Propagation Channel Characterization, Parameter Estimation, and Modeling for Wireless Communications E-Book
Xuefeng Yin
123,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Wissenschaft und neue Technologien
- Serie: Wiley - IEEE
- Sprache: Englisch
A comprehensive reference giving a thorough explanation of propagation mechanisms, channel characteristics results, measurement approaches and the modelling of channels
Thoroughly covering channel characteristics and parameters, this book provides the knowledge needed to design various wireless systems, such as cellular communication systems, RFID and ad hoc wireless communication systems. It gives a detailed introduction to aspects of channels before presenting the novel estimation and modelling techniques which can be used to achieve accurate models.
To systematically guide readers through the topic, the book is organised in three distinct parts. The first part covers the fundamentals of the characterization of propagation channels, including the conventional single-input single-output (SISO) propagation channel characterization as well as its extension to multiple-input multiple-output (MIMO) cases. Part two focuses on channel measurements and channel data post-processing. Wideband channel measurements are introduced, including the equipment, technology and advantages and disadvantages of different data acquisition schemes. The channel parameter estimation methods are then presented, which include conventional spectral-based estimation, the specular-path-model based high-resolution method, and the newly derived power spectrum estimation methods. Measurement results are used to compare the performance of the different estimation methods. The third part gives a complete introduction to different modelling approaches. Among them, both scattering theoretical channel modelling and measurement-based channel modelling approaches are detailed. This part also approaches how to utilize these two modelling approaches to investigate wireless channels for conventional cellular systems and some new emerging communication systems. This three-part approach means the book caters for the requirements of the audiences at different levels, including readers needing introductory knowledge, engineers who are looking for more advanced understanding, and expert researchers in wireless system design as a reference.
- Presents technical explanations, illustrated with examples of the theory in practice
- Discusses results applied to 4G communication systems and other emerging communication systems, such as relay, CoMP, and vehicle-to-vehicle rapid time-variant channels
- Can be used as comprehensive tutorial for students or a complete reference for engineers in industry
- Includes selected illustrations in color
- Program downloads available for readers
- Companion website with program downloads for readers and presentation slides and solution manual for instructors
Essential reading for Graduate students and researchers interested in the characteristics of propagation channel, or who work in areas related to physical layer architectures, air interfaces, navigation, and wireless sensing
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 657
Veröffentlichungsjahr: 2016
Ähnliche
Table of Contents
Cover
Title Page
Copyright
Preface
References
List of Acronyms and Symbols
List of Acronyms
List of Symbols
Chapter 1: Introduction
1.1 Book Objective
1.2 The Historical Context
1.3 Book Outline
Bibliography
Chapter 2: Characterization of Propagation Channels
2.1 Three Phenomena in Wireless Channels
2.2 Path Loss and Shadowing
2.3 Multipath Fading
2.4 Stochastic Characterization of Multipath Fading
2.5 Duality of Multipath Fading
2.6 WSSUS Assumption of Multipath Fading
2.7 A Review of Propagation Channel Modeling
Bibliography
Chapter 3: Generic Channel Models
3.1 Channel Spread Function
3.2 Specular-path Model
3.3 Dispersive-path Model
3.4 Time-evolution Model
3.5 Power Spectral Density Model
3.6 Model for Keyhole Channel
Bibliography
Chapter 4: Geometry-based Stochastic Channel Modeling
4.1 General Modeling Procedure
4.2 Regular-shaped Geometry-based Stochastic Models
4.3 Irregular-shaped Geometry-based Stochastic Models
4.4 Simulation Models
4.5 Simulation Models for Non-isotropic Scattering Narrowband SISO V2V Rayleigh Fading Channels
Bibliography
Chapter 5: Channel Measurements
5.1 Channel-sounding Equipment/System
5.2 Post-processing of Measurement Data
5.3 Impact of Phase Noise and Possible Solutions
5.4 Directional Radiation Patterns
5.5 Switching-mode Selection
Bibliography
Chapter 6: Deterministic Channel-parameter Estimation
6.1 Bartlett Beamformer
6.2 The MUSIC Algorithm
6.3 The ESPRIT and Propagator Methods
6.4 Maximum-likelihood Method
6.5 The SAGE Algorithm
6.6 A Brief Introduction to the RiMAX Algorithm
6.7 Evidence-framework-based Algorithms
6.8 Extended Kalman-filter-based Tracking Algorithm
6.9 Particle-filter-based Tracking Algorithm
Bibliography
Chapter 7: Statistical Channel-parameter Estimation
7.1 A Brief Review of Dispersive Parameter Estimators
7.2 Dispersive Component Estimation Algorithms
7.3 PSD-based Dispersive Component Estimation
7.4 Bidirection-delay-Doppler Frequency PSD Estimation
Bibliography
Chapter 8: Measurement-based Statistical Channel Modeling
8.1 General Modeling Procedures
8.2 Clustering Algorithm based on Specular-path Models
8.3 Data Segment-length Selection
8.4 Relay and CoMP Channel Modeling
Bibliography
Chapter 9: In Practice: Channel Modeling for Modern Communication Systems
9.1 Scenarios for V2V and Cooperative Communications
9.2 Channel Characteristics
9.3 Scattering Theoretical Channel Models for Conventional Cellular MIMO Systems
9.4 Scattering Theoretical Channel Models for V2V Systems
9.5 Scattering Theoretical Channel Models for Cooperative MIMO Systems
Bibliography
Appendix A
A.1 Influence of Neglecting Doppler Shift within the Sensing Periods
A.2 Simplification of the Noise Component in an Objective Function
A.3 Derivations of Equations (7.6)–(7.8)
A.4 Derivation of Eqs (4.20a) and (4.20b)
A.5 Derivation of the CF
A.6 Probability Density Functions
A.7 Computation of the Gerschgorin Radii
A.8 Derivations for Chapter 9
Bibliography
Index
End User License Agreement
Pages
xi
xii
xiii
xiv
xv
xvi
xvii
xviii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
379
380
381
382
383
384
385
386
387
388
Guide
Cover
Table of Contents
Preface
Begin Reading
List of Illustrations
Chapter 2: Characterization of Propagation Channels
