同济大学：《传播信道特征估计和建模》课程教学资源（教案讲义）Chapter 01 Introduction.pdf_大学文库

1 Introduction 1.1 Book Objective The characteristics of the propagation channel is of great importance for designing wireless communication systems, analyzing the 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, such as the path loss, shadowing, multipath fading, and present some models in the standards. Since the procedures of 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 on the applicability of models, and questions arise on the appropriate implementation of the models in channel simulations. Furthermore, the fast-growing wireless communication networks and services, pose more demanding on the high spectral efficiency. Numerous techniques have been used which are essentially exploiting the resources from propagation channels. For example, parallel spatial channels are resolved and utilized by the multiple-input multipleoutput (MIMO) techniques for diversity or multiplexing. Similar MIMO techniques in other domains, such as in polarizations, and in wavefronts, are invented. It is no-doubt that the future wireless system design will be more and more adaptive to the environments. Network architecture design is also getting more and more complicated in order to take the most usage of the channel. For example, the techniques of distributed antennas, massive MIMO, relay, cooperative transmission, and joint processing, all require detailed knowledge of channels in both stochastic sense and site-specific scenarios. Therefore, channel characterization based on theoretical approaches and real measurements will become more critical in the future. Considering the multiple aspects of channel, it is actually a mission impossible to write a book that is so comprehensive that every topic of channel studies is included. This book is written with the aim to cover only some aspects of the propagation channel, namely the high-resolution approach to analyze channels based on measurement data, and stochastic channel modeling using either the empirical parameters or based on simulation of scattering. The objectives of this book are three folded: first, the book provides with the fundamentals for both empirical measurementbased and theoretical scattering-based channel modeling. The topics covered are widely spread in the fields of wideband channel measurements, model parameter extraction, stochastic model generation, and theoretical channel modeling. Second, the book provides some updated channel models that can be practically used for simulations. Thus, engineers in wireless communication industry can use them to evaluate their system performance. Thirdly, this book highlights the on-going trends with some fresh research results that might be interesting for researchers when designing new systems. 1.2 The Historical Context Importance of channel characterization The statistical characteristics of the channels can significantly influence the design of wireless communication systems. For example, the path loss model established 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. The shadowing models can be used to determine the maximum and the minimum transmission power in order to avoid the blindspots in the coverage. The multipath fading models that include the fading rate and fading duration characteristics can be used to determine the packet length and the transmission rate. The delay spread models can be used to evaluate the frequency selectivity of the environment, so as to determine the coherence frequency bandwidth This is a Book Title Name of the Author/Editor c XXXX John Wiley & Sons, Ltd

14 Introduction or the separation of the orthogonal channels in the frequency domain. The 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, e.g. the cluster-based bidirectional models can be applied to determine the antenna beamwidth in beamforming applications, or calculate the degree of freedom for the channels with MIMO configurations. Stochastic models themselves are established based on extensive measurements in many environments categorized into specific types, such as outdoor, indoor, urban/suburban etc. So they are valid for the regions with similar kinds of environments. The model parameters can be used to determine many thresholds used in communication systems. For example, for frequency hopping multiple access (FHMA) systems, the frequency offsets due to the Doppler effect of channel, and the timing problems due to the multipath arriving at different time instants, can cause a certain portion of the desired signal’s energy to appear in spurious adjacent frequency bins, and consequently, the detection of 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 channel in certain environments and scenarios, the threshold level of envelop detectors can be appropriately selected. Furthermore, if the instantaneous knowledge of the channel dispersion characteristics