Aalborg Universitet

Reproducing Standard SCME Channel Models for Massive MIMO Base StationRadiated TestingFan, Wei; Zhang, Fengchun; Jämsä, Tommi; Gustafsson, Mattias ; Kyösti, Pekka; Pedersen,Gert F.Published in:2017 11th European Conference on Antennas and Propagation (EUCAP)

Publication date:2017

Link to publication from Aalborg University

Citation for published version (APA):Fan, W., Zhang, F., Jämsä, T., Gustafsson, . M., Kyösti, P., & Pedersen, G. F. (2017). Reproducing StandardSCME Channel Models for Massive MIMO Base Station Radiated Testing. In 2017 11th European Conferenceon Antennas and Propagation (EUCAP)

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

? Users may download and print one copy of any publication from the public portal for the purpose of private study or research. ? You may not further distribute the material or use it for any profit-making activity or commercial gain ? You may freely distribute the URL identifying the publication in the public portal ?

Take down policyIf you believe that this document breaches copyright please contact us at vbn@aub.aau.dk providing details, and we will remove access tothe work immediately and investigate your claim.

Downloaded from vbn.aau.dk on: April 30, 2017

1

Reproducing Standard SCME Channel Models for

Massive MIMO Base Station Radiated TestingWei Fan, Fengchun Zhang, Tommi Jämsä, Mattias Gustafsson, Pekka Kyösti and Gert F. Pedersen

Abstract—Massive MIMO is a multi-user technology, whereradio base stations (BSs) are equipped with a large number ofantennas to simultaneously serve many terminals in the sametime-frequency resource. Performance evaluation of such large-scale antenna systems in the design and development stageis challenging. In this paper, we propose to evaluate massiveMIMO BSs with a sectorized multi-probe anechoic chamber(MPAC) setup. A sectorized MPAC setup with 16 probe antennasdistributed uniformly within [−60o, 60o] in azimuth domain isutilized to reproduce target channel models. A 8×8 and a 16×16uniform planar array at 3.5 GHz are selected as the BS underevaluation, respectively. Radio channel emulation accuraciesin terms of power-angular spectrum, spatial correlation andbeamforming pattern are investigated for the proposed MPACsetup and desired channel models.

Index Terms—Anechoic chamber, massive MIMO base sta-tions, over-the-air evaluation, radio channel emulation

I. INTRODUCTION

Massive multiple-input multiple-output (MIMO) techniques

have arisen as enabling technologies to deal with the high

data and reliable communication requirement in 5G cellular

systems [1]. Massive MIMO is a multi-user technology, where

radio base stations (BSs) are equipped with a large number of

antennas to simultaneously serve many terminals in the same

time-frequency resource. The promising features of utilizing

excessive number of antenna elements at the BS side have

been discussed extensively from a theoretic point of view in

many works in the literture, see e.g. [1], [2]. Meanwhile, there

has been a strong ongoing effort to design practical massive

MIMO systems both from academia and industry, see e.g.

promising results reported in [1], [3].

Radio performance evaluation is required in different phases

of product development from early research prototypes, de-

sign optimization, to actual product approval for roll-out.

Conventional BSs are often evaluated by connecting coaxial

cables to the BS antenna ports, i.e. with antennas bypassed

[4]. However, this becomes problematic for massive MIMO,

where hundreds of cable connections to the antenna ports and

respective hardware are expected. In addition, antenna ports

might be unaccessible for BSs in future 5G systems. Thus,

there is a strong need to replace conductive testing.

Historically, over-the-air (OTA) testing has been used to

evaluate radio performance of handset antenna systems [5]–

[7]. The testing is done in a wireless manner (i.e. over-the-air),

Wei Fan, Fengchun Zhang and Gert F. Pedersen are with the Antennas,Propagation and Radio Networking section at the Department of ElectronicSystems, Aalborg University, Denmark (email: wfa@es.aau.dk).

Tommi Jämsä and Mattias Gustafsson are with Huawei TechnologiesSweden AB, Gothenburg, Sweden.