Figure 2.1 A wireless communication system consisting of three parts
Figure 2.2 Three phenomena in wireless channels
Figure 2.3 Fourier relationship between the system functions of non-directional channels
Figure 2.4 Fourier relationship between the system functions of directional channels
Figure 2.5 A diagram of a signal system represented for a wireless system with multipath fading channels
Figure 2.6 Typical wireless communication scenarios: (a) fixed Tx and Rx; (b) moving Rx; (c) multiple antennas
Figure 2.7 Fourier relationship between the correlation functions of the non-directional system functions
Figure 2.8 Fourier relationship between the correlation functions of the directional system functions
Figure 2.9 Duality relationship between selectivity and dispersion of multipath fading
Figure 2.10 Summarization of duality of multipath fading
Figure 2.11 Fourier relationship between the correlation functions of the non-directional system functions for WSSUS channels
Figure 2.12 Fourier relationship between the correlation functions of the directional system functions for WSSUS channels
Figure 2.13 Classification of MIMO channel models
Figure 2.14 The sketches of three important GBSMs: (a) one-ring model; (b) two-ring model; (c) elliptical model
Figure 2.15 Tradeoff in MIMO channel models. DM, deterministic models; SM, stochastic models; PM, physical models; GBM, geometry-based models; NGM, non-geometrical models; AM, analytical models; CBM, correlation-based models; PMM, propagation-motivated models
Figure 2.16 A typical V2V environment and the corresponding geometrical description of the GBDM
Chapter 3: Generic Channel Models
Figure 3.1 Estimated propagation paths in an outdoor environment
Figure 3.2 Channel dispersion in multiple dimensions
Figure 3.3 Direction of incidence characterized by
Figure 3.4 Estimated power DoD spectra for the channel response observed at different delays: (a) for ns (23 samples); (b) for ns (29 samples)
Figure 3.5 Contribution of the th wave to the received signal in a MIMO system incorporating dual-antenna arrays
Figure 3.6 Timing structure of sounding and sensing windows
Figure 3.7 Two propagation scenarios in which the keyhole or pinhole effect may occur
Figure 3.8 Keyhole channel that can be split into Tx and Rx subchannels
Chapter 4: Geometry-based Stochastic Channel Modeling
Figure 4.1 A general modeling procedure for the geometry-based stochastic modeling approach
Figure 4.2 A typical F2M cellular propagation environment for macro-cell scenarios
Figure 4.3 Geometrical configuration of a narrowband one-ring channel model with local scatterers around the mobile user
Figure 4.4 Geometrical description for an RS-GBSM of the typical V2V environment of Figure 2.16: SB, single-bounced; DB, double-bounced
Figure 4.5 Geometrical description for an IS-GBSM of the typical V2V environment of Figure 2.16
Figure 4.6 Relationship between the reference model and simulation model
Figure 4.7 Filter method of channel waveform generation
Figure 4.8 SoS method of channel waveform generation
Figure 4.9 Geometrical two-ring model for SISO V2V channels
Figure 4.10 Comparison between the introduced stochastic model and the ARA stochastic model
Figure 4.11 Squared error in the CF of the introduced deterministic simulation model with for different non-isotropic scattering V2V Rayleigh fading channels: (a) and ; (b) ; (c) , , , and
Figure 4.12 Mean squared error in the CF of the introduced stochastic simulation model with for different non-isotropic scattering V2V Rayleigh fading channels: (a) and ; (b) ; (c) , , , and
Figure 4.13 Comparison of the CF of the reference model and that of the simulation models with and for various values of the mean AoD and mean AoA
Figure 4.15 Comparison of the CF of the reference model and those of the simulation models with for various values of , , the mean AoD , and mean AoA
Figure 4.14 Comparison of the CF of the reference model and those of the simulation models with and for various values of and
Chapter 5: Channel Measurements
Figure 5.1 MIMO channel sounding systems with (a) a switched architecture and (b) a parallel architecture
Figure 5.2 The diagram of a measurement of phase noise, as conducted by Taparugssanagorn et al. [2007a]
Figure 5.3 Allan deviation
Figure 5.4 Scheme for investigating the impact of phase noise
Figure 5.5 The DoA and DoD estimates of multiple propagation paths using 50 × 32 MIMO-TDM sounder: (a) without use of sliding window and (b) applying the sliding window
Figure 5.6 The azimuth-elevation pattern of array gain generated by a 4 × 1 patch-antenna array in a node-B of a Long-Term-Evolution (LTE) system
Figure 5.7 The difference between synthetic and estimated channels, denoted by , versus the dynamic range of array gain
Figure 5.8 Elektrobit/Technology University of Vienna measurement campaign: (a) antenna array used in measurements; (b) indices of the array element; array gain for (c) the omnidirectional and (d) directional Rx array
Figure 5.9 Average PDP of channels observed in a cycle considered
Figure 5.10 Path-parameter estimates obtained using the SAGE algorithm from the measurement data with the following settings: (a) and (b), data collected for 50 × 32 subchannels; (c) and (d), data collected for 50 × 10 subchannels, and dB; (e) and (f), data collected for 50 × 10 subchannels, and dB. AoA, azimuth of arrival; AoD, azimuth of departure, EoA, elevation of departure; EoD, elevation of departure
Figure 5.11 The considered TDM measurement mode
Figure 5.12 The graphs of (dotted), (dash-dotted), (dashed) and (solid) for the parameter settings reported in Table 5.1
Figure 5.13 Objective functions for the joint Doppler frequency and DoA ML estimates in the case study. The following switching modes are selected: (c) (d) ; (e) a randomly selected cycle-dependent switching mode. (a) and (b) depict the factors of the objective function (see Eq. (5.54)) for the identity switching mode
Figure 5.14 RMSEEs of versus the SNR with the repetition rate as a parameter for Cases (1) and (2). The thin solid straight lines denote the root Cramér–Rao lower bound. The empirical RMSEEs are calculated using 100 Monte-Carlo simulation runs. RMSEE, root mean-square estimation error
Figure 5.15 RMSEEs of versus the SNR with the true Doppler frequency as a parameter. CRLB, Cramér–Rao lower bound