is available, the channel can be equalized accordingly. SISO channel models The channel investigation starts at the end of 1960’s Okumura et al. (1968). At that moment, wireless systems were built for voice communications using the frequency division multiple access (FDMA). Thus, the channel characteristics of interesting is the fading distributions at particular frequencies when the single-input single-output system is considered. For the outdoor scenarios, as reported in Okumura et al. (1968),Lee and Yeh (1972),?,Schmid (1970) , 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. In Suzuki (1977), more distributions, including the Nakagami and lognormal distributions are considered to fit the empirical data. It was found that the Rayleigh distribution does not always show a good fit for most data, and the lognormal distribution seems better than the Rayleigh. A possible reason for this observation, as given in Suzuki (1977), 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. For indoor propagation environments, the SISO channel models have been established for the line-of-sight (LoS) and obstructed (OBS) scenarios, as in factory and open plan office cases Saleh and Valenzuela (1987) Rappaport et al. (1991) Rappaport and Seidel (1989) Yegani and McGillem (1991) Yegani and McGillem (1989b) Yegani and McGillem (1989a) Kozlowski et al. (2008) Seidel et al. (1989). The motivation for investigating the indoor SISO channels is to provide models for deploying radio systems in indoor which accommodate data rates up to 1 Mb/s. Such systems include the Digital European Cordless Telephone 802.41, the WLAN, i.e. IEEE 802.11 standards, as well as the communication systems for autonomous guided vehicles (AGV ˛a´rs) Rappaport et al. (1991). The interesting characteristics of the channels in indoor environments include the path loss, delay spread. It has been found that the delay spread can be several times greater in unpartitioned factory buildings as compared to partitioned office buildings Hawbaker and Rappaport (1990b). Besides the large-scale parameters, the detailed wideband characteristics of the channel, e.g. the dispersion of the channel in delay domain has been paid attention too. For example, in Hawbaker and Rappaport (1990a), the so-called “pulse overlapping” phenomenon was found which reveals that even in the line-of-sight (LoS) scenarios, the obstructed (OBS) path components can be added to the LoS path components within the resolution of transmitted pulse, resulting in the so-called multipath fading. Furthermore, resolvable rays in the time domain have been applied to modeling channels. This kind of models was called discrete models. 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), which is large when the channel delay profile exhibits several dominating discrete peaks, and is 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, 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 was also computed and used for modeling the channel Cox (1973). In addition, the trend of describing the channel properties in two dimensions appears in literature Cox (1973). The Doppler spectra

Introduction 15 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, i.e. the channel property at specific delays, became an important subject for modeling. Some important observations were obtained through extensive measurements. For outdoor urban environments, the excess delay of a channel in 900 Mhz can be up to 9 to 10 µs Cox (1973), delay spread, defined as the square root of the second central moment of the power delay profile, is 2 to 2.5 µs. The path resolved with 0.1 µs resolution exhibits Rayleigh distribution, inferring that the fading coefficients for the first arrival path can be modeled as Gaussian random process. The uncorrelated scattering among paths with different delays is confirmed by the observation that the Doppler frequency power spectra are quite different for the paths with different delays. The conclusion that the paths with different delays are uncorrelated seems more appropriate for the urban clutter environments. Some literature proposed to use correlated paths to construct the discrete models, which is contradictory with the observations in Cox (1973). For indoor manufacturing environments, Yegani and McGillem (1991) provided with the statistics of the channels in different sites in a factory under four scenarios among the settings of LoS/OBS, light/heavy clutter. It was found that the interarrival times of the paths were modeled by the Weibull distribution, the number of paths the modified Beta distribution, path gain coefficients the Rayleigh, Rician and Lognormal distributions. The values of the parameters of these distributions are reported. 