shows how the path loss PL varies with the Tx–Rx distance d. The "A-B" model is employed in this work to fit the PL: (1)where d is the distance between the Tx and the Rx, A is the slope, B is the intercept, and Xσ is the shadow fading (SF), which can be expressed as a zero-mean Gaussian random variable with an STD of σ. The fitting results are compared in Fig. 8 as well. As can be seen, a breakpoint at a distance dbp exists in both scenarios. When d ≤ dbp, the region is defined as a near region. On the contrary, the far region is where d > dbp. The extracted parameters are provided in Table 2. dbp of the straight tunnel is five times that of the curved tunnel. In both scenarios, Anear is less than Afar. It is noteworthy that the train body, which is made of metal, blocks the direct transmission path at the very beginning of the near region. Thus, only the reflected and scattered rays can reach the Rx, which results a great difference between the simulated path loss and the free-space path loss. In the far region, the path loss in the curved tunnel is more severe than that in the straight tunnel, and Acurved is more than two times Astraight. σcurved is smaller than σstraight in both the near and far regions. The correlated distance of SF is defined as λSF, and the units are meters. λSF of the near region is smaller than that of the far region, and λSF of the curved tunnel is at least four times shorter than that of the straight tunnel. This observation indicates that the variation in the straight tunnel is less than that of the curved tunnel. Figure 8Open in figure viewerPowerPoint Path losses of both scenarios: (a) path loss of the straight tunnel and (b) path loss of the curved tunnel. Table 2. Extracted parameters for the PL PL Straight tunnel (dbp = 257.2 m) Curved tunnel (dbp = 50 m) Near Far Near Far A 0.64 16.50 0 33.40 B 113.56 76.06 107.76 51.45 σ (dB) 5.93 4.96 4.53 4.83 λSF (m) 1.10 1.70 1.00 1.10 3.2 Delay Spread and Rician K-Factor Figure 9 shows the CDFs of the root-mean-square (RMS) delay spreads (DSs) στ of both scenarios. στ can be fitted by a lognormal distribution function (Table 3). In the straight tunnel, the mean RMS delay spread in the near region is larger than in the far region. In the curved tunnel, the values of στ for the near region and far region are similar and are fitted with the same model. The mean value and the range of variation of the straight tunnel are smaller than those of the curved tunnel. A Rician distribution indicates the presence of a specular dominant component in the channel over other very weak paths, and its probability density function (PDF) is expressed as (2) (3)where I0 is the modified Bessel function of the first kind and zeroth order, A is the amplitude of the dominant path, and σn is the STD of all other weak path amplitudes. The amplitudes are computed by using the peak values of the channel impulse responses. The Rician K-factor (KF) is expressed by (3). The CDFs are compared in Fig. 10, and the fitting results are summarized in Table 4. The mean KFs of both scenarios are greater than 24 dB. In the far region of the straight tunnel, the mean value is greater than that in the near region. The mean KF of the curved-tunnel scenario is the lowest, indicating that multipath components contribute more significantly compared to the straight tunnel. The correlated distance λKF of the curved tunnel is much less than that of the straight tunnel. Figure 9Open in figure viewerPowerPoint Delay spreads of both scenarios: (a) DS of the straight tunnel and (b) DS of the curved tunnel. Table 3. Extracted parameters for the DS DS Straight tunnel Curved tunnel Near Far Mean log10 (ns) –9.19 –9.24 –9.14 σDS log10 (ns) –9.29 –10.67 –9.53 λDS (m) 6.80 20.30 4.60 rDS (m) 2.83 2.83 2.57 Figure 10Open in figure viewerPowerPoint Rician K-factors: (a) KF of the straight tunnel and (b) KF of the curved tunnel. Table 4. Extracted parameters for the KF KF Straight tunnel Curved tunnel Near Far Mean (dB) 25.70 32.95 24.02 σKF (dB) 14.23 19.15 12.24 λKF (m) 11.20 14.60 2.10 3.3 Angular Domain According to the 3GPP definition 24, the conventional angular spread (AS) calculation for the composite signal is given by where Pn,m is the power for the subpath of the nth path, θn,m,μ is defined as μθ is defined as and θn,m is the angle of arrival/departure of the mth subpath of the nth path. m is 1 for RT simulation results. The RMS ASs of both scenarios are shown in Fig. 11. The fitting parameters are listed in Table 5. ASA, ASD, ESA, and ESD are the angular spreads of the azimuth angle of arrival (AoA), the azimuth angle of departure, the elevation angle of arrival (EoA), and the elevation angle of departure, respectively. The ASs of the straight tunnel in the far region are the smallest with the smallest variation. λAS of the far region of the straight route is the largest compared to the others (Table 6). Figure 11Open in figure viewerPowerPoint CDFs of the ASs. Table 5. Extracted