Performance Evaluation of OFDM Based Wireless System Working in the Frequency Band of 60 GHz. CHAPTER 3. PROPAGATION AND CHANNEL MODELING OF 60 GHz

Transcription

1 CHAPTER 3 PROPAGATION AND CHANNEL MODELING OF 60 GHz 27

2 3.1 Introduction A communication channel represents a physical medium between the transmitter and the receiver. The channel model is a representation of the input-output relationship of the channel in mathematical form. The channel imposes constraints on the communication quality measures, affecting the system s transmission rate, error probability and the distance over which the system can operate. One of the challenges in channel modeling is the translation of a detailed physical propagation model into a form that is suitable for simulation. This chapter introduces 60 GHz frequency band and discusses various aspects such as standardization, front end technology. This chapter elaborates various aspects of channel modeling and recommends a suitable channel model for 60 GHz band. 3.2 Basics of Radio wave Propagation There are many different types of interactions between electromagnetic waves, the antennas which transmit and receive them, and the environments through which they propagate. All of these effects must be considered in order to understand and analyze the performance of wireless communication systems. The indoor radio channel differs from the traditional mobile radio channel in two ways. The distances are much smaller, and the environment can vary much more for a smaller range of transmitter and receiver separation. Indoor propagation is strongly influenced by specific features such as the layout of the building, construction materials, and the building type. Still, indoor propagation is governed by the same mechanisms as outdoors: reflection, transmission, diffraction, scattering, etc Reflection, Diffraction and Scattering Reflection occurs when an electromagnetic wave impinges upon an object that has very large dimensions compared to the wavelength of the propagating wave. Reflection occurs from the surface of the ground, from walls and from furniture. While reflecting the wave may also be partially refracted. The coefficients of reflection and refraction are functions of the material properties of the medium and generally depend on the wave polarization, the angle of incidence and the frequency of the propagating wave. 28

3 Diffraction occurs when the propagation path between the transmitter and receiver is obstructed by a surface that has a sharp edge. The obstructing surface causes waves to bend around the obstacle, even if there is no line-of-sight path between transmitter and receiver. Diffraction depends on the geometry of the object, as well as the polarization of the incident wave at the point of diffraction. The phenomenon of diffraction can be understood by using Huygens Principle, which states that each element of a wave front can be considered as a point source for the production of secondary wavelets, and that these wavelets produce a new wave front in the direction of propagation. Diffraction is caused by the propagation of secondary waves into areas shadowed by the obstacle [44]. Scattering is a process where a wave is forced to deviate from a straight path by one or more localized non-uniformities in the medium through which it passes. This also includes deviation of reflected radiation from the angle predicted by the law of reflection. Reflections that undergo scattering are often called diffuse reflections, while reflections following the Snell's law are called specular reflections. Scattering causes the energy of a radio wave to spread out in many directions and the wave is also depolarized. Scattered waves are produced by rough surfaces, small objects or other irregularities in the propagation path, when the number of these irregularities is large. The degree of scattering depends on the angle of incidence and the roughness of the surface in comparison to the wavelength Multi path Channel Characteristics The propagation channel is the part of a communication link between the transmitter and receiver antennas. Millimeter-wave (mm-wave) indoor radio channel, which is considered in this work, is essentially a multipath channel. If a single pulse is transmitted over a multipath channel then the received signal will appear as a pulse train, with each pulse in the train corresponding to a line-of-sight component, a distinct multipath component from a distinct scatterer and diffuse multipath from a group of scatterers. The time delay spread of a multipath channel can result in significant distortion of the received signal. Multipath channel parameters such as path loss, delay spread and angle spread describe the statistical properties of the channel. 29

4 3.2.3 Path Loss Path loss is the average of the ratio of the transmitted power to the received power between two antennas, usually expressed in decibels. It includes all kinds of losses associated with interactions between the propagating wave and any objects between the transmitting and receiving antennas. One of the main goals of propagation modeling is to predict path loss as accurately as possible. The path loss exponent, n is a measure of decay in signal power with distance, d, according to 1. Path loss obeys the distance power law, given in decibels as ( ) = ( ) +10 log( ) (3.1) where is the reference distance which is determined from measurements close to the transmitter. ( ) is the average measured energy at the reference distance [44] and ( ) depends mainly on the frequency. Figure 3.1: Free space path loss against distance for frequencies commonly used At 60 GHz path loss is several orders greater than at lower frequencies that are commonly used in wireless communications. Figure 3.1 shows theoretical free-space path loss for 3.5, 5 and 60 GHz Fading In a real propagation environment, path loss is not constant for a given transmitter and receiver distance. Objects along a propagation path at a given distance will be different for every path, causing variation with respect to the value given by simple path loss models. Some paths will suffer increased loss, while others will be less obstructed 30