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, light- and heavy-clutter scenarios. When the threshold is considered, when the threshold is greater than −10 dB, the path gain distribution follows the Rician pdf and for threshold 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 of channels for SISO has been evolved into multiple areas. For example, the polarization characteristics of channel has been investigated in 1970’s since polarization diversity was used for combating with the multipath fading property of channel. Employing orthogonally polarized channels over the same microwave link for satellite communications can obtain two times system capacity as much as 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 wireless communication system by polarization in scattering-rich environment. Channel models have been proposed that can be used to generate the channel responses with arbitrary pair of vertical and horizontal polarizations in both transmitter and receiver sites Spatial channel model for Multiple Input Multiple Output (MIMO) simulations (Release 7) (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. In ?, the correlation coefficients for both co-polarized and cross-polarized channels are studied. Some phenomena have been reveals, i.e. polarization decorrelation outperforms the spatial decorrelation in strong LOS scenario; the horizontally polarized channels are more correlated than the vertically polarized channels; the correlation of the copolarized channels increases as the Ricean K factor increases; channels have very higher correlation in the elevation domain. A strong conclusions was made that is, the cross-correlation of the cross-polarized channels are not affected by the environment, while the performance of the co-polarized channels is scenario dependent. To be added Spatial channel models Estimating the directions or bearings of incoming signals has been a research topic for years. The original objective for the study is for signal detection and estimation, including radar target tracking, component separation. The methods used for estimating directions of arrival are similar with the time-series spectral analysis methods. The former is applied specifically with the samples obtained from spatially distributed array of sensors, including antennas for receiving electromagnetic waves, and microphones for acoustic signals. The study of the arrival angles of the signals for design of the communication systems can be traced back to 1970’s. For example, in Lee and Brandt (1973) it was found from field measurements of mobile radio signals that signal arrival is concentrated in elevation angles lower than 16◦ . Based on this finding, omnidirectional antenna with vertical directivity is usually selected to increase average received signal strength

16 Introduction There are also many practical concerns which require the knowledge of spatial characteristics of channel. For example, when the 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 direction of arrival, e.g. how the energy is concentrated, what the spread of the energy in the dominant path is. Additionally, with the directional parameters, the propagation of the waves can be easily visualized when the actual constellation of the scatterers is presented for specific environments. The geometry-based channel modeling (GBSM) became flourishing in the last decade. One major reason is that the channel dispersion in the directional domains can be obtained by spectral analysis of the measurement data. The spatial-spectral analysis methods can be categorized into two classes, i.e. the spectral-based methods and the parametric-model-based methods. Theoretically, the conventional methods, such as periodogram Schuster (1898) and correlogram Chatfield (1989), belonging to the category of the spectral-based methods are not applicable in many cases due to the limit spatial aperture of the sensor array and the responses of the sensors. So the eigen-structure based methods have been widely adopted, which include the MUSIC algorithm and the variants thereof Kaveh and Barabell (1986), Stoica and Nehorai (1989), Rao (1990) Friedlander (1990), Krim et al. (1992), Jäntti (1992), Rao (1993), Krim and Proakis (1994), Asztély and Ottersten (1998), De Jong and Herben (1999),Wang et al. (2001), Salameh and Tayem (2006) and other subspace-based methods, such as the propagator method Marcos et al. (1994) Marcos et al. (1995) Tayem and Kwon (2005) Tayem and Kwon (2005), and ESPRIT (which does not result in a spectrum, but provides analytically the solutions for parameter estimates)Paulraj (1986), Jäntti (1992), Asztély and Ottersten (1998). In 1990’s, algorithms derived based on the parametric models of channel were applied to extracting the channel model parameters from the measurement data. The maximum likelihood estimator and the approximation of it with iterative estimate updating procedure can be used to estimate both the deterministic parameters and the statistical parameters of channels depending on the generic model