parameters for the ASs ASD Straight tunnel Curved tunnel Near Far Mean log10 (˚) 0.63 9.65 × 10–4 0.02 σASD log10 (˚) 0.64 –0.52 0.44 λASD (m) 6.00 19.40 2.90 ESD Straight tunnel Curved tunnel Near Far Mean log10 (˚) 0.65 0.03 0.87 σESD log10 (˚) 0.56 –0.38 0.57 λESD (m) 2.90 17.50 4.80 ASA Straight tunnel Curved tunnel Near Far Mean log10 (˚) 0.37 –0.04 0.47 σASA log10 (˚) 0.62 –0.76 0.26 λASA (m) 6.95 19.90 4.00 ESA Straight tunnel Curved tunnel Near Far Mean log10 (˚) 0.28 –0.07 –0.07 σESA log10 (˚) 0.47 –0.94 0.12 λESA (m) 15.45 20.00 4.20 Table 6. Cross-correlation of the straight tunnel in the near region DS KF SF ASD ASA ESD ESA DS 1 0.52 0.20 0.87 0.76 0.83 0.78 KF 1 –0.12 0.65 0.23 0.64 0.21 SF 1 0.13 0.18 0.08 0.17 ASD 1 0.59 0.76 0.57 ASA 1 0.53 0.92 ESD 1 0.61 ESA 1 3.4 Cross-Correlation between the Key Parameters The cross-correlations are computed for the aforementioned parameters. Tables 7 and 8 summarize the parameters for the near and far regions of the straight tunnel, respectively. The cross-correlation of the DS and SF is positive in both regions, whereas the correlation between the DS/SF and other parameters are negative. Further, the cross-correlations of the other parameters significantly increase in the far region. When the diversity of multipath components (MPCs) increases, the variation in the SF increases as well. When the DS increases/decreases, the multipath components are more/less diverse in the time domain. The DS and SF are positively correlated, indicating that the SF is likely to vary with a similar trend as the DS. However, the angles are constrained within the narrow propagation region, and the range of ASs in the far region is less than that in the near region of the straight tunnel. Thus, the variations in the delays of rays barely influence the angles, especially in the far region of the straight tunnel. In the curved tunnel (see Table 8), the KF and SF are barely correlated with the other parameters. In both scenarios, the ASs are strongly correlated. Table 7. Cross-correlation of the straight tunnel in the far region DS KF SF ASD ASA ESD ESA DS 1 –0.65 0.11 –0.75 –0.68 –0.72 –0.64 KF 1 0.01 0.96 0.98 0.89 0.99 SF 1 –0.01 –0.01 –0.03 0.01 ASD 1 0.97 0.94 0.96 ASA 1 0.94 0.99 ESD 1 0.89 ESA 1 Table 8. Cross-correlation of the curved tunnel DS KF SF ASD ASA ESD ESA DS 1 0.12 0 0.45 0.84 0.69 0.57 KF 1 –0.03 0.07 –0.17 0.05 0.01 SF 1 0.03 –0.03 0.02 0.07 ASD 1 0.47 0.52 0.67 ASA 1 0.60 0.69 ESD 1 0.47 ESA 1 3.5 Polarization The cross-polarization ratio (XPR) refers to the field received in the vertical (horizontal) copolarization relative to the field transmitted in the vertical polarization (horizontal) and received horizontal polarization (vertical). The XPR is expressed as The XPRs are fitted as a normal distribution (Fig. 12), and Table 9 summarizes the parameters. The smallest XPRs are negative in both scenarios with a probability of less than 20%, indicating that the copolarization configuration outperforms the cross-polarization configuration with a very large chance. The depolarization is the most severe in the far region of the straight tunnel, as the mean XPR and minimum XPR are the smallest compared with the others. Thus, a dual polarization configuration at the Tx and Rx is suggested. Figure 12Open in figure viewerPowerPoint Comparison of the XPRs. Table 9. Extracted parameters for the XPR XPR Straight tunnel Curved tunnel Near Far Mean (dB) 13.65 4.29 14.86 σDS (dB) 10.31 5.14 16.45 3.6 Clustering The clustering procedure is realized by using a K-power-means algorithm clustering algorithm by considering the power, delay, AoA, and EoA. Twenty clusters are generated for both scenarios. The per-cluster parameters are summarized in Table 10. The per-cluster parameters in the far region of the straight tunnel are the smallest compared to the others. Table 10. Per-cluster parameters Straight tunnel Curved tunnel Near Far SF (dB) 3 3.0 1.0 ASD (˚) 2 0.5 0.5 ESD (˚) 3 0.4 4.0 ASA (˚) 1 0.4 2.0 ESA (˚) 1 0.4 0.4 3.7 Validation of the Channel Models With all of the provided parameters in this work, stochastic channels can be generated by using a 3GPP-like channel generator. The QuaDRiGa channel generator 25, which includes all of the features of the 3GPP channel framework and supports the evolution of time-variant channel parameters, is used to realize and validate the channels. The deployment configuration and mobility patterns of the Tx and Rx are exactly the same as those in the RT, as shown in Fig. 13. Two segments are defined on the basis of the different values of dbp of the two scenarios. Examples of the values of PL of the generated channels are shown in Fig. 14. By running QuaDRiGa 105 times, the mean absolute error of PL is 2.5 dB, and the error in the PL coefficient is 0. The corresponding examples of the DS and KF are shown in Fig. 15. The mean absolute errors of the means and STDs are also 0. The validated parameters and the channel generator maintain consistency in the distribution of the large-scale parameters. As a result, the results of this work can be used to practically evaluate link/system-level technologies for similar tunnel scenarios. Figure 13Open in figure viewerPowerPoint Generated mobility patterns for both scenarios: (a) staight tunnel layout and (b) curved tunnel layout. Figure 14Open in figure viewerPowerPoint Generated values of PL using the extracted parameters: (a) PL of the straight tunnel and (b) PL of the curved tunnel. Figure 15Open in figure viewerPowerPoint Generated RMS delay spread and Rician K-factors for both scenarios using the extracted parameters. 