close

Вход

Забыли?

вход по аккаунту

?

el%3A20000921

код для вставкиСкачать
fore fdT, is fixed. (i) shows a 128-subcarrier system. ro and rN-,
have higher correlation since the OFDM symbol duration is
shorter (128TJ. (ii) shows a 256-subcarrier system. r, and rN-,
have lower correlation. If we use few subcarriers (implying an
even shorter symbol duration), the correlation between r, and rNapproaches 1 (0' = 0.5). The PDF retrogrades to the distribution of a single Rayleigh RV, see (iii). This indicates slow fading
during one OFDM symbol period. Another important finding is
that, with decreasing correlation, the main lobe of PDF curve
moves towards the right, leaving a light tail near the origin .
0.8
x
.-
U)
5 0.6
x
._
._
a
2 0.4
U
0.2
-0
0
1.0
0.5
1.5
2.0
3.0
2.5
1896/11
a
Fig. 1 PDF of sum of two correlated Rayleigh RVs
(02=
JT
where Q(x) = 1 4 2 ~ cr2I2dt,x 2 0, No is the one-sided power
spectral density of the AWGN, and E, is the BPSK symbol
energy.
The theoretical and simulation (using Jakes' channel model)
results of the BER performance are shown in Fig. 2. The tail of
&(a)near the origin plays an important role in eqn. 8. Increasing
the number of subcarriers decreased the correlation between ro
and rN-, . Therefore,fa(a)moves to the right, resulting in a much
smaller P, value. Eqn. 8 can be computed using Monte Carlo
integration. We can see from Fig. 2 that the theoretical results
match very well with the simulation results. It also shows that the
BER performance of an OFDM system is better than that of a single camer system when the OFDM symbol duration is comparable to the length of the fades. Modelling the multiple distortion as
Rayleigh distribution is not applicable here.
Conclusions: The assumption that the channel impulse response
varies in a linear fashion during one OFDM symbol is a good
assumption, since not only does it simplify the analysis, but the
resulting P, formula accurately describes the system performance.
Under this assumption, the PDF derived from the sum of N ( N 2
3) correlated Rayleigh RVs reduces to the sum of two correlated
Rayleigh RVs. The BER results calculated using the derived formula are accurate.
0.5)
0 IEE 2000
I 1 May 2000
Electronics Letters Online No: 20000944
DOI: 10.1049/el:20000944
Lei Wan and V.K. Dubey (School of Electrical and Electronic
Engineering, Nunyang Technological University, SI, Nanyang Avenue,
639798 Republic of Singapore)
(i) U = 0.9037, 128 subcarriers
= 0.6425, 256 subcarriers
--C (iii) single Rayleigh variable
-A- (ii) U
References
0
0
.A.
....A
*!'..a
5
lolo'
'
"
15
I
'
"
20
"
'
'
25
'
"
"
Eb/No,dB
'
30 '
1896/21
"
Fig. 2 BER performance of OFDM over time-variant Rayleigh fading
channels
Data ratef, = 128kbit/s,fd = 100Hz, o2 = 0.5
0 (i) analysis: SC Rayleigh channel
.... .... (ii) simulation: 128 subcarriers
(iii) simulation: 256 subcarriers
A (iv) analysis: 128 subcarriers
(v) analysis: 256 subcarriers
~
CIMINI, L.J.: 'Analysis and simulation of a digital mobile channel
using orthogonal frequency division multiplexing', ZEEE Trans.
Commun., 1985, COM-33, (7), pp. 665-675
JEON, WON GI, CHANG,KYUNG HI, and CHO,YONG soo: 'An
equalization technique for orthogonal frequency-division
multiplexing systems in time-variant multipath channels', ZEEE
Trans. Commun., 1999, 41, (l), pp. 27-32
KIM, YUN HEE: 'Performance analysis of a coded OFDM system in
time-varying multipath Rayleigh fading channels', ZEEE Trans.
Veh. Technol., 1999, 48, (5), pp. 1610-1615
RUSSELL, M.,and STUBER, G L.: 'Interchannel interference analysis of
OFDM in a mobile environment'. Proc. IEEE Veh. Technol.
Conf., Chicago, IL, July 1996, Vol. 2, pp. 820-824
PROAKIS, J.G.:'Digital communications' (McGraw-Hill, New York,
1995), 3rd edn.
JAKES, w.C., Jr. (Ed.): 'Microwave mobile communications' (Wiley,
New York, 1974)
PAPOULIS, A.: 'Probability, random variables, and stochastic
processes' (McGraw-Hill, New York, 1991), 3rd edn.
+
The extra additive noise pk could be approximately modelled as
Gaussian noise, and the variance of pk is [4]
i= 1
\
-
-
Comparison of ISMA and DECT-DCS
channel access policies for indoor wireless
communications systems
R.C. Atkinson and J. Dunlop
Two channel allocation schemes for the indoor environment are
considered: a mobile-controlled scheme representative of DECT's
dynamic channel selection (DCS), and basestation-controlled
inhibit sense multiple access (ISMA). It is shown that the mobilecontrolled scheme is subject to inferior performance in packetbased voice transmission due to the higher degree of front end
clipping which is introduced.
where the envelope correlation is E{ lhll Ih/+il}= (d2)o2fl-:, -1; 1;
X;?) [6], .F( ) is the hypergeometric function, E{ IXkI2}is normaksed
to 1, Xi = J,(2KfDTS . 21, E{lhi12} = 202.
The overall bit error rate for BPSK modulated OFDM can be
written as [5]
Introduction: In this Letter, we assess channel access mechanisms