applied. These algorithm are also called super resolution methods, as they may achieve higher resolution than the conventional spectral-based methods. Typical examples of these algorithms are the expectation-maximization (EM) algorithm Moon (1997) Frenkel and Feder (1999) Nielsen (2000) Zhang (2001), the space-alternating generalized expectation-maximization (SAGE) algorithm Fessler and Hero (1994) Fleury et al. (1999) Yin et al. (2006b) Taparugssanagorn et al. (2007) Yin et al. (2007), the Richter’s Maximum likelihood (RiMAX) algorithm Richter (2004) Richter (2005) Richter et al. (2000) Richter et al. (2003), and the variants of the SAGE algorithm by adopting models different from the widely used resolvable specular path model Bengtsson and Ottersten (2000) Yin et al. (2006a). These literature covered many aspects of the algorithms, including the impact of the antenna arrays used for data collection, the influence of the model mismatch between the usually applied resolvable specular-path model and the true scattering effect, also the analysis and comparison of these algorithms. These algorithms are applied to extract multi-dimensional parameters of channels from measurement data. The parameters include the direction of arrival, direction of departure, delay, Doppler frequency and polarization matrix of individual propagation paths. The estimates are used to construct the stochastic geometry-based or ray-based channel models, such as the well known 3GPP TR 25.996 models Spatial channel model for Multiple Input Multiple Output (MIMO) simulations (Release 7) (2007), the WINNER II spatial channel model-enhanced (SCME) WINNER II Channel Models (IST-WINNER2. Tech. Rep., 2007), the IMT-advanced channel models REPORT ITU-R M.2135 Guidelines for evaluation of radio interface technologies for IMT-Advanced (2007). In the spatial channel model, clustering of multiple paths is considered as a necessary step for generating the small-scale parameters of the channel, such as the cluster delay spread, cluster angular spread, and the time-variant behavior of the clusters. How to appropriately cluster the multiple propagation paths has been discussed in literature Czink et al. (2005a) Czink et al. (2005b). First visual-inspection-based clustering methods were proposed Czink et al. (2007c), which is impractical for a large amount of measurement data. Moreover, the clustering results may not be unique when the users have different opinion about the clusters. The automatic clustering methods were alternatively designed which require minimum interactions of users. These methods make use of the so-called multipath component distance measure, or environment characterization metric to group the paths into cluster Czink et al. (2005c) Czink et al. (2005b) Czink et al. (2006). Readers are referred to Czink (2007) for the detailed description of various clustering methods and their performance. The multipath clustering concept has also been extended to the modeling of time-variant channelsCzink et al. (2007a) Czink et al. (2007b) Xiao et al. (2007) Xiao and Burr (2008). The objective of introducing the timevariant clusters is to reduce the computational complexity when generating spatial-correlated time-variant channel realizations or channel matrices. The parameters of the clusters, especially the centroid of clusters are tracked through

Introduction 17 consecutive channel snapshots Czink and Galdo (2005) Czink et al. (2007b). It was found in Czink et al. (2007b) that for both outdoor and indoor scenarios, clusters can be easily tracked. The histogram of the logarithmic cluster lifetime follows an exponential distribution of the cluster lifetime for the outdoor scenarios. Other channel models The future generations of wireless communication systems employ new techniques that rely on channel modeling for more complex network constellations. For example, designing of the distributed antenna, cooperative relay and joint processing systems or algorithms require the models of co-existing multi-link channels. Some preliminary works have been done on the multi-link correlation channel models recently Yin et al. (2012b) Yin et al. (2012c) Yin et al. (2012d) Yin et al. (2012a) Yin et al. (2011). Other channel models also exist for the non-stationary scenarios, the distributed scenarios. Some models focus on specific behavior of the channels, such as the reciprocity behavior of channels with respect to time and frequency, polarization characteristics, the interferences, and the LoS and NLoS probabilities, etc.. To be added Books and PhD theses describing the channels There are already some books dedicated on channel investigations, such as Saunders (1999), Parsons (2000), Pätzold (2002), Durgin (2003) and Koivunen (Dec 2007). A brief review of these books is provided below. Besides books, some PhD theses focus on different perspectives of the channels. These theses are