1. Submission Title: [IMST time-angular characteristics analysis]
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs)
Submission Title: [IMST time-angular characteristics analysis]
Date Submitted: [March 2006]
Source: [Ali Sadri, Alexander Maltsev, Alexei Davydov]
Company: [Intel Corporation]
Address: [Intel Corporation, Evening Creek Drive ,San Diego , CA ,USA ]
Abstract: [IMST time-angular characteristics analysis]
Purpose: [To provide results for angular characteristics of the IMST data]
Notice: This document has been prepared to assist the IEEE P It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.
Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P
Ali Sadri (Intel Corporation)
<author>, <company>

2. Review of Previous Results on IMST Data Analysis
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Review of Previous Results on IMST Data Analysis
In J. Kunish et al. “MEDIAN 60GHz Wideband Indoor Radio Channel Measurements and Model” [1], VTC 1999, Vol. 4, pp , Sept and A. Sadri et al. “IMST data analysis” IEEE c” [2], basic analysis of the IMST measured data has been performed.
Omni- and directional antennas were considered
Simple time-domain parameters of the channel were calculated
Exponential parameter of multipath region
K-factor for the strongest path
Distribution of rays amplitudes in multipath region
Based on the results single cluster Saleh-Valenzuela model has been proposed in [1] with rate of multipath ray arrival = 1ns-1
Ali Sadri (Intel Corporation)
<author>, <company>

3. Goals of this Presentation
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Goals of this Presentation
Extend analysis of IMST data to characterize time-angular parameters of millimeter wave channel
Evaluate cross polarization discrimination (XPD) for linear polarization
Determine diffraction loss model
Propose enhanced channel model based on measured data
Ali Sadri (Intel Corporation)
<author>, <company>

4. Measurement scenarios plan
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Measurement scenarios plan
LOS
NLOS
Edge
Ali Sadri (Intel Corporation)
<author>, <company>

5. Measurements Scenarios
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Measurements Scenarios
Library environment with tables, chairs and metal bookshelves with books
3 main types of measurement scenarios
LOS: unobstructed line of sight conditions
Edge: partially obstructed line of sight by the edge of a metal bookshelf
NLOS: non line of sight obstructed by a densely filled bookshelf
3 types of RX antennas (horn, wideband dipole array antenna, biconical)
Fixed TX lens antenna position at the suspended ceiling, RX measurements range ~2-5m
Time resolution is 1/960MHz ≈ 1ns
Ali Sadri (Intel Corporation)
<author>, <company>

6. Antennas Used in Experiment
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Antennas Used in Experiment
Transmitting antenna
Receiving antennas
Omni directional in horizontal dielectric lens antenna; 8dBi 76º from vertical
20 dBi standard gain horn antenna
Printed array of 8x8 pentagonal dipoles with 22 dBi gain
Omni directional in horizontal biconical antenna with 9 dBi gain
Ali Sadri (Intel Corporation)
<author>, <company>

7. Data Processing Methodology
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Data Processing Methodology
Original measurements were performed in frequency domain using vector network analyzer
The measured frequency response was windowed by Blackman-Harris window and converted to time domain using IFFT
Originally the Channel frequency response was measured over several RX positions. For each position virtual uniform linear array was synthesized by shifting RX antenna along horizontal straight line
For determination of DoA characteristics, Fourier transform with spatial windowing procedures were used
Ali Sadri (Intel Corporation)
<author>, <company>

8. Cross Polarization Discrimination (XPD) Analysis
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Cross Polarization Discrimination (XPD) Analysis
Polarization may be considered as a way of multipath component rejection
Reflection and diffraction may change polarization orientation and increase cross-polarized component
For IMST experiment only linear vertical polarization was used at transmitter
Ali Sadri (Intel Corporation)
<author>, <company>

9. Some XPD Observations from the Measurements
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Some XPD Observations from the Measurements
Co-polarized Rx antenna
Cross-polarized Rx antenna
Significant attenuation of LOS component (~20 dB) for cross polarized antenna
Almost the same attenuation profile for all rays in multipath region (~10 dB)
Ali Sadri (Intel Corporation)
<author>, <company>

10. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Diffraction Loss
Due to diffraction radio waves may propagate in region of shadow
The signal attenuation in shadow region is angle dependent
Diffraction loss was measured for ‘edge’ scenario, when metal bookshelf wall plays a role of “knife” edge
The difference between two signal levels for free space and for shadow region was taken as the diffraction loss
Ali Sadri (Intel Corporation)
<author>, <company>

11. Diffraction Loss Model
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Diffraction Loss Model
Ali Sadri (Intel Corporation)
<author>, <company>

12. Observation of Diffraction Loss
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Observation of Diffraction Loss
LOS component change from experiment
Theoretical LOS component change
~6 dB
Diffraction loss shows linear decay in dB scale for shadow region; Slope of the curve is distance dependent
Ali Sadri (Intel Corporation)
<author>, <company>

13. DoA Parameters for mmWave Channel
<month year>
doc: IEEE c
Tuesday, March 07, 2006
DoA Parameters for mmWave Channel
DoA parameters of the channel are important for characterization of multi-antenna systems and systems with directional antennas, since these models may take into account the effects of specific Rx antenna radiation pattern G(φ,θ)
Determination of DoA/ToA parameters is required for accurate description of reflected waves/multipath components in channel model at 60 GHz
Ali Sadri (Intel Corporation)
<author>, <company>

14. Direction of Arrival Diagrams
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Direction of Arrival Diagrams
Ali Sadri (Intel Corporation)
<author>, <company>

15. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Direction of Arrival vs. Time of Arrival (LOS scenario, horn/biconical antennas)
Pulses in the first part of PDP mostly come from single scattered (reflected) rays
Magnitude of response is different due to spread angular distribution of the reflected rays
Slope is almost the same for all angles in the second part of PDP
Ali Sadri (Intel Corporation)
<author>, <company>

16. Direction of Arrival vs. Time of Arrival (LOS scenario, horn antenna)
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Direction of Arrival vs. Time of Arrival (LOS scenario, horn antenna)
Multipath components are largely attenuated by directional antenna
Ali Sadri (Intel Corporation)
<author>, <company>

17. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Direction of Arrival vs. Time of Arrival (LOS scenario, biconical antenna)
K ~ 7 dB
K ~ 15 dB
K < -50 dB
Several strongly localized rays in time/angular dimensions for delays < 50 ns and angles < +-700
Strong multipath components may have Rician distribution or may be almost deterministic with random phases
Ali Sadri (Intel Corporation)
<author>, <company>

18. Direction of arrival vs. Time of arrival (NLOS scenario, horn antenna)
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Direction of arrival vs. Time of arrival (NLOS scenario, horn antenna)
Broadening of direction of arrival angles due to scattering during the propagation of waves through the books
Ali Sadri (Intel Corporation)
<author>, <company>

19. DoA for Different Delays
<month year>
doc: IEEE c
Tuesday, March 07, 2006
DoA for Different Delays
Ali Sadri (Intel Corporation)
<author>, <company>

20. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Main Conclusions
The most useful and valuable are the measurements performed with the help of omni-directional (bi-conical antenna) that allowed, through the synthesis of large antenna array, to resolve the arriving rays and to estimate their parameters (DoA, ToA and power)
Measurements and signals processing performed in such a way showed that the received signal has specific “speckle-structure” which manifests itself as the number of strong single reflected rays in the received signal (“bright speckles” not forming the clusters in time-angle plane)
Ali Sadri (Intel Corporation)
<author>, <company>

21. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Summary
The channel response function for indoor environment in mmWave band is the result of three components: the LOS ray, the number of single reflected rays, and the small background diffuse multipath component ( dB).
Free space propagation and first order reflections are the dominant propagation mechanisms. The detailed analysis in time-angle domain shows that these components may be of almost the same order (see slides 14, 16, 18).
Strong multipath components are results of strong reflectors, such as walls, floor, metallic furniture, and other plane surfaces. Single reflected waves are strongly angle/time localized, have magnitude comparable to the magnitude of LOS component, and have Rician distribution with large K-factor (~7 dB)
Note: The similar values of magnitudes of multipath rays and LOS component in the first part of power delay profile may be explained by the difference in heights of the transmit and the receive antennas and dependence of the antenna pattern on elevation angle
Ali Sadri (Intel Corporation)
<author>, <company>