Pekka Kyösti is with Keysight Technologies and Oulu University, Finland

without a need to break the device. Various OTA method-

ologies have been proposed to evaluate UE systems with

multiple antennas, where the multi-probe anechoic chamber

(MPAC) method has been selected in the standardization for

UE systems with multiple antennas [6]. The MPAC method

is attractive, since realistic propagation environments can be

physically synthesized and controlled in a shielded anechoic

chamber, via controlling the signals radiated from multiple

probe antennas. Some works have discussed the applicabil-

ity of MPAC method for massive MIMO BS performance

evaluation. In [8], [9], possible testing systems are discussed

for BS performance evaluation. In [10], suitable measurement

distances and physical dimensions of the MPAC setups for the

performance evaluation of massive MIMO BSs in anechoic

chambers are investigated. Standardization work on radiated

conformance testing of BSs, equipped with active antenna

systems, is ongoing [7].

In this paper, we first discussed OTA testing for massive

MIMO BSs in sectorized MPAC setups in Section II. After

that, we demonstrate the simulation results for reproducing

standard SCME channel models for massive MIMO BS testing

in a sectorized setup in Section III and Section IV. Section V

concludes the paper.

II. MASSIVE MIMO OTA TESTING IN MPAC SETUPS

An illustration of typical cellular propagation environment

is shown in Figure 1. For the UE side, since the scatterers

are often nearby and located around the UE, the angle profiles

are less specular. For example, a root mean square (RMS)

angle spreads (ASA) of 35o are specified for each cluster for

the SCME UMa and SCME UMi channel models at the UE

side, as shown in Figure 2. Further, multipath components

can impinge on the UE from arbitrary directions. Note that

the UMa and UMi channel models are essentially the same

SCME models, but with different channel parameters.

An illustration of the typical MPAC configuration for UE

testing is shown in Figure 3, where the system consists of a

BS emulator, one or several channel emulator and multiple

probe antennas. The UE is often located in the center of the

OTA ring. For UE OTA testing, a uniform configuration of

the OTA probes over the azimuth plane is often adopted [12],

since it offers the possibility to recreate any spatial channel

model without relocation of the probe antennas.

A sectorized MPAC setup would be more appropriate for

massive MIMO BS testing, as shown in Figure 4:

• From point of view of propagation environments. The

angle profiles at the BS side have small angle spread due

2017 11th European Conference on Antennas and Propagation (EUCAP)

/17/$31.00 ©2017 IEEE 3669

2

Massive

MIMO

Base

station

UE 1

UE 2

Figure 1. Real multipath environments for cellular systems.

0

6000

0.005

0.01

200

Pow

er

[lin

ear]

4000

0.015

SCME UMa model, UE

Delay [ns]Angle [

o]

0.02

02000

0 -200

0

1000

0.005

0.01

200

Pow

er

[lin

ear] 0.015

SCME UMi model, UE

Delay [ns]

500

Angle [o]

0.02

0

0 -200

Figure 2. Power-angle-delay profiles for the SCME UMa channel model (left)and SCME UMi channel model (right) at the UE side.

to the fact that the BS is placed higher and further away

from scatterers. Furthermore, paths might be blocked due

to practical installations and therefore field of view of the

array might be limited, see e.g. Figure 1.

• From point of view of BS antenna array. The target

coverage area of a BS antenna is smaller than that of a UE

antenna. While a UE antenna by practical reasons needs

to receive and transmit from/to all angles, a BS antenna

normally is restricted to receive and transmit power in a

desired angular zone. This angular zone normally is about

60o in azimuth and about 20o in elevation.

Test Area

Base Station

Emulator

Channel

Emulator

Amplifier Box

Downlink

Uplink

K K

Anechoic Chamber

Uplink

Communication

Antenna

K Probe

Antennas

UE

Figure 3. An illustration of the MPAC setup for UE performance evaluation.K OTA antennas are connected to K radio frequency (RF) interface channels.

Radio

channel

emulator

UE 1

Massive MIMO BS

OTA probe antenna

To UE 1

To UE 2

downlink

uplink

UE 2

Figure 4. Sectorized MPAC setups for radiated testing of massive MIMO BS.

• From point of view of setup cost. A sectorized MPAC

configuration has the potential to reduce the system cost.

Firstly, a sectorized probe configuration has the potential

to save cost, via reducing the required number of fad-

ing channel emulators and respective hardware resource.