either open for downloading or can be obtained by contacting the authors. A brief introduction of these theses is provided in the following: To be added 1.3 Book Outline This book contains totally ten chapters. Chapter 2 “Characterization of Propagation Channels” begins with introducing three phenomena of fading in wireless channels, i.e. path loss, shadowing, and multipath fading. Then the stochastic characterizations of path loss, shadowing and multipath fading are described. Following that, we proceed to emphasize the duality relationship between the selectivity and dispersion of multipath fading, and also explain the definition of the wide sense stationary uncorrelated-scattering (WSSUS) assumption and its applicability in reality. In this chapter, a review of propagation channel modeling is provided which generally describes different modeling approaches with the aim of accurately and/or efficiently generate channel impulse responses with desired channel characteristics. These channel models are categorized as two main classes, i.e. MIMO channel models and V2V channel models. Chapter 3 “Generic channel models” introduces the basic mechanisms of radio propagation, the representation of channels in terms of multi-dimensional spread functions, and the generic models usually applied for channel parameter estimation. These generic models include the specular-path model, dispersive-path model, time-evolution model for the path parameters, power spectral density models for individual components. Furthermore, the influence of system configurations e.g. the amplify-and-forward relay systems on the format of the generic models is also described. Finally, the model of received signal in e.g. channel sounding context is given. Chapter 4 “Geometry based stochastic channel modeling” introduce the difference between the geometry-based deterministic modeling approach and geometry-based stochastic modeling approach, the details of the latter for the regular-shaped and irregular-shaped scenarios. Furthermore, the simulation methods of the theoretical/mathematical reference model in reality are introduced. Chapter 5 “Channel measurements” introduces the methodologies, equipments, and procedures of measuring the impulse responses of propagation channels. Besides the general description of channel measurements, we go one step further to discuss the influences of the imperfections occurring during the calibration of the equipments on the measurement results. For example, we analyze the impact of the existence of time-variant phase noises and the inconsistency between the measured and real radiation pattern measurements, and also for the first time, provide experimental analysis for how the directionality of the radiation of antennas influences the parameter estimation results. All these studies are based on experimental data. Of course, we focus on introducing the phenomena, and only briefly describe several solutions available at present. The readers are encouraged to discover more solutions for these important issues

18 Introduction Chapter 6 “Deterministic channel parameter estimation” focuses on the high-resolution parameter estimation algorithms derived based on the generic specular-path model for extracting the parameters of individual path components from measurement data. Besides the traditional SAGE and RiMAX algorithms, which have been used extensively for parameter estimation in MIMO measurement-based channel modeling, we also introduce some newly developed estimation methods, which are promising to be used in the nearest future for accurate channel modeling. For the time-varying scenarios where the path parameters evolve with respect to time, a tracking scheme based on the particle filter concept is elaborated. For each method introduced, we include the results of performance evaluation that were carried out by processing the real measurement data. Chapter 7 “Statistical channel parameter estimation” describes another category of parameter estimation methods which make use of the generic models for the power spectral density of the channel, to estimate the statistical parameters, such as the second moments of the channel/channel components. Two methods are introduced, which are the generalized array manifold (GAM) model-based approach and the power spectral density-based approach. These methods, although not been widely adopted, can result in more accurate estimates of channel statistics. We also describe the practical limitation of the methods when being used in reality. Chapter 8 “Measurement based statistical channel modeling”systematically describes the modeling procedures based on the measurements in details. Both the common issues in the modeling, such as the clustering algorithms, data segmentation, and specific approaches for some new topics of channel modeling, such as the non-stationarity modeling, relay and CoMP channel