5 and has increased signal strength. This phenomenon is called shadowing or slow fading. The effects of shadowing can be included in the path loss model in Equation (3.1) by expanding it to ( ) = ( ) +10 log( ) + (3.2) where is a zero mean Gaussian-distributed random variable in db with standard deviation equal to (also in db). Even after path loss and shadowing have been predicted for particular locations there is still significant variation in the received signal when an antenna is moved over relatively small distances. The reason for this is that the received signal is a sum of many contributions coming from different directions. Since the phases are random, the sum amplitude varies in a wide range, as much as 30 to 40 db. This phenomenon is called fast fading and the signal variation is so rapid that it can only be usefully predicted by statistical means Channel Impulse Response Impulse response is a wideband channel characteristic that can be used to derive the information necessary to simulate and analyze any type of radio transmission through the channel. This is because a mobile radio channel may be modeled as a linear system with a time-variant finite impulse response. For a given time t, let ( ) represent the input signal. Then the received signal ( ) can be expressed as a convolution of ( ) with the impulse response (, ) where is the delay variable. ( ) = ( ) (, ) (, ) (, ) (3.3) It is useful to discretize the multipath delay axis of the impulse response into equal time delay segments called excess delay bins, where each bin has a time delay width of =. All multipath signals received within the i th bin are represented by a single multipath component having delay. = 0 is the time delay of the first multipath component to arrive to the receiver, after normalizing the 31

6 propagation delay between the transmitter and receiver. is the excess delay of the i th multipath component as compared to the first arriving component Power Delay Profile The power delay profile (PDP) gives the intensity of a signal received through a multipath channel as a function of time delay. It is defined as ( ) = [ (, ) ] (3.4) 2 where E is the ensemble average. Different multipath components can be detected as peaks in a PDP. The PDP can be used to define the following parameters: Total excess delay: the difference between the delay of the first and last arriving bin. The last arriving bin can be defined as the greatest excess delay during which multipath power is at a certain level above the noise floor, or the delay during which power falls to X db below the maximum. Mean excess delay: the first moment of the power delay profile, corresponding to the mean value of the PDP is mean excess delay. RMS delay spread: the second moment, or variance, of the profile is RMS delay spread. This considers the relative powers of the bins and their delays, making it a better indicator of system performance. It has been shown that under some circumstances, the error probability due to delay dispersion is proportional to the RMS delay only, however, this does not apply to all cases. Coherence Bandwidth: the parameter used to characterize the channel in the frequency domain. The rms delay spread is inversely proportional to coherence bandwidth, although their exact relationship is a function of the exact multipath structure. Coherence bandwidth is a statistical measure of the range of frequencies over which the channel can be considered "flat", a channel which passes all spectral components with approximately equal gain and linear phase. Spectral analysis techniques and simulation are required to determine the exact impact that time varying multipath has on a particular transmitted signal. Therefore, accurate multipath channel models must be used in the design and simulation of wireless systems. 32

7 Coherence Time: The transmission channel can vary over time due to movements of objects and persons in the environment or moving antennas at the transmitter and/or receiver which results in a spectrum broadening. Doppler spread is a measure of the spectral broadening caused by the time rate of change of the mobile radio channel. The amount of spectral broadening depends on the relative velocity of the mobile, and the angle between the direction of motion of the mobile and direction of arrival of the scattered waves. If the baseband signal bandwidth is much greater than Doppler spread, the effects of Doppler spread are negligible at the receiver. Coherence time, is the time domain dual of Doppler spread and is used to characterize the time varying nature of the frequency dispressiveness of the channel in the time domain. The Doppler spread and coherence time are inversely proportional to one another. 3.3 Channel Models Channel models are used to predict the behavior of radio channels in specific environments without having to measure them. Channel models can be used to simulate the channels for the testing of the communication system prototypes. The channel models are available for different environments, both indoor and outdoor. Indoor channel models can be divided based on the type of building, such as residential, industrial and office environments, or based on the scale of the propagation channel, for example, within a room, within a floor or between floors. There are three kinds of models for radio wave propagation as follows, Empirical (or stochastic) models are based on measurement data. They are described with only a few parameters using statistical properties. This has the advantage of taking into account all propagation factors, both known and unknown [46]. Deterministic models are site-specific. They estimate radio wave propagation analytically. They require huge amounts of accurate geometry information about the location and very high computational effort. Deterministic modeling methods include the finite difference time domain (FDTD) [46], the method of moments (MoM) [47], and ray-tracing (RT) [47]. Semi-deterministic models are based on empirical models with deterministic aspects, combining both computer simulation results and measurements [48] 33