Figure 5.16 RMSEEs of (solid curves) and (dotted curves) versus computed using the settings given in Table 5.2 for different switching modes. The dashed and the dash-dotted lines represent the CRLBs of and respectively. The curves with symbols ⋄, □, ○, ▵ have been obtained using three cycle-independent switching modes and one cycle-dependent switching mode, leading to , and 0.28, respectively
Figure 5.17 Normalized (top) and pseudo-envelope (bottom) computed from the measurement data obtained in Scenario I (dashed lines) and Scenario II (solid lines). The marks and ⋄ denote the maxima of in Scenario I when the DFER is respectively and extended to . The mark denotes the maximum of in Scenario II (DFER )
Chapter 6: Deterministic Channel-parameter Estimation
Figure 6.1 DoD power spectra calculated using the beamforming method, with 50 Tx antennas and the No. 1 Rx antenna
Figure 6.2 The signal subspace and the noise subspace. Vectors and represent two steering vectors. The eigenvectors and are an orthonormal basis of the range space of the matrix
Figure 6.3 Flow graph of the SAGE algorithm
Figure 6.4 Contribution of the
l
th wave to the received signal in a MIMO system incorporating dual-antenna arrays
Figure 6.5 Timing structure of sounding and sensing windows
Figure 6.6 Parameter estimates of specular paths obtained using the conventional SAGE algorithm and the proposed EF-MEM. The stems with the markers “” and “” represent the true paths and the estimated paths respectively
Figure 6.7 (a) RMSEE of obtained by using the conventional SAGE algorithm and the proposed EF-MEM; (b) mean and spreads of the evidence of the specular-path model and the GAM-based dispersive-path model
Figure 6.8 The CDFs of the true, estimated delay spreads and the theoretical CDF of a log-normal distribution fitted to the empirical CDF
Figure 6.9 Normalized square error (NSE) of the approximation to the effective signal model using the first-order Taylor-series expansion versus (a) the Doppler frequency difference, (b) the phase difference with s, and (c) the number of observation samples in
Figure 6.10 A comparison of the performance of the EKFs for (a) tracking the Doppler frequency trajectory in each observation snapshot and (b) the estimation errors obtained by using respectively EKF without and with considering the phase aggregation due to the non-zero Doppler frequency
Figure 6.11 A comparison of the performance of the EKFs for (a) tracking the azimuth of arrival trajectory in each observation snapshot and (b) the estimation errors obtained by using respectively EKF without and with considering the phase aggregation due to the non-zero Doppler frequency
Figure 6.12 Photographs and plan of the environment
Figure 6.13 Average power-delay profiles of the received signals from 50 bursts
Figure 6.14 Performance of the PF in tracking three time-variant paths. The legend given in (b) applies to (a)–(f). In (a), the PDPs of the signals in 100 bursts are shown in the background
Figure 6.15 Geometries and characteristics of reconstructed propagation paths in the investigated environment
Chapter 7: Statistical Channel-parameter Estimation
Figure 7.1 Empirical (Emp.) and approximate (Approx.) CCDFs of the absolute normalized estimation error of the SS-ML azimuth estimator applied in an SDS scenario: (a) with as a parameter, ; (b) with as a parameter,
Figure 7.2 Empirical CCDF of the absolute normalization error of the nominal azimuth estimators with , dB,
Figure 7.3 RMSEEs of nominal azimuth estimators versus the input SNR with ,
Figure 7.6 RMSEEs of the NA and AS estimators versus the NA separation in a two-SDS scenario, , , dB and dB. The legend given in (a) applies to all subfigures
Figure 7.4 RMSEEs of the NA estimators versus the true AS with dB,
Figure 7.5 AEE (a) and RMSEE (b) of the AS estimators versus the true AS with the fixed array size , and dB
Figure 7.8 Pseudo-code representation of the steps of the power spectrum estimator
Figure 7.7 Updating the estimate of a specific parameter of the center of gravity of individual components
Figure 7.9 The power delay profile of the channel averaged over 16 observation snapshots
Figure 7.10 The power-delay profile of the channel
Figure 7.11 Power delay profiles computed from original data and reconstructed using the estimated parameter estimates
Figure 7.12 Comparison of the DoD Bartlett power spectra for the delay bin no. 1
Figure 7.13 Comparison of the DoA Bartlett power spectra for delay bin 1
Figure 7.14 Comparison of the DoD Bartlett power spectra for delay bin 2
Figure 7.15 Comparison of the DoA Bartlett power spectra for delay bin 2
Figure 7.16 Constellation of the dispersive components in AoA–AoD
Chapter 8: Measurement-based Statistical Channel Modeling
Figure 8.1 The channel modeling approach
Figure 8.2 The components in an RF channel
Figure 8.3 (a) Concatenated power delay profiles observed in a LoS scenario from channel measurements at 28 GHz; (b) the scatter plot of multipath components estimated by using the SAGE algorithm. A virtual array formed by rotating an Rx horn antenna in 36 steps along a circle was used to collect the data. The Tx horn antenna was fixed
Figure 8.4 A time-domain cluster model proposed in the IEEE 802.11ad standards [Erceg et al. 2004]
Figure 8.5 Constellation of the paths in the six clusters: Left, in Doppler frequency and delay domains; right, in DoA.
Figure 8.6 versus the burst indices when applying the Kolmogorov–Smirnov test. The solid line represents the critical value for at a significance level of 5%
Figure 8.7 The histogram of the number of bursts per segment and a fitted log-normal PDF multipled by the total number of segments
Figure 8.8 Cooperative relay system with a mobile station, a base station, and a relay station
Figure 8.9 Ilsan City measurement campaign: (a) environment where the measurements were conducted; (b) routes of the mobile station at one of the measurement sites
Figure 8.10 (a) Empirical distribution of for ; (b) comparison of fitted regression lines and the empirical PDF of
Figure 8.11 Empirical PDFs of the cross-correlation coefficients of the SSF of the BS-MS and RS-MS links vs. the candidates of geometrical variables: (a) the angle of separation in degrees; (b) the average distance in meters; (c) the relative distance represented in logarithm; (d) the ratio between and the RS-BS distance; (e) the BS-RS-MS triangle's area ; (f) the composite parameter defined to be
Figure 8.12 The angle of separation : (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. ; (b) the standard deviation of vs.
Figure 8.13 The average distance : (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. ; (b) standard deviation of vs.
