Radio Wave Propagation from LOS to Street-Level in an Urban Area

Radio Wave Propagation from LOS to Street-Level in an Urban Area Peter Nysten Master of Science Thesis Ericsson Research Royal Institute of Technology Advisor Jan-Erik Berg, Expert, Propagation Ericsson Research, Corporate Unit Ericsson Radio Systems AB Kista, Sweden Examiner Björn Ottersten, Professor Department of Signal Processing, S3 Royal Institute of Technology Stockholm, Sweden

2 (87) Abstract The radio wave propagation from LOS to street-level in an urban area with "high-raised" buildings has been investigated in this thesis project. Measurements with different MS-antenna heights have been conducted in the city of Aalborg with a wide-band (UMTS) signal at 1712 MHz. The data have been analyzed and the DOA has been determined in order to find the places where diffraction, reflection, and scattering occur. Two different models, describing the mean signal strength dependence of the MS-antenna height, have been developed based on the local geographical environment. Acknowledgements I am very grateful to my advisor Jan-Erik Berg at Ericsson Radio Systems AB for his valuable consultation and support throughout my thesis work. I would also like to express my gratitude to my parents, my fiancée Jessica Öberg and my son Philip Nysten. They bless me with abundant love and support all the time. Peter Nysten nysten@kth.se March 14, 2002 Stockholm, Sweden.

3 (87) 1 INTRODUCTION... 5 1.1 BACKGROUND... 5 1.2 OBJECTIVES... 6 1.3 REPORT OUTLINE... 6 2 INTRODUCTION TO RADIO WAVE PROPAGATION... 7 2.1 DIFFRACTION... 7 2.2 REFRACTION... 8 2.3 REFLECTION... 8 2.3.1 Specular reflection... 8 2.3.2 Diffuse scattering or Rough surface scattering... 8 2.4 SCATTERING... 10 3 MEASUREMENT SETUP... 11 3.1 GEOGRAPHICAL DATA FOR THE MEASUREMENT POSITIONS... 12 3.2 ENVIRONMENT... 14 3.3 MOBILE STATION ANTENNA... 15 3.4 BASE STATION ANTENNA... 15 4 ANALYSIS OF DATA... 16 4.1 SIGNAL TO NOISE RATIO... 16 4.2 AVERAGE POWER DELAY PROFILE... 17 4.3 FADING...18 5 MEAN SIGNAL STRENGTH... 20 5.1 MEAN SIGNAL STRENGTH AT LOS... 20 5.2 EXCESS PATH LOSS FROM LOS TO 5M... 21 5.3 MEAN SIGNAL STRENGTH DEPENDENCE OF HEIGHT FOR DIFFERENT LOCAL GEOGRAPHICAL ENVIRONMENTS... 22 5.3.1 LOS-boundary due to building BF, measurement position 1,5,7,8 and 11... 22 5.3.2 LOS-boundary due to building TOB, measurement position 3,4,12... 25 5.3.3 LOS-boundary due to building BF, measurement position 2,6 and 10... 27 5.3.4 LOS-boundary due to building BF, measurement position 9 and13... 29 5.3.5 Summary of the mean signal strength dependence of height... 31

4 (87) 6 DIRECTION OF ARRIVAL (DOA)... 32 6.1 SIGNAL ANALYSIS... 32 6.2 BEAM-FORMING... 32 6.3 EVALUATION OF THE RESULTS FROM THE BEAM-FORMING... 34 6.4 SUMMARY OF THE DOA ANALYSIS... 36 6.4.1 LOS... 36 6.4.2 LOS to the eaves... 37 6.4.3 The Eaves to street-level, 5m... 38 6.4.4 Comparison with earlier made measurements at street-level... 39 6.5 DOA FOR DIFFERENT LOCAL GEOGRAPHICAL ENVIRONMENTS... 40 6.5.1 LOS-boundary due to building BF, measurement position 1... 40 6.5.2 LOS-boundary due to building TOB, measurement position 4... 42 6.5.3 LOS-boundary due to building BF, measurement position 6... 44 6.5.4 LOS-boundary due to building BF, measurement position 13... 46 7 PATH LOSS MODEL FOR THE MEAN SIGNAL STRENGTH... 48 7.1 KNIFE-EDGE DIFFRACTION MODEL... 49 7.2 PROPOSED PROPAGATION MODELS... 51 7.2.1 Model 1... 51 7.2.2 Model 2... 55 7.3 CONCLUSIONS FROM THE PROPOSED MODELS... 60 7.4 EXTENSION OF THE PROPOSED MODEL IN 2D... 61 8 CONCLUSIONS... 64 9 FUTURE WORK... 65 10 REFERENCES... 66 11 APPENDIX... 67 11.1 AVERAGE POWER DELAY PROFILE... 67 11.2 FADING...74 11.3 RESULTS FROM THE PROPOSED MODELS... 81

1 Introduction 5 (87) 1.1 Background Most cellular radio systems operate in urban areas where there is no direct LOS (Line Of Sight) path between the transmitter and receiver. Due to the highly complicated terrain profile of an urban area the radio waves travels along different paths of varying lengths. Radio wave propagation is a complicated process including three basic propagation mechanisms called diffraction, reflection and scattering. The prediction of path loss is a very important step in planning a mobile radio system, and accurate prediction methods are needed to determine the parameters of a radio system, which will provide efficient and reliable coverage of a specified service area. Unlike wired channels that are stationary and predictable, radio channels are extremely random and do not offer easy analysis. Modeling the radio channel has historically been one of the most difficult parts of mobile radio system design, and is typically done in a statistical fashion, based on measurements made specifically for an intended communication system. In this thesis project, radio wave propagation from LOS to street-level in an urban area with high-raised buildings has been investigated, by analyzing earlier made measurements with different MS (Mobile Station) height. The measurements have been done in the Danish City of Aalborg by CPK (Center for Personkommunikation) at Aalborg University with help from Ericsson Radio Systems AB in the year of 1999. The paper Prediction path Loss in Environments with High-Raised Buildings [1] investigates the accuracy of prediction models estimating the path loss, is an outcome of these measurements. Previous measurements have been done in similar urban environments in order to investigate how radio waves propagate in an urban area, but they have been done at the street level, like [5] and [6]. The area between LOS and street-level has, what we know of, never been investigated before but it does have importance in models used for cell planning and of course to increase the knowledge of how radio waves propagate in an urban area.