4 Conclusion In this study, the channel characteristics are analyzed and modeled for railway tunnel scenarios in the mmWave band. Straight and curved route shapes are defined, and the 3D environment models are available online for free to download. Measurements are conducted inside the tunnel of a Seoul subway line. A 3D ray tracer is calibrated to ensure that the environment model and material parameters are practically close to reality. The transmission loss of the front window is 5 dB, and the dielectric parameters of the considered materials are obtained. The path loss, RMS delay spread, Rician K-factor, 3D angular spreads, and polarizations are extracted from the calibrated RT simulation, and the distributions and correlations are modeled. On the basis of a 3GPP-like channel modeling framework and the QuaDRiGa channel generator, the provided parameters are validated. The work of this paper indicates that the RT simulator can be integrated with channel measurements to analyze more channel characteristics. Moreover, the 3GPP-like framework and QuaDRiGa generator are suitable for describing high-mobility scenarios. With the provided parameters in this work, researchers and engineers can practically realize channels to evaluate the designed communication technology in similar scenarios. Moreover, the main observations from this work are summarized as follows: A break point exists in both scenarios, and dbp of the straight tunnel is larger than that of the curved tunnel. Thus, two sets of parameters should be extracted, and special attention should be paid when implementing a stochastic channel generator. The correlated distances of the key parameters in the far region are larger than those in the near region, meaning that the channel characteristics vary more drastically in the near region. Owing to the drastic variation in the multipath components, the PL slopes in the near region of both scenarios are approximately 0, which makes PL barely vary with the distance. In the far region, the curved tunnel suffers more path loss than that in the straight tunnel. The DS and SF are less correlated with the other parameters in the far region of the straight tunnel, whereas the other parameters are more correlated with each other. In the curved tunnel, the KF and SF are barely correlated with the other parameters. The ASs are strongly correlated in both scenarios. The copolarization configuration outperforms the cross-polarization configuration with more than 80% probability. The depolarization in the far region of the straight tunnel is the most severe compared to those of the others. Thus, a dual polarization configuration at the Tx and Rx is suggested. In future work, different Tx/Rx configurations and multilink channel simulations will be explored. The influence of different configurations on link-level and system-level technologies will be evaluated. Acknowledgements This work was supported by a grant from the Institute for Information & Communications Technology Promotion (IITP) funded by the Korean government (MSIT) (No. 2014-0-00282, Development of 5G Mobile Communication Technologies for Hyper-connected Smart Services). Biographies Danping He received her BS degree from Huazhong University of Science and Technology, Wuhan, China in 2008; her MS degree from the Universite Catholique de Louvain, Belgium and Politecnico di Torino, Piemonte, Italy in 2010; and her PhD degree from Universidad Politécnica de Madrid, Spain in 2014. In 2012, she was a visiting scholar at the Institut national de recherche en informatique et en automatique, France. She worked at Huawei Technologies from 2014 to 2015 as a research engineer, and she is currently conducting postdoctoral research at the State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University. Her current research interests include radio propagation and channel modeling, ray-tracing simulator development, and wireless communication algorithm design. Bo Ai received his MS and PhD degrees from Xidian University, Xian, China in 2002 and 2004, respectively. He graduated with great honors as an Excellent Postdoctoral Research Fellow at Tsinghua University, Beijing, China in 2007. He is now working at Beijing Jiaotong University, China, as a professor and advisor of PhD candidates. He is a deputy director of the State Key Lab of Rail