Indoor MIMO WLAN Channel Models

1. Indoor MIMO WLAN Channel Models Speaker: Chi-Yeh Yu Advisor: Tzi-Dar Chiueh chiyeh@analog.ee.ntu.edu.tw Nov 17, 2003

2. Outline • Motivation and Goal • Existing SISO Model • Cluster Modeling Approach • Saleh- Valenzuela’s statistical model • Proposed MIMO Model • IEEE 802.11 indoor MIMO WLAN channel model (draft document in progress) • Conclusion

3. Motivation and Goal

4. Motivation • Why Need Channel Model? • Channel models can be used to evaluate new WLAN proposals before built. • Why MIMO? • IEEE 802.11a and HiperLAN/2 require rate up to 54 Mbps, and the key enabling technology of high data rate is to adopt “antenna array.”

5. MIMO system capacity • MIMO theoretic capacity [REF2]

6. Goal • To develop a set of multiple-input multiple-output (MIMO) channel models backwards compatible with existing 802.11 channel models (Developed by Medbo and Schramm [1]).

7. Existing SISO Model

8. Existing SISO Models (1/3) • Five delay profile models for single-input single-output (SISO) systems were proposed in [1] for different environments (A-E): • Model A for a typical office environment, non-line-of-sight (NLOS) conditions, and 50 ns rms delay spread. • Model B for a typical large open space and office environments, NLOS conditions, and 100 ns rms delay spread. • Model C for a large open space (indoor and outdoor), NLOS conditions, and 150 ns rms delay spread. • Model D, same as model C, line-of-sight (LOS) conditions, and 140 ns rms delay spread (10 dB Ricean K-factor at the first delay). • Model E for a typical large open space (indoor and outdoor), NLOS conditions, and 250 ns rms delay spread.

9. Existing SISO Models (2/3) • The resulting models are revised and proposed as follows: • Model A, flat fading model with 0 ns rms delay spread (one tap at 0 ns delay model). • Model B for a typical residential environment, line-of-sight (LOS) conditions, 15 ns rms delay spread, and 10 dB Ricean K-factor at the first delay. • Model C for a typical residential or small office environment, LOS/NLOS conditions, 30 ns rms delay spread, and 3 dB Ricean K-factor at the first delay. • Model D for a typical office environment, NLOS conditions, and 50 ns rms delay spread. • Model E for a typical large open space and office environments, NLOS conditions, and 100 ns rms delay spread. • Model F for a large open space (indoor and outdoor), NLOS conditions, and 150 ns rms delay spread.

10. Summary (3/3)

11. Cluster Modeling Approach

12. Saleh- Valenzuela’s statistical model • Pioneer work done by Saleh and Valenzuela [2] and further elaborated and extended upon by many researchers [3-8]. • Channel impulse response • Independent inter-arrival exponential probability density functions • Average power gains

13. Channel Model-D Example • Three clusters can be clearly identified. Cluster 1 Cluster 2 dB Cluster 3

14. Spatial Representation of 3 Clusters Cluster 1 Cluster 2 R1 R2 LOS Tx Antennas Rx Antennas R3 Cluster 3

15. Modeling Approach • Only time domain information from A-E SISO models can be determined (delay of each delay within each cluster and corresponding power using extrapolation methods). • In addition, for the MIMO clustering approach the following parameters have to be determined: • Power azimuth spectrum (PAS) shape of each cluster and tap • Cluster angle-of-arrival (AoA), mean • Cluster angular spread (AS) at the receiver • Cluster Angle-of-departure (AoD), mean • Cluster AS at the transmitter • Tap AS (we assume 5o for all) • Tap AoA • Tap AoD

16. Cluster and Tap PAS Shape • Cluster and tap PAS follow Laplacian distribution. Example of Laplacian AoA (AoD) distribution, cluster, AS = 30o

17. Cluster AoA and AoD • It was found in [3,4] that the relative cluster mean AoAs have a random uniform distribution over all angles.

18. Cluster AS (1/3) • We use the following findings to determine cluster AS: • In [3] the mean cluster AS values were found to be 21o and 25o for two buildings measured. In [4] the mean AS value was found to be 37o. To be consistent with these findings, we select the mean cluster AS values for models A-E in the 20o to 40o range. • For outdoor environments, it was found that the cluster rms delay spread (DS) is highly correlated (0.7 correlation coefficient) with the AS [9]. It was also found that the cluster rms delay spread and AS can be modeled as correlated log-normal random variables. We apply this finding to our modeling approach.

19. Cluster AS (2/3) • The mean AS per model was determined using the formula (per model mean AS values in the 20o – 40o range) where DS is cluster delay spread. • Cluster AS variation within each model was determined using 0.7 correlation with cluster DS and assuming log-normal distributions.