6 (87) 1.2 Objectives The main tasks of this master thesis project is to: 1. Analyze data from previously made measurements in an urban area with highraised buildings and from these data try to evaluate how the radio waves propagate from LOS down to the street level by finding the DOA (Direction Of Arrival). What is most important in this process diffraction, reflection or scattering or is it a combination of these? 2. Develop a model describing the mean signal strength dependence of MS (Mobile Station) height. Is this model useful for determining the mean signal strength in a planning tool? The reason for doing this is of course to increase the knowledge of how radio waves propagate in an urban area and whether it is possible to describe this with an accurate model of the mean signal strength dependence of MS (Mobile Station) height. In order to investigate the Direction Of Arrival (DOA) Beam-forming have been used and the reason for using this method is that it is a relatively simple method and do not require that much of computer resources. 1.3 Report outline Chapter 2 is a short introduction to radio wave propagation, Chapter 3 contains the measurement setup and the geographical data for the urban area there the measurement took place. Chapter 4 presents a brief analysis of the data. Chapter 5 presents the mean signal strength, first the mean signal strength at LOS and later the mean signal strength dependence of height relative the mean signal strength at LOS. Chapter 6 presents the Direction Of Arrival and is together with chapter 7 the most important chapters in this report. Chapter 7 first explains knife-edge diffraction and later the two proposed models describing the mean signal strength dependence of MS-heights, this chapter is an outcome of the earlier chapters. Chapter 8 presents a summary of the most important conclusions and chapter 9 gives proposals for future work in this area of interest. Chapter 11 is the appendix and contains the Average Power Delay Profile (APDP), the fading characteristics and the results from the two proposed models for all measurement positions.

7 (87) 2 Introduction to radio wave propagation This is a short introduction to radio wave propagation, to make this report shorter only the most important processes are explained briefly, for those who want to read more about radio wave propagation, I would like to recommend you to read [2]. 2.1 Diffraction Diffraction occurs when an object (obstacle) obstructs the radio path between the transmitter and receiver. The diffracted waves are present throughout the space and even behind the obstacle, even when a line-of-sight (LOS) path does not exist between transmitter and receiver. Diffraction depends on the geometry of the object, as well as amplitude, phase, polarization, and the wavelength of the incident wave. One way to explain why the field in the shadowed regions is nonzero would be through Huygen s principle [2]. In simple terms, the principle suggests that each point on a wave-front acts as the source of secondary wavelet and that these wavelets combine to produce a new wave-front in the direction of propagation. The usual assumption that an obstacle can be represented by an ideal, straight, perfectly absorbing screen (knife-edge) is often a very rough approximation. Neither hills nor buildings can be truly represented by a knife-edge (assumed infinitely thin, long and perfectly absorbing). The knife-edge diffraction formula depends on carrier frequency and the geometrical dimensions (see figure 2.1), the higher the frequency is the less will be diffracted. A simple example is visible light (high frequency) that does not diffract as much as sound that has a lower frequency. The knife-edge formula is explained in section 7.1. Figure 2.1 The geometry of the knife-edge The knife-edge diffraction formula depends on the following geometrical data: T is the transmitters position at the distance d t from the screen R is the receivers position at the distance d r from the screen h is here the height of the screen, relative a straight line between T and R. Θt The angle to the diffraction screen at the transmitter (T) Θ The angle to the diffraction screen at the receiver (R) r

8 (87) 2.2 Refraction Refraction is the phenomenon when a wave that passes through a media changes direction. In an outdoor environment with large objects, the effects of waves passing through the objects are negligible [2]. 2.3 Reflection Reflection occurs when a propagating wave impinges upon an object, which has very large dimensions compared to the wavelength. For radio waves reflections occur from roofs, walls and other large objects. 2.3.1 Specular reflection Specular reflections occur when a wave impinges upon a surface that is flat or smooth compared to the wavelength of the impinging wave. The wave gets reflected at the surface and is reflected in a specific direction, see figure 2.3.1. Figure 2.3.1 Specular reflection The wave has its source at point A and is reflected at the surface to point B, the point B considers the point A as the source of the wave, note that the surface does not have to be a plane it could also be a curved object. 2.3.2 Diffuse scattering or Rough surface scattering Specular reflection is not the general case since most surfaces are not absolutely flat or smooth (compared to the wavelength), this could cause reflection in other directions, see figure 2.3.2-1. Figure 2.3.2-1 Reflection and scattering from a rough surface The wave has its source at point A and is reflected at the surface to point B (specular reflection), the point B considers point A as the source of the wave. There are also reflections in other directions (diffuse scattered or diffuse reflected), as to point C, the point C consider point A as the source of the wave. The rough surface at point A acts as a new source.

