154 ◾ Advances in Communications-Based Train Control Systems
just the hando latency in this chapter, should be as small as possible and the
error-free period should be as large as possible. ese two QoS indexes, together
with the FER of the hando signalings, will be used to evaluate the QoS perfor-
mance in CBTC communication system in this chapter.
8.2.2 Features of Handoff in CBTC Communication System
e APs in the CBTC system in urban rail transit system are located with a small
distance of several hundred meters between each other. As the APs are arranged
along the rail in xed positions, the mobile station in the train would switch from
one AP to another predetermined AP. erefore, the hando algorithm can be
simplied and optimized to avoid ping-pong hando and inappropriate hando.
Another distinct feature of CBTC system is that the train can acquire high-
resolution train location determination [19] by speed sensors and Doppler radar
together with the transponders, which are aside the rail and provide the absolute
reference position information for the train. e train transmits the location infor-
mation to the wayside device by the wireless communication system. According to
the requirements of the CBTC system, the location resolution would be in the limit
of 5–10m. Because the location of the train can be got in real time, we propose
the location-based hando algorithm in [5] to reduce the happening of ping-pong
hando caused by the random eects of the wireless channel.
In CBTC systems, it is important to maintain communication link availability
in order to guarantee train operation safety and eciency. In this chapter, on the
basis of the location-based hando scheme, we present a “make-with-break” hand-
o scheme to provide high link availability in CBTC systems by applying MIMO
multiplexing and STBC interference cancelation algorithms.
8.3 Proposed MAHO Scheme
Two or four antennas can be congured in each AP and the MS on the train
according to [12]. We suppose that the MAHO scheme with 2×2 antenna con-
guration be applied in the chapter, and the method can be easily expanded to the
case of four antennas as dened in 802.11n standard [12].
In the proposed MAHO scheme, the hando procedure is triggered by the loca-
tion of the train. When the hando is triggered, the station would keep connection
with old AP by one antenna on the train and set up a connection with the new AP
by another antenna at the same time. e interference at the MS between two APs
is canceled by MIMO signal detection and interference canceling algorithms. In
CBTC environments, the wireless networks are planned to assure that even in the
border of coverage area the received power exceeds the minimum receiver sensitiv-
ity by more than 10dB [5], which guarantees the reliability of the data transmis-
sion and make the signal detection algorithm feasible. With the strong enough
Novel Handoff Scheme with MIMO ◾ 155
SNR, the MIMO signal detection and interference cancelation algorithms can be
reliably applied and the “make-with-break” scheme is realized without multiple
radio transceivers.
8.3.1 MIMO Transmission in the HandoffProcedure
in the MAHO Scheme
When the train reaches the hando point, the MS will initiate the hando proce-
dure by transmitting the probe request packet. e APs receiving the request will
send probe response to the MS. In our proposed MAHO scheme, the candidate AP
is predetermined along the rail, and the MS can decode the probe response signal
from the candidate AP. e MS can learn whether the candidate works well by
receiving the periodic Beacon frame from the candidate AP.
According to the regulations in the standards, the probe response packets have
to be transmitted in sequence and the delay in this stage is the largest part of the
hando latency. In the MAHO scheme, hando signaling and uplink data are
transmitted in MIMO multiplexing mode in the uplink, whereas in the downlink
the multiuser STBC transmission modes are used to reduce the latency.
e physical layer-related topics are rst introduced in Section 8.3.1.1, then the
related synchronization schemes are given.
8.3.1.1 Physical Layer Processing
As shown in Figure8.2a, if at the time instant of sending the probe request there
is an uplink train control-related information to send, the MS sends the uplink
(UL) data together with probe request, where the hando signaling and the data
packet are transmitted by dierent antennas, respectively. And the decoding of the
MIMO multiplexing packets can apply the well-known V-BLAST algorithm[15].
In such a way, the data communication will not be interrupted by the hando
procedure. When the serving AP receives the UL data multiplexed with the
probe request packet, it returns the ACK packet sending at one antenna and the
proberesponse at another antenna, which are transmitted in MIMO multiplex-
ing mode. e packets are transmitted after the period of short interframe space
(SIFS), which is required before sending the ACK packet as dened in 802.11
standard.
However, the candidate AP senses the channel status and returns a probe
response after the period of distributed interframe space (DIFS), plus the period
that the contention window (CW) timer expires if the channel is idle, where the
transmission mode is the MIMO diversity by using Alamouti STBC [20].
e MIMO transmission mode of the transmission is indicated in high through-
put signal eld part of the frame header in 802.11 packets, where the number of spacial
streams, N
SS
, and space time streams, N
STS
, are included. For the MIMO multiplexing
mode, N
SS
= 2 and N
STS
= 2, and for the MIMO STBCs, N
SS
= 1 and N
STS
= 2.
156 ◾ Advances in Communications-Based Train Control Systems
If there are no uplink train control-related data to send at the time instant of
sending the probe request as shown in Figure8.2b, concurrent transmission of the
probe response packets in the downlink is implemented to reduce the latency in the
scanning stage.
After the transmission of probe response packets, the MS will transmit other
hando signaling packets together with UL data packets in the MIMO multi-
plexing mode if there are UL packets to send at that same time. e serving AP
and the candidate AP will return the ACK packet for the UL data packet and the
hando signaling packets simultaneously as shown in Figure8.2b, which is the
same as the transmission of probe response packets. Otherwise, if there are no
data packets to send at the time of transmitting the hando signaling packets,
the signaling packets are transmitted at both antennas in the MIMO diversity
mode.
