iskandar - presentation
TRANSCRIPT
Study on Channel Characteristic and Its Performance for Wireless Communication Employing Stratospheric Platform
Graduate School of Global Informationand Telecommunication Studies
Waseda University
Iskandar
Supervisor:Prof. Shigeru Shimamoto
Doctor Defense
2
Outline
Research BackgroundBrief Introduction to Stratospheric PlatformResearch MotivationMain Research Content
Part 1Part 2Part 3
ConclusionFuture Work
3
Research BackgroundCompared to wired line, the demand in wireless mobile communication has been increasing exponentially in the last decade because of o Users mobilityo Flexibility
Nowadays, we have two well-established method of delivering information through wireless channelo Terrestrial systemo Satellite system
However, the fundamental problem of wireless communication is thato Multipath propagation problemo How to share the common transmission medium by as many users as
possible with a good quality of service
Researchers in communication community are now continuously solving the problems:o Various new technologies (diversity, channel coding, advanced modulation,
MIMO, etc…)o And at the same time they are looking for another alternative of wireless
delivery method
4
What is the Stratospheric Platform?
In ITU, the Stratospheric Platform (SPF) is called as a High Altitude Platform Station (HAPS) which is defined as:
“a station located on an object at an altitude of20 to 50 km and at a specified, nominal, fixedpoint relative to the earth”
The definition does not mention if the object is piloted or unmanned or how it is powered.
In WRC 2000, ITU has allocated spectrum:o Fixed communication : - 28/31 GHz (mostly in Asia)
- 47/48 GHz (worldwide)o Mobile communication : - 2 GHz
5
Position at the Atmosphere
10km
20km
50km
RainTroposphere
0km
AirplaneCloud
SPF
Stratosphere
Tropopause
Stratopause
6
o Small coverage
o Low propagation delay
o Low power requirement
o Huge numbers of base station for global coverage
o Rayleigh fading channel
Terrestrial Satellite700 - 36.000 km
above the ground
o Global coverage
o Large propagation delay
o High power due to large distance
o Free-space-like channelwith Ricean fading
Comparison among the SystemsStratospheric Platform
o Medium coverage
o Low propagation delay
o Low power like in terrestrial
o Free-space-like channel with Ricean fading
20 km
dα
7
SPF Advantages
Advantages compared with terrestrialo Better propagation in many scenarioso Rapid deploymento Eliminate huge number of existing BTSo Large system capacity
Advantages compared with satelliteo Close range → good link budget and low delayo Lower cost (no launch vehicle)o Rapid and incremental system deploymento Larger overall system capacity
o Environmentally friendly (no launch vehicle or rocket)
8
Major Projects and System Examples
Japanese SPF program → NICT
Korean program → KARI and ETRI
US program → NASA, Aerovironment and Skytower
European program → Helinet and Capanina
Japan Korea USA-HELIOS USA-HALO
USA-Pathfinder USA-Lockheed UK-StratSat ESA
9
Research MotivationOne of the first problems encountered in designing a novel wireless communications system is that the channel characterization and the propagation modeling need to be defined.
There are many researches have been done in channel characterization and modeling for either terrestrial or satellite system.
In contrary, there has not been much reported for the case of SPF.
Therefore motivation in this study are:
o Part 1
Try to evaluate the SPF channel characteristic in semi-urban environment based on experiment.
o Part 2
Evaluate propagation model in low-rises urban environment based on ray-tracing simulation.
o Part 3
Examine its performance and estimate the system capacity based onthe result of the proposed channel model.
Channel characterization and performance evaluation for wireless communication employing stratospheric platform (SPF)
Part 1
11
Objectives
Propose a definition and describe an analysis of wireless channel in SPF communication in a wide range of elevation angles.
Investigate channel parameters such as Rice factor (K) and local mean received power.
Describe channel performance based on the proposed channel parameters for a particular modulation scheme.
12
How to Model the Wireless Channel?
Statistical modelo Based on measurementso Specific for an intended communication system, spectrum allocation
or areao Less computational burden
Site−specific or deterministic modelo Based on theory of electromagnetic waveo Do not rely on measurementso Provide accurate predictiono Complicated mathematical operationo Time consuming
13
Methodology
Multipath power experimento LOS situationo 1.2 and 2.4 GHzo Elevation angles from 100 to 900
o Power level measurement
Data processingo Row data is a power level apply to each elevation angleo Data conversion from power level to amplitude levelo Generation of Cumulative Distribution Function (CDF)o Fading characterization
Channel modeling using best-fit test approach and channel performance evaluationo Rice factor ( K )o BER performance
14
Experimental Setup
900800700100
Remote carriercontrol machine
Balloon control
Stratospheric Platform
15
Data Analysis
N samples of received instantaneous power over 900 elevation angle were first collected
ii rr PP =90
Then other data from other elevation angles (α = 800, 700, 600…) are normalized to RMS value of the data in 900.