Secondly, the DUT can be placed further away from OTA

antennas to utilize the full chamber dimension.

An illustration of the proposed sectorized MPAC setup is

shown in Fig. 4. For massive MIMO BS testing, BS and UE

positions are swapped compared to Figure 3, which requires

a larger test area. UEs or UE emulators are used to emulate

the user ends of the radio link. Multiple UEs might be needed

for performance testing. Radio channel emulators and OTA

antennas are utilized to mimic spatial channels for the desired

user and interfering users.

III. CHANNEL EMULATION IN SECTORIZED MPAC SETUPS

A. Target channel models

In this paper, two representative 2D channel models, i.e.

SCME UMa and SCME UMi [11], are selected for the sake

of simplicity. The channel parameters, e.g. delay, power, angle

of departure (AoD), angle spread of departure (ASD), angle of

arrival (AoA), angle spread of arrival (ASA) for each cluster

can be found in [11]. Note that both channel models are

assumed to be vertically polarized for discussion for simplicity.

The mean AoDs of the composite power angle spectra at

the BS side can be calculated as in (1)

θ̄ =

∑piθi∑pi

, (1)

where pi and θi denote the power and azimuth angle of the

i-th cluster, respectively. In the SCME UMa and SCME UMi

channel models, the mean AoD of the composite power angle

spectra at the BS side is θ̄ = 84.3o and θ̄ = 21.7o, respectively.

In this paper, the mean AoDs of both channel models are

shifted to 0o for simplicity.

The power-angle-delay profiles for the modified SCME

UMa channel model and SCME UMi channel model at the

BS side are shown in Figure 5. As we can see, each path is

quite specular, with cluster RMS angle spread 2o for the UMa

and with 5o for the UMi case, respectively. Further, the spatial

profiles of both channel models are spatially confined, within

[−60o, 60o] angle region, as shown in Figure 5.

B. Sectorized MPAC configuration and DUT

In this paper, a 2D sectorized configuration is utilized

to reproduce the 2D SCME channel models. Note that a

2017 11th European Conference on Antennas and Propagation (EUCAP)

3670

3

0

1000

Delay [ns] 0

0.05

50

Pow

er

[lin

ear]

0.1

SCME UMi model, BS

500

Angle [o]

0.15

0

-50Delay [ns] 0

0

6000

0.1

50

Pow

er

[lin

ear]

4000

0.2

SCME UMa model, BS

Angle [o]

0.3

02000

-50

Figure 5. Power-angle-delay profiles for the modified SCME UMa channelmodel (left) and SCME UMi channel model (right) at the BS side. Note thatthe angle axis is limited to [−60o, 60o].

-0.20

0.2

4

3

2

1

[m]

0

-14

-2 3

[m]

2-3

1-4

OTA antenna array

BS antenna array

BS array boresight

direction 0o

Figure 6. An illustration of the probe configuration.

3D configuration is required to reproduce 3D channel mod-

els, i.e. with elevation domain, see e.g. discussions in [10].

K = 16 probe antennas uniformly distributed in azimuth

angle [−60o, 60o] are utilized on a circular probe configuration

with R = 5m. 16 OTA probe antennas are selected as an

example in the simulation, since this configuration can be

supported by a single radio channel emulator (e.g. Propsim

F32). The antenna array boresight direction is defined to be

0o. The carrier frequency is set to be 3.5 GHz, as it is

widely used for massive MIMO analysis in the literature.

Isotropic antenna patterns are assumed for the OTA probes and

mutual coupling is not considered for the sake of simplicity.

Note that the sectorized MPAC configuration can only support

modeling channel models whose spatial profiles are confined

within the sectorized angle region. To have more flexibility,

a more uniform MPAC configuration might be needed. Two

BS antenna arrays are considered as DUT, i.e. one 8 × 8uniform planar array (i.e. 64 antenna elements) and one 16×16(i.e. 256 antenna elements) uniform planar array with half

wavelength element spacing, as illustrated in Figure 7. The

physical size of the antenna arrays are around 0.3m × 0.3mand 0.64m × 0.64m, respectively. Isotropic antenna patterns

for antenna elements are used in the simulations.