9 (87) Figure 2.3.2-2 Rough surface I S D Incoming wave Specular component The amplitude of the diffused scattered field When the surface roughness increases the amplitude of the specular reflection will decrease and the diffused reflected field would be more spread out (compare the left and the right part of figure 2.3.2-2 above). When the surface roughness is increased so much that the specularly reflected part disappears, the diffused reflected field will still be present. The specularly reflected component s amplitude and phase can be determined deterministically but the diffused reflected field can only be determined statistically. The Rayleigh criterion is commonly used to evaluate the apparent smoothness of a surface [2], which is λ h < Equation 2.3.2-1 8 cosθi Where h is the height difference between two points on the surface (see the right part of figure 2.3.3-2). Θi is the incident angle of the wave λ is the wavelength When the Rayleigh criterion is fulfilled, the surface is regarded as smooth and specular reflection is dominating. This leads to the fact that a surface could be considered smooth in one frequency range and rough in others. And if the incident angle is increased, the Rayleigh criterion can be fulfilled for a greater h, see figure 2.3.2-3.

10 (87) Figure 2.3.2-3 The Rayleigh criterion for different incident angles for λ =0.175m Figure 2.3.2-3 shows the Rayleigh criterion for the specific wavelength used in this report. The horizontal axis shows the incident angle and the vertical axis shows the degree of roughness. 2.4 Scattering Scattering occurs when the medium through which the wave travels consists of objects with dimensions that are small compared to the wavelength. The number of scattering objects per unit volume can be large. Scattered waves are produced by small objects or by other irregularities in the environment. In practice, foliage, streetsigns, street-lamps and even trees can cause scattering.

3 Measurement setup 11 (87) The measurements have been conducted in an urban area with high-raised buildings. The Base Station was mounted on a tall building while the Mobile Station (MS) was moved to 13 different positions (measurement positions). For every measurement position the height of the MS was changed from LOS (Line Of Sight) to NLOS (Non Line Of Sight). The goal for this measurement campaign was to investigate how radio waves propagate from LOS to street-level in an urban area and to measure the signal strength dependence of MS height. The MS (Mobile Station) was mounted on a lorry with a lift, the height of the MS was changed from 25m down to 5m with a step of 2.5m. For each height the MS antenna (transmitter) was moved around in a circular path (horizontal) with a radius of 0.53 m (see figure 3-1). The MS transmits a PN (Pseudo Noise) sounding sequence and the receiving unit (BS) estimates the impulse response of the radio channel in 8 parallel channels. Both the BS and MS used rubidium standards for the reference signals to ensure phase stability during measurements, [1]. The carrier frequency was 1712MHz and the bandwidth was approximately 5MHz. Further information about the antennas is given in section 3.3 and 3.4. Figure 3-1 Measurement setup

12 (87) 3.1 Geographical data for the measurement positions The measurements have been conducted in the Danish City Aalborg with about 100000 inhabitants. Figure 3.1-1 Map over the city of Aalborg

13 (87) For all 13-measurement positions were the following geometrical data gathered, see figure 3.1-2 and the table below. Figure 3.1-2 Definition of the geometrical data gathered at all measurement positions. TOB BF BB BS MS Tallest Obstructing Building in the front Building in the Front Building in the Back Base Station Mobile Station Geometrical data Height of the BS (Receiver) above the ground (h BS =48m) Height of the MS (Transmitter) (h MS =5,7.5,,,25m) Distance between the MS and the center of the tallest obstructing building in the front (TOB) (d 1 =65-444m) Distance between the MS and the center of the building in the front (BF) (d 2 =6.6-10.1m) Distance between the MS and the building in the front (BF) (d 3 =1.9-4.7m) Distance between the MS and the building in the back (BB) (d 4 =8.6-34.9m) Height of the tallest obstructing building in the front (TOB) (h 1 =17.8-20.3m) Height of the building in the front (BF) (rooftop) (h 2 =12.3-19.7m) Height of the building in the front (BF) (the eaves) (h 3 =10-15m) Height of the building in the back (BB) (the eaves) (h 4 =3-14.6m) Average height of the buildings TOB and BF 17.9m Separation distance between the BS and MS (d BS = 1455-3466m) Note: The area between the base station and the measurement positions was fairly flat so the topography according to figure 3.1-2 is quite close to the real conditions. The specific geometrical data for all the 13 measurement positions can be found in the sections 5.3 and 6.5

14 (87) Figure 3.1-3 Geometrical definitions TOB Tallest Obstructing Building in the front BF Building in the Front BB Building in the Back LOS Line Of Sight NLOS Non Line Of Sight BS Base Station MS Mobile Station 3.2 Environment The area where the measurements took place is close to a fjord (see figure 3.1-1 above). The buildings along each side of the alley are attached, forming a street canyon (see figure 3.1-3). The buildings are mostly made of brick and the buildings have high-raised (saddle-shaped) rooftops. The buildings in the area do not have a uniform height and shape. Many of the walls and roofs have obstacles stretching out and there are also obstacles in the street canyons like street lamps hanging between the buildings, trees and other things at the street-level like parked cars etc. Most of the measurement positions are located between the street crossings about 50 meters to each crossing. Some of the measurements have been done close to street corners. For most of the measurement positions, building BF (see figure 3.1-3) will obstruct the MS-BS LOS-path first when the MS is lowered from 25m to 5m (building BF gives rise to the LOS-boundary). However when the distance d 1 (see figure 3.1.2) is relatively small and h 1 >h 2, the building TOB (at the distance d 1 from the MS) will be the first building to obstruct the MS-BS LOS-path when the height is lowered (building TOB gives rise to the LOS-boundary). See the sections 5.3 and 6.5 for more information about the different measurement positions.