22. <month year>
doc: IEEE c
Tuesday, March 07, 2006
Summary (continued)
Diffuse multipath component is caused by multiple scattering. Multiple reflected rays are largely attenuated and arrived isotropically from all azimuth angles. The power delay profile of these rays follows the simple exponential law. These rays are substantially Rayleigh distributed
Directional antennas provide good angular resolution of multipath components and significantly attenuate most of multipath rays. At the same time small misalignment of TX/RX antenna pointing may result to significant attenuation of the LOS ray by ~10dB [1]
Ali Sadri (Intel Corporation)
<author>, <company>

23. Suggestions for modeling (to be discussed)
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Suggestions for modeling (to be discussed)
The “speckle” channel model in time-angle domain may be proposed as a basic channel model for indoor mmWave multipath propagation. Models of such a type are known in optics, and it has been discovered that mmWave signals in indoor environment behave more like optic wave rather than radio waves of GHz bands.
The speckle channel model may be described as two dimensional tapped delay/angle model with specific power profile P(φm,τn) consisting of three components: the LOS ray, the number of single reflected rays, and the small background diffuse multipath component.
Open issue
Need to agree on the statistical characteristics of single reflected rays in typical indoor environment and include them into the speckle channel model
Ali Sadri (Intel Corporation)
<author>, <company>

24. <month year>
doc: IEEE c
Tuesday, March 07, 2006
References
J. Kunish et al. “MEDIAN 60GHz Wideband Indoor Radio Channel Measurements and Model”, VTC 1999, Vol. 4, pp , Sept. 1999
A. Sadri et al. “IMST data analysis” IEEE c
C. Loyez et al. “Indoor 60 GHz radio channel sounding and related T/R module considerations for high data rate communications”. Electronics Letters, Vol. 37, No. 10, May 2001, pp
Ali Sadri (Intel Corporation)
<author>, <company>

25. Appendix <month year> doc: IEEE 802.15-05-0394-00-003c
Tuesday, March 07, 2006
Appendix
Ali Sadri (Intel Corporation)
<author>, <company>

26. <month year> doc: IEEE 802.15-05-0394-00-003c
rxpos
scen rx
mean excess delay, ns
delay spread, ns
delay window, W90%
XPD, dB
czk
los
h01
1.19
1.35
1.9
13.5
cyf
1.72
1.8 
14.1 
cxe
1.23 2 
11.6 
cxh
p04
1.13
1.9 
 N/A
cyj
1.14
1.08
N/A  
czn
1.18
1.1
ewg
1.25
1.24
2.2 
exj
eym
1.21
2.1 
ezo
1.11
ezl
1.5
10.2 
eyj
1.47
7.7 
exg
1.26
1.75
11.9 
ewd
1.88 
17.9 
fzg
edge
1.31
2.22 
13.2 
dyi
nlos
1.61
3.19
dym
1.3
czm
b01
21.35
23.77 
30.2
cym
12.6
16.83 
27 
cxm
12.5
17.08 
28.2 
fzm
14.6
16.65 
25.3 
ewm
8.47
12.83 
26.4 
exm
12.85
14.64 
21.7 
16.67
18.45 
26.7 
ezm
15.17
17.73 
25.7 
Tuesday, March 07, 2006
Ali Sadri (Intel Corporation)
<author>, <company>

27. Typical time-domain impulse response structure
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Typical time-domain impulse response structure
First part of PDP is composed from LOS component and first-order reflections. Pulses may have Rician distribution and are usually largely attenuated when directional antenna is used.
LOS component
Second part of PDP is composed from diffuse rays arriving from all directions. The power delay profile of this part can be accurately described by exponential function. The fading of multipath components can be characterized by Raleigh distribution.
Ali Sadri (Intel Corporation)
<author>, <company>

28. Statistics of multipath amplitudes
<month year>
doc: IEEE c
Tuesday, March 07, 2006
Statistics of multipath amplitudes
Ali Sadri (Intel Corporation)
<author>, <company>