Project: IEEE P82.15 Working Group for Wireless Personal Area Networks N (WPANs( WPANs) Title: [UWB Channel Model for Indoor Residential Environment] Date Submitted: [2 September, 24] Source: [Chia-Chin Chong, Youngeil Kim, SeongSoo Lee] Company [Samsung Advanced Institute of Technology (SAIT)] Address [RF Technology Group, Comm. & Networking Lab., P. O. Box 111, Suwon 44-6, Korea.] Voice:[+82-31-28-6865], FAX: [+82-31-28-9555], E-Mail: [chiachin.chong@samsung.com] Re: [Response to Call for Contributions on IEEE 82.15.4a Channel Models] Abstract: [This contribution describes the UWB channel measurement results in indoor residential environment based in several types of high-rise apartments. It consists of detailed characterization of the frequency-domain parameters, temporal-domain parameters, small-scale amplitude statistics and S-V clustering multipath channel parameters of the UWB channel with bandwidth from 3 to 1 GHz.] Purpose: [Contribution towards the IEEE 82.15.4a Channel Modeling Subgroup.] Notice: This document has been prepared to assist the IEEE P82.15. 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 P82.15. Slide 1
UWB Channel Model for Indoor Residential Environment Chia-Chin Chong, Youngeil Kim, SeongSoo Lee Samsung Advanced Institute of Technology (SAIT), Korea Slide 2
Outline Measurement Setup & Environment Data Analysis & Post-Processing Measurement Results Large-Scale Parameters Small-Scale Parameters Conclusion Slide 3
Measurement Setup (1) Frequency domain technique using VNA Center frequency, f c : 6.5GHz Bandwidth, B : 7GHz (i.e. 3-1GHz) Delay resolution, τ : 142.9ps (i.e. τ=1/b) No. frequency points, N : 161 Frequency step, f : 4.375MHz (i.e. f=b/(n-1)) Max. excess delay, τ max : 229.6ns (i.e. τ max =1/ f) Sweeping time, t sw : 8ms Max. Doppler shift, f d,max : 1.25Hz (i.e. f d,max =1/t sw ) Slide 4
Measurement Setup (2) UWB wideband planar dipole antennas Measurement controlled by laptop with LabVIEW via GPIB interface Calibration performed in an anechoic chamber with 1m reference separation Static environment during recording Both large-scale & small-scale measurements Large-scale: different RX positions local point Small-scale: 25 (5x5) grid-measurements around each local point spatial point At each spatial point, 3 time-snapshots of the channel complex frequency responses are recorded Slide 5
Measurement Setup (3) TX antenna RX antenna Propagation Channel Coaxial Cables Vector network analyzer (Agilent 8722ES) Low Noise Amplifier (Miteq AFS5) Power Amplifier (Agilent 832A) Laptop with LabVIEW GPIB Interface Attenuator (Agilent 8496B) Slide 6
UWB Planar Dipole Antenna Slide 7
Measurement Environment Measurements in various types of high-rise apartments based on several cities in Korea typical types in Asia countries like Korea, Japan, Singapore, Hong Kong, etc. 3-bedrooms (Apart1) 4-bedrooms (Apart2) Both LOS and NLOS configurations TX-RX antennas: Separations: up to 25m Height: 1.25m (with ceiling height of 2.5m) TX antenna: always fixed in the center of the living room RX antenna: moved around the apartment (i.e. 8-1 locations) 12, channel complex frequency responses are collected (i.e. 2 apartments x 8 RX local points x 25 spatial points x 3 time snapshots 2x8x25x3=12,) Slide 8