15 (87) 3.3 Mobile station antenna The mobile station antenna consists of an omni directional antenna (vertical polarized dipole antenna) and a directional antenna (patch antenna) mounted on a rotor arm with a radius of 0.53m (see figure 3.3-1). Measurements have been recorded at the BS from each MS antenna for at least one revolution (each revolution takes about 8s and about 800 samples are recorded at the BS). Note that the two antennas did not transmit simultaneously see further information below. Unfortunately the results from the patch antenna are not reliable due to errors in the measurement equipment, therefore only the results from the dipole antenna have been used for further analysis. Figure 3.3-1 Top view of the rotor arms with the two MS antennas The rotor arm through the measurements Step 1. The Antenna 1 (The Dipole) was pointing north ➀ and started to rotate clockwise. Step 2. After 2s antenna 1 was pointing east ➁ and the BS started to record from Antenna 1 (The dipole). Step 3. The antenna 1 was pointing south ➂. The BS has now been recording from antenna 1 for more than one revolution (about 1000 samples). Step 4. The antenna 1 was pointing west and antenna 2 east ➃. The BS started to record from antenna 2. Step 5. The antenna 1 was pointing west and antenna 2 east ➄. The BS has now been recording from antenna 2 for one revolution (about 800 samples). Step 6. The rotor arm is stopped and starts to rewind from ➅ to ➀ 3.4 Base station antenna The base station antenna contains 4 dual polarized patch antennas (the polarization angles are ±45 degrees), such that it is possible to simultaneously measure the channel characteristics in two polarization s (totally 8 parallel channels). The results from the first patch antenna have been used. The phase difference between the two polarization in this antenna where determined to be 1.245 radians and the cross correlation between these two was about 0.9. The investigated signal is the sum from these two polarizations.

16 (87) 4 Analysis of data This section evaluates the signal from this measurement campaign, first the Signal to Noise Ratio is investigated, and this is only done to make sure that the gathered data can be used for further analysis. The Average Power Delay Profile is later presented in this chapter and the narrow and wide-band fading characteristics. 4.1 Signal to noise ratio The Signal to Noise Ratio (SNR) is defined as SNR = E E S N 2 2 σ = σ 2 s 2 n Equation 4.1-1 The signal to noise ratio is the averaged noise floor compared with the signal in the impulse response. The SNR is decreasing while the height is decreased and has similar features for all measurement positions. At LOS the SNR is between 42 to 33dB, the lowest SNR is 13dB at 5m. The measurements at the two lowest heights at measurement position 7 have some errors and those have been excluded.

17 (87) 4.2 Average Power Delay Profile The Average Power Delay Profile (APDP) is the averaged impulse response from positions around the revolution of the MS antenna (there are about 800 impulse responses). Unfortunately the time window is shifted through all measured heights at each measurement position (there are no absolute time in the measurements) this could have caused that some late arrived signals may have been excluded. The time window can at least show the APDP for 9.2 microseconds (counted from the first part of received signal to the end of the time window). Which in this case corresponds to the fact that a signal could have traveled at least 80% longer distance than the direct transmitter receiver (T-R) separation distance. The APDP have similar features for all measurement positions. At the heights above LOS/rooftop level, there is only one dominating peak but when the height is lowered below LOS does other late received parts of the signal show up from reflections far away (see the DOA, section 6.4 and 6.5). The first part of the signal is usually the strongest but for some of the lower heights the late received part of the signal is the strongest. It can also be seen that the amplitude of the later part generally increases compared to the first received part when the height of the buildings around the measurement position is lower than the average building height and the width of the street is larger than average. The APDP has exponential decay, which according to [5] is typical for scenarios where local scattering dominates. In the appendix (section 11.1) all the APDP s are presented for all measurement positions and heights. More information can be found in the sections 6.4 and 6.5. Figure 4.2 The APDP for measurement position 1 Figure 4.2 is an example of the APDP, note that from 17.5m (right under the LOSboundary) and below does the amplitude of the late received part of the signal increase, the amplitude of these late received parts increases relative the first part while the height decreases.

18 (87) 4.3 Fading The small scale fading (or simply fading) variation is greatest when the mobile station is well below the surrounding buildings, where there is no LOS path between the BS and MS. Propagation in this area is mainly due to scattering from the surfaces of buildings and other obstacles and diffraction over and around them. Substantial variations occur in the signal amplitude. Fading is caused by interference between two or more versions of the transmitted signal, which arrives at the receiver at slightly different times and with different phases. The signal amplitude fluctuations are known as fading and the short-term fluctuations caused by the local multi-path is known as fast fading or Rayleigh fading to distinguish it from the much longer-term variations in mean signal-level, known as slow fading, [2]. The Rayleigh probability density function (pdf) can be written as 2 r () 2 r r r = e 2 2 p Equation 4.3-1 r where 2 r Is the averaged power of the short-term fading 2 r is the power Figure 4.3-1 The fading variations for the narrow-band signal at measurement position 1 for different MS antenna heights.

19 (87) Figure 4.3-1 describes the fading characteristics for the narrow-band signal, for all heights at measurement position 1, the numbers at each line is the measurement height (MS height). The measured fading is here compared with the Rayleigh fading. Note that in figure 4.3-1 and 4.3-2 the power level for the Rayleigh fading is chosen arbitrary, and that these figures could also be referred as the CDF (Cumulative Distribution Function). The fading characteristics have similar features for all measurement positions. When the measurement height is lowered below the rooftop level, the fading is close to Rayleigh fading. The fading variations for the narrow-band signal, for all measurement positions can be found in the appendix (section11.2). Figure 4.3-2 The fading variations for the wide-band signal at measurement position 1 for different MS antenna heights. Figure 4.3-2 describes the fading characteristics for the wide-band signal (5MHz), for all heights at measurement position 1, the numbers at each line is the measurement height. The measured fading is here compared with the Rayleigh fading. The fading characteristics have similar features for all measurement positions and when the measurement height is lowered below the rooftop level, does the fading increase. The fading for the wide-band signal is lower (or smaller) than for the narrow-band signal.

