Graph-based stochastic model for high-speed
railway cutting scenarios
ISSN 1751-8725
Received on 9th December 2014
Revised on 29th May 2015
Accepted on 7th July 2015
doi: 10.1049/iet-map.2014.0827
www.ietdl.org
Tao Zhou
1
✉
, Cheng Tao
1,3
, Sana Salous
2
, Zhenhui Tan
1
, Liu Liu
1
, Li Tian
4
1
Institute of Broadband Wireless Mobile Communications, Beijing Jiaotong University, Beijing 100044, People’s Republic of China
2
School of Engineering and Computing Sciences, Durham University, Durham DH1 3LE, UK
3
National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, People’s Republic of China
4
College of Electronics and Information Engineering, Tongji University, Shanghai 200092, People’s Republic of China
✉ E-mail: 11111037@bjtu.edu.cn
Abstract: This study proposes a stochastic channel model based on the propagation graph theory for high-speed railway
(HSR) cutting scenarios. Single-input single-output wideband measurements are conducted under a cutting for acquiring
realistic channel data. The propagation graph covering line of sight, single bounced and double bounced conditions is
used to describe the measured propagation scenario and generate the virtual channel impulse responses. The graph
model i s validated focusing on small-scale characteristics such as Ricean K-factor, de lay and frequency dispersions,
and a close agreement is achieved between model and measurement results. In particular, the verifie d model is
extended to the multiple-input multipl e-output case for simulation analysis of MIMO performance of the measured
environment, concentrating on the spatial correlation and the ergodic capacity. The proposed model can facilitate t he
reliable simulation and evaluation of MIMO systems in the HSR cutting scenario.
1 Introduction
The fast and tremendous development of high-speed railway (HSR)
has been attracted world-wide attention in recent years. Conventional
train control communications systems have been able to guarantee
the secure operation of HSR, whereas future HSR broadband
communication systems aim at providing high quality and high
data rate services for passengers. Accurate characterisation of the
radio channel is vital to the research and development of a mobile
communication system as well as design and evaluation of some
advanced transmission technologies, such as multiple-input
multiple-output (MIMO), cooperative relay and smart antennas.
Different from traditional rural and suburban environments, the
HSR environment has the unique propagation characteristic
because of its special structures like cutting, viaduct and tunnel
[1]. Thus, realistic channel measurements are inevitable to be
carried out in these typical HSR scenarios, and corresponding
reliable channel models should be developed for facilitating
researches of wireless communications on HSR.
Thus far, a number of works have concentrated on the channel
models in the HSR scenarios. For the empirical models, path loss
(PL), delay and Doppler characteristics in a HSR tunnel have been
measured and analysed in [2]. WINNER II model [3] has given
the results of PL, shadow fading, and other small-scale parameters
related to K-factor, delay spread as well as angle spread for HSR
conditions. He et al. [4–8] have carried out a series of narrowband
measurements in the HSR viaduct and cutting scenarios at
930 MHz. Empirical PL models and partial small-scale fading
characteristics covering viaduct height and cutting width were
originally presented. Based on the wideband data collected under a
viaduct, a position-based channel model at 2.35 GHz has been
especially proposed in [9]. Regarding the theoretical models, the
ray-tracing deterministic approach to wave propagation modelling
for HSR outdoor and tunnel environments was presented in [10,
11]. To numerically analyse the space-time-frequency correlation
characteristics, some researchers [12–14] have developed the
geometry-based stochastic model (GBSM) for HSR MIMO
channels, which are based on the assumption that the rail is
surrounded by scatterers randomly distributed on a regular shape
such as a ring, a sphere or an ellipse.
In fact, there are considerable difficulties to carry out the field
measurements in the HSR scenarios, for example, strict limitation
of frequency bands, extremely lower measurement efficiency and
costly reservation for the test train, which leads to the less research
of the empirical HSR channel models. On the other hand, existing
theoretical HSR models such as ray-tracing models and
regular-shaped GBSMs have the drawbacks of high simulation
complexity and unrealistic scatterers distribution, respectively.
Recently, a simple but realistic channel modelling approach
according to the propagation graph theory, applied as a
simulation-based method for predicting channel characteristics in
indoor scenarios [15–17], has been employed in HSR
environments [18, 19]. Unlike the GBSM, graph modelling is a
frequency domain method. Modelling the propagation channel in
frequency domain is simpler than that in time domain. It can
easily simulate the realistic propagation scenario to generate the
virtual time-varying wideband channel impulse response (CIRs)
and can be used to extract the channel characteristics in time,
frequency and spatial domains. However, the present graph-based
HSR models lack the validation of experimental data. Moreover,
we have not yet found research results of propagation graph
modelling for the cutting scenario.
The primary contribution of this paper is to develop a stochastic
channel model using the propagation graph theory for the special
cutting scenario on HSR. The objective of the model is for studies
of HSR communication systems based on synthetic realisations of
the CIRs. In particular, the model is verified according to
single-input single-output (SISO) channel measurement results in a
specific cutting scenario. Furthermore, the proven graph-based
stochastic model is extended to the MIMO case to analyse the
MIMO performance for the measured propagation environment.
The remainder of this paper is outlined as follows. Section 2
describes the measurement campaign. In Section 3, a graph-based
stochastic model is established. Section 4 provides the model
validation. MIMO performance analysis is presented in Section 5.
Finally, conclusions are drawn in Section 6.
IET Microwaves, Antennas & Propagation
Research Article
IET Microw. Antennas Propag., 2015, Vol. 9, Iss. 15, pp. 1691–1697
1691
&
The Institution of Engineering and Technology 2015