8 For outdoor environment commonly used channel models are Hata model (150 MHz to 1500MHz), COST 321 model (1 GHz to 3 GHz) or Erceg Model (1 GHz to 3 GHz.) Commonly used indoor propagation models are Log-distance path loss model, Ericson Multiple Break point model, site specific two dimensional or three dimensional ray tracing model and Saleh-Valenzuela model. For 60 GHz band literature supports Saleh- Valenzuela model as a generic statistical model for indoor multipath propagation. 3.4 Saleh-Valenzuela Model: Generic Channel Model Small-scale fading (i.e. fast fading) is caused by the multipath signals that arrive at the receiver with random phases. It causes rapid changes in signal amplitude over a small distance (less than 10 wavelengths). Over this small local area, the smallscale fading is approximately superimposed on the constant large-scale fading. The generic channel model representation for the 60 GHz channel is based on clustering phenomenon. The clustering observed in both the temporal and spatial domains from measurement data available [63, 67, 70, 71], a generic 60 GHz channel model that takes clustering into account is reasonable choice since it can always be reduced to the conventional single cluster channel model[68, 72, 73]. The proposed cluster model is based on the Saleh Valenzuela (SV) model is based on a clustering phenomenon observed in their experimental data. In all of their observations, the received signal rays (the arrivals) arrived in one or two large groups (clusters) within a 200 ns observation window. It was observed that the second clusters were attenuated in amplitude, and that rays within a single cluster decayed with time too. Thus, a channel consists in a random spatial distribution of ray clusters. Figure 3.2 illustrates this, showing the decay envelope of channel. A cluster comes from various reflection area: a wall, a piece of furniture, a board, a computer or anything else in the environment of propagation; as a consequence, the size of the clusters depends on the nature and layout of the objects distributed in the room. 34

9 Figure 3.2: A typical power delay profile based on the Saleh-Valenzuela model The complex baseband directional channel impulse response (CIR) is given by (,, ) =,,,, (3.5) where (. ) is the Dirac delta function, is the total number of clusters and is the total number of rays in the l th cluster. The scalars,,,,, and, denote the complex amplitude, time of arrival (ToA), angle of arrival (AoA), angle of departure (AoD) of the k th ray of the l th cluster, respectively. The AoA and AoD can consist either of azimuth domain, elevation domain, or both domains. Similarly, the scalars, and represent the mean ToA, mean AoA and mean AoD of the of the l th cluster. The key assumption used in deriving Equation (3.5) is that the spatial and temporal domains are independent and thus uncorrelated. The measurement have shown a correlation between these two domains, which can be modeled by two joint probability density functions. [76] Each multipath in Equation (3.5) will experience distortions due to the frequency dependence of the scatterers. The frequency dependence and correlation of spatial and temporal domains are so far not considered in any 60 GHz channel model due to lack of measurement results. For very high-speed transmission, the channel described in Equation (3.5) can be assumed to remain approximately static over tens to thousands of symbols. This is also called block fading channel. This is mainly due to the low speed (up to 1.5 m/s) movement in indoor environment. This yields a maximum Doppler frequency ( ) of 300 Hz. The coherence time yields approximately 0.6 msec, considering the relationship between coherence time ( ) and, = 9 16 [64]. The coherence time calculation assumes uniform 35

10 multipath arrival over [0, 2 ], which is usually valid for cellular systems but might be doubtful at 60 GHz. These two schemes use vastly different antenna and propagation characteristics. Measurement results show that the coherence time where the correlation starts to fall below 0.5 is no greater than 50 msec for all the measurements scenarios in which people walked along a clear LOS path at a speed of 1.7 m/sec. Such a coherence time is significantly larger than the symbol time and corresponds to at least a few superframes per beacon interval in IEEE and IEEE [64] Number of Clusters The number of clusters is an important parameter for the channel models. Different definitions of clustering appear in the literature. Clustering is defined is a group of rays arriving at approximately the same time and angle. Various methods have been derived for cluster identification, ranging from simple visual inspection [74, 75] to advanced signal processing such as kernel density estimation [76]. Different results have been obtained to describe the numbers of clusters for wideband systems. The number of clusters is Poisson distributed and can be fully characterized by a mean number of clusters, L. [66] The available literature on 60 GHz reports diverse findings for number of clusters. Analysis of the measurement results indicates that only a single cluster can be observed. [68, 69, 72, 73] Furthermore, IEEE c measurement data for various environments and scenarios shows that L does not follow a specific distribution; mean number of clusters can be calculated by visual inspection. Values range typically from 3 to 14. These different findings are due to number of objects causing scattering of wave in the environment. When the environment under consideration has more furniture, a larger number of clusters would be expected as a result of superstructure (such as walls, furniture, computers and doors) [77]. Higher bandwidth provides higher resolution that can resolve more multipath components. For higher bandwidths clusters can propagate more easily in the channel Arrival Times Under the assumption that the delay and angular domains can be modeled independently, the ToA of the generic channel model described in Equation (3. 5) relies 36

11 on two sets of parameter. Inter-cluster parameters, {, } characterizes the cluster and intra-cluster parameters,,,, characterizes the multipath components. The cluster arrival and ray arrival time distributions are described by two Poisson processes. According to this model, cluster inter-arrival times and ray intra-arrival times are given by two independent exponential joint probability density functions. The cluster arrival time for each cluster is an exponentially distributed random variable conditioned on the cluster arrival time of the previous cluster, which can be expressed as ( ) = [ ( )] for l >0 (3.6) where is the cluster arrival rate. Similarly, the ray arrival time for each ray is an exponentially distributed random variable conditioned on the ray arrival time of the previous ray given by ( ) = ( ), for k >0, (3.7) where is the ray arrival rate. In the classical SV model, T 0 and 0, l is assumed to be zero and all arrival times are relative with respect to the delay of the first path Power Delay Spectrum The PDS of a channel, more commonly called the power delay profile (PDP), is the average power of the channel as a function of the excess delay with respect to the first arrival path. A number of important parameters can be derived by analyzing the PDP. In particular, the mean excess delay, root mean squared (rms) delay spread, timing jitter [67] can be obtained from the PDP of the channel. The most commonly used parameter is the rms delay spread, which is a second moment of the PDP that statistically measures the time dispersion of a channel. The rms delay spread, denoted is inversely proportional to the coherence bandwidth of a channel, which determines whether a system is narrowband or wideband with respect to its channel. A narrowband system occurs when is less than the symbol period (or when is larger than the signal bandwidth) of the system. This results in a flat fading channel. The reverse is true for a wideband system when >Ts (or when is less than the signal bandwidth), which results in a frequency selective channel. A flat fading channel reduces the received signal-to-noise ratio due to 37