20 (87) 5 Mean signal strength In this section the mean signal strength for all measurement positions and heights is presented, first the mean signal strength at LOS as a function of the separation distance between the BS and MS and later the mean signal strength dependence of MS height is shown. For all the following sections when the mean signal strength is presented it is the mean signal strength is considering the narrow-band signal. 5.1 Mean signal strength at LOS Mean signal strength dependence of separation distance between the BS and MS at LOS. Figure 5.1 The relative mean signal strength at LOS (the mean of 25 to 20m) for each measurement position compared with the free space propagation model (in db). Figure 5.1 shows the relative mean signal strength at LOS (the mean of the mean signal strength at 25 to 20m). Note that the measurement positions are reversed order from the BS, position 13 is closest to the BS and position 1 is at the largest distance from the BS. The mean signal strength follows the free space propagation model relatively well (within 3.7dB for all measurement positions). The model in figure 5.1 is the free space propagation model relative the measured mean signal strength. P ( d) = C L db Equation 5.1-1 where LdB = 32.4 + 20 log10 ( d ) + 20 log10 ( f GHz ) Equation 5.1-2 d is the distance between BS & MS in meters fghz is the carrier frequency in GHz The constant C was estimated to: C = 159. 1 db

21 (87) 5.2 Excess path loss from LOS to 5m Excess path loss is here referred as the total path loss from LOS to 5m, for every measurement position the mean signal strength at 5m is relative the mean signal strength at LOS, see section 5.1. Figure 5.2.1 Excess path gain from LOS to 5m For every measurement position, the mean signal strength at 5m is relative the mean of the mean signal strength at LOS (25 to 20m), see section 5.1. The mean of the excess path loss from LOS to 5m is 32 db.

22 (87) 5.3 Mean signal strength dependence of height for different local geographical environments The 13 measurement positions have been divided into four groups with respect to the local geographical environment (distances and height of the buildings in the front and the back), see the following sections for more information about these different groups. For every measurement position, the mean signal strength at each height is relative the mean of the mean signal strength at LOS (25 to 20m) at the specific measurement position. 5.3.1 LOS-boundary due to building BF, measurement position 1,5,7,8 and 11 In these cases is h 3 > h and 4 h1 h2 Figure 5.3.1-1 Local geographical environment Measurement positions 1,7 and 8 all have in common that they have a building in the back (BB) that is lower than the ones in the front. At the same time as the building in front of the measurement position (BF), is the building that gives rise to the LOSboundary, see figure 5.3.1-1 and figure 3.1-3. Figure 5.3.1-2 Relative mean signal strength for the measurement positions 1,7 & 8

23 (87) Figure 5.3.1-2 shows that the mean signal strength decreases faster for measurement position 1 than for position 8. Position 8 has a building in the back that is higher than the corresponding building for position 1. The difference in signal strength at lower heights is probably due to the difference in heights of these buildings, the lower building in the back (BB) is the smaller will the reflections from that building be. Table 5.3.1-1 (m) Position 1 Position 7 Position 8 Tallest Obstructing h 1 17.8 19.8 20.3 Building in the front d 1 411 100 444 (TOB), (rooftop) Building in the front, h 2 17.9 19.8 17.3 BF (rooftop) d 2 9.3 7.7 7.6 Building in the front, h 3 13.0 14.6 13.0 BF (eaves) d 3 3.7 2.5 2.5 Building in the Back h 4 3 6.6 10.0 (BB), (eaves) d 4 34.9 15.0 16.0 Base station (h BS=48m) d BS 3466 2624 2530 Notes. Position 1. There are some trees in the back with the height of 10m, 20 meters behind the measurement position. Position 7. The measurements at 7.5 & 5 m are damaged, the lane of buildings in the back have irregular heights and at least one of them have the same heights as the building in the front. There are some small trees in the back with a height of 5m, about 12m behind the measurement position. Position 8. There are some trees in the back with the height of 10m, 12 meters behind the measurement position.

Measurement position 5 and 11 24 (87) These two measurement positions also have a lower building in the back (BB) than in the front of the MS, but the difference is not as big as for measurement position 1,7 and 8 Figure 5.3.1-3 Relative mean signal strength for the measurement positions 5 & 11 Figure 5.3.1-3 shows that the mean signal strength is about the same for these two measurement positions. The height of the building BB for these positions is approximately the same as at position 8, which has the same relative mean signal strength at the height 5m. Table 5.3.1-2 (m) Position 5 Position 11 Tallest Obstructing h 1 17.8 ---- Building in the front d 1 65 ---- (TOB), (rooftop) Building in the front, h 2 17.8 19.7 BF (rooftop) d 2 9.5 8.0 Building in the front, h 3 12.6 15.0 BF (eaves) d 3 4.7 3.3 Building in the Back h 4 11.1 12.6 (BB), (eaves) d 4 18.3 9.3 Base station (h BS=48m) d BS 3055 1896 Notes. Position 5. There are trees in the back with the height of about 5m, 10 meters behind the measurement position. Position 11. The measurement position is close to a street corner (about 4m). The other street in this crossing is in the direction of the BS. There is a tree with the height of about 7m between the corner and the measurement position. There is no taller building between the BS and the building in front of the measurement position (BF).