modeling are discussed. Chapter 9 “Practices: channel modeling for modern communication systems”, as the last chapter of the book, provides the examples of models established by using the methods introduced throughout the book. Those examples cover many typical scenarios which have been popular for channel modeling in recent years. The readers can use this chapter as a collection of models recently developed for MIMO, vehicular, relay, CoMP, multilink channels. Students can also make their own practice for other scenarios based on the procedures presented in these examples. Bibliography Andrews MR, Mitra PP and deCarvalho R 2001 Tripling the capacity of wireless communications using electromagnetic polarization. Nature pp. 316– 318. Asztély D and Ottersten B 1998 The effects of local scattering on direction of arrival estimation with MUSIC and ESPRIT Proceedings of IEEE International Conference on Acoustics, Speech, Signal Processing (ICASSP), vol. 6, pp. 3333–3336. Bengtsson M and Ottersten B 2000 Low-Complexity Estimators for Distributed Sources. Signal Processing, IEEE Transactions on 48(8), 2185–2194. Cox D 1973 910 mhz urban mobile radio propagation: Multipath characteristics in new york city. Communications, IEEE Transactions on 21(11), 1188–1194. Cox D and Leck R 1975 Correlation bandwidth and delay spread multipath propagation statistics for 910-mhz urban mobile radio channels. Communications, IEEE Transactions on 23(11), 1271 – 1280. Cox DC 1972 Delay doppler characteristics of multipath propagation at 910 mhz in a suburban mobile radio environment. IEEE Transactions on Antennas and Propagation AP-20(5), 625U635. ˝ Czink N 2007 The Random-Cluster Model - A Stochastic MIMO Channel Model for Broadband Wireless Communication Systems of the 3rd Generation and Beyond PhD thesis Technology University of Vienna, Department of Electronics and Information Technologies. Czink N and Galdo GD 2005 Validating a novel automatic cluster tracking algorithm on synthetic ilmprop time-variant mimo channels. Technical Report TD-05-105, COST273. Czink N, Bonek E, Yin X and Fleury BH 2005a Cluster angular spreads in a MIMO indoor propagation environment Proceedings of the 16th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC’05), vol. 1, pp. 664–668, Berlin, Germany. Czink N, Cera P, Salo J, Bonek E, Nuutinen J and Ylitalo J 2005b Automatic clustering of MIMO channel parameters using the multi-path component distance measure Proceedings of Wireless Personal Multimedia Communications WPMC’05, Aalborg, Denmark. Czink N, Cera P, Salo J, Bonek E, Nuutinen J and Ylitalo J 2005c Automatic clustering of MIMO channel parameters using the multi-path component distance measure Proceedings of Wireless Personal Multimedia Communications WPMCâA˘Z05 ´ , Aalborg, Denmark. Czink N, Galdo GD, Yin X and Meklenbrauker C 2006 A novel environment charaterisation metric for clustered MIMO channels used to validate a SAGE parameter estimator Proceedings of the 15th IST Mobile & Wireless Communication Summit, Myconos, Greece. Czink N, Tian R, Wyne S, Eriksson G, Zemen T, Ylitalo J, Tufvesson F and Molisch A 2007a Cluster parameters for time-variant mimo channel models Antennas and Propagation, 2007. EuCAP 2007. The Second European Conference on, pp. 1 –8. Czink N, Tian R, Wyne S, Tufvesson F, Nuutinen JP, Ylitalo J, Bonek E and Molisch A 2007b Tracking time-variant cluster parameters in mimo channel measurements Communications and Networking in China, 2007. CHINACOM ’07. Second International Conference on, pp. 1147 –1151. Czink N, Yin X, Özcelik H, Herdin M, Bonek E and Fleury B 2007c Cluster characteristics in a MIMO indoor propagation environment. IEEE Transactions on Wireless Communications 6(4), 1465–1476. De Jong Y and Herben M 1999 High-resolution angle-of-arrival measurement of the mobile radio channel. Antennas and Propagation, IEEE Transactions on 47(11), 1677–1687. Durgin GD 2003 Space-Time Wireless Channels. Pearson Education, Upper Saddle River, NJ. (ed. Chatfield C) 1989 The Analysis of Time Series: An Introduction fourth edition edn. Chapman and Hall. (ed. Pätzold M) 2002 Mobile Fading Channels. Wiley. Fessler JA and Hero AO 1994 Space-alternating generalized expectation-maximization algorithm. IEEE Trans. on Signal Processing 42(10), 2664– 2677. Fleury BH, Tschudin M, Heddergott R, Dahlhaus D and Pedersen KL 1999 Channel parameter estimation in mobile radio environments using the SAGE algorithm. IEEE Journal on Selected Areas in Communications 17(3)(3), 434–450. Frenkel L and Feder M 1999 Recursive Expectation-Maximization algorthms for time-varying parameters with applications to multiple target tracking. IEEE Transactions on Signal Procesing 47(2), 306–320