8

16

7-0.15

24

15

6

0.2

32

23

14-0.1

5

40

31

22

13

4

-0.05

48

39

30

21

12

30.1 1

0

56

47

38

29

20

11

2

0.05

64

55

46

37

28

0.5

19

10

10

0.1

63

54

45

36

27

18

9

0.15

0

62

53

44

35

26

17

-0.1

61

52

43

34

25

-0.5

60

51

42

33

59

50

41

-0.2 -1

58

49

57

-1

-0.5

0

0.5

-0.4

-0.2

0

0.2

0.4

-0.4

-0.2

0

0.2

0.4

256

241

1

16

Figure 7. An illustration of the massive MIMO BSs in the simulations: 8×8uniform planar array (left) and 16× 16 uniform planar array (right).

0

1000

0.2

50

Pow

er

[lin

ear]

0.4

SCME UMi model, BS, OTA

Delay [ns]

500

Probe angular position [o]

0.6

0

-500

0

6000

0.2

0.4

50

Pow

er

[lin

ear]

4000

0.6

SCME UMa model, BS, OTA

Delay [ns]

Probe angular position [o]

0.8

02000

-500

Figure 8. Power-angle-delay profiles for the SCME UMa channel model (left)and SCME UMi channel model (right) at the BS side. The ASA of the clusterat the BS is 2o for the UMa and 5o for the UMi, respectively. Note that theangle axis is limited to [−60o, 60o].

C. Channel emulation technique

The PFS technique has gained its popularity in the industry

due to its capability to emulate all dimensions of the standard-

ized channel models [13], with only power calibration required

for OTA antennas. With the PFS technique, each cluster is

emulated by several probe antennas. For each cluster, the BS

side spatial characteristics are reconstructed by allocating ap-

propriate power weights to the fading signals from the probes.

To obtain the optimal power weights allocated to the probe

antennas, the objective is to minimize the deviation between

target spatial correlation and emulated spatial correlation [13].

The obtained power weights for the target channel models are

shown in Figure 8. As we can see, the emulate power angle

delay profiles match target profiles well. The deviation is due

to limited OTA antennas.

IV. CHANNEL EMULATION ACCURACY WITH MPAC

SETUPS

A. Spatial correlation

The correlation between signals received on the antenna ele-

ments can be calculated according to the correlation definition.

The correlations between signals received by the first antenna

and the other antennas under target and emulated channel

models for both channel models for the 8× 8 uniform planar

array is shown in Figure 9. As we can see, good agreement

2017 11th European Conference on Antennas and Propagation (EUCAP)

3671

4

10 20 30 40 50 60

Antenna Index

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Sp

atia

l co

rre

latio

n

UMa, emulated

UMa, target

UMi, emulated

UMi, target

Figure 9. Target and emulated spatial correlation between 1st antenna elementand other antenna elements on the 8× 8 uniform planar array, as illustratedin Figure 7

0 5 10 15 20 25 30

Antenna Index

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Sp

atia

l co

rre

latio

n

Uma, emulated

Uma, target

Umi, emulated

Umi, target

Figure 10. Target and emulated spatial correlation between 1st antennaelement and other antenna elements on the 16 × 16 uniform planar array,as illustrated in Figure 7. Note that only antenna elements in two bottomcolumns are shown.

between the spatial correlations calculated under target channel

models and under emulated channels can be achieved for

both channel models. The agreement is better for the SCME

UMi channel models due to the larger cluster angle spread in

the target channel. For the SCME UMa model, a deviation

up to 0.1 exists in the high correlation region, which might

slightly impact the system performance. Note that the periodic

pattern in spatial correlation curves are due to the fact that

the target and emulated channels are concentrated on a 2D

azimuth plane. Therefore, we expect same correlation values

for antenna elements on different elevation columns.

The correlation values for target and emulated channel

models for the 16 × 16 uniform planar array are shown

in Figure 10. We observe good agreement for both SCME

UMa and UMi channel models up to around 5λ (i.e. antenna

separation between element 1 and element 11). Large deviation

exists in spatial correlation accuracy for antenna separation

above 5λ. The deviation is due to that fact that a larger test

object requires more probe antennas. Note that the deviation

due to limited physical dimension of the MPAC (i.e. R = 5m)

is negligible in the simulation.