12 deep fading, while a frequency selective channel causes inter-symbol interference that leads to irreducible bit error rate (BER) performance. Hence, determines the maximum transmission data rate in the channel without equalization and serves as a key design parameter for orthogonal frequency division multiplexing systems. For indoor scenario the rms delay spread is strongly decided by the room superstructure. The rms delay spread increases as the size of the room and the density of objects present in the room increase. For larger rooms with more objects, it takes longer for the multipath to reach the receiver, prolonging the rms delay spread. The increment in rms delay spread becomes more pronounced in the presence of more reflective objects such as metal than absorptive objects such as wood. Secondly, the rms delay spread decreases as the directivity of the transmitting and receiving antennas increases. As the misalignment increases, the rms delay spread will start to increase even if directive antennas are used at transmitter and receiver. Ray tracing results show that in the case of perfect alignment, rms delay spread decreased by more than 50% [45]. 3.5 Introduction to 60 GHz Band Since the first wireless transatlantic radio wave transmission demonstration by Marconi, in 1901 (based on long wave), wireless communications have undergone tremendous growth. Wireless communication was first used mainly by military and shipping companies. Later quickly expanded into commercial use such as commercial broadcasting services (such as shortwave, AM and FM radio, terrestrial TV), cellular telephony, and global positioning service (GPS), wireless local area network (WLAN), and wireless personal area network (WPAN) technologies. Today, these wireless communications systems have become an integral part of daily life and continue to evolve in providing better quality and user experience. One of the recent emerging wireless technologies is mm-wave technology. The mm-wave technology has been known for many decades, but has mainly been deployed for military applications. Over the past decade, advances in process technologies and low cost integration solutions have made mm-wave a technology to watch and begun to attract a great deal of interest from academia, industry and standardization bodies. In very broad terms, mmwave technology is concerned with that part of the electromagnetic spectrum between 30 and 300 GHz, corresponding to wavelengths from 10 mm to 1 mm. This research focuses 38

13 specifically on 60 GHz radio which enables many new applications that are difficult to offer by wireless systems at lower frequencies Comparison 60 GHz Band with other Unlicensed Band 60 GHz technology offers various advantages over existing communications systems. One major reason for the recent interest in 60 GHz technology is the huge unlicensed bandwidth. As shown in Figure 3.3, at least 5 GHz of continuous bandwidth is available in many countries worldwide. Comparing to the unlicensed bandwidth allocated for ultra-wideband (UWB) purposes [61], the 60 GHz bandwidth is continuous and less restricted in terms of power limits. UWB system is an overlay system and subject to very strict and different regulations [62]. The large bandwidth at 60 GHz is one of the largest unlicensed bandwidths ever been allocated. This huge bandwidth represents great potential in terms of capacity and flexibility, making 60 GHz technology particularly attractive for gigabit wireless applications. Figure 3.3: Worldwide Frequency Spectra available around 60 GHz. The 60 GHz regulation allows much higher transmit power; equivalent isotropic radiated power (EIRP). Table 3.1 shows examples of typical 60 GHz, UWB and IEEE systems that operate near the US Federal Communications Commission (FCC) regulatory limit. Table 3.1: Comparison of Typical Systems Technology Frequency (GHz) PA output (dbm) Antenna gain (dbi) EIRP output (dbm) 60 GHz UWB IEEE n 2.4/