Introduction 19 Friedlander B 1990 A sensitivity analysis of the music algorithm. Acoustics, Speech, and Signal Processing [see also IEEE Transactions on Signal Processing], IEEE Transactions on 38(10), 1740–1751. Hawbaker D and Rappaport T 1990a Indoor wideband radio propagation measurement system at 1.3 ghz and 4.0 ghz Vehicular Technology Conference, 1990 IEEE 40th, pp. 626 –630. Hawbaker D and Rappaport T 1990b Indoor wideband radiowave propagation measurements at 1.3 ghz and 4.0 ghz. Electronics Letters 26(21), 1800 –1802. Jäntti TP 1992 The influence of extended sources on the theoretical performance of the MUSIC and ESPRIT methods: narrow-band sources. Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing, pp. II–429–II–432. Jeon K, Hui B, Chang K, Park H and Park Y 2012 Siso polarized flat fading channel modeling for dual-polarized antenna systems Information Networking (ICOIN), 2012 International Conference on, pp. 368 –373. Joo J, Moon S, Yoon Y and Kim K 2003 Effects of fast frequency hopping multiple access systems due to the frequency and timing offset under rayleigh fading Wireless Communications and Networking, 2003. WCNC 2003. 2003 IEEE, vol. 1, pp. 126 –131 vol.1. Kaveh M and Barabell AJ 1986 The statistical performance of the MUSIC and the minimum-norm algorithms in resolving plane waves in noise. IEEE Transactions on Acoustics, Speech, and Signal Processing ASSP-34(2), 331–342. Koivunen J Dec 2007 Characterization of MIMO Propagation Channel in Multilink Scenarios. Helsinki University of Technology, Dissertation. Kozlowski S, Szumny R, Kurek K and Modelski J 2008 Statistical modelling of a wideband propagation channel in the factory environment Wireless Technology, 2008. EuWiT 2008. European Conference on, pp. 190 –193. Krim H and Proakis J 1994 Smoothed eigenspace-based parameter estimation. Automatica 30(1), 27–38. Krim H, Forster P and Proakis J 1992 Operator approach to performance analysis of root-music and root-min-norm. IEEE Transactions on Signal Processing 40, 1687 – 1696. Lee W and Yeh Y 1972 Polarization diversity system for mobile radio. Communications, IEEE Transactions on [legacy, pre - 1988] 20(5), 912–923. Lee WY and Brandt R 1973 The elevation angle of mobile radio signal arrival. Vehicular Technology, IEEE Transactions on 22(4), 110 – 113. Marcos S, Marsal A and Benidir M 1994 Performances analysis of the propagator method for source bearing estimation Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), vol. IV, pp. 19–22. Marcos S, Marsal A and Benidir M 1995 The propagator method for source bearing estimation. Signal Processing 42, 121–138. Moon T 1997 The expectation-maximization algorithm. IEEE Signal Processing Mag. 13(6), 47–60. Nielsen SF 2000 The stochastic em algorithm: estimation and asymptotic results. Bernoulli 6(3), 381–570. Okumura Y, Ohmori E, Kawano T and Fukuda K 1968 Field strength and its variability in vhf and uhf land-mobile radio services. Review of the Electrical Comm. Lab. Parsons JD 2000 The Mobile Radio Propagation Channel 2 edn. John Wiley and Sons, Ltd., Chichester, England. Paulraj, A.; Roy RKT 1986 A subspace rotation approach to signal parameter estimation. Proceedings of the IEEE 74, 1044–1046. Rao, B.D.; Hari K 1990 Effect of spatial smoothing on the performance of noise subspace methods. Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing 5, 2687 –2690. Rao, Bhaskar D.; Hari K 1993 Weighted subspace methods and spatial smoothing. analysis and comparison. IEEE Transactions on Signal Processing. Rappaport T and Seidel S 1989 Multipath propagation models for in-building communications Mobile Radio and Personal Communications, 1989., Fifth International Conference on, pp. 69 –74. Rappaport T, Seidel S and Takamizawa K 1991 Statistical channel impulse response models for factory and open plan building radio communicate system design. Communications, IEEE Transactions on 39(5), 794 –807. Richter A 2004 RIMAX - a flexible algorithm for channel parameter estimation from channel sounding measurements. Technical Report TD-04-045, COST 273, Athens, Greece. Richter A 2005 Estimation of Radio Channel Parameters: Models and Algorithms PhD thesis Technische Universität Ilmenau, ISBN 3-938843-02-0 Ilmenau, Germany. Richter A, Hampicke D, Sommerkorn G and Thoma R 2000 Joint estimation of DoD, time-delay, and DoA for high-resolution channel sounding Proceedings of IEEE 51st Vehicular Technology Conference (VTC-Spring), vol. 2, pp. 1045 –1049. Richter A, Landmann M and Thomä RS 2003 Maximum likelihood channel parameter estimation from multidimensional channel sounding measurements Proceedings of the 57th IEEE Semiannual Vehicular Technology Conference (VTC), vol. 2, pp. 1056 –1060. Salameh A and Tayem N 2006 Conjugate music for non-circular sources Proceedings of 2006 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Saleh A and Valenzuela R 1987 A statistical model for indoor multipath propagation channel. IEEE Journal of Selected Areas in Communications 5(2), 128–137. Saunders SR 1999 Antennas and Propagation for Wireless Communication Systems. John