3-Bedroom Apartment Grid-Measurement Slide 9
4-Bedroom Apartment Slide 1
Data Analysis & Post-Processing All measurement data are calibrated with the calibration data measured in anechoic chamber to remove effect of measurement system Perform frequency domain windowing to reduce the leakage problem Complex passband IFFT is deployed to transform the complex frequency response to complex impulse response Perform temporal domain binning before extract channel parameters Slide 11
Complex Passband IFFT Slide 12
Channel Model Description Large-Scale Parameters: Path loss and Shadowing Frequency Decaying Factor Small-Scale Parameters: Temporal Domain Parameters S-V Multipath Channel Parameters Small-Scale Amplitude Statistics Slide 13
Path Loss and Shadowing Path loss (PL) vs. Distance (d): d PL ( d) = PL + 1 n log 1 + S; d d d d = 1m PL : intercept n : path loss exponent S : Shadowing fading parameter Perform linear regression to the above equation with measured data to extract the required parameters Slide 14
Path Loss vs. Distance LOS 6 Path Loss under LOS Scenario in 3-Bedroom Apartment Data Linear Regression 58 56 Path Loss (db) 54 52 7 Path Loss under LOS Scenario in 4-Bedroom Apartment 5 65 48 1 2 3 4 5 6 7 8 1log (Distance) (m) 1 Path Loss (db) 6 55 5 Data Linear Regression 45 1 2 3 4 5 6 7 8 1log (Distance) (m) 1 Slide 15
Path Loss vs. Distance NLOS 65 Path Loss under NLOS Scenario in 3-Bedroom Apartment Data Linear Regression 6 Path Loss (db) 55 5 45 4 1 2 3 4 5 6 7 8 9 1log (Distance) (m) 1 8 75 7 Path Loss under NLOS Scenario in 4-Bedroom Apartment Data Linear Regression Path Loss (db) 65 6 55 5 45 1 2 3 4 5 6 7 8 9 1log (Distance) (m) 1 Slide 16
Frequency Decaying Factor Path loss (PL) vs. Frequency (f): PL f ( ) exp( δ f ) 1 (Method 1) or δ2 PL ( f ) f (Method 2) Slide 17
Frequency Decaying Factor LOS September 24 Frequency Dependence Path Loss under LOS Scenario (3-Bedroom Apartment Measurement Data Method 1-1 Method 2 Path Loss [db] -2-3 -4-5 -6 3 4 5 6 7 8 9 1 Frequency [GHz] Frequency Dependence Path Loss under LOS Scenario (4-Bedroom Apartment Measurement Data -5 Method 1 Method 2-1 Path Loss [db] -15-2 -25-3 -35-4 Slide 18-45 3 4 5 6 7 8 9 1 Frequency [GHz]
Frequency Decaying Factor NLOS September 24 requency Dependence Path Loss under NLOS Scenario - (3-Bedroom Apartme Measurement Data Method 1-1 Method 2 Path Loss [db] -2-3 -4-5 Frequency Dependence Path Loss under NLOS Scenario (4-Bedroom Apartmen -5-1 -6 3 4 5 6 7 8 9 1 Frequency [GHz] Path Loss [db] -15-2 -25-3 -35-4 Measurement Data -45 Method 1 Method 2-5 3 4 5 6 7 8 9 1 Frequency [GHz] Slide 19
Large-Scale Parameters Slide 2
Temporal Domain Parameters These parameters were obtained after taking frequency domain Hamming windowing, passband IFFT & temporal domain binning Slide 21