14 The output power of a power amplifier for 60 GHz is typically limited to 10 dbm because the implementation of efficient power amplifiers at this frequency is very challenging though FCC regulations allow up to 27 dbm. Large antenna gain up to 40 dbi has boosted the allowable EIRP limits. While, UWB systems which are required to meet the strict power spectrum mask of 41.3 dbm/mhz offer only very limited EIRP of the order of 10 dbm. This makes the UWB system a very short-range and low-power device. In contrast, the design of power amplifiers for 2.4/5.0 GHz is simpler and can deliver much higher power than the 60 GHz system. However, the EIRP limit is typically confined to 30 dbm due to the crowded Industrial, Scientific and Medical band. It can be seen from Table 3.1 that the EIRP of the 60 GHz system is approximately ten times larger than the IEEE n and 30,000 times larger than the UWB system. The higher path loss at 60 GHz limits the operation to within a room in an indoor environment. In addition to path loss the high penetration loss for most materials in the 60 GHz band perfectly isolates closed rooms without causing any interference from adjacent rooms. The effective interference levels for 60 GHz are less severe than those systems located in the congested GHz and GHz regions. The huge bandwidth available for 60 GHz simplifies the system design. A system with much lower spectral efficiency can be designed to deliver a Gbps transmission to provide low cost and simple implementation Table 3.2 shows the spectral efficiency required by the 60 GHz, UWB and IEEE n systems to achieve 1Gbps transmission as well as spectral efficiency of the actual deployment of such systems. Technology Table 3.2: Spectral Efficiency Comparison Efficiency Target at 1 Gbps data rate (bps/hz) (Mbps) Bandwidth (MHz) 60 GHz UWB IEEE n Efficiency Required (bps/hz) A typical 60 GHz system requires only 0.4 bps/hz to achieve 1Gbps, making it an ideal candidate to support very high data rate applications using simple modulation. Though the UWB system only requires 2 bps/hz to achieve 1Gbps, its actual deployment is limited to 400 Mbps at 1m operating range. IEEE n-alike systems will require 25 40

15 bps/hz in order to achieve 1Gbps, making the extension of such system to beyond 1Gbps unappealing in terms of cost and implementation Regulation and Standardization Issues of 60 GHz Band Since 1996, several European consortiums have been formed to define 60 GHz. WPAN/WLAN systems. Within the IST MAGNET project [ ] ( millimeter wavelength UWB-OFDM systems have been designed and evaluated by France Telecom R&D to achieve bit rates up to 1 Gbps over 528 MHz subchannels. France Telecom R&D designed an advanced multi-user access MC-SS scheme involving 200 MHz channels and UWB channels for 60 GHz transmissions. More recently, IHP Microelectronics, Germany and France Telecom R and D have designed a dual-band UWB-OFDM prototype operating at 60 GHz and in the { } GHz band for down link and up link transmissions respectively. In 2001, the FCC allocated 7 GHz in the GHz band for unlicensed use. This is the largest contiguous block of radio spectrum ever allocated. FCC rules allow 10 W of EIRP in this band, which complies with a maximum power density of 9 mw/cm 2 at 3 m distance. This means that 20 dbm transmit power would be the legal power limit with an antenna having 20 dbi gain. In Japan there was a new regulation in July 2000 for high speed data communication. The frequency range is GHz. The limitation of power to the feeder of the antenna is 10 dbm, whereas the antenna gain should be less than 47 dbi. In July 2003, a new IEEE SIG was created to deal with 60 GHz radio communications. The first investigations on 60 GHz regulations in Europe prove that a minimum of 7 GHz bandwidth should be addressed in the {59 66} GHz band to licenseexempt short-range applications. In 2007, following the SE-24 meeting held in September 2007, it was decided to translate the 7 GHz exempted license band into {57 64} GHz in considering the use of {57 59} GHz band. This decision has been motivated by the high oxygen absorption in the {57 64} GHz band which is foreseen as an advantage in the frequency reuse of resource allocation between adjacent WPAN cells. In 2003, Newlans introduced Gi-Fi and Giga Ethernet to the Desktop (GTTD) applications to illustrate the relevance of 60, 70 and 90 GHz frequency band use for P2P high data rate transmissions. Motorola and Oki initiated the SIG group activity over 41

16 several mm- wave WPAN concepts. In November 2005, IHP microelectronics, Germany proposed the first RF front end for 60 GHz WPAN applications Front End Technology The assignment of the large bandwidth around 60 GHz created new opportunities for 60 GHz front-end technology. In particular, gallium arsenide (GaAs) field effect transistor (FET) technology has evolved to the point where 60 GHz GaAs MMICs (mmwave integrated circuits) are production-ready [47]. GaAs-based 60 GHz devices such as low-noise amplifiers, high-power amplifiers, multipliers, and switches can nowadays be ordered in large quantities in die form at prices on the order of $10 20 apiece [48]. For application in WLAN equipment, however, this might still be too expensive. An alternative technology based on silicon germanium (SiGe) promises to provide truly low-cost mm-wave front-end MMICs while simultaneously maintaining the favorable performance of GaAs. Coplanar wire bond interconnects between chips could be low loss at 60 GHz, whereas multichip module technologies could well accommodate mm-wave components along with intermediate frequency (IF) and baseband circuits [49]. The challenge will be to achieve high-volume production of high-performance compact 60 GHz transmitter/receiver modules (e.g., like those reported in [50]). A further improvement would be monolithic integration of antennas with MMIC chips in order to avoid significant interconnection losses Channel Properties At 60 GHz there is much more free space loss than at 2 or 5 GHz since free space loss increases quadratically with frequency. In principle this higher free space loss can be compensated for by the use of antennas with more pattern directivity while maintaining small antenna dimensions. When such antennas are used, however, antenna obstruction (e.g., by a human body) and mis-pointing may easily cause a substantial drop of received power, which may nullify the gain provided by the antennas. This effect is typical for mm-waves because the diffraction of mm-waves (i.e., the ability to bend around edges of obstacles) is weak. Regarding blocking effects, omnidirectional antennas have an advantage in a reflective (e.g., indoor) environment since there they have the ability to still collect contributions of reflected power in the event of line of sight (LOS) obstruction. 42