Figure 8.14 Relative distance in logarithm (): (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. ; (b) standard deviation of vs.
Figure 8.15 The ratio : (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. the ratio; (b) standard deviation of vs. the ratio
Figure 8.16 Area of the BS-RS-MS three-node triangle: (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. area ; (b) standard deviation of vs. area
Figure 8.17 The composite parameter : (a) absolute value of the slope of the first-order regression line fitted to the empirical vs. ; (b) standard deviation of vs.
Chapter 9: In Practice: Channel Modeling for Modern Communication Systems
Figure 9.1 A new wideband multiple-ring based MIMO F2M channel model
Figure 9.2 The frequency CFs (reference model) and (simulation model, ) for different values of the parameter
Figure 9.3 The SF CFs (reference model) and (simulation model, ) for different values of the parameter
Figure 9.4 (a) The SF CF versus the frequency separation and the normalized antenna spacing at the MS with ; (b) the SF CF versus the frequency separation and the normalized antenna spacing at the BS with
Figure 9.5 (a) The time CF of the reference model and the time CF of the corresponding simulation model with ; (b) the frequency CF of the reference model and the time CF of the corresponding simulation model with
Figure 9.6 (a) The space CF of the reference model and (b) the space CF of the corresponding simulation model with
Figure 9.7 A generic channel model combining a two-ring model and an ellipse model with LoS components and single- and double-bounced rays for a MIMO V2V channel ()
Figure 9.8 A generic channel model combining a two-ring model and an ellipse model with LoS components, single- and double-bounced rays for a SISO V2V channel
Figure 9.9 Comparison of the LCRs , , and
Figure 9.10 The error between the LCRs and
Figure 9.11 Space CFs of the single-bounce (SB) ellipse (EL) model, double-bounce (DB) two-ring (TR) model, and SB TR model for different scenarios (, , and ). SD: same direction (); OD: opposite direction ( and ); Scenario (): (isotropic environments); Scenario (): (non-isotropic environments), , and
Figure 9.12 Frequency CFs of the single-bounce (SB) ellipse (EL) model, double-bounce (DB) two-ring (TR) model, and SB TR model for different scenarios (, ): SD, same direction (); OD, opposite direction. Scenario (), (isotropic environments); Scenario (), (non-isotropic environments), , and
Figure 9.13 Normalized Doppler PSDs of the single-bounce (SB) ellipse (EL) model, double-bounce (DB) two-ring (TR) model, and SB TR model for different scenarios (, ): SD, same direction (); OD, opposite direction ( and ). Scenario , () (isotropic environments); Scenario (), (non-isotropic environments), , and
Figure 9.14 Normalized space-Doppler PSDs of the single-bounce (SB) ellipse (EL) model, double-bounce (DB) two-ring (TR) model, and SB TR model for different antenna element spacings in a V2V non-isotropic scattering environment (, , ) with the Tx and Rx moving in the same direction ()
Figure 9.15 Normalized frequency-Doppler PSDs of the single-bounce (SB) ellipse (EL) model, double-bounce (DB) two-ring (TR) model, and SB TR model for different frequency separations in a V2V non-isotropic scattering environment (, , ) with the Tx and Rx moving in the opposite direction ( and )
Figure 9.16 Normalized Doppler PSDs of the proposed adaptive model for different SISO picocell scenarios (, ): (a) Tx and Rx are moving in opposite directions; (b) Tx and Rx are moving in the same direction. VTD: vehicular traffic density
Figure 9.17 (a) LCRs and (b) AFDs of the developed V2V channel model with a low or high vehicular traffic density (VTD) when the Tx and Rx are moving in the same direction
Figure 9.18 A geometry-based stochastic channel model combining a two-ring model and a multiple confocal ellipse model with LoS, single- and double-bounced rays for a wideband MIMO V2V channel
Figure 9.19 Geometrical description of the LoS, single-, and double-bounced rays for different taps in the proposed wideband MIMO V2V GBSM. SB, single-bounce; DB, double-bounce; T1, tap 1; T2, tap 2
Figure 9.20 Squared error in the ST CF of the deterministic simulation model obtained using MMEA, MMEA2, and IMMEA for a non-isotropic scattering MIMO V2V channel: , , and
Figure 9.21 Squared error in the ST CF of the deterministic simulation model obtained using MMEA, MMEA2, and IMMEA for a non-isotropic scattering MIMO V2V channel:
,
, and
Figure 9.22 Squared error in the ST CF of the deterministic simulation model obtained using MMEA, MMEA2, and IMMEA for a non-isotropic scattering MIMO V2V channel: , , and
Figure 9.23 Absolute value of the ST CF of the reference model, deterministic simulation models (MMEA model, MMMEA2 model, and IMMEA model), and stochastic simulation model for a non-isotropic scattering MIMO V2V channel; ( and ), with antenna spacings (a) and (b)
Figure 9.24 Normalized (space-)Doppler PSDs of (a) first tap and (b) second tap of the proposed wideband MIMO V2V channel model with low and high VTDs when the Tx and Rx are moving in opposite directions on an expressway
Figure 9.25 Normalized (space-)Doppler PSDs of (a) first tap and (b) second tap of the proposed wideband MIMO V2V channel model with low and high VTDs when the Tx and Rx are moving in the same direction on an expressway
Figure 9.26 Absolute value of the (space-)time CFs of (a) first tap and (b) second tap of the proposed wideband MIMO V2V channel reference model and corresponding simulation models with low and high VTDs when the Tx and Rx are moving in opposite directions on an expressway
Figure 9.27 Absolute value of the FCF for the reference model and the deterministic and stochastic simulation models. SD: same direction; OD: opposite directions
Figure 9.28 Geometry of a unified cooperative MIMO channel model framework
Figure 9.29 The proposed cooperative MIMO GBSM
Figure 9.30 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the first single-bounced component ; (b) the third single-bounced component ; (c) the third double-bounced component ; and (d) the triple-bounced component
Figure 9.31 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the single-bounced components and (b) the double- and triple-bounce components
Figure 9.32 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the first single-bounced component ; (b) the third double-bounced component ; and (c) the triple-bounced component with different values of parameters and ()
Figure 9.33 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the first single-bounced component ; (b) the third double-bounced component ; and (c) the triple-bounced component with different values of parameters , , and ()
Figure 9.34 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the second single-bounced component ; (b) the second double-bounced component ; and (c) the triple-bounced component with different values of parameters and ()
Figure 9.35 Absolute values of spatial correlation functions between the BS–RS link and BS–MS link for (a) the outdoor macrocell MS cooperation scenario and (b) the indoor MS cooperation scenario with different LSDs
Appendix A
Figure A.1 Relationship between the AoA and AoD for single-bounced rays.