20. Cluster AS (3/3) • Resulting cluster AS (at the receiver) and DS for all five models (A-E) is shown in the figure below

21. Tap AS, AoA, and AoD (1/2) • We assume that each tap PAS shape is Laplacian with AS = 5o. • Following constraints that satisfy cluster AS and AoA (AoD), tap AoA and AoD can be determined using numerical methods. where li is a zero-mean, unit-variance Laplacian random variable, bi is a scaling parameter related to the power roll-off coefficient of the cluster, D is a parameter that is determined using numerical global search method to satisfy the required AS and mean AoA of each cluster; ao is the mean cluster AoA; s2tot is cluster AS, and s2a,i is tap AS.

22. Tap AS, AoA, and AoD(1/2) • Example: Distribution of taps within a cluste.

23. MIMO Channel Model A: Table of Parameters

24. Next Steps • So far we have completely defined PAS of each tap (AS and Laplacian AoA distribution) and AoA of each tap. These parameters were determined so that the cluster AS and mean cluster AoA requirements are met (experimentally determined published results). • Next, we show how we use tap AoA and AS information to calculate per tap transmit and receive antenna correlation matrices and from that finally the MIMO channel matrices H.

25. Proposed MIMO Model

26. MIMO Channel Matrix Formulation (1/3) • Example 4 x 4 MIMO matrix H for each tap is as follows where Xij (i-th receiving and j-th transmitting antenna) are correlated zero-mean, unit variance, complex Gaussian random variables as coefficients of the Rayleigh matrix HV, exp(jfij) are the elements of the fixed matrix HF, K is the Ricean K-factor, and P is the power of each tap.

27. MIMO Channel Matrix Formulation (2/3) • To correlate the Xij elements of the matrix X, the following method can be used • 4x4 MIMO channel transmit and receive correlation matrices are

28. MIMO Channel Matrix Formulation (3/3) • Correlation coefficients for each tap can be determined using tap PAS (represented by Laplacian distribution and corresponding AS) and tap AoA (AoD) where D = 2pd/l (for linear antenna array) and RXX and RXY are the cross-correlation functions between the real parts (equal to the cross-correlation function between the imaginary parts) and between the real part and imaginary part, respectively.

29. MIMO Channel Matrix H Generation • Use power, AS, AoA and AoD tap parameters from tables A-D. • Per tap, calculate transmit and receive correlation matrices. • Using correlation matrices and Hiid generate instantiations of channel matrices H, as many as required by simulation.

30. Conclusion • WLAN MIMO channel models were developed based on extensive published experimental data and models. • The models are based on per tap correlation matrices determined from tap AS and AoA.

31. References [1] J. Medbo and P. Schramm, “Channel models for HIPERLAN/2,” ETSI/BRAN document no. 3ERI085B. [2] A.A.M. Saleh and R.A. Valenzuela, “A statistical model for indoor multipath propagation,” IEEE J. Select. Areas Commun., vol. 5, 1987, pp. 128-137.  [3] Q.H. Spencer, et. al., “Modeling the statistical time and angle of arrival characteristics of an indoor environment,” IEEE J. Select. Areas Commun., vol. 18, no. 3, March 2000, pp. 347-360. [4] R.J-M. Cramer, R.A. Scholtz, and M.Z. Win, “Evaluation of an ultra-wide-band propagation channel,” IEEE Trans. Antennas Propagat., vol. 50, no.5, May 2002, pp. 561-570.  [5] A.S.Y. Poon and M. Ho, “Indoor multiple-antenna channel characterization from 2 to 8 GHz,” submitted to ICC 2003 Conference. [6] G. German, Q. Spencer, L. Swindlehurst, and R. Valenzuela, “Wireless indoor channel modeling: Statistical agreement of ray tracing simulations and channel sounding measurements,” in proc. IEEE Acoustics, Speech, and Signal Proc. Conf., vol. 4, 2001, pp. 2501-2504. [7] J-G. Wang, A.S. Mohan, and T.A. Aubrey,” Angles-of-arrival of multipath signals in indoor environments,” in proc. IEEE Veh. Technol. Conf., 1996, pp. 155-159. [8] Chia-Chin Chong, David I. Laurenson and Stephen McLaughlin, “Statistical Characterization of the 5.2 GHz Wideband Directional Indoor Propagation Channels with Clustering and Correlation Properties,” in proc. IEEE Veh. Technol. Conf., vol. 1, Sept. 2002, pp. 629-633. [9] K.I. Pedersen, P.E. Mogensen, and B.H. Fleury, “A stochastic model of the temporal and azimuthal dispersion seen at the base station in outdoor propagation environments,” IEEE Trans. Veh. Technol., vol. 49, no. 2, March 2000, pp. 437-447. [10] L. Schumacher, Namur University, Belgium, (laurent.schumacher@ieee.org). [11] IEEE 802.11-03/161r1 Document