17 Walls may considerably attenuate mm-waves. The transmissivity strongly depends on material properties and thickness. At 60 GHz, transmissivity of glass may range from 3 to 7 db, whereas transmission through a 15 cm thick concrete wall can be as high as 36 db [51]. We may therefore expect concrete floors between stocks of a building to act as reliable cell boundaries. This helps to create small indoor cells for hot spot communications. A typical/moderate inner wall consisting of multiple partitions of different materials (e.g., windows and doors), on the other hand, may be considered neither a reliable cell boundary nor a transparent medium. Due to the possible significant attenuation of inner walls, it will generally be necessary to have at least one access point per indoor environment (room, hall, corridor, etc.) to create a reliable shared medium. A consequence of the confinement to smaller cells is that channel dispersion is smaller than values encountered at lower frequencies because echo paths are shorter on average. RMS delay spread may range from a few to 100 ns. It is expected to be highest if omnidirectional antennas are used in large reflective indoor environments [51, 52]. When, instead, high gain antennas are used, rms delay spread may be limited to a few nanoseconds only. [53, 54] Attenuation Due to Atmospheric Gases and Rain Fall One of the major reasons of degradation of the density of electromagnetic energy during propagation is the progressive absorption of the wave by the medium. This energy absorption is mainly due to molecular absorption by atmospheric gases and absorption or scatter by liquid and solid particles in the atmosphere particularly by raindrops Oxygen Absorption Since Nitrogen has no absorption bands in radio frequency range, absorption is almost entirely due to oxygen and water vapor. In frequency range of 57.5 GHz to 62.5 GHz in the lower part of the atmosphere the specific attenuation due to oxygen is typically 14.7 db/km. 43

18 Figure 3.4: Specific Attenuation Along a terrestrial path it can generally be assumed that the atmosphere is homogeneous and total attenuation can be calculated by multiplying the length of the path by the sum of specific attenuation due to oxygen and water vapor. For many applications the effect of temperature can be complexly neglected Rain/Moisture Absorption Wireless communication systems operating at frequencies below 10 GHz are least affected by rain. However at millimeter frequencies, rain induced attenuation (due to absorption and scattering) is one of the principal factors that increases the overall path loss, limits the coverage area and consequently degrades the system performance. Extensive qualitative and quantitative analysis of effect of rain on the propagation of electromagnetic waves have been published in literature. Most rain attenuation models are based on statistical data of rain rate. This section briefly presents two popular rain models. The two rain models are Crane Global Rain Attenuation Model and Modified SAM/CCIR (Simplified Attenuation Model) Rain Model. Crane Global Rain Attenuation Model: The Crane models, after Robert K. Crane, are popular for satellite earth links but also have terrestrial models. The Global Crane model, developed in 1980, is rigorously treated in [54]. The theoretical prediction model, based on the rain attenuation model, can be summarized by the following equations. In these Equations, is the rain attenuation in db, R is the point rain rate in mm/hr, and d 44

19 is the distance in km. Constants a and b are rain attenuation coefficients that are functions of frequency and polarization and are tabulated in [59]. = 1 0 (3.8) = 1 + (3.9) For 22.5 (3.10) Where = ( ), =2.3., = ( ) and = For paths longer than 22.5 km, the attenuation is calculated for a 22.5 km path, and the resulting rain attenuation is multiplied by a factor of ( 22.5) Simplified Attenuation Model (SAM/CCIR): The Simplified Attenuation Model (SAM/CCIR) was developed for NASA to provide a simplified technique for hand calculation. This model [60] is based on nominal water droplet sizes and distribution and allows calculation of attenuation rate (db/km) due to a specified rainfall rate. The attenuation rate can be approximately expressed by Equation 3.11, where R represents the rainfall rate in millimeters per hour and parameters a and b are approximated by Equations = (3.11) = = = = (3.12) 45

20 3.5.6 Channel Measurements in 60 GHz Band IEEE c channel models are mainly derived based on wideband measurement results conducted in office, residential, library and desktop environments. For each environment, LOS and NLOS scenarios are defined. Some NLOS scenarios are generated from their LOS counterparts by appropriately removing the LOS component from the model. IEEE c adopted the concept of generic SV channel modeling as described in Equation (3.5). Measurement results indicated that when directive antennas are used, especially in the LOS scenario, there appears a distinct and strong LOS path on top of the clustering phenomena. The IEEE c channel model only characterizes the AoA. The 15.3c Task Group (TG3c) adopted a uniform distribution over [0, 2 ] for the cluster mean AoA, conditioned on the first cluster mean AoA,.On the other hand, the ray AoAs within each cluster can be modeled either by zero-mean Gaussian or zero-mean Laplacian distributions. The presence of a strong LOS component the values of and, re no longer zero since the reference zero point has been moved to the direction arrival of the LOS component. With proper normalization with respect to the strong LOS, the modified model behaves in a similar way to the directional SV model expressed in Equation ( ) =, ( ) (3.13) Table 3.3: SV Channel Parameters for Different Indoor Scenario Parameter Scenario Residential Office Library [1/ns]: Cluster arrival rate [1/ns]: Ray arrival rate : Mean Number of Clusters The channel parameters enlisted in Table 3.3 are IEEE a channel model parameters employed for simulation purpose. The scenarios like residential, office 46