In our transmission scheme, both APs use Alamouti code and the pack-
ets are transmitted to the MS simultaneously, which will cause interference
between the two APs at the receiver. However, the interference can be can-
celed with simple linear algorithms [21,22]. Let
c
=
()
12
cc
T
and
s
=
()
12
ss
T
be the
codewords transmitted by the rst and second APs, respectively.
yyy
jjj
T
=
()
12
is the received signal vector at the antenna j, where the components of
y
j
are
the signals received at the antenna j over two consecutive symbol periods.
erefore, we have
MS
Probe request
Probe response
Probe response
CW
UL data
ACK
G
1
= SIFS
G
3
= DIFS
Candidate AP
(a)
(b)
Serving AP
G
1
G
3
MS
Probe request
Probe response
CW
UL data
Authentication request
ACK
G
1
= SIFS
G
3
= DIFS
Authentication response
Probe response
Candidate AP
Serving AP
G
1
G
1
G
3
Figure8.2 Proposed handoff scheme. (a) Multiplexing of UL data with probe
request. (b) Synchronized downlink transmission.
Novel Handoff Scheme with MIMO ◾ 157
y
j
j
j
jj
jj
y
y
hh
hh
c
c
g
=
=
−
⋅
+
1,
2,
1, 2,
2, 1,
1
2
1
,, 2,
2, 1,
1
2
1
2
jj
jj
g
gg
s
s
n
n
−
⋅
+
(8.1)
where:
h
i,j
is the channel coecient from the antenna i of AP
1
to the j receive antenna
of MS
g
i,j
is the channel coecient from the antenna i of AP
2
to the j receive antenna
of MS
e noise samples n
1
, n
2
are independent samples of a zero-mean complex
Gaussian random variable
And we dene the channel matrix as
HG
j
jj
jj
j
jj
jj
hh
hh
gg
gg
=
−
=
−
1, 2,
2, 1,
1, 2,
2, 1,
,
A simple array signal processing results in the following signal:
Gy
G
Gy
G
GG
G
HH
H
22
2
2
11
1
2
21
2
2
21
2
2
1
2
||||
||||
HH
HH
c
c
−
=−
+
′
′
n
n
1
2
(8.2)
In the above equation, the signal of the AP
2
has been canceled, and the decoding
of the signal of AP
1
is straightforward. is array processing technique provides a
diversity of degree 2 for each AP.
In the case that the train and APs have four antennas each, multiplexing of
hando signaling with data packets can be implemented in the uplink, where some
antennas are used to transmit signaling and the others are used for the transmission
of data packets. e decoding of the multiplexed packets can apply the V-BLAST
algorithm [15] as well. In the downlink that APs transmit packets to the train,
because orthogonal STBC for complex signal transmission does not exist [23],
aquasiorthogonal STBC (QOSTBC) transmission for the AP in the downlink can
be implemented, where the MS on the train can cancel the interference of other
APs through array processing [24].
In the proposed MAHO scheme, the hando signaling and train control-related
data packets are transmitted at dierent antennas to reduce the hando latency.
However, the concurrent transmission of data and signaling packet will cause
more packet loss compared with the transmission of signaling or data packets by
158 ◾ Advances in Communications-Based Train Control Systems
transmit diversity mode, even with the interference cancelation algorithm employed.
erequired data rate in CBTC communication system is less than 1Mbps [25];
therefore, the modulation and coding scheme (MCS) of 6.5Mbps in 802.11n is
commonly applied, which is also the most robust transmission scheme in 802.11n
system with orthogonal frequency division multiplexing (OFDM) modulation. e
moving speed of the train in the subway is not more than 80km/h in general, so
the maximum Doppler frequency shift of the 2.4GHz WLAN system is at most
200Hz. Although the subcarrier bandwidth in 802.11g system is 312.5KHz, the
normalized frequency oset in the OFDM system is about
6.4 10
4
×
−
, which has
nearly no impacts on the BER performance of the system [26]. In [27], it is proposed
that when the zero forcing algorithm is adopted at the receiver, the BER is approxi-
mated as
P
b
≈
−
αγ
β
(8.3)
where:
γ is the average received SNR
α and β are related with the achieved channel degrees of freedom (DoF)
With 6.5 Mbps MCS in CBTC communication system, where binary phase shift
keying (BPSK) modulation with coding rate of 1/2 is adopted, the DoF is 2, and
α is 0.0272 and β is 7.528 when MIMO multiplexing mode is used with two
streams multiplexed, which is the case of the uplink transmission mode in our
scheme. In the scheme of concurrent transmission in the downlink by two APs,
2×2 Alamouti STBC codes are used at both APs and two antennas at the receiver
at the MS and the DoF is 4. e corresponding coecients of α and β are 0.0031
and 9.853, respectively. When 2×2 Alamouti STBC codes are adopted in single
user’s transmission, which is the case of normal data transmission with trans-
mit diversity and receiver diversity, the values are 0.0008 and 9.540, respectively.
ese give an upper bound of the BER of the packet transmission in the proposed
hando MAHO scheme.
en the BER results are used to calculate the FER P
fr
:
PP
L
fr b
fr
=− −1(
1)
(8.4)
where:
P
b
is the BER
e curves of FER P
fr
versus
EN
s
/
0
, with dierent data rates and channel DoF are
given in Figure8.3 with the frame length of 150 bytes.
To apply the interference cancelation algorithm, the packets from dierent APs
have to arrive at the MS at the same time instant and synchronized at the receiver.
e synchronization methods are introduced in Section 8.3.1.2.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.