S-V Multipath Channel Parameters Saleh-Valenzuela model described by the following parameters: Γ : cluster decay factor γ : ray decay factor Λ : mean cluster arrival rate λ : mean ray arrival rate σ a : standard deviation of lognormal distributed path powers (db) L mean : Mean number of clusters µ Kl : Mean of the exponential distributed number of MPCs per cluster Slide 22
Normalized Cluster Relative Amplitude September 24 Cluster Decay Factor, Γ LOS Normalized Cluster Relative Power vs. Cluster Relative Delay (Apart1-LOS) 1 1 1 1-1 1-2 1-3 Cluster amplitude Linear least-squares fit Γ = 22.1 ns 1 2 3 4 5 Cluster Relative Delay, [ns] Normalized Cluster Relative Amplitude Normalized 1 Cluster Relative Power vs. Cluster Relative Delay (Apart2-LOS) 1-1 1-2 1-3 Cluster amplitude Linear least-squares fit Γ = 23.95 ns 1-4 1 2 3 4 5 6 Cluster Relative Delay, [ns] Slide 23
Cluster Decay Factor, Γ NLOS September 24 Normalized 1 2 Cluster Relative Power vs. Cluster Relative Delay (Apart1-NLOS) Normalized Cluster Relative Power 1 1-2 1-4 1-6 Cluster power Linear least-squares fit Γ = 51.47 ns 2 4 6 8 1 12 Cluster Relative Delay, [ns] Normalized 1 1 Cluster Relative Power vs. Cluster Relative Delay (Apart2-NLOS) Normalized Cluster Relative Power 1 1-1 1-2 1-3 1-4 Cluster power Linear least-squares fit Γ = 36.86 ns 1-5 1 2 3 4 5 6 7 Cluster Relative Delay, [ns] Slide 24
Ray Decay Factor, γ LOS Slide 25
Ray Decay Factor, γ NLOS Slide 26
Cluster Arrival Rate, Λ LOS September 24 Cluster Inter-Arrival Times CCDF (Apart1-LOS) -1-2 ln(ccdf) -3-4 1/Λ = 8.69 ns Cluster Inter-Arrival Times CCDF (Apart2-LOS) -5 Cluster inter-arrival times Linear least-squares fit -6 5 1 15 2 Clsuter Inter-Arrival Times, T [ns] ln(ccdf) -1-2 -3-4 1/Λ = 11.79 ns -5 Cluster inter-arrival times Linear least-squares fit -6 1 2 3 4 5 6 Clsuter Inter-Arrival Times, T [ns] Slide 27
Cluster Arrival Rate, Λ NLOS September 24 Cluster Inter-Arrival Times CCDF (Apart1-NLOS) -1-2 ln(ccdf) -3-4 -5 1/Λ = 21.45 ns Cluster Inter-Arrival Times CCDF (Apart2-NLOS) -6 Cluster inter-arrival times Linear least-squares fit -7 1 2 3 4 5 6 7 8 Clsuter Inter-Arrival Times, T [ns] ln(ccdf) -1-2 -3-4 1/Λ = 15.65 ns -5 Cluster inter-arrival times Linear least-squares fit -6 5 1 15 2 25 3 35 4 Clsuter Inter-Arrival Times, T [ns] Slide 28
Ray Arrival Rate, λ LOS Ray Intra-Arrival Times CCDF (Apart1-LOS) -5 ln(ccdf) -1-15 1/λ =.51 ns Ray Intra-Arrival Times CCDF (Apart2-LOS) -2 Ray intra-arrival times Linear least-squares fit -25 2 4 6 8 1 12 Ray Intra-Arrival Times, τ [ns] ln(ccdf) -2-4 -6-8 -1-12 -14 1/λ =.86 ns -16 Ray intra-arrival times Linear least-squares fit -18 5 1 15 Ray Intra-Arrival Times, τ [ns] Slide 29
Ray Arrival Rate, λ NLOS Ray Intra-Arrival Times CCDF (Apart1-NLOS) ln(ccdf) -5-1 -15-2 -25-3 -35-4 1/λ =.72 ns -45 Ray intra-arrival times Linear least-squares fit -5 5 1 15 2 25 3 35 Ray Intra-Arrival Times, τ [ns] ln(ccdf) -5-1 -15-2 Ray Intra-Arrival Times CCDF (Apart2-NLOS) 1/λ =.56 ns Slide 3-25 Ray intra-arrival times Linear least-squares fit -3 2 4 6 8 1 12 14 16 Ray Intra-Arrival Times, τ [ns]
Mixture Poisson Distribution Fitting the ray arrival times to a mixture of 2 Poisson distributions similar to [1]: ( ) = exp[ ( )] p τ τ βλ λ τ τ kl, ( k 1), l 1 1 kl, ( k 1), l ( β 1) λ exp[ λ ( τ τ )] + 2 2 kl, ( k 1), l β: mixture probability λ 1 & λ 2 : ray arrival rates Slide 31