Figure A.2 Graphical description of (a)
Scenario1
and (b)
Scenario2
.
Figure A.3 Comparison of the Doppler PSDs of
Scenario1
and
Scenario2
based on the CF definitions in Eqs (9.49) and (9.50).
List of Tables
Chapter 4: Geometry-based Stochastic Channel Modeling
Table 4.1 Important V2V channel models
Chapter 5: Channel Measurements
Table 5.1 Parameter settings used for simulations
Table 5.2 Case study: Settings of the TDM-SIMO system and parameters of the incident wave
Table 5.3 Parameter settings used for simulations
Table 5.4 Settings of the channel sounder for measurement Scenarios I and II
Chapter 6: Deterministic Channel-parameter Estimation
Table 6.1 Parameter values computed from the geometrical figure
Chapter 7: Statistical Channel-parameter Estimation
Table 7.1 Final estimates of the parameters of dispersive components
Chapter 8: Measurement-based Statistical Channel Modeling
Table 8.1 Parameter settings in the SAGE algorithm and the clustering algorithm
Chapter 9: In Practice: Channel Modeling for Modern Communication Systems
Table 9.1 Important V2V channel measurements
Table 9.2 Definition of parameters in Figure 9.29
Table 9.3 Main features of the proposed cooperative mimo GBSM
Appendix A
Table A.1 The values of versus
Propagation Channel Characterization, Parameter Estimation and Modeling for Wireless Communications
Xuefeng Yin
Tongji University, Shanghai, China
Xiang Cheng
Peking University, Beijing, China
This edition first published 2016
© John Wiley & Sons Singapore Pte Ltd. All rights reserved.
Registered office
John Wiley & Sons Singapore Pte Ltd, 1 Fusionopolis Walk, #07-01 Solaris South Tower, Singapore 138628
For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com.
The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.
Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent professional should be sought.
Library of Congress Cataloging-in-Publication data applied for
ISBN: 9781118188231
A catalogue record for this book is available from the British Library.
Preface
The investigation of the propagation channel is becoming more and more important in modern wireless communication. The demand for spectral efficiency motivates exploitation of all channels that can possibly be used for communications. Nowadays, a common trend for designing physical layer algorithms is to adapt the transceiving strategy, either by maximizing the diversity gains or by utilizing the coherence of the channels to improve the signal-to-noise power ratio.
Dr. Xiang Cheng and I have been working on topics relevant to channel characterization for years. My major research has been focused on measurement-based stochastic channel modeling using high-resolution estimates of the channel parameter from real measurement data. Xiang's work concentrates more on accurate yet easy-to-use channel modeling and simulation based on geometry-based stochastic channel modeling approach. This book is intended to cover both theoretical and experimental studies of channels by merging Xiang's and my own study results, obtained in the last decade. Most of the content has been published in journals and conference proceedings. New results that are still under review for publication are also addressed in order to give a complete presentation of specific topics. In general, the book can be viewed as a collection of the latest results in the field of theoretical and experimental channel characterization. The contributions of Dr. Xiang Cheng and myself to this book are equivalent.
There are already several books dedicated to channel investigations (Durgin 2003, Koivunen 2007, Parsons 2000, Saunders 1999, Pätzold 2002). Our book includes more of an introduction to the methods used for the steps of channel characterization than these earlier studies, instead of presenting only the final results. From this point of view, our book tells more complete stories about channels, linking the methods applied in the different stages of channel analysis. Furthermore, combining the description of theoretical and the empirical methods in one book helps the reader conceive more clearly the merits of these methods. Another feature of the book is that we also cover the methods used for extracting the parameters of generic models. Normally, books address channels in one or two chapters and do not describe and comment on the methods used for characterizing them. An example is the book by (Correia 2001). However, since readers usually want to know how realistic the models are, it is important to give a clear view of the underlying methods. Although some of the methods described in this book have also been described in the literature for general cases (McLachlan and Krishnan 1997, Stoica and Moses 1997), we focus on the adaptation of methods applied for analyzing the propagation channels.
This book can be used as a textbook for the courses dedicated for propagation channel characterization, or for parts of courses that focus on wireless communication systems and networks. We organize the book in such a way that the chapters are self-contained and can be selected individually for specific topics. We start in Chapter 2 by introducing the phenomena of propagation in wireless communication channels and the terminologies, and also the parameters used to characterize their properties. Then, in Chapter 3, the generic parametric models applied for representing multiple components in channel impulse responses are introduced. For stochastic behaviors of channels represented by these model parameters, statistical models are needed. We, therefore, review the approaches adopted in channel characterization and modeling; from their theoretical aspects in Chapter 4 and by using measurements in Chapter 5. The impacts of measurement equipment on the observations and model accuracy are also discussed in Chapter 5. Chapters 6 and 7 introduce the high-resolution channel-parameter estimation methods for extracting the parameters of the generic channel models from measurement data, based on deterministic specular-path models and statistical models, respectively. Chapter 8 elaborates the modeling procedure and key techniques for constructing stochastic models based on parameter estimates. At the end of the book, Chapter 9 illustrates specific channel models for different communication systems as examples of the methods and techniques introduced in this book.
Acknowledgements
This book is based on the publications written by the authors and their colleagues in recent decades. We are indebted to colleagues and students who have made valuable suggestions and comments leading to many important changes. In this regard, we are particularly grateful to the co-authors of the publications relevant to this book: Prof. Bernard Fleury, Aalborg University; Prof. Troels Pedersen, Aalborg University; Attaphongse Taparugssanagorn, Asian Institute of Technology; Dr. Li Tian, ZTE Technology; Stan X. Lu, Huawei Technology Company; Dr. Zhimeng Zhong, Huawei Technology Company; and Nicolai Czink. We also wish to acknowledge the valuable comments of the manuscript reviewers: Prof. Bo Ai, Beijing Jiaotong University and Prof. Jianhua Zhang, Beijing University of Post & Telecommunication.