90i
i
ir
rFader P
PP
α
=
{ }iF FNF max=∆
Compute cumulative probability by dividing range interval of the fade data into NF power bins of the size ∆F.
Fi : Fade level
16
Measured Instantaneous Received Power
1.2 GHz 2.4 GHz
17
Statistical Property
The envelope statistics of received signal (R) can be described by Ricean distribution in the presence of dominant line of sight component.
0,2
exp)(202
22
2≥⎟
⎠⎞
⎜⎝⎛
⎥⎦
⎤⎢⎣
⎡ +−= RARIARRRp
σσσ
average power of multipath component
Modified Bessel function of the first kind and zeroth
order
Amplitude of LOS component
2
2
2σAK = Distribution of the
envelope Receivedsignal
K
Rayleigh distribution
Gaussian Normal distributian
small
large
K factor is defined as the power ratio of the line of sight (LOS) component to the multipath rayleigh component.
18
Method of Moment (1)
The moments of original Rice distribution can be expressed
⎟⎠⎞
⎜⎝⎛ ++Γ= − KnFenRE Knn ;1;1
2)1
2()2(][ 11
2/2σ
First and second moment and then can be expressed as
⎟⎠⎞
⎜⎝⎛Γ= − KFeRE K ;1;
23)
23(2][ 11
2σ
( ) )1(2
2;1;2)2(2][ 2
22
1122 +=
+=Γ= − K
AKFeRE K σ
σσ
⎟⎠⎞
⎜⎝⎛ −
+
Γ=
2exp
1)2/3(
][][2
KKRE
RE⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛+⎟
⎠⎞
⎜⎝⎛+
22)1( 10
KIKKIK
K factor can be obtained from ratio of the first and second moment
19
Method of Moment (2)
0 10 20 30 40 500.88
0.9
0.92
0.94
0.96
0.98
1
K factor
E[R
]/sqr
t(E[
R2 ])
R (envelope statistic of received signal) was obtained from measurement, so we can fit to the above curve to find measured K factor.
20
Rice factor (K)
10 20 30 40 50 60 70 80 900
5
10
15
20
25
Elevation angle [deg]
K fa
ctor
[dB]
Frequency 1.2 GHzFrequency 2.4 GHz
2.4 GHz
1.41-16.77 dB
1.2 GHz
0.94-18.60 dB
From a measurement, K factor for SPF communication is in the range of 0 – 20 dB.
21
Local Mean Power
1.2 GHz 2.4 GHz Elevation angle K
factor [dB]
Local mean
receivedpower [dBm]
Standard deviation of local mean
received power [dB]
K factor [dB]
Local mean
receivedpower [dBm]
Standard deviation of local mean
received power [dB]
100 0.94 -88.59 5.15 1.41 -89.79 7.61 200 1.51 -84.08 2.65 1.99 -84.80 6.96 300 2.20 -84.38 1.75 2.33 -81.41 5.02 400 4.07 -78.62 3.90 2.66 -78.22 5.06 500 8.85 -74.85 1.46 4.61 -74.31 3.26 600 11.39 -74.50 1.96 6.35 -73.52 2.91 700 13.50 -74.03 2.75 9.21 -73.22 3.64 800 15.23 -69.43 1.31 12.15 -72.28 1.59 900 18.60 -67.31 0.47 16.77 -70.10 0.48
It is found that local mean power is an increasing function of elevation angle.
Standard deviation of local mean power is a decreasing function of elevation angle because the multipath power become smaller in high elevation angle.
22
Performance Evaluation
Bit error probability evaluation is performed under the case of DPSK and DQPSK modulation based on values of measured K factor.
DPSK :⎟⎟⎠
⎞⎜⎜⎝
⎛++ΓΓ−
++Γ+
=K
KK
KP DPSKe 1exp
)1(2)1(
,
DQPSK : ∫ ∑∞ ∞
=
−
−+Γ−++
+=
0 1,
1)12()cos22(1
.)1(21
m
mE
DQPSKe dK
eKPπ
θθπ
∫∞ −
−+++
0 )cos22(1.)(cos.)1( θ
Γθθ d
KemKx
E
Γ−++
Γ−=
)cos22(1)cos22(
θθ
KKE
Eb/N0
Numerical integration by trapezoidal method over 1 million sample was carried out to obtain the bit error probability.