21 and library environment are considered. The parameters selected are the derived parameters from measurements performed with omnidirectional transmitting antenna and a directive receiving antenna with 3 db beam-widths of [63] 3.6 Justification of Selection of 60 GHz Channel Model This section verifies the correctness of the proposed channel model through a review of measurement campaign reported in literature. A major advantage of 60 GHz band is due to a physical property of the propagation channel at this frequency. Oxygen absorption leads to attenuation of 14.7 db/km at 60 GHz. This attenuation in addition to path loss enables a shorter reuse distance in cellular systems because it inherently reduces cochannel interference. For short distance of a typical indoor scenario this additional attenuation can be neglected. Millimeter wave exhibit a general property that the behavior of propagation rays is characterized by geometric optics. These waves do not penetrate through walls and reflection is the main mechanism leading to multipath. Scattering and diffraction are considered far less important. [86] A Review of Measurement Results at 60 GHz band Most of the work has been done on indoor channels and their modeling; because of the range limitation of 60 GHz band. A major activity in the field of mm-wave propagation has been conducted in the framework of the RACE-MBS (Research into Advanced Communications systems in Europe project 2067, Mobile Baseband Systems) [87]. The measurements reported include material characterization and indoor and outdoor propagation studies. For predicting propagation parameters ray tracing models were used. A major contribution to indoor propagation originates from the research of P. Smulders. [51] The main parameters of interest of the channel model are NRP (Normalized Received Power), the Rician K-factor and the RDS (rms delay spread). Most studies report results of NRP and RDS. Various research groups had conducted measurements that diversified in the following parameters: Measurement equipment and method used. Antenna characteristics and configurations. Environments investigated. 47

22 Parameters measured and presented. This comparison is organized by discussing the parameters of interest and elaborating on the impact of some of the above listed factors. Only wideband measurements are considered, because of the significance of characterizing the timedispersive nature of the channel Measurement Setups and Techniques Most indoor measurements use vector network analyzers to scan the channel transfer function [83]-[90]. The conditions to use vector network analyzer are short distances, because a phase reference must be provided between the transmitting and the receiving sides. In a (quasi) static channel it takes time to acquire the frequency transfer function. These conditions are reasonable in indoor scenarios. A narrowband (continuous wave) signal is transmitted in which the entire transmit power is concentrated. The delaytime resolution reported is normally around 1, corresponding to a scanning bandwidth of 1 2 GHz. Correlation type channel sounders were developed for the measurements performed in the RACE-MBS project [86]. A separate indoor channel sounder is based on the transmission of a pseudo random binary sequence and a sliding-correlator on the receiver s side ([91]) Discussion of Channel Parameters The RDS, NRP and Rician K-factor are the three most significant parameters for specifying the channel s frequency-selective nature. The RDS determines the number of fades per bandwidth, while the K factor specifies the depth of the fades. The normalized received power (NRP) determines the average signal-to-noise ratio (SNR). The SNR is usually considered as a variable in any kind of system studies RDS: RMS Delay Spread The following are main features of the propagation environment that influence the RDS. The mentioned properties are applicable for indoor channels only. Room size The work by Smulders reported the RDS increase with the room size [87], who measured typical values of RDS between 15 and 45 ns in small rooms ( m 3 ) 48

23 and between 30 and 70 ns in larger rooms. These values are large, compared to the other indoor measurements found in the literature. The large values are governed by the antenna design implemented. Antenna Directivity Directive antennas attenuate parts of the reflected waves. Therefore, the RDS usually decreases when more directive antennas are in use. This is clearly seen from measurements and ray-tracing simulations performed. In a room with dimensions of m 3, measured RDS-values are 18, 14, 5 and 1, respectively for, an omnidirectional antenna ( /2-dipol), and antennas with 3 db beam-widths of ~60,~10, and ~5.[87]. Based on ray-tracing method for a room of m 3, different antenna configurations lead to various RDS-values of for the less directional antenna, and values lower than 5 for the most directional ones. [86]. Smulders has conducted additional measurements using a 15 circular-horn antenna on one side of the link [57]. The results confirm the behavior if median values of RDS are considered (RDS decreases from ~40 to ~25 ). The maximum RDS values observed were larger than for the standard antenna configuration (increase from ~48 to ~60 ). The more directive antenna, which also has the higher gain, may emphasize some reflected paths with a rather large delay time. Such paths contribute strongly to the RDS. Building Material The reflectivity of building material is another important factor influencing the RDS. Smulders, measured higher RDS values in a small room with metal walls (room dimensions ~ m 3 ; ~ 45 ), than in a much larger auditorium room with walls covered by wood and acoustically soft material (room dimensions ~ ; ~ 35 ). In a small room (~ ) with wood-covered walls, RDS-values of ~20 were measured. [58] Rician K-factor While statistics of the RDS are found in most propagation studies, the Rician K- factor is often not investigated. Some studies assume Rayleigh fading amplitude distributions, i.e., K-factor of zero. When the line-of-sight between transmitter and receiver is obstructed, this assumption may be reasonable. K-factors below -3 db can be 49