References
Durgin GD 2003
Space-Time Wireless Channels
. Pearson Education.
(ed. Correia L) 2001
Wireless Flexible Personalised Communication (COST 259: European Co-Operation in Mobile Radio Research)
. John Wiley & Sons.
(ed. Pätzold M) 2002
Mobile Fading Channels
. Wiley.
Koivunen J 2007 Characterization of MIMO propagation channel in multilink scenarios PhD thesis Helsinki University of Technology.
McLachlan GJ and Krishnan T 1997 The EM Algorithm and Extensions vol. 1. Wiley Series in Probability and Statistics.
Parsons JD 2000
The Mobile Radio Propagation Channel
2nd edn. John Wiley and Sons.
Saunders SR 1999
Antennas and Propagation for Wireless Communication Systems
. John Wiley & Sons.
Stoica P and Moses RL 1997
Introduction to Spectral Analysis
. Prentice Hall.
List of Acronyms and Symbols
List of Acronyms
3GPP
Third Generation Partnership standards bodies
5G
Fifth generation wireless communications
AEE
average estimation error
AFD
average fading duration
AGV
autonomous guided vehicles
AoA
azimuth of arrival
AoD
azimuth of departure
ARA
acceptance rejection algorithm
ARIMA
autoregressive integrated moving average
AS
azimuth spread
ASA
array size adaptation
CBM
correlation-based model
CCDF
complementary cumulative distribution functions
CLEAN
an iterative beam removing technique
COMET-EXIP
covariance matching estimator-extended envariance principle
CoMP
cooperative multipoint
COST
Commission of Science and Technology
CRLB
Cramér–Rao lower bound
DECT
Digital European Cordless Telephone
DER
direction estimation range
DFER
Doppler frequency estimation range
DI
diffuse component
DML
deterministic maximum likelihood
DoA
direction of arrival
DoD
direction of departure
DRAG
dynamic range of array gain
ECM
environment characterization metric
EM
expectation-maximization
Emp.
empirical
EoA
elevation of arrival
EoD
elevation of departure
ESPRIT
estimation of signal parameters via rotational invariance techniques
Est.
estimated
E-step
expectation step
F2M
fixed to mobile
FB
Fisher–Bingham
FFHMA
fast frequency hopping multiple access
FHMA
frequency hopping multiple access
GAM
generalized array manifold
GBDM
geometry-based deterministic model
GBSM
geometry-based stochastic modeling
GR
Gerschgorin radii
HFB
higher frequency band
HRPE
high-resolution parameter estimation
IMT
international mobile telecommunications
IS-GBSM
irregular-shaped GBSM
ISI
improved initialization and search
ISM
industrial, scientific and medical bands
JADE
joint angle and delay estimation
LCR
level cross rate
LoS
line-of-sight
MD
mobile scatterer
MEA
method of equal area
MEDS
method of exact Doppler spread
METIS
Mobile and Wireless Communications Enablers for the Twenty-Twenty Information Society
MIMO
multiple-input, multiple-output
ML
maximum likelihood
MMEA
modified method of equal area
MODE
method of direction estimation
M-step
maximization step
MUSIC
multiple signal classification
NC-ML
non-coherent-maximum-likelihood
NFD
Newton forward difference
NGSM
non-geometric stochastic models
NLoS
non-line-of-sight
NSL
normalized side-lobe level
OLOS
obstructed line-of-sight
OMUSIC
orthonormal-basis MUSIC
OSM
orthogonal stochastic measure
probability density function
PDP
power delay profile
PE
pseudo-envelope
PMM
propagation-motivated model
PN
pseudo-noise
PSD
power spectral density
PSM
parametric stochastic models
RF
radio frequency
RIMAX
Richter's maximum likelihood estimation
RMSEE
root mean square estimation error
RS-GBSM
regular-shaped GBSM
Rx
receiver
SAGE
space-alternating generalized expectation-maximization
SCM
spatial channel model
SCME
spatial channel model enhanced
SD
static scatterer
SIOD
space-invariance of determinant
SISO
single-input, single-output
SML
stochastic maximum likelihood
SNR
signal-to-noise ratio
SoS
sum of sinusoids
SS
specular-scatterer
ST
space time
SVD
singular-value decomposition
SW
switch
TDL
tap-delay line
TDM
time-division-multiplexing
TEM
transverse electric and magnetic wave
Tx
transmitter
ULA
uniform linear array
V2V
vehicle to vehicle
Vec-MUSIC
vector-MUSIC
vMF
von-Mises–Fisher
VTD
vehicular traffic density
WINNER
wireless world initiative new radio
WSS
wide-sense stationary
WSSUS
wide-sense stationary uncorrelated-scattering
XPD
cross-polarization discrimination
List of Symbols
real line
complex plane
-dimensional sphere
domains
Kronecker product
Hadamard – that is, element-wise – product
real part of the complex number given as an argument
imaginary part of the complex number
Frobenius form of the vector or matrix given as an argument
absolute value of the given argument
determinant of the matrix given as an argument
trace of the matrix given as an argument
Hermitian of the vector or matrix given as an argument
transpose of the vector or matrix given an argument
complex conjugate of the scalar given as an argument
pseudo-inverse of the matrix given as an argument
scalar product of the given arguments
the Kronecker delta
Dirac delta function
an identity matrix of dimension given as an index
diagonal matrix with diagonal elements listed as argument
the
th eigenvalue of the matrix given as an argument
the projection operator onto the column space of the matrix given as an argument
array response
the first derivative of the array response
the gamma function
the modified Bessel function of the first kind and order
standard deviation of random variable given as an argument
parameter vector
parameters with indices specified in a set
in a subset of
parameters with indices listed in the complement of
intersected with
hidden-data space for
index set in the
th iteration of the SAGE algorithm
output signal from an antenna array
noise vector
total number of path components
the region where the Tx array is confined
the region where the Rx array is confined
parameter vector associated with the
th path component
the polarization matrix of the
th propagation path
the field pattern of the
th element of array
for polarization
the input signal vector
a unit direction vector
a sounding period
a sensing period of a Rx antenna
a period separating two consecutive sensing intervals
separation between the beginnings of two consecutive measurement cycles
the guard interval
noise variance
wavelength
the number of antennas in the transmitter site
the number of antennas in the receiver site
the location of the
th element of array
weight of the polarization component with Tx polarization
and Rx polarization
log-likelihood function of the parameter(s) given as an argument
Chapter 1Introduction
1.1 Book Objective
The characteristics of the propagation channel are of great importance for designing wireless communication systems, analyzing communication qualities, and simulating the performance of networks. However, in most books on wireless communications, propagation channels are usually presented in only one or two chapters, which describe the fundamental characteristics of channels – for example path loss, shadowing, and multipath fading – and present some standard models. Since the procedures for measuring the wireless channels, the methodologies adopted for parameter estimation, and the modeling approaches implemented are neglected in these books, it is impossible for readers to understand how the models are established for specific scenarios. This also results in suspicions about the applicability of models, and questions also arise about the appropriateness for implementation in channel simulations.