23
Performance under DPSK
Frequency 1.2 GHz Frequency 2.4 GHz
0 5 10 15 20 25 3010
-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
Eb/No [dB]
Bit
erro
r pr
obab
ility
, Pe
K=010 [deg]20 [deg]30 [deg]40 [deg]50 [deg]60 [deg]70 [deg]80 [deg]90 [deg]
0 5 10 15 20 25 3010
-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
Eb/No [dB]
Bit
erro
r pr
obab
ility
, Pe
K=010 [deg]20 [deg]30 [deg]40 [deg]50 [deg]60 [deg]70 [deg]80 [deg]90 [deg]
24
Performance under DQPSK
Frequency 1.2 GHz Frequency 2.4 GHz
0 5 10 15 20 25 3010
-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
Eb/No [dB]
Bit
erro
r pr
obab
ility
, Pe
K=010 [deg]20 [deg]30 [deg]40 [deg]50 [deg]60 [deg]70 [deg]80 [deg]90 [deg]
0 5 10 15 20 25 3010
-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
100
Eb/No [dB]
Bit
erro
r pr
obab
ility
, Pe
K=010 [deg]20 [deg]30 [deg]40 [deg]50 [deg]60 [deg]70 [deg]80 [deg]90 [deg]
25
SPF channel has been characterized in semi-urban environment in the condition of LOS.
In that condition, SPF channel has been found to be the Ricean fading model due to the presence of LOS signal.
It is observed that channel parameters (i.e. K factor and local mean power) are an increasing function of the elevation angle.
An analysis shows that elevation angle higher than 400 yields better channel performance.
Summary of Part 1
Radio propagation evaluation using ray tracing algorithm for wireless communication based on SPF.
Part 2
27
Objective
Evaluate propagation model not only in LOS but also in NLOS environment in the city low-rises urban environment.
Estimate power consumption that is required at the SPF in many situations of different elevation and azimuth angles.
Describe the downlink channel performance of the SPF link for delivering IMT-2000 services.
28
Geographical data survey to find:o Building density and height distributiono Visibility
Based on the survey, we develop building block model as an area for propagation evaluation.
Develop ray-tracing algorithm for various elevation and azimuth angles.
Apply ray-tracing scheme to the building block model.
Evaluate propagation parameters, power requirement and the downlink channel performance.
Methodology
29
Experimental Configuration
00
100
200
300
400
500600
700 800 90000
300
600
900
Side view Top view
The car is equipped by fish-eye lens and moves in various directions about 5 km in the city. The target iso Building densityo Building heighto Visibility
30
Building Block Model
SPF
dm
hb
hwb
ws ws
x
z
ySide view
x
y
z
θ =900
D
70
70
θ =450
θ =600
55
35
35
85
25
Top view 8 buildings block model.
Three different azimuth: 900, 600, and 450.
Elevation angle varies from 50 to 900 in a step of 10.
The buildings in the model are assumed to have:o height = 20 mo width = 25 mo vary in lengtho street width = 35 m and is
assumed to be equal in the model
o MS height = 1.5 m
31
Ray launching method.Based on GO and GTD.We employ 9 categories of rays involved in the simulation.Each ray undergoes up to 4 bounces is considered in the simulation.
Rays outside the above category will be terminated from the simulation environment.
Ray Tracing Scheme
single building reflection
double building to street reflection
singlediffraction
single streetreflection
double building to building reflection
multipath ray
direct ray
diffractionand reflection
32
1. Platform height (h) : 20 km2. MS height (hm) : 1.5 m3. Building height (hb) : 20 m4. Frequency : 2 GHz5. Street width (ws) : 35 m6. MS position (dm) : ws /27. Azimuth angle (θ ) : 900, 600, 450
8. εr Building : 39. εr Street : 1510. σ Building : 0.005 W-1m-1
11. σ Street : 7 W-1m-1
10
Simulation parameters
33
Electric field of the ray arriving at the MS is calculated using the following formulas, E0 is the transmit electric field at the transmitter and k is wave number (2π/λ).