Mixture Poisson Distributions LOS September 24 Slide 32
Mixture Poisson Distributions NLOS September 24 Slide 33
Number of Clusters.4 Apart1-LOS.4 Apart1-NLOS Probability.3.2.1 Probability.3.2.1.8 1 2 3 4 5 6 No. of Clusters Apart2-LOS.8 2 3 4 5 6 7 8 No. of Clusters Apart2-NLOS Probability.6.4.2 Probability.6.4.2 1 2 3 4 No. of Clusters 1 2 3 4 5 No. of Clusters Slide 34
Number of MPCs per Cluster 1 Apart1-LOS 1 Apart1-NLOS µ Kl = 24.1 µ Kl = 87.19 CDF.5 Empirical Data Exponential Fit CDF.5 Empirical Data Exponential Fit 5 1 15 2 1 No. of MPCs per Cluster Apart2-LOS 5 1 1 No. of MPCs per Cluster Apart2-NLOS µ Kl = 3.47 µ Kl = 117.36 CDF.5 Empirical Data Exponential Fit CDF.5 Empirical Data Exponential Fit 1 2 3 No. of MPCs per Cluster 5 1 15 No. of MPCs per Cluster Slide 35
S-V Multipath Channel Parameters Slide 36
Small-Scale Amplitude Statistics Comparison of empirical path amplitude distribution with the following four commonly used theoretical distributions: Lognormal Nakagami Rayleigh Ricean Weibull The goodness-of-fit of the received signal amplitudes is evaluated using Kolmogorov-Smirnov (K-S) test & Chi-Square (χ 2 ) test with 5% and 1% significance level, respectively. Slide 37
Goodness-of-Test: LOS Slide 38
Goodness-of-Test: NLOS Slide 39
CDF of Path Amplitude LOS Small-scale amplitude CDFs at different exces delays (3-bedroom apartment, LOS) 1.9.8.7 Empirical Lognormal Nakagami Rayleigh Weibull.6 CDF.5.4.3 Excess delay 1ns Excess delay 5ns.2.1 Excess delay 5ns -12-1 -8-6 -4-2 Amplitude in db Small-scale amplitude CDFs at different exces delays (4-bedroom apartment, LOS) 1.9.8.7 Empirical Lognormal Nakagami Rayleigh Weibull CDF.6.5 Excess delay 5ns.4.3 Excess delay 1ns.2.1 Excess delay 5ns Slide 4-12 -1-8 -6-4 -2 Amplitude in db
CDF of Path Amplitude NLOS Small-scale amplitude CDFs at different exces delays (3-bedroom apartment, NLOS) 1.9.8.7 Empirical Lognormal Nakagami Rayleigh Weibull CDF.6.5.4.3 Excess delay 5ns.2.1 Excess delay 1ns Excess delay 5ns -12-1 -8-6 -4-2 2 Amplitude in db Small-scale amplitude CDFs at different exces delays (4-bedroom apartment, NLOS) 1.9.8.7.6 Empirical Lognormal Nakagami Rayleigh Weibull CDF.5.4.3 Excess delay 5ns.2.1 Excess delay 1ns Excess delay 5ns -14-12 -1-8 -6-4 -2 2 Amplitude in db Slide 41
Small-Scale Amplitude Statistics Parameters The results demonstrate that lognormal, Nakagami and Weibull fit the measurement data well. Parameters of these distributions (i.e. standard deviation of lognormal PDF, m-parameter of Nakagami PDF and b-parameter of Weibull PDF) can be modeled by a lognormal distribution These parameters are almost constant across the excess delay Slide 42
Standard Deviation of Lognormal PDF LOS September 24 1 Lognormal Standard Deviation CDF (Apart1-LOS) CDF.9.8.7.6.5.4.3.2 Empirical CDF Lognormal CDF 1.9 Lognormal Standard Deviation CDF (Apart2-LOS).1.5 1 1.5 2 2.5 Lognormal Standard Deviation CDF.8.7.6.5.4.3.2.1 Empirical CDF Lognormal CDF.5 1 1.5 2 2.5 3 Lognormal Standard Deviation Slide 43