Furthermore, fast-growing wireless communication networks and services bring greater demands for high spectral efficiency. Numerous techniques have been used, all essentially exploiting the resources from propagation channels. For example, parallel spatial channels are resolved and utilized by multiple-input, multiple-output (MIMO) techniques for diversity or multiplexing. Similar MIMO techniques in other domains, such as in polarizations and in wavefronts, have been developed. It is of no doubt that future wireless system design will be more and more adaptive to the environments in which they are used. Network architecture design is also becoming increasingly complicated in order to make the most use of specific channels. For example, the techniques of distributed antennas, massive MIMO, relay, cooperative transmission, and joint processing all require detailed knowledge of channels in both a stochastic sense and in site-specific scenarios. Therefore, channel characterizations based both on theoretical approaches and real measurements are going to become critical in the future.
Considering the multiple aspects of a channel, it is actually a “mission impossible” to write a book that is sufficiently comprehensive that every topic of channel studies is included. This book is written with the aim of covering only some aspects of the propagation channel:
the high-resolution approach of analyzing channels based on measurement data
stochastic channel modeling either using empirical parameters or based on simulation of scattering.
The objectives of this book are threefold. First, the book provides the fundamentals of both empirical measurement-based and theoretical-scattering-based channel modeling. The topics covered are widely spread, touching on the fields of wideband channel measurements, model parameter extraction, stochastic model generation, and theoretical channel modeling. Second, the book provides some updated channel models, which can be used for practical simulations. Engineers in the wireless communication industry can therefore use them to evaluate their system performance. Thirdly, this book highlights ongoing trends, revealing some fresh research results that might be interesting for researchers when designing new systems.
1.2 The Historical Context
1.2.1 Importance of Channel Characterization
The statistical characteristics of channels can significantly influence the design of wireless communication systems. For example, the path-loss model, based on the measurements in specific regions, can be used to determine the appropriate value of the separation between cells, in order to keep the interference below a certain threshold. Shadowing models can be used to determine the maximum and the minimum transmission power in order to avoid blindspots in the coverage. Multipath fading models, which include the fading rate and fading-duration characteristics, can be used to determine the packet length and the transmission rate. Delay spread models can be used to evaluate the frequency selectivity of the environment, so as to determine the coherence frequency bandwidth or the separation of the orthogonal channels in the frequency domain. Doppler frequency spread models can be used to calculate the coherence time of the channel, and therefore determine the cycle duration to renew the estimate of channel coefficients. The models in the spatial domains, for example the cluster-based bidirectional models, can be applied to determine the antenna beamwidth in beamforming applications, or to calculate the degrees of freedom for channels with MIMO configurations. Stochastic models themselves are based on extensive measurements in many environments categorized into specific types, such as outdoor, indoor, urban/suburban, and so on; they are therefore valid in similar environments.
The model parameters can be used to determine the many thresholds used in communication systems. For example, for frequency hopping multiple access systems, the frequency offsets due to the Doppler effect of the channel, and the timing problems due to the multipath arrivals at different time instants, can cause a certain portion of the desired signal's energy to appear in spurious adjacent frequency bins; consequently the detection of the desired signal becomes difficult [Joo et al. 2003], and the detection matrix may have erroneous entries [Yegani and McGillem 1993]. With the knowledge of the delay-Doppler frequency dispersion behavior of channels in certain environments and scenarios, the threshold level of envelope detectors can be appropriately selected. Furthermore, if the instantaneous knowledge of the channel dispersion characteristics is available, the channel can be equalized accordingly.
1.2.2 Single-input, Single-output Channel Models
Channel investigation started at the end of the 1960s [Okumura et al. 1968]. At that time, wireless systems were built for voice communications using frequency division multiple access technique. The channel characteristics of interest when the single-input, single-output (SISO) system was considered was therefore the fading distributions at particular frequencies.
For outdoor scenarios, it has been found that the fading distribution is Rayleigh in a local geographical area with diameter of less than a few hundred wavelengths, and lognormal over large geographical areas [Lee and Yeh 1972, Okumura et al. 1968, Schmid 1970, Turin et al. 1972]. Suzuki [1977] considered more distributions, including the Nakagami and lognormal distributions, to fit the empirical data. It was found that the Rayleigh distribution is not always a good fit for most data, and that the lognormal distribution is often better. A possible reason for this observation is that the distribution – actually a mixture of Rayleigh distributions with a lognormal mixing distribution – is an intermediate distribution between the Rayleigh and the lognormal distributions [Suzuki 1977].
For indoor propagation environments, the SISO channel models have been established for the line-of-sight (LoS) and obstructed LoS (OBS) scenarios, as in the factory and open-plan office cases [Kozlowski et al. 2008, Rappaport and Seidel 1989, Rappaport et al. 1991, Saleh and Valenzuela 1987, Seidel et al. 1989, Yegani and McGillem, 1989a,b, 1991]. The motivation for investigating indoor SISO channels is to provide models for indoor deployment of radio systems that accommodate data rates up to 1 Mb/s. Such systems include the Digital European Cordless Telephone (DECT) 802.41 and WLAN (IEEE 802.11) standards, as well as the communication systems for autonomous guided vehicles (AGVs) [Rappaport et al. 1991]. The interesting characteristics of the channels in indoor environments include the path loss and delay spread. It has been found that the delay spread can be several times greater in unpartitioned factory buildings than in partitioned office buildings [Hawbaker and Rappaport 1990b]. Besides the large-scale parameters, the detailed wideband characteristics of the channel, for example the dispersion of the channel in the delay domain, has been examined. For example, Hawbaker and Rappaport [1990a] found the so-called “pulse overlapping” phenomenon, which revealed that even in the LoS scenario the OBS path components can be added to the LoS path components within the resolution of transmitted pulse, resulting in so-called multipath fading.