for indoor 3G mobile communications systems offering packetbased services. One such service in which the channel access mechanism will have a significant impact is voice over internet protocol
(VOIP).
ELECTRONICS LETTERS
20th July2000
Vol. 36
No. 15
1307
At present most indoor coverage is provided by illumination
from external macrocellular base stations. However, as the
number of subscribers continues to grow this approach may not
meet the required capacity since active users of all buildings
within the beamwidth of the BS antenna must share the limited
channels on the basestation. A solution to this problem is the
deployment of low power indoor picocellular basestations, to
provide local coverage.
Operators have begun installation of GSM indoor base stations, and the DECT cordless technology has been established in
Europe for many years. It is widely expected that cellular and
cordless systems will merge in future wireless communications systems and that indoor base stations will be required to support a
wider range of services than their exterior counterparts. The channel access policies employed in such systems must be able to
accommodate the requirements of the services which they support.
lead
frame
talksDurt
interior of a building was simulated, comprising a number of
rooms separated by internal walls. The Motley indoor propagation model [4 - 61 was used to determine the signal power of the
various mobiles at the basestation. The mobiles were u r i f o d y
distributed around the building, changing position periodically.
A collision occurs if two or more mobiles attempt to transmit on
the same channel. In this event, one or more packets may be lost,
depending on whether one mobile was able to capture the channel.
To enable proper comparison between the two techniques, the
GSM speech frame length (20ms) and frame structure (8 :< 5771.1s)
was common to both techniques. Call traffic was generated via a
Poisson process (for equivalent circuit-switched traffic loads in
erlangs). Throughout the duration of a call, voice activity is modelled by on-off periods of talkspurts and silences. During a talkspurt, speech frames are produced with a constant arrival rate.
Each speech frame is decomposed into speech packets and interleaved over a number of slots. The speech frame is inlerleaved
over four slots as used in the GPRS system. In the evaluation of
packet-speech systems two packet level metrics are commonly used;
95th percentile values for overall packet loss probabdity and
packet loss probability at the beginning of talkspurts (packets
lost from the first speech frame of a talkspurt constitute packet
losses at the start of the talkspurt), shown in Fig. 1. The metric
values were extracted from the CDFs of various runs, averaged,
and 95% confidence limits produced.
5r
packet
Fig.
Packet decomposition
4.5 -
4.08
(0
3.5 -
3.0-
0
._
,X 2.5 .-
n
20
2.0-
'
1.5-
5
0
1.0-
0.5 -
0;
10
15
20
traffic, erlangs
I
d
25
30
@iz
Fig. 3 Packet loss middle/end talkspurt
I
I
I
5
10
15
traffic, erlangs
I
20
25
30
-0- ISMA
-+--
DCS
(946/21
Fig. 2 Overall packet loss
-0- ISMA
-+--
DCS
Access policies: The DECT system operates a DCS (dynamic
channel selection) scheme whereby mobiles monitor the channelset and transmit on a channel that they perceive to be unoccupied
(based on measurements of interference). The channel selection is
based on a least interfered channel (LIC) algorithm which is
designed to avoid the simultaneous use of a particular channel by
more than one mobile. The operation of the LIC algorithm is satisfactory in a circuit-switched environment; however this Letter
indicates that performance may not be satisfactory when supporting packet voice transmission.
Inhibit sense multiple access (ISMA) has shown promise in the
indoor environment due to its ability to resolve the hidden terminal problem which occurs when mobile-controlled schemes erroneously perceive that a channel is unoccupied. In contrast to
mobile controlled schemes, the basestation informs each mobile
which channels are vacant, hence reducing the potential of collision. A number of studies [l - 31 have shown ISMA to provide
high throughput values. However, its ability to carry voice traffic
has not been fully examined.
Approach: Simulations have been conducted to compare the performance of the DCS and the ISMA schemes for voice traffic. The
1308
traffic, erlangs
Biz
Fig. 4 Packet loss start talkspurt
-0- ISMA
-+--
DCS
Results: Fig. 2 shows the average overall packet loss metric. It
indicates similar performance of the two schemes, with B M A providing superior performance below 20 erlangs, and DCS providing superior performance above this value.
ELECTRONICS LETTERS
20th July2000
Vol. 36
No. 15
When the packets are decomposed into separate groups (packets at start of talkspurts and packets elsewhere), a clearer picture of
