Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Saleh-Valenzuela Channel Model Parameters for Library Environment] Date Submitted: [July 2006] Source: [Alexei Davydov,Alexander Maltsev, Ali Sadri] Company: [Intel Corporation] Address: [Intel Corporation, 603950 Turgeneva 30, Nizhny Novgorod, Russia], E-mail: [alexei.davydov@intel.com, alexander.maltsev@intel.com, ali.sadri@intel.com] Abstract: [This contribution contains the parameters for Saleh-Valenzuela channel model with direction-of-arrival extension extracted from IMST data for library environment] Purpose: [Contribution to 802.15 TG3c at July 2006 meeting in San-Diego, USA] Notice: This document has been prepared to assist the IEEE P802.15. 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 P802.15. Alexei Davydov (Intel Corporation)

2. Goals • Present estimated parameters for Saleh-Valenzuela model with direction-of-arrival extension (currently considered by 802.15.3c channel model subgroup) for library environment • Minimize the impact of TX/RX antenna patterns in IMST measurements data on extracted parameters of the channel model Alexei Davydov (Intel Corporation)

3. 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 • Two types of virtual uniform antenna arrays for direction of arrival analysis • 501x1 uniform linear array with 1mm antenna spacing (los scenarios) • 301x51 uniform planar antenna array with 1mm antenna spacing (edge scenario) Alexei Davydov (Intel Corporation)

4. Measurements Scenarios Plan LOS NLOS Edge Measurements data for biconical RX antenna in pure NLOS scenario is not available Alexei Davydov (Intel Corporation)

5. 40 35 30 KLOS 25 Relative Power, [dB] 20 KMP 15 10 5 0 0 10 20 30 40 50 60 70 80 Delay, [ns] Saleh-Valenzuela Channel Model • L – Number of clusters • Kl – Number of MPC in the lthcluster • KLOS – LOS K factor • KMP – Cluster K factor • αk,l – MPC complex amplitude • Tl – Time of arrival of lthcluster • τk,l – Relative time of arrival for kth MPC within lthcluster • θk,l – Relative direction of arrival for kth MPC within lthcluster • Θl – Direction of arrival of lthcluster • δ(·) – Delta function Alexei Davydov (Intel Corporation)

6. 0.015 Empirical pdf Uniform pdf 0.01 Probability Density 0.005 0 -150 -100 -50 0 50 100 150 Cluster DoA, [deg] 1 Empirical cdf Uniform cdf 0.9 0.8 0.7 0.6 Cumulative Density 0.5 0.4 0.3 0.2 0.1 0 -150 -100 -50 0 50 100 150 Cluster DoA, [deg] Inter-Cluster Parameters (1) 00 – reference direction: RX antenna pointed in the TX antenna direction. Uniform distribution may be used for cluster DoA modeling. Alexei Davydov (Intel Corporation)

7. 0.25 1 Empirical cdf Empirical cdf Exponetial cdf Exponetial cdf 0.9 0.2 0.8 0.7 0.15 0.6 Cumulative Density Probability Density 0.5 0.1 0.4 0.3 0.05 0.2 0.1 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Cluster inter-arrival time, [ns] Cluster inter-arrival time, [ns] Inter-Cluster Parameters (2) Poisson process (Λ = 0.25, [ns-1]) may be used to model cluster arrival times Alexei Davydov (Intel Corporation)

8. 20 20 Empirical data Empirical data Linear LS fit Linear LS fit 10 10 0 0 -10 -10 Relative Power, [dB] Relative Power, [dB] -20 -20 -30 -30 -40 -40 -50 -50 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 Cluster arrival time, [ns] Cluster arrival time, [ns] Inter-Cluster Parameters (3) Flat-Exp aprx of cluster PDP Exp aprx of cluster PDP KLOS= 8, [dB] KLOS= 8, [dB] Δ= 11, [dB] τ= 30, [ns] Cluster decay (Γ= 12 [ns])was estimated from clusters arriving with delays > 30 ns Note: Amplitudes of cluster arriving with delay < 30 [ns] are affected by TX beam shaped antenna. Two approximations of cluster PDP are possible. Alexei Davydov (Intel Corporation)

9. Empirical cdf Empirical pdf Normal cdf Normal pdf Laplacian cdf Laplacian pdf Cumulative Density Probability Density 0 MPC normalizedDoA MPC normalizedDoA Intra-Cluster MPC parameters (1) Various empirical pdf’s for MPC DoA estimated from the measurements data may be approximately described by common Gaussian distribution with fixed angle spread (σAS = 100). Alexei Davydov (Intel Corporation)