24 well represented by the Rayleigh distribution. When the dominant path carries greater or equal power than all the reflected paths, the Rician model must be used. Also usages of more directive antennas yield higher K-factors. Adaptive antennas (beamforming) can avoid the pointing the antenna manually. Larger K-factors reduce RDS, when the decay exponents of the average power delay profile remain constant. Smulders model parameters [57] imply that even for the 15 dbi directive antenna, and in the presence of a LOS path, the K-factor would be less or equal to therefore well described by the Rayleigh model Review of Saleh-Valenzuela Model The channel models proposed for mm-wave channels are based on the indoor propagation model presented by Saleh and Valenzuela. (Section 3.4) Several authors have applied a number of amendments to SV channel model to match it to mm-wave channels. Most of the literature recommends reduction in the number of clusters to one [84, 85]. This simplification is for a typical indoor mm-wave channel. Since 60 GHz frequencies hardly penetrate through building material, all the reflections initiate within one room, leading to a single, dense cluster of ray arrivals. Smulders [57] proposes a composite average PDP. The exponentially decaying part of the single cluster is leaded by a constant-level part. The reason for introducing this part is to better describe first-order reflections arriving at similar strength due to the antenna design chosen. The channel model used for system simulation exemplifies the single cluster versions of the SV model introduced above. The Delay Power Spectrum of the FD-model is equivalent to the average power delay profile of the single-cluster SV model Comparison of Analytical Rayleigh Channel Model with SV Channel Model for OFDM System We could not succeed in writing a very correct analytical model for BER performance of OFDM system with SV channel. Smulders recommended that at 60 GHz frequency and indoor scenario the K-factor would be less or equal to -6 db although a LOS path exists. [58] Therefore the channel is well described by Rayleigh model. Hence 50

25 analytical model of Rayleigh channel is used in this research for validation of SV channel simulation model. The analytical BER expressions for BPSK signaling and OFDM scheme in Rayleigh channel is given as = Where M denote modulation order which is (3.14) The modified single cluster SV-model in OFDM environment shows similarities with a Rayleigh fading channel with additional shadowing [81]. Hence validation of SV channel model is accomplished with reference to the analytical performance of Rayleigh fading channel. The performance of both the channel models at 60 GHz frequency is indicated in Figure 3.3. The simulation scheme generates realizations of channel transfer functions with well-defined channel parameters. With an OFDM system data symbols can be transmitted independently over multipath fading radio channels. The simulation model used assumes that the channel s maximum excess delay is shorter than the guard interval, and the system has been synchronized sufficiently. The simulated OFDM system uses 512 subcarriers with a cyclic prefix to retain orthogonality of the subcarriers over a dispersive channel. This simulation model do not consider insertion of pilot subcarriers and hence any the channel estimation technique. (See Chapter OFDM Section: Validation). This simulation model assumes BPSK modulation scheme. While for SV channel the parameters are referred from the IEEE a channel model, the same are enlisted in Table 3.3. [82] The computer simulation results of the SV channel model and analytical results of Rayleigh channel are compared to test appropriateness of the SV channel model. 51

26 10 0 BER for OFDM with BPSK modulation Analytical Simulation 10-1 Bit Error Rate Eb/No(dB) Figure 3.5: Validations of Simulation Model and Analytical Design at 60 GHz. Results indicate that the BER performance of simulation model of BPSK modulated OFDM system in SV channel for 60 GHz is in harmony with the analytical results of Rayleigh fading channel for the same frequency. Hence we successfully negotiate that simulation model for SV channel can be said validated because nature of the BER curve remains close to the analytical results of Rayleigh fading channel 3.7 Summary In this chapter, we have presented stimulus in using the 60 GHz frequency band. The chapter began with a discussion of the basics of radio wave propagation and channel modeling. A generic channel model used at 60 GHz, SV model was then reviewed. In particular, the path loss and shadowing effects impose huge losses to the communication link. Channel parameters depend on a number of features of the propagation environment and of the antenna set-up. The high path loss and high penetration loss for most materials in the 60 GHz band almost perfectly isolates closed rooms and limits interference at 60 GHz, restricting the operation to within a room in an indoor environment. As a modulation scheme, OFDM allows high spectral efficiency in multipath environments and was therefore chosen as the preferred candidate at 60 GHz. 52

27 The simulation scheme introduced directly generates realization of SV channel model transfer function with selected channel parameters at 60 GHz frequency band. The performance comparison of analytical results of Rayleigh channel OFDM system and simulation model of the OFDM system with SV channel model has confirmed the correctness of the simulation model designed. The basic model introduced assumes perfect synchronization. The system model presented can be utilized in further studies of various methodologies for generation of OFDM signal, as, for instance, wavelet based OFDM system. 53