5.3.2 LOS-boundary due to building TOB, measurement position 3,4,12 25 (87) In these cases h3 h and 4 h 1 > h2 Figure 5.3.2-1 Local geographical environment Measurement positions 3,4 and 12 all have in common that the tallest obstructing building in the front (TOB) gives rise to the LOS-boundary. At the same time as the building in the back (BB) has about the same height as the building in the front (BF), see figure 5.3.2-1 and figure 3.1-3. Figure 5.3.2-1 Relative mean signal strength for the measurement positions 3,4,12 Figure 5.3.2-1 These results differ from the rest of the measurement positions, the mean signal strength decreases almost linear from LOS down to 5m.

26 (87) Table 5.3.2 (m) Position 3 Position 4 Position 12 Tallest Obstructing h 1 17.8 17.8 19.3 Building in the front (TOB), (rooftop) d 1 200 94 118 Building in the Front h 2 12.6 12.3 13.8 (BF), (rooftop) d 2 8.5 10.1 7.2 Building in the Front h 3 10.0 10.0 10.0 (BF), (eaves) d 3 3.0 4.6 2.7 Building in the Back h 4 9.8 10.2 11.5 (BB), (eaves) d 4 19.1 17.7 8.6 Base station (h BS=48m) d BS 3255 3149 1572 Notes. Position 3. There are some trees 15m behind the measurement position with the height of 7 meters, there are also trees on both sides of the measurement position (about 5m) with the height of 7 meters. The measurement position is about 10m from the corner of the building (There is not a street corner). Position 4. There are trees on both sides of the measurement position (about 5m) with the height of 8 meters. Position 12. There are no trees in the street canyon.

27 (87) 5.3.3 LOS-boundary due to building BF, measurement position 2,6 and 10 In these cases h 4 > h and 3 h 1 > h2 Figure 5.3.3-1 Local geographical environment Measurement positions 2,6 and 10 all have in common that they have a building in the back (BB) that is higher than the building in the front (BF). At the same time as the building in front (BF) is the building that gives rise to the LOS-boundary, see figure 5.3.3-1 and figure 3.1-3. Figure 5.3.3-2 Relative mean signal strength for the measurement positions 2,6,10 In this case, when building BB is slightly taller than building BF, there is no correlation between the relative mean signal strength at the height 5m and the height of building BB.

28 (87) Table 5.3.3 (m) Position 2 Position 6 Position 10 Tallest Obstructing h 1 17.8 19.8 19.7 Building in the front d 1 268 181 190 (TOB), (rooftop) Building in the Front h 2 16.5 16.6 16.3 (BF),(rooftop) d 2 8.9 8.4 6.6 Building in the Front h 3 11.0 12.6 11.6 (BF),(eaves) d 3 3.0 3.6 1.9 Building in the Back h 4 13.7 13.6 14.6 (BB), (eaves) d 4 16.0 12.4 9.9 Base station (h BS=48m) d BS 3322 2805 2086 Notes Position 2,6 and 10. There are no trees in the street canyons at any of these measurement positions. At measurement position 2 and 10 the buildings in the front and in the back do not have uniform heights.

29 (87) 5.3.4 LOS-boundary due to building BF, measurement position 9 and13 In these cases h 4 = h and 3 h1 h2 Figure 5.3.4-1 Local geographical environment Measurement positions 9 and 13 have in common that they have a building in the back (BB) that have about same height as the building in the front (BF). At the same time as the Building in the front (BF) is the building that gives rise to the LOSboundary, see figure 5.3.4-1 and figure 3.1-3. Figure 5.3.4-2 Relative mean signal strength for the measurement positions 9 &13 In this case, when building BB and BF have about the same height. The relative mean signal strength at the height 5m is lower for measurement position 13 than for measurement position 9 this could be explained by the fact that the buildings BF and BB are taller at measurement position 13 than at measurement position 9.

30 (87) Table 5.3.4 (m) Position 9 Position 13 Tallest Obstructing h 1 20.3 ---- Building in the front d 1 444 ---- (TOB), (rooftop) Building in the front h 2 17.3 19.3 (BF) (rooftop) d 2 7.6 8.4 Building in the front, h 3 13.0 14.0 BF (eaves) d 3 2.5 3.1 Building in the Back h 4 10.0 14.2 (BB), (eaves) d 4 13.5 9.3 Base station (h BS=48m) d BS 2530 1455 Notes. Position 9. The building next to the building in front (about 5m away) is half as tall as the one in the front, there are no trees in the street canyon. Position 13. There are no trees in the street canyon, there is no taller building between the BS and the building in front of the measurement position (BF).

31 (87) 5.3.5 Summary of the mean signal strength dependence of height. Figure 5.3 Relative mean signal strength for the measurement positions 1,4,6,13 These four measurement positions have been chosen to represent the four different types of local geographical environment (see the sections 6.5.1 to 6.5.4 for more information). Measurement position Building that gives rise to the LOS-boundary Relation between the height of the eaves, building BF and BB 1 BF h 3 >> h 4 13 BF h 3 h 4 6 BF h 3 <h 4 4 TOB h 3 h 4 Table 5.3.5-1 There is a correlation between the local geographical environment and how the relative mean signal strength decreases from LOS down to 5m. But it is difficult to predict the total loss in mean signal strength from LOS to 5m since the mean signal strength has large variations at LOS. If the height of the building behind the measurement position (BB) increases, the mean signal strength for the lower heights (below the height of the eaves of the building in the back) will increase. The measurement positions 1 and 13 can be compared in figure 5.3 where measurement position 1 have a building in the back that is very low and position 13 has a building in the back that has about the same height as the building in the front. In those cases when the tallest obstructing building in the front (TOB), gives rise to the LOS-boundary (see figure 3.2). The mean signal strength decreases almost linear from LOS to 5m (compare position 4 with 1,6 and 13). This means that the variations in mean signal strength in height depends highly on the distance to the building that gives rise to the LOS-boundary.