10. 4.5 1 Empirical pdf Exponetial pdf 0.9 4 0.8 3.5 0.7 3 0.6 2.5 Probability Density Cumulative Density 0.5 2 0.4 1.5 0.3 1 0.2 0.5 0.1 0 0 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 MPC inter-arrival time, [ns] MPC inter-arrival time, [ns] Intra-Cluster MPC parameters (2) Poisson process (λ = 4, [ns-1]) may be used to model intra-cluster MPC arrival times Alexei Davydov (Intel Corporation)

11. 0 Empirical Data Linear LS fit -5 -10 -15 Relative Power, [dB] -20 -25 -30 -35 -40 0 5 10 15 MPC relative arrival time, [ns] Intra-Cluster MPC parameters (3) Ray decay constant (γ= 7 [ns]) and cluster K-factor (KMP= -13 [dB]) were estimated for cluster group arriving in interval from 0 [ns] to 50 [ns] Alexei Davydov (Intel Corporation)

12. 1.4 1 Empirical pdf Empirical cdf 0.9 Log-normal pdf Log-normal cdf 1.2 Rayleigh pdf Rayeigh cdf 0.8 1 0.7 0.6 0.8 0.5 Cumulative Density Probability Density 0.6 0.4 0.3 0.4 0.2 0.2 0.1 0 0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 MPC amplitude MPC amplitude Intra-Cluster MPC parameters (4) Log-normal distribution (σ2 = 6[dB]) may be used for intra-cluster MPC amplitude modeling. Rayeligh distribution gives worse approximation of empirical data. Alexei Davydov (Intel Corporation)

13. -90 -95 -100 -105 Relative Power, [dB] -110 -115 -120 -125 -130 0 100 200 300 400 500 Spacing, [mm] Edge Scenario Shadow region with blocked LOS component LOS region The scenario with blocked LOS component (“partial NLOS”) can covered with LOS model by dropping the direct path by 10-15 dB and keeping the remaining parameters unchanged. Alexei Davydov (Intel Corporation)

14. Extracted channel model parameters • Inter-cluster Power Delay Profile parameters • Exponential PDP:K factor KLOS= 8 [dB], cluster decay Γ = 12 [ns] • Flat-Exponential PDP: K factor KLOS= 8 [dB], cluster decay Γ = 12 [ns], Δ = 11 [dB], τ = 30 [ns] • Inter-cluster DoA – Uniform distribution • Intra-cluster DoA – Gaussian (Angles Spread (AS) σAS = 10 [0]) • Inter-cluster ToA / Intra-cluster ToA – Poisson (Λ = 0.25 [ns-1] / λ = 4 [ns-1]) • Cluster / Ray amplitude – Log-normal (σ1 = 5 [dB] / σ2 = 6 [dB]) • Intra-cluster K factor KMP= -13 [dB] • Intra-cluster ray decay γ = 7 [ns] Note: Two approximations of cluster PDP are proposed: Exponential PDP - for TX omni-directional antenna mounted at the suspended ceiling. Flat-exponential PDP - for TX beam-shaped antenna mounted at the suspended ceiling. Alexei Davydov (Intel Corporation)

15. Summary • Based on IMST measurements data the parameters of Sale-Valenzuela channel models with DoA extension were extracted for the LOS scenario in library environment • Exponential PDP - for TX omni-directional antenna • Flat-exponential PDP - for TX beam-shaped antenna • NLOS scenario (with blocked direct path) may be covered by LOS model by dropping the direct path by 10-15dB • Different RX antenna patterns may be taken into account in the framework of the proposed channel models Alexei Davydov (Intel Corporation)

16. Back up • Channel generation procedure • Matlab code • Examples Alexei Davydov (Intel Corporation)

17. Channel Generation Procedure • Initialize the channel parameters • Generate cluster’s parameters • Cluster time of arrival / angle of arrival / amplitude • For each cluster generate MPC’s parameters • Ray time of arrival / angle of arrival / amplitude Alexei Davydov (Intel Corporation)

18. Matlab code Example of impulse response generation routine Alexei Davydov (Intel Corporation)

19. 50 45 40 35 30 40 Relative Power, [dB] 25 Relative Power, [dB] 20 20 0 150 0 15 100 100 10 200 5 50 300 0 Delay, [ns] 0 50 100 150 0 Azimuth, [deg] Delay, [ns] Modeled channel impulse responses 2-D Instantaneous impulse response 1-D Instantaneous impulse response Alexei Davydov (Intel Corporation)

20. -115 -20 -25 -120 -125 -30 -35 -130 -40 -135 Relative Power, [dB] Relative Power, [dB] -45 -140 -50 -145 -55 -150 -60 -155 -65 -160 -50 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 450 Delay, [ns] Delay, [ns] Modeled / Measured channel impulse responses Ray amplitudes are affected by TX/RX antenna elevation patterns LOS component Individual echoes Exponential decaying Noise floor Modeled Average Power Delay Profilesfor flat-exponential PDP Measured Average Power Delay Profiles for different measurement scenarios Alexei Davydov (Intel Corporation)