deEE
dkj
LOS
−
= 01. Direct ray
2. Reflected ray21
)(
0
21
.ss
eREEsskj
R +=
+−
)(cos)sin(
)(cos)sin(2
2
βεβ
βεβ
−+
−−=
r
rR
3. Diffracted ray )(
3
3
3
0 3
)()(. sskj
D esss
sDsEE +−
+= γ
γγ
πγ
sin2cos1
21)( +
−=k
D
Analytical Model (1)
34
Respective rays for each ray category were added at the MS and expressed as :
∑=
=n
jji EE
1
Ei : total electric field for ith categories of rayEj : electric field for jth ray
∑=
=M
iiTot EE
1
The total electric field contribution consists of vector summation of M ray categories and can be expressed by :
⎟⎟⎠
⎞⎜⎜⎝
⎛=
04log20
EE
L tot
πλ
Finally, the total path loss formulation is :
Analytical Model (2)
35
Azimuth 900 Azimuth 600
Propagation Loss (1)
36
Azimuth 450 Very good agreement between ray tracing and Physical statistical model for the scenario of azimuth 900.In LOS, the result obtain by ray tracing always about 3 dB lower than that by Physical statistical model for all scenarios.Good agreement between two model for NLOS situation middle low elevation angle.In NLOS very low elevation angle (for azimuth 600 and 450), propagation loss calculated by ray tracing is smaller than that calculated by Physical statistical model.
Propagation Loss (2)
37
Required transmitted power at the SPF for IMT-2000 application is calculated based on the following expression and specification.
Parameters SpecificationFrequency [Gz] 2Information Rate [kbps] 8, 32, 64, 384, 2000SPF Antenna Gain [dBi] 30MS Antenna Gain [dBi] 0Blotzmann’s constant [J/K] 1.38 x 10-23
Temperature’s Chamber [K] 290Link Margin [dB] 15.4Cable, Connector, and Other Loss [dB] 2Eb/N0 [dB] Max 7.9
Lb
rtTb
MLLTkRGGP
NE
000 ),( θα=
IMT-2000 Specification
38
Average propagation path loss [dB] versus elevation angle
Region 1 Region 2 Region 3≥ 45 45 > α ≥ 15 15 > α ≥ 5
90 121.6 142.3 172.860 121.1 138.3 152.645 121.2 135.6 151.4
Elevation angle [deg]Azimuth [deg]
Region 1
Required Transmit Power (1)
39
Region 2 Region 3
In region 1 (high elevation angle), the required transmit power by SPF is almost similar for all scenarios. This means required power by SPF is not sensitive to the azimuth angle.However in Region 2 and 3, the required transmit power is a function of azimuth.The worst scenario is observed in Region 3 for 900 azimuth angle. Such high power requirement may could not be implemented in SPF system.
Required Transmit Power (2)
40
We have demonstrated the prediction of propagation loss in a low-risesurban environment for mobile communication using SPF by means ofray tracing algorithm.
The comparison with physical-statistical model has been performed for verification and the result is in a good agreement in some cases.
Estimations of required transmitted power for IMT-2000 application based on SPF have also been evaluated.
The results clearly show a critical limitations of mobile communication IMT-2000 by using the concept of SPF.
Summary of Part 2
CDMA capacity analysis for multibeamand multiple SPF communication
Part 3
42
Objectives
Analyze the interference mechanism in a multibeam and multiple SPF system.Evaluate CDMA system based on the proposed channel model.Describe analysis under fading, shadowing and power control imperfection.Demonstrate the SPF system capacity.
43
Methodology
Develop multibeam and multiple SPF model.
Perform an analysis of interference mechanism
Evaluate system capacity in terms of outage probability
44
Interference Mechanism in Terrestrial
Usually fourth power law of the distance is assumed in terrestrial system due to multipath.
( ) 4−≈ dI
Interfering cell
Reference cell
45
Reference SPF
Interference Mechanism in SPFAdjacent SPF
BS 1 BS 2
( ) 2−≈ dI
Because of LOS, square power law of the distance is assumed in SPF system, higher interference would be produced.
Reference cell Interfering cell Interfering cell
46
Proposed Model
reference cell
SPF
user (i,j)
ijq0ijq
coverage
ijl0ijl
Single SPF model
reference cell coverage ofreference SPF
overlapped region between two SPFs
reference SPF adjacent SPF
coverage ofadjacent SPF
ijkqkij
0q
ijkl¢ijkl
user (i,j,k)
Multiple SPF model
Overlapped region in multiple SPF model is a region outside the coverage but still seen by the reference platform.
