&v TECHNISCHE UNIVERSITÄT ILMENAU Fakultät für Elektrotechnik und Informationstechnik CRLp W MIMO Channel Modeling in Wireless Communications and its Applications Marko Milojevic Dissertation zur Erlangung des akademischen Grades Doktoringenieur (Dr.-Ing.) Anfertigung im: Gutachter: Vorgelegt am: Verteidigt am: Fachgebiet Nachrichtentechnik Institut für Informationstechnik Fakultät für Elektrotechnik und Informationstechnik Univ.-Prof. Dr.-Ing. Martin Haardt Prof. Dr.-Ing. Albert Heuberger Prof. Dr.-Ing. Armin Dekorsy 09.11.2010 19.04.2011
Contents Preface Abstract Zusammenfassung Contents List of Figures List of Tables i Hi v vii xi xvii 1. Introduction 1 1.1. State of the Art 1 1.2. Overview and Contributions 4 2. MIMO Systems and MIMO Channels 7 2.1. MIMO Systems in Wireless Communications 7 2.1.1. Spatial Processing Techniques 8 2.2. The Deterministic Description of the MIMO Channels 12 2.2.1. The First Set of Bello Functions 12 2.2.2. Bello Functions Extended to the Spatial Domain 15 2.2.3. The Full MIMO Channel Description 19 2.2.4. Polarization 23 2.2.5. The Sampled Frequency-Selective Time-Variant MIMO Channel 24 2.3. Stochastic Characterization of Multi-path Channels 24 2.3.1. Stochastic Process Theory 24 2.3.2. Stochastic Description of the Full MIMO Channel 25 2.3.2.1. The Power Delay Spectrum 27 2.3.2.2. The Autocorrelation Function in Frequency Domain 29 2.3.2.3. Doppler Power Spectrum 31 2.3.2.4. The Autocorrelation Function in Time Domain 32 2.3.2.5. Power Angular Spectrum 33 2.3.2.6. The Autocorrelation Function in Spatial Domain 35 vii
2.3.3. The Sampled Autocorrelation Function of Frequency-Selective Time-Variant MIMO Channel 36 3. Modeling of the MIMO Channels 37 3.1. Channel Models Classification 37 3.2. Deterministic Channel Models 41 3.3. Analytical Channel Models 42 3.3.1. Full Spatial Correlation Matrix Channel Model for Frequency-Non-Selective Time-Invariant Channels 43 3.3.2. The Kronecker Channel Model 44 3.3.3. The Weichselberger Channel Model 46 3.3.4. A Subspace-Based Channel Model for Frequency-Selective Time-Variant MIMO Channels 47 3.3.4.1. The Channel Model Validity and Channel Parametrization... 50 3.3.4.2. A Physical Interpretation of the Eigenmodes 54 3.3.4.3. Correlation in Frequency and Time 56 3.3.5. A Correlation Tensor-Based Model for Frequency-Selective Time-Variant MIMO Channels 59 3.4. Hybrid Channel Models 62 3.4.1. COST 259 Directional Channel Model 62 3.4.2. IST METRA Channel Models 64 3.4.3. IEEE 802.11n Channel Model 65 3.4.4. The IlmProp Channel Model 66 3.4.5. 3GPP Spatial Channel Model (SCM) 67 3.4.6. WINNER Spatial Channel Model Extended (SCME) 71 3.4.6.1. Bandwidth Extension 73 3.4.6.2. Extension of the Frequency Range 74 3.4.6.3. LOS Option and Time Evolution 75 4. WINNER Channel Models (WIM) 79 4.1. Introduction to the WINNER Channel Models 80 4.2. WIM Modeling Approach 81 4.2.1. The Network Layout 86 4.2.2. Modeling the Time Evolution 86 4.2.3. Inter-Dependence of Large-Scale Parameters 88 4.2.4. Modeling of the Nomadic Channel Condition 90 4.2.5. WIM 3D Antenna Array Model 91 4.2.5.1. The Description of 3D Antenna Arrays 91 4.2.5.2. Integration of 3D Antenna Arrays in the WIM Channel Model. 96 viu