Furthermore, resolvable rays in the time domain have been applied to modeling channels. This kind of model was called a discrete model. For outdoor environments, discrete channel models consist of discrete rays or discrete peaks of the power-delay profiles [Cox and Leck 1975, Turin et al. 1972]. The magnitude of each ray can be set to follow the lognormal distribution [Suzuki 1977]. The correlation bandwidth is also applied as a model parameter for channels [Cox and Leck 1975], and this is large when the channel-delay profile exhibits several dominating discrete peaks, but small when multipath is severe [Cox 1972]. Since the channel-impulse response in the delay domain is available, the distribution of the number of paths, and the mean and standard deviations of logarithmic path strength are considered for channel characterization [Cox 1972]. Furthermore, by using multiple observations of the channel, the Doppler frequency spectrum has also been computed and used for modeling the channel [Cox 1973]. In addition, the trend of describing the channel properties in two dimensions has appeared in the literature [Cox 1973]. The Doppler spectra versus delay and the distribution of path strength versus delay have been studied for outdoor channels in urban environments [Cox 1973]. The small-scale characteristics of the channel – the channel property at specific delays – have become important for modeling.
Some important observations have been obtained through measurement. For outdoor urban environments, the excess delay of a channel at 900 MHz can be up to 9–10 µs [Cox 1973]; the delay spread, defined as the square root of the second central moment of the power-delay profile, is 2–2.5 µs. The path with 0.1 µs resolution exhibits a Rayleigh distribution, inferring that the fading coefficients for the first arrival path can be modeled as a Gaussian random process. For paths with different delays, uncorrelated scattering is confirmed by the observation that their Doppler frequency power spectra are quite different. The conclusion that paths with different delays are uncorrelated seems more useful for urban environments. Some authors have proposed to use correlated paths to construct discrete models, but this contradicts the observations of Cox [1973].
For indoor manufacturing environments, Yegani and McGillem [1991] provided the statistics for channels in different sites in a factory under four scenarios with different settings of LoS/OBS, and sparsely or densely distributed scatterers. It was found that the interarrival times of the paths were well modeled by the Weibull distribution, the number of paths by the modified beta distribution, and the path-gain coefficients by the Rayleigh, Rician, and lognormal distributions. The values of the parameters of these distributions were reported by the authors. It is interesting to observe that the average number of paths for different sites at a fixed threshold of signal strength is about the same, an indication that the statistics of the number of paths arriving at the receiver is not very sensitive to the topography of the factory site. Furthermore, the geometry of the factory and the layout of the working area have a strong influence on the distribution of the path-gain coefficients. There are also some new findings, for example when the dynamic range is not selected the path-gain coefficients follow the lognormal distribution regardless of the LoS, OBS, or how densely distributed the scatterers are. As for the threshold, when this is greater than −10 dB the path-gain distribution follows the Rician PDF, but when lower than −10 dB, the Rayleigh distribution provides a better fit. Thus, the estimated PDF for gain coefficients depends on the level of the dynamic range set at the receiver.
The research into channels for SISO has evolved into multiple areas. For example, polarization characteristic have been investigated since the 1970s, when polarization diversity was used to combat multipath fading. Employing orthogonally polarized channels over the same microwave link for satellite communications can result in twice the system capacity as when using single-polarized antennas [Lee and Yeh 1972]. In 2001, Andrews et al. [2001] pointed out that six channels without any correlation can immensely improve the transmission rate and system capacity of a wireless communication system, by polarization in a scattering-rich environment. Channel models have been proposed that can be used to generate the channel responses with an arbitrary pair of vertical and horizontal polarizations at both transmitter and receiver sites [3GPP 2007, Jeon et al. 2012]. Besides the cross-polarization discriminations (XPDs) of individual propagation paths, these models also involve the responses of antennas in different polarizations.
Jiang et al. [2007] studied the correlation coefficients for both copolarized and crosspolarized channels. They found that:
polarization decorrelation outperforms spatial decorrelation in the strong LOS scenario
horizontally polarized channels are more correlated than vertically polarized channels
the correlation of copolarized channels increases as the Rician K factor increases
channels have much higher correlation in the elevation domain.
A strong conclusion was that the crosscorrelation of crosspolarized channels is not affected by the environment, while the performance of copolarized channels is scenario dependent.
1.2.3 Spatial Channel Models (SCMs)
Estimating the direction or bearings of incoming signals has been a research topic for years. The original objective was for signal detection and estimation, including radar target tracking and component separation. The methods used for estimating direction of arrival are similar to those used in time-series spectral analysis and they are applied specifically with the samples obtained from spatially distributed arrays of sensors, including antennas for receiving electromagnetic waves and microphones for acoustic signals.
The study of the arrival angles of signals as part of the design of communication systems can be traced back to the 1970s. For example, Lee and Brandt [1973] found from field measurements of mobile radio signals that the signal arrival is concentrated at elevation angles lower than . Based on this finding, an omnidirectional antenna with vertical directivity is usually selected to increase the average received signal strength.
There are also many practical concerns that require a knowledge of the spatial characteristics of a channel. For example, when MIMO techniques are used in communication systems, the spatial diversity and/or multiplexing gains need to be evaluated based on realistic modeling of the covariance of the spatial channels. Furthermore, in the case where the beamforming technique is used in a base station, it is necessary to know the distribution of the energy in the direction of arrival; in other words how the energy is concentrated and what its spread is in the dominant path. Additionally, with directional parameters, the propagation of the waves can be easily visualized when the actual constellation of the scatterers is presented for specific environments. Geometry-based channel modeling (GBSM) has flourished in the last decade. One major reason is that channel dispersion in the directional domains can be obtained by spectral analysis of the measurement data.