47
System capacity and Outage Probability
The transmission quality for CDMA system is describe in terms of Eb /N0
η+=
IC
RW
NEb
0
W : channel bandwidthR : single user information bit rateC : received carrier powerη : AWGN power
Outage probability is defined as the probability of failing to achieve the required (Eb /N0 )req
⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
⎟⎟⎠
⎞⎜⎜⎝
⎛≤=
req
bbout N
ENEP
00
Pr
Interference
48
Reverse Link Interference Analysis
Even though power control must be employed in CDMA system because all users are contending the same bandwidth at the same time.However it is rather impractical to assume that there is perfect power control.Thus Eb/N0 can be expressed as
η
δ
++=
erra
b
IIeP
RW
NE kji
intint
0
0
000
P0 : nominal received power with ideal power controld : zero mean Gaussian random variable to model power control
imperfection with standard deviation sdIintra : Interference originated from users within the reference cell Iinter : Interference originated from users outside the reference cell
49
Interference from users within the reference cell is expressed as
∑−
==
1
10int
N
ira
iePI δυ
υ : voice activity factor
δi : Gaussian random variable of the received power of the ith user
N : number of user per cell including the user of interest
Imperfect power control
Intra Cell Interference
50
Interference from users within the other cell is expressed as :
M : number of SPFJ : number of cell for each paltform
∑∑∑∑∑= = == =
+=M
k
J
j
N
iijkijk
J
j
N
iijer PePI ij
2 1 1
220
1 1
20int εβυβυ δ
2ijkε
2ijkβ : power discrimination due to spot beam antenna radiation pattern
: power control factor for users in the overlapped region
)()(
002
ijkj
kijjijk G
Gθθ
β = θ : boresight angle relative to the reference spotbeam
102 100
ijkijk
kij
ijkijk l
l ξξµ
ε−′
⎟⎟⎠
⎞⎜⎜⎝
⎛ ′=
ijkl ′ : distance from the users to their own serving platformijkl : distance from the users to reference platform
ijkξ ′ : random variable modeling the shadowing effectcorresponding to these two paths
ijkξ
Inter Cell Interference
51
Outage Probability
Outage probability of reverse link can therefore be expressed as :
⎭⎬⎫
⎩⎨⎧
≥+=
⎪⎪⎭
⎪⎪⎬
⎫
⎪⎪⎩
⎪⎪⎨
⎧
⎟⎟⎠
⎞⎜⎜⎝
⎛≤
++=
ς
η
δ
0
int
0
int
0
00
int
0
int
Pr
Pr000
PI
PI
NE
PPI
PI
eRWP
erra
req
b
erraout
kji
where
⎥⎥⎦
⎤
⎢⎢⎣
⎡−=
00 /1
)/(
000
ης
δ
breqb ENEe
RW kji
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛ −=
2221
I
Iout erfcP
σµςδ
Therefore
where µI and is the mean and variance of Gaussian distribution of (I /P0)2Iσ
52
Simulation Parameters
5 MHz20 km200 km100 and 200
3/820 dB0.3
1 dB0.9 – 18.6 dB
Channel bandwidth (W)SPF heightDistance between SPFMin. elevation angleVoice activity factor (ν)Eb/h0Shadowing probability (A)Power control error (PCE) for unshadowed users (sus)K factor
ValueParameters
53
Capacity of Single Platform ModelVoice, R = 12.2 kbps (Eb/N0 )req= 5.0 dB
54
Capacity of Multiple Platform ModelVoice, R = 12.2 kbps (Eb/N0 )req = 5.0 dB
o Compared with the result obtained for single SPF model, if perfect power control can be achieved, the number of users supported at Pout = 10-2 would reduce by at least 14% for speech services.
55
Summary of Part 3
It is found that because of the power control imperfection, the system capacity in SPF CDMA system is significantly decreased.
In multiple SPF model, multiple access interference produced by the users within an overlapped region is a nontrivial reduction of the system capacity.
Therefore, the capacity reduction caused by these users has to be compensated.
One solution is to increase the minimum elevation angle defined for each platform’s coverage.
For the model we consider in this work, with the setting of minimum elevation angle is 200, the system capacity can be improved so as nearly as the capacity brought by a single SPF model.
56
Conclusion
We have evaluated channel characteristic and propagation model for SPF communication both in semi-urban and in low-rises urban environment.
In SPF communication, we found that Ricean fading channel is a proper model for the SPF because of dominant LOS situation in many places in the coverage.
K factor is observed to be between 0 and 20 dB depending on the elevation and azimuth angle.
Propagation loss is found much lower than that in satellite system or terrestrial system except for very low elevation angle such as below 100
in the area of low-rises urban environment.
SPF downlink channel performs better and sufficient to support IMT-2000 services if elevation angle can be made higher than 400.
CDMA system capacity in the SPF communication is found to be a bit higher than that in terrestrial system.
57
Future Work
Several issues that have not been considered in this study and therefore need to be further investigated.
o K factor in Part 1 is estimated using the method that ignores thenoise, however in real implementation it needs to include the noise in estimating K factor.
o Wideband channel modeling such as power delay profile in stratospheric platform communication has not been investigated.
o Inter-platform communication link.
o Integrated network among terrestrial, satellite and stratospheric platform.