the relative performance of the two schemes emerges. The packets
lost elsewhere dominate the overall packet loss metric due to their
greater numbers relative to packets lost at the start of talkspurts.
Fig. 3 shows the relative performance of the two schemes considering packets lost elsewhere and is remarkably similar to those of
Fig. 2 for both schemes. Fig. 4 shows the packet loss metric for
packets lost at the start of a talkspurt; the ISMA scheme provides
vastly superior performance compared to the DCS scheme.
Packets lost at the start of a talkspurt have a greater impact on
perceived speech quality than those lost elsewhere. This is due to
the inability to interpolate missing packets at the start of a talkspurt from preceding packets.
Conclusion; This Letter has examined two channel allocation
schemes: mobile-controlled @CS scheme) and basestation-controlled (ISMA scheme). It has been shown that the ISMA system
provides superior speech quality than the DCS scheme. Whilst
collisions that occur in circuit-switched operation of the DCS
scheme are tolerable, the number of collisions becomes unacceptably high for packet access techniques. The high number of collisions is the result of repeated access attempts at the start of
talkspurts.
0 IEE 2000
15 May 2000
Electronics Letters Online No: 20000921
DOI: 10.1049/e1:20000921
R.C. Atkinson and J. Dunlop (Department of Electronic and Electrical
Engineering, University of Strathclyde, Glasgow GI I X W , United
Kingdom)
E-mail: robert.atkinson@ieee.org
capacity networks, the propagation conditions are such that the
transmission is often confined to a single room or to a small
number of adjacent rooms [2]. To assess and understand the
advantages of such a move, extensive measurements are conducted
in two different rooms at 37.2GHz and 893MHz. The comparison
concerns radio coverage performance, rapid fading variations,
and data rate limitations.
CW measurement scenario
I
,
/
i
im
I
. \
m
u
u
laboratory table
1
1564/11
References
Fig. 1 Measurement site
ZDUNEK, K.J., UCCI, D.R., and LOCICERO, J.L.: ‘Throughput of
wnpersistent inhibit sense multiple access with capture’, Electron.
Lett., 1989, 25, (l), pp. 30-32
PRASAD, N.R.,and NIJHOF, JAM.: ‘Indoor wireless communications
using slotted non-persistent isma, 1-persistent isma and nonpersistent isma\cd’. Proc. IEEE Veh. Tech. Conf. (VTC-97), 1997,
Vol. 3, pp. 1513-1517
PRASAD, R.,
and
LUI, c.Y.:
‘Throuput analysis of some mobile
packet radio protocols in Rician fading channels’, IEE Proc. I,
1992, 139, (3), pp. 297-302
KEENAN, J.M., and MOTLEY, A.J.: ‘Radio coverage in buildings’,
British Telecom Technol, J., 1990, 8, (l), pp. 19-24
MOTLEY, A.J., and KEENAN, J.M.P.:‘Personal communication radio
coverage in buildings at 900MHz and 1700MHz’, Electron. Lett.,
1988, 24, (12), pp. 763-764
GOLLREITER,R.: ‘Channel models issue 2’. ATDMA deliverable
R2084/ESG/CC3/DS/P/O29/bl,1994, pp. 77-81
-40
r
-45
-50
gE
-55
d -60
4
-65
L
Q
-”
-‘‘?!5
Effect of frequency carrier on indoor
propagation channel
2.0
2:O
2.5
3.0
3.5
2:5
30
3:5
4:O
distance, m
b
4.0
4.5
5.0
5.5
6.0
4:5
5:O
5:;
6.b
Fig. 2 Typical CW measurement results
a Signal obtained at 893MHz
b Signal obtained at 37.2GHz
L. Talbi
Experimental results obtained for narrowband and wideband
indoor radio channels at 37.2GHz and 893MHz, respectively, are
presented. Under the same line-of-sight (LOS) conditions, the
multipath channel characteristics for both frequency bands are
compared in terms of path loss gradient and cumulative
distribution functions (CDF) of the CW measurements,and rootmean-square (RMS) delay spread of the impulse response
measurements.
Introduction: The design of digital communication receivers for
universal digital portable communications is a challenging task.
The receivers must work in an indoor environment which has
multipath propagation and moderate motion [11. These factors
combine to give deep fades, pulse spreading, and hence intersymbo1 interference. Furthermore, if one leaves the UHF band and
wishes to use the higher microwave frequency bands between 20
and 60GHz, as would likely be required for high-speed and high-
ELECTRONICS LETTERS
20th July 2000
Vol. 36
Table 1: Loss gradient and fast fading
Loss gradient
Fast fading
Frequency
Room 2
Rayleigh
Measurements: Measurements were performed using the EHF
experimental system described in [2]. The receiver was mounted on
an automatic tracking and drive system cart guided by a wire
placed on the floor that determined the path. The receiver moved
at a speed of 5 d s and collected CW data at a rate of 2samples/mm
with the transmitter kept at a fmed location. Direct time domain
sounding using a radar-like system of a 200MHz bandwidth was
used to conduct wideband measurements. Only few modifications
No. 15
1309
Документ
Категория
Без категории
Просмотров
1
Размер файла
337 Кб
Теги
3a20000921
1/--страниц
Пожаловаться на содержимое документа