Standard Deviation of Lognormal PDF NLOS September 24 1 Lognormal Standard Deviation CDF (Apart1-NLOS).9.8.7 Empirical CDF Lognormal CDF CDF.6.5.4.3.2.1.5 1 1.5 2 2.5 3 Lognormal Standard Deviation CDF 1.9.8.7.6.5.4.3.2.1 Lognormal Standard Deviation CDF (Apart2-NLOS) Empirical CDF Lognormal CDF.5 1 1.5 2 2.5 3 Lognormal Standard Deviation Slide 44
m-nakagami Parameter LOS 1 Nakagami-m Values CDF (Apart1-LOS).9.8.7 Empirical CDF Lognormal CDF CDF.6.5.4.3.2.1.5 1 1.5 2 Nakagami-m Values CDF 1.9.8.7.6.5.4.3.2.1 Nakagami-m Values CDF (Apart2-LOS) Empirical CDF Lognormal CDF Slide 45.5 1 1.5 2 2.5 3 Nakagami-m Values
m-nakagami Parameter NLOS 1 Nakagami-m Values CDF (Apart1-NLOS).9.8.7 Empirical CDF Lognormal CDF CDF.6.5.4.3.2.1 1 2 3 4 5 6 7 Nakagami-m Values CDF 1.9.8.7.6.5.4.3.2.1 Nakagami-m Values CDF (Apart2-NLOS) Empirical CDF Lognormal CDF Slide 46.5 1 1.5 2 Nakagami-m Values
b-weibull Parameter LOS 1 Weibull-b Values CDF (Apart1-LOS).9.8.7 Empirical CDF Lognormal CDF CDF.6.5.4.3.2.1.5 1 1.5 2 2.5 3 Weibull-b Values CDF 1.9.8.7.6.5.4.3.2.1 Weibull-b Values CDF (Apart2-LOS) Empirical CDF Lognormal CDF Slide 47.5 1 1.5 2 2.5 3 3.5 4 Weibull-b Values
b-weibull Parameter NLOS 1 Weibull-b Values CDF (Apart1-NLOS).9.8.7 Empirical CDF Lognormal CDF CDF.6.5.4.3.2.1 1 2 3 4 5 6 7 8 Weibull-b Values CDF 1.9.8.7.6.5.4.3.2.1 Weibull-b Values CDF (Apart2-NLOS) Empirical CDF Lognormal CDF.5 1 1.5 2 2.5 3 Weibull-b Values Slide 48
Variations of Lognormal-σ with Delay LOS September 24 Lognormal Standard Deviation vs. Excess Delay (Apart1-LOS) 2.5 Lognormal Standard Deviation 2 1.5 1.5 Lognormal Standard Deviation vs. Excess Delay (Apart2-LOS) 2.5 1 2 3 4 5 6 Excess Delay, τ [ns] Lognormal Standard Deviation 2 1.5 1.5 1 2 3 4 5 6 7 8 Excess Delay, τ [ns] Slide 49
Variations of Lognormal-σ with Delay NLOS September 24 Slide 5
Variations of Nakagami-m with Delay LOS September 24 2 Nakagami-m Values vs. Excess Delay (Apart1-LOS) 1.8 1.6 Nakagami-m Values 1.4 1.2 1.8 3 Nakagami-m Values vs. Excess Delay (Apart2-LOS).6.4 2.5.2 1 2 3 4 5 6 Excess Delay, τ [ns] Nakagami-m Values 2 1.5 1.5 1 2 3 4 5 6 7 8 Excess Delay, τ [ns] Slide 51
Variations of Nakagami-m with Delay NLOS September 24 Slide 52
Variations of Weibull-b with Delay LOS September 24 3 Weibull-b Values vs. Excess Delay (Apart1-LOS) 2.5 Weibull-b Values 2 1.5 4 Weibull-b Values vs. Excess Delay (Apart2-LOS) 1 3.5.5 1 2 3 4 5 6 Excess Delay, τ [ns] Weibull-b Values 3 2.5 2 1.5 1.5 1 2 3 4 5 6 7 8 Excess Delay, τ [ns] Slide 53
Variations of Weibull-b with Delay NLOS September 24 Slide 54
Small-Scale Amplitude Statistics Parameters Slide 55
Conclusion Frequency domain technique UWB measurement campaign has been carried out in indoor residential environment (high-rise apartments) covering frequencies from 3-1 GHz. Measurement covered both LOS & NLOS scenarios. Channel measurement results which characterize both the large-scale and small-scale parameters of the channel are reported. Slide 56
Reference 1. B. Kannan et. al., Characterization of UWB Channels: Small- Scale Parameters for Indoor and Outdoor Office Environment, IEEE 82.15-4-385--4a, July 24. Slide 57