6 Direction Of Arrival (DOA) 32 (87) 6.1 Signal analysis The direction of arrival can be determined by using beam-forming, but before this do we need to do some further analysis of the signal. The MS antenna was turned around in a circular path for more than one revolution (1.25 rev) and the overlap can be used to determine possible phase drift between the transmitting unit (MS) and the receiving unit (BS), investigations showed that there was almost no phase drift. Both the BS and the MS use rubidium standards for reference signal to ensure phase stability during measurements [1]. The resolution in the time domain is 122ns. 6.2 Beam-forming The beam-forming technique is a fairly simple method for estimating the angle of arrival, see [11], [12]. The dipole antenna was rotated in a circular horizontal path (see section 3.3 and figure 6.2). One revolution gives about 822 samples and these together can form a Synthetic Uniform Circular Array (SUCA) with 822 elements with a radius of 0.53m. Problems with mutual coupling are avoided since only one antenna element was used. The Nyquist criterion imposes in this case that the two measurement positions have to be separated less than or equal to half a wavelength, which is about 8.8cm and it is fulfilled. From the beam-forming, the angle of arrival can be fully determined in the azimuthal (horizontal) plane. In the elevation plane, the angle of arrival can be determined, but unfortunately not if the angle is positive or negative. The resolution in the azimuth plane is uniform, but not in the elevation plane, the resolution is increasing with increasing elevation angle. Due to the pattern of the omnidirectional dipole antenna, the amplitude of an incident wave with extremely high elevation angle will be underestimated.

33 (87) Figure 6.2 The geometry of the Synthetic Uniform Circular Array (SUCA) U ( Φ, Θ ) = 20 log10 W ( Φ ) B R ( Φ, Θ ) Equation 6.2-1 k m n k n n n k m where ( Φ, Θ ) R n = e k m i cos 2 π π π ( Θm ) r cos cos( Φk ) + r sin sin( Φk ) λ N N 2 n 2 n Equation 6.2-2 N =Number of antenna positions n = 0,1,,,, ( N 1) (The index of the specific antenna element) λ = Wavelength in meters r = radius of the circle in meters n number of steps in the elevation cut (in this case set to be 50) Θ nφ number of steps in the azimuth cut (In this case set to be 200) k = 0,1,, n Φ m = 0,1,, n Θ 2 π k Φk = (The azimuth angle to each antenna element) ( n Φ + 1) 2 π m Θm = (The elevation angle to each antenna element) nθ The measurement (M) are complex and comes in this case as a matrix with the size of NxT where the rows correspond to the different antenna positions and T to the number of steps in the time domain. To be able to use the beam-forming the inputs needs to be a vector and this could be done by summation in the time domain like: A 1 e d = + qx M qxk or simply choosing one of the columns (d=0). l= e T B n = [ A qx 1] Equation 6.2-3 To minimize the side lobes has this window function (W) beenappliedtothedata[1]. W ( Φ ) = cos( Φ ) k n k, cos Φ k 2 π 1 2 π ( N 1),,,,,, cos Φ k Equation 6.2-4 N N

34 (87) 6.3 Evaluation of the results from the beam-forming The beam-forming equation has been used through all measurement positions and heights and also in the time domain, the results have been plotted as contour plot with the local geographical data included (see the example below). Due to the large number of figures, those cannot be presented in this report. Figure 6.3-1 Example of the outcome from the beam-forming Figure 6.3-1 shows the DOA at measurement position 11 at 7.5m with the local geographical information included (The Buildings has here been filled to make it clearer where they are located). The horizontal axis shows the azimuth angle in degrees (0 degrees correspond to the MS-BS direction) and vertical axis the elevation angle in degrees. See figure 6.2-3 for more information about the measurement position. In this case the whole part of the signal is included (in the time domain). To make it easier to identify the waves with the strongest amplitude have a threshold of 6dB from the maximum value been used in this case. The DOA in this figure is in the area between 150 to 50 degrees along the horizontal axis (azimuth) and between 0 to 40 degrees along the vertical axis (elevation).

35 (87) Figure 6.3-2 Figure 6.3-2 shows the geometrical data for measurement position 11 The MS-BS direction is almost eastwards (see figure 3.1-1) so to make it a little bit easier for understanding have the approximate direction of north, south, west and east been marked out in figure 6.3-1 and 6.3-2.

36 (87) 6.4 Summary of the DOA analysis This section is a summary of the results from all measurement positions, to show where reflection and diffraction generally occur depending on the height of the MS. For most measurement positions, the building in the front (BF) gives rise to the LOSboundary and the building in the back (BB) has about the same height as (BF), therefore this scenario has been used in this summary, see figure 6.4.1. Local geographical environment For all measurement positions the distance to the building in the front (d 3 ) is really small (1.9 to 4.7m with the average of 3.15m), this makes the diffraction loss really high especially for the lowest heights. Since the step between each height is large (2.5m), it is difficult to get the same scenario for every measurement position. And of course, the measurement positions do not have the same environment, some of them are close to a street crossing and some have trees in the street canyon and the buildings have irregular walls and roofs. For all measurement positions the orientation of the street is perpendicular to the MS-MS direction. Figure 6.4.1 The local geographical environment 6.4.1 LOS LOS (Section A in figure 6.4.1) If the propagation path takes place in the LOS-region (over the LOS-boundary), the DOA in the azimuth plane is in the MS-BS direction, close to the LOS-boundary the DOA is more spread out, but still close to the MS-BS direction, these fluctuations in azimuth angle are lower than ± 8 degrees.