4.2.6. WIM Models With Reduced Complexity 97 4.2.6.1. WIM Clustered Delay Line Models 98 5. Channel Modeling for Multiple Mobile Satellite Broadcasting Systems 99 5.1. Satellite Digital Audio Radio Systems ХМ Radio and Sirius 102 5.1.1. XM Radio 103 5.1.2. Sirius Satellite Radio 103 5.1.3. Measurements of XM Radio and Sirius S-DARS Systems 104 5.2. Channel Model for Single and Two Satellite Systems Based on the First Order Markov Chains 106 5.2.1. Channel State Model for Single Satellite Systems 106 5.2.2. Channel State Model for the Two Satellite Systems 108 5.2.3. The Channel Amplitude Modeling Ill 5.2.4. Correlation and State Duration Properties of the Measured S-DARS Channels 112 5.2.4.1. Correlation Coefficient Between Two S-DARS Satellites 113 5.2.4.2. Spatial Autocorrelation Function of S-DARS Satellites 113 5.2.4.3. Channel State Duration of the S-DARS Systems 115 5.2.5. Environmental Influence on the Noise Power 117 5.3. The State Duration of the Markov Chain Based Channel Models 117 5.3.1. The PDFSD Modeling Based on Dynamic Order Markov Chains 121 5.3.2. Dynamic SPTM Model Approximations 123 5.3.2.1. Partial Dynamic SPTM Model 124 5.3.2.2. Approximated Partial Dynamic SPTM Model 125 5.3.3. PDFSD Based Channel State Generation Algorithm 127 5.4. The PDFSD Modeling for Multiple Satellite Systems 128 6. Spatio-Temporal Availability in Satellite-to-lndoor Broadcasting 131 6.1. Introduction 132 6.2. Satellite-to-lndoor Channel Model 133 6.2.1. System Model for Simulation Setup 133 6.2.2. The Ray Tracing Engine WinProp 134 6.2.3. Geometry-Based Channel Modeling 135 6.2.4. Polarization and Different Antenna Configurations 139 6.2.5. Frequency Selectivity of Studied Channels 139 6.2.6. Temporal Variation of the Satellite Link 140 6.2.7. Signal Availability 141 6.3. Performance Comparison of Different Receive Antenna Schemes 142 6.3.1. Ray Tracing Results for Vertical Dipole 142 6.3.2. Spatial Correlation of Multi-Antenna Configurations 143
6.3.3. Signal Availability for Different Receive Antenna Configurations 146 6.3.4. Influence of Windows on the Availability 153 6.3.5. User Interaction 155 6.3.6. Influence of Temporal Variation on Signal Availability 156 7. Applications of the Channel Modeling on the Channel Prediction and Channel Interpolation 159 7.1. On the Channel Prediction 159 7.2. Subspace-Based Framework for the Prediction of Frequency-Selective Time-Variant MIMO Channels 160 7.2.1. Subspace-Based Channel Prediction Framework 161 7.2.1.1. Filter Structures 163 7.2.2. Propagation Scenario for Simulations 164 7.2.3. Simulation Results 165 7.3. Tensor-Based Framework for the Prediction of Frequency-Selective Time-Variant MIMO Channels 168 7.3.1. Tensor-Based Channel Prediction Framework 169 7.3.2. Simulation Results 172 7.4. Interpolating the Channel 174 8. Conclusions 177 Appendix A. Mathematical Operators 179 A.l. The Scalar Product 179 A.2. The Vector Product 179 A.3. The Kronecker Product 180 Appendix B. Higher-Order Signal Processing Calculus 181 B.l. Tensor Properties 181 B.2. The Outer, «-Mode, and»-mode Inner Tensor Products 183 B.3. Higher Order Singular Value Decomposition (HOSVD) 185 Appendix С Generation of WIM Channel Coefficients and WIM Parameters 187 C.l. Generation of WIM Channel Coefficients 187 C.2. WIM Parameters 194 Appendix D. Glossary of Acronyms, Symbols, and Notation 197 Bibliography 203