B. Beamforming power pattern

The beamforming power patterns of the planar arrays under

the target and emulated channel models for both channel

-30

1000

-25

-20

60

-15

Pow

er

[dB

]

40

-10

SCME UMi model, 8 X 8

Delay [ns]

500 20

-5

Angle [o]

0

0

-20-40

0 -60

cluster 1, emulated

cluster 2, emulated

cluster 3, emulated

cluster 4, emulated

cluster 5, emulated

cluster 6, emulated

cluster 1, target

cluster 2, target

cluster 3, target

cluster 4, target

cluster 5, target

cluster 6, target

Power-AoA

Figure 11. Beamforming power pattern of the target and emulated SCMEUMi channel models with the 8× 8 planar array.

-30

5000

-25

-20

4000 60

-15

Pow

er

[dB

]

403000

-10

SCME UMa model, 8 X 8

Delay [ns]

20

-5

Angle [o]

2000 0

0

-201000-40

0 -60

Figure 12. Beamforming power pattern of the target and emulated SCMEUMa channel models with the 8× 8 planar array.

models are shown in Figure 11 (SCME UMi channel model

and 8×8 planar array), Figure 12 (SCME UMa channel model

and 8×8 planar array), Figure 13 (SCME UMi channel model

and 16×16 planar array) and Figure 14 (SCME UMa channel

model and 16 × 16 planar array), respectively. As we can,

good match in terms of the main beam, sidelobes and beam

shape can be achieved for the SCME UMi channel model

with 8×8 planar array. As for the SCME UMa channel model,

though we can still achieve good agreement for the main beam

and beam shape, some deviations exist in the sidelobes. This

is introduced by the small deviations in spatial correlation

emulation accuracy, as discussed earlier in Figure 9. Note that

the mean AoD of each target cluster is plotted in the figure as

a reference for visualization purpose.

As for the 16× 16 planar array cases, though good agree-

ment can be observed for the beam shape, as shown in Figure

13 and Figure 14, some deviations exist in the main beam and

sidelobes. This is due to the fact that the spatial resolution of

the planar array is higher than the probe angular separations.

This effect is clearly observed in Figure 15, where a peak

around target angle-of-arrival exists in the beamforming power

pattern under target channel models, while two peaks exist at

the two dominant probe directions for the beamforming pattern

under emulated channel models.

2017 11th European Conference on Antennas and Propagation (EUCAP)

3672

5

-40

1000

-30

800 60

-20

Pow

er

[dB

]

40600

SCME UMi model, 16 X 16

-10

Delay [ns]

20

Angle [o]

400 0

0

-20200-40

0 -60

Figure 13. Beamforming power pattern of the target and emulated SCMEUMi channel models with the 16× 16 planar array.

-30

5000

-25

-20

4000 60

-15

Po

we

r [d

B]

40

-10

3000

SCME UMa model, 16 X 16

Delay [ns]

20

-5

Angle [o]

2000 0

0

-201000-40

0 -60

Figure 14. Beamforming power pattern of the target and emulated SCMEUMa channel models with the 16× 16 planar array.

V. CONCLUSION

This paper aims to reproduce SCME channel models with

a sectorized MPAC configuration for massive MIMO BS per-

formance evaluation. Channel emulation accuracy in terms of

power angle delay profile, spatial correlation and beamforming

power pattern are investigated. Simulation results demon-

strate that with 16 OTA antennas uniformly distributed within

[−60o, 60o], the radio channels can be accurately reproduced

for a 8 × 8 planar array BSs with half wavelength spacing

at 3.5GHz. For a 16× 16 planar array BSs, though a similar

-60 -40 -20 0 20 40 60

Angle [o]

-40

-35

-30

-25

-20

-15

-10

-5

0

Pow

er

[lin

ear]

SCME UMa model, first cluster, 16 X 16

Beampattern, emulatred channel

Power-AoA, target

Beampattern, target channel

Power, AoA, emulated

Figure 15. Beamforming power pattern of the first cluster in the target andemulated SCME UMa models with the 16× 16 planar array.

shape for beamforming power pattern can be still achieved,

deviations exist in terms of spatial correlation, main beam and

sidelobes for the beamforming power pattern. Therefore, more

OTA antennas are needed for such large scale antenna systems.