37 (87) 6.4.2 LOS to the eaves Figure 6.4.2 Top view of the measurement positions, DOA for the strongest received waves for the heights between the LOS-boundary to the eaves. Note that this figure is schematic. LOS to the eaves (Section B in figure 6.4.1) In this vertical section the diffracted waves over the building in the front (BF) are dominant and belong to the first part of the received signal in the time domain. When the height decreases does the amplitude of the diffracted part decreases. While the height is decreased, the places where diffraction occurs spreads out along the rooftop (the DOA gets spread out and is here called angular spread). See figure 6.4-2 where sector ➀ denotes the upper heights and sector ➀ and ➁ for the lower heights. The trend that the places where diffraction occurs gets spread out along the rooftop while the height is decreased could be explained by the fact that the diffraction loss depends on the elevation angle (see figure 2.1) to the diffracting edge (in this case the rooftop). The elevation angle to the rooftop will decrease when the azimuth angle relatively the MS-BS direction increases. This means that the diffraction loss will decrease when the azimuth angle relatively the MS-BS direction increases. The angular spread is generally not uniform (so as the rooftop). Right below the LOS-boundary the mean signal strength decreases rapidly and some delayed received parts of the signal (3.5 to 5 microseconds after the first received part of the signal, see figure 6.4.2-2) contribute to the mean signal strength. Generally the delayed received parts of the signal are oriented in the direction of the street, in most cases these parts of the signal comes from the north where the fjord is. The DOA and amplitudes of these parts are relatively unchanged through these heights due to the fact that the streets are straight. Close to the height of the eaves, the azimuthal DOA is spread out and almost all directions appear. Nevertheless the strength of the diffracted field (the first received part of the signal) is always strongest at these heights.

6.4.3 The Eaves to street-level, 5m 38 (87) Figure 6.4.3-1Top view of the measurement position, DOA for the strongest received waves for the heights between the eaves and street level. The eaves to street-level, 5m (Section C in figure 6.4.1) Reflections from the buildings in the back are generally dominant at these heights and belong to the first part of the received signal in the time domain, but there is still a part diffracted over the obstructing buildings in the front. When the height is decreased the area where the reflections occur gets wider (in the direction of the street), see figure 6.4.3 above, the azimuth DOA angle gets more spread out when the height is decreased. Sector ➀ denotes the upper heights and ➀ to 3 denotes the lower heights. For the lowest height, most DOA can be seen due to scattering from obstacles in the street canyon. The points of reflection (where the strongest reflection occurs) are not the same through the different heights, but the area right below the eaves of the building in the back seams to be the place where most reflections occur. Some of the measurement positions have trees in the street canyon (both close to the measurement position and in the back) according to their positions, trees are causing shadowing and scattering.

39 (87) Figure 6.4.3-2 Example of the APDP The delayed received part of the signal (3.5 to 5 microseconds after the first received part of the signal, see figure 6.4.3-2) is oriented in the direction of the street, in most cases from the north where the fjord is. The delayed received part of the signal can be seen at most of the measurement positions, especially those with a large separation distance between the BS and MS, and with a straight and wide street which ends in the fjord (See the map in earlier section, figure 3.3-1). For some of the measurement positions, the delayed received part of the signal are dominant at the lower heights but seldom at the lowest height due to the increasing number of obstacles in the street-canyon when the height is decreased. In those cases when the distance to the corner of the building in the front is shorter than the height to the rooftop, the strongest received signal comes from a diffraction in that direction. Scattering effects can be seen at all heights, for the upper heights, the scattered waves mostly come from the front and the back while for lower height, the scattering is spread out in almost all directions. The scattered waves can be seen through all parts of the signal (in the time domain). 6.4.4 Comparison with earlier made measurements at street-level Measurements at street-level (NLOS and LOS) have been done in downtown Paris [5],[6] (an urban area similar to the one in Aalborg but with taller buildings from 25 to 35m, the width of the streets is 5 to 27m and the distance between BS and MS is about 700m to 2100m). In a classical street scenario (like the measurement positions in Aalborg), where the orientation of the street is perpendicular to the BS-MS direction. Short-delayed waves come from almost all directions (uniform local scatters around the MS) and longer delayed waves confined to the direction of the street. The long-delayed waves (about 5 microseconds) have, in this case, been reflected at a very high building in the direction of the street and then coupled into the street canyon. Or if there is no LOS path between the MS and the reflecting building, the waves have been diffracted over the building roofs or at street corners in the orientation of the street. Propagation over the roofs dominates because 50% of the power have elevation angle larger than 16 degrees. It have also been shown that the shorter and narrower the street canyon gets, the more of the power are incident from the direction of the street. The overall results from all the measurement positions in Paris, including measurements done close to street corners and in streets with an orientation perpendicular to the BS-MS direction or in the BS-MS direction. The general trend is that the strongest components travel horizontally, while waves traveling over the roofs significantly contribute to the overall signal typically (about 65% of the power is incident with elevation angle over 10 degrees).

40 (87) 6.5 DOA for different local geographical environments The 13 measurement positions have been divided into four groups with respect to the local geographical environment. To make this report short, only one of the measurement positions from each group will be presented. See the section 5.3 for more information about these different groups of measurement positions. 6.5.1 LOS-boundary due to building BF, measurement position 1 Figure 6.5.1-1 Local geographical environment Measurement positions 1,5,7,8 and 11 all have in common that they have a building in the back that is lower than the ones in the front. At the same time as the building in the front (BF) gives rise to the LOS-boundary (see figure 6.5.1-1). Position 1 is most extreme of them all with a building in the back that is only 3m tall and has therefore been chosen to represent this group of measurement positions. See section 5.3.1 for more information about the local geographical environments.