Wiley & Sons, Chichester, England. Schmid H 1970 A prediction model for multipath propagation of pulse signals at vhf and uhf over irregular terrain. Antennas and Propagation, IEEE Transactions on 18(2), 253 – 258. Schuster A 1898 On the investigation of hidden periodicities with application to a supposed 26 day period of meteorological phenomena. Terrestrial Magnetism and Atmospheric Electricity 3, 13–41. Seidel S, Takamizawa K and Rappaport T 1989 Application of second-order statistics for an indoor radio channel model Vehicular Technology Conference, 1989, IEEE 39th, pp. 888 –892 vol.2. Stoica P and Nehorai A 1989 Music, maximum likelihood, and cramer-rao bound. Acoustics, Speech and Signal Processing 37, 720 –741. Suzuki H 1977 A statistical model for urban radio propagtion channel. IEEE Transactions on Communication Systems 25, 673–680. Taparugssanagorn A, Alatossava M, Holappa VM and Ylitalo J 2007 Impact of channel sounder phase noise on directional channel estimation by space-alternating generalised expectation maximisation. Microwaves, Antennas Propagation, IET 1(3), 803 –808. Tayem N and Kwon H 2005 L-shape 2-dim. arrival angle estimation with propagator method. AP 53, 1622–1630. REPORT ITU-R M.2135 Guidelines for evaluation of radio interface technologies for IMT-Advanced REPORT ITU-R M.2135 Guidelines for evaluation of radio interface technologies for IMT-Advanced 2007. Spatial channel model for Multiple Input Multiple Output (MIMO) simulations (Release 7) Spatial channel model for Multiple Input Multiple Output (MIMO) simulations (Release 7) 2007. WINNER II Channel Models WINNER II Channel Models IST-WINNER2. Tech. Rep., 2007. Turin G, Clapp F, Johnston T, Fine S and Lavry D 1972 A statistical model of urban multipath propagation channel. IEEE Transactions on Vehicular Technology 21, 1–9. Wang Y, Chen J and Fang W 2001 Tst-music for joint doa-delay estimation. IEEE Trans. Signal Processing 46, 721–729. Xiao H and Burr A 2008 Reduced-complexity cluster modeling for time-variant wideband mimo channels Wireless Conference, 2008. EW 2008. 14th European, pp. 1 –5

20 Introduction Xiao H, Burr A and de Lamare R 2007 Reduced-complexity cluster modelling for the 3gpp channel model Communications, 2007. ICC ’07. IEEE International Conference on, pp. 4622 –4627. Yegani P and McGillem C 1989a A statistical model for line-of-sight (los) factory radio channels Vehicular Technology Conference, 1989, IEEE 39th, pp. 496 –503 vol.2. Yegani P and McGillem C 1989b A statistical model for the obstructed factory radio channel Global Telecommunications Conference, 1989, and Exhibition. Communications Technology for the 1990s and Beyond. GLOBECOM ’89., IEEE, pp. 1351 –1355 vol.3. Yegani P and McGillem C 1991 A statistical model for the factory radio channel. Communications, IEEE Transactions on 39(10), 1445 –1454. Yegani P and McGillem C 1993 Fh-mfsk multiple-access communications systems performance in the factory environment. Vehicular Technology, IEEE Transactions on 42(2), 148 –155. Yin X, Fu Y, Liang J and Kim MD 2011 Investigation of large- and small-scale fading cross-correlation using propagation graphs Proceedings of International Symposium on Antennas and Propagation, SB01-1004, Jeju, Korea. Yin X, Liang J, Chen J, Park JJ, Kim ˛E MD and Chung HK 2012a Empirical models of cross-correlation for small-scale fading in co-existing channels Proceedings of Asia-Pacific Communication Conference 2012, Jeju, Korea. Yin X, Liang J, Fu Y, Yu J, Zhang Z, Park JJ, Kim MD and Chung HK 2012b Measurement-based stochastic modeling for co-existing propagation channels in cooperative relay scenarios Future Network Mobile Summit (FutureNetw), 2012, pp. 1 –8. Yin X, Liu L, Nielsen D, Pedersen T and Fleury B 2007 A sage algorithm for estimation of the direction power spectrum of individual path components, pp. 3024–3028. Yin X, Pedersen T, Czink N and Fleury B 2006a Parametric characterization and estimation of bi-azimuth and delay dispersion of individual path components, pp. 1–8. Yin X, Pedersen T, Czink N and Fleury B 2006b Parametric characterization and estimation of bi-azimuth dispersion path components Signal Processing Advances in Wireless Communications, 2006. SPAWC ’06. IEEE 7th Workshop on, pp. 1 –6. Yin X, Zeng Z, Cheng X and Zhong Z 2012c Empirical modeling of cross-correlation for spatial-polarimetric channels in indoor scenarios Personal Indoor and Mobile Radio Communications (PIMRC), 2012 IEEE 23st International Symposium on. Yin X, Zhou X, Zhang Z, Kim MD and Chung HK 2012d Parametric modeling of the cross-correlation for large-scale-fading of propagation channels Vehicular Technology Conference (VTC Spring), 2012 IEEE 75th, pp. 1 –5. Zhang, Y.; Brady MSS 2001 Segmentation of brain mr images through a hidden markov random field model and the expectation-maximization algorithm. Medical Imaging, IEEE Transactions on 20(1), 45–57