The sectorized MPAC setup is a cost-effective solution, com-

pared to conventional conducted testing, since the number of

required probe antennas might be significantly less than the

number of antenna elements on the DUT. As demonstrated

in the paper, both SCME channel models can be accurately

reproduced within a test area of 5λ×5λ with 16 probes, where

a 64 element planar array with half wavelength spacing can be

supported. Note that 2D MPAC configuration is adopted for

the sake of simplicity and selected 2D target channel models.

For actual BS OTA testing, a 3D MPAC setup might be more

suitable in the future. 3GPP 3D channel models and channel

models above 6 GHz emphasize the importance of modeling

elevation domain. Further, planar arrays, not linear arrays, are

typically adopted for massive MIMO BS systems.

ACKNOWLEDGMENT

This work has been partially supported by Huawei Tech-

nologies Sweden AB, Gothenburg, Sweden. Dr. Wei Fan

would like to acknowledge the financial assistance from

Danish council for independent research (grant number: DFF

611100525). The authors appreciate the valuable suggestions

and comments from Dr. Gerhard Steinbock.

REFERENCES

[1] J. Vieira, S. Malkowsky, K. Nieman, Z. Miers, N. Kundargi, L. Liu,I. Wong, V. Owall, O. Edfors, and F. Tufvesson, “A flexible 100-antennatestbed for massive mimo,” in Globecom Workshops (GC Wkshps), 2014,Dec 2014, pp. 287–293.

[2] T. Marzetta, “Massive MIMO: An Introduction,” Bell Labs Technical

Journal, vol. 20, pp. 11–22, 2015.[3] Huawei, “Massive mimo takes prominent role

in 5g developments,” 2015. [Online]. Avail-able: http://www.rethinkresearch.biz/articles/massive-mimo-takes-prominent-role-5g-developments/

[4] “Multi-Standard Radio (MSR) Base Station (BS) radio transmission andreception,” 3GPP, TS 37.104 V12.6.0, December. 2014.

[5] “Test Plan for Wireless Device Overthe-Air Performance: Method ofMeasurement for Radiated RF Power and Receiver Performance,” CTIACertification, Tech. Rep. Version 3.5.2, September 2015.

[6] “Test Plan for 2x2 Downlink MIMO and Transmit Diversity Over-the-Air Performance,” CTIA Certification, Tech. Rep. Version 1.0, August2015.

[7] “Active Antenna System (AAS) Base Station (BS) conformance testing;Part 2: radiated conformance testing ,” CTIA Certification, Tech. Rep.Version V0.0.1, Release 13, February 2016.

[8] M. Gustafsson, T. Jämsä, A. Ahlström, M. Högberg, and B. Sihlbom,“Ota testing of base stations,” The International Wireless Industry

Consortium (IWPC), pp. 1–21, September 2015.[9] M. Gustafsson, T. Jämsä, A. Ahlstrom, M. Högberg, and B. Sihlbom,

“Over-the-Air (OTA) Testing of Base Stations ,” COST IC1004, Tech.Rep. TD(15)13039, September 2015.

[10] P. Kyösti, W. Fan, G. F. Pedersen, and M. Latva-aho, “On dimensionsof ota setups for massive mimo base stations radiated testing,” IEEE

Access, vol. 1, pp. 335–349, 2016.[11] D. S. Baum, J. Hansen, J. Salo, G. Galdo, M. Milojevic, and P. Ky{osti,

“An interim channel model for beyond-3g systems: extending the 3gppspatial channel model (scm),” in 2005 IEEE 61st Vehicular Technology

Conference, vol. 5, May 2005, pp. 3132–3136 Vol. 5.[12] “Verification of radiated multi-antenna reception performance of User

Equipment,” 3GPP, TR 37.977 V1.0.0, Sep. 2013.[13] P. Kyösti, T. Jämsä, and J.-P. Nuutinen, “Channel modelling for multi-

probe over-the-air mimo testing,” International Journal of Antennas and

Propagation, vol. 2012, 2012.

2017 11th European Conference on Antennas and Propagation (EUCAP)

3673