1. The Mobile MIMO Channel and Its Measurements
Jack H. Winters
[Carol Martin, Nelson Sollenberger (Mobilelink)]
AT&T Labs - Research
Middletown, NJ

2. OUTLINE
Introduction
Test Setup
Performance Measures
Results
Conclusions

3. MIMO Capacity Increase
4/10/2017
MIMO Capacity Increase
Multiple antennas at both the base station and terminal can significantly increase data rates if the multipath environment is rich enough
With M antennas at both the base station and the mobile, M independent channels can be provided in the same bandwidth
sufficient multipath  low correlation  high spectral efficiency
With 4 transmit and receive antennas, 4 independent data channels can be provided in the same bandwidth
Data rates as high as 1.5 Mbps (4x384 kbps) may be possible for EDGE, 216 Mbps WLAN (802.11a), or 20 Mbps for Wideband OFDM

4. Mobile MIMO Radio Channel Measurements
Objectives
Characterize the mobile MIMO channel to determine feasibility of MIMO approach in a typical cellular environment
sufficient multipath  low correlation  high spectral efficiency
With 4 transmit/receive antennas, theoretically up to a 3.77-fold increase is possible
Approach
Conduct field tests to show the potential increase in capacity using 4 transmit and 4 receive antennas at both the base station and terminal
For reliable measurement at mobile speeds, collect data continuously and simultaneously on all 4 transmit and 4 receive antennas
Collect data on drive routes at low to moderate speeds plus pedestrian and indoor tests in suburban environment
Collect data for different antenna configurations
Measure 30 kHz complex channel with IS-136 Smart Antenna Test Bed

5. MIMO Channel Testing Mobile Transmitter W1 Tx Rx W2 Tx Rx W3 Tx Rx W4
Test Bed Receiver with Rooftop Antennas
Base Station Antenna Configurations W1 Tx Rx
11.3 ft
Perform timing recovery and symbol synchronization
Record 4x4 complex channel matrix
Evaluate capacity and channel correlation W2 Tx Rx W3 Tx Rx W4 Tx Rx
1.5 ft
1.25 ft
Synchronous
test
sequences LO LO
Space diversity
Space / polarization diversity
Space / polarization / pattern diversity
Terminal Antenna Configurations

6. MIMO Channel Measurement System
Transmitter
Receive System
4 antennas mounted on a laptop
4 coherent 1 Watt 1900 MHz transmitters with synchronous waveform generator
Dual-polarized slant 45° PCS sector antennas separated by 11 feet and fixed multibeam antenna with ° beams
4 coherent 1900 MHz receivers with real-time baseband processing using 4 TI TMS320C40 DSPs

7. Test Bed Receivers with Rooftop Antennas Terminal Antennas on a Laptop
MIMO Channel Testing
Mobile Transmitters
Test Bed Receivers with Rooftop Antennas W1 Tx Rx
Perform timing recovery and symbol synchronization
Record 4x4 complex channel matrix
Evaluate capacity and channel correlation W2 Tx Rx W3 Tx Rx
Terminal Antennas on a Laptop W4 Tx Rx
Prototype Dual Antenna Handset
Synchronous
test
sequences LO LO
Rooftop Base Station Antennas
11.3 ft
Mobile Transmitters

8. MIMO Testing Drive Routes
Drive routes within coverage of test sector
Non line-of-sight conditions along route
Suburban environment with gently rolling terrain
Maximum downrange distance of 2.5 miles
Peak speed of 45 mph, average speed of 30 mph in residential area
Peak speed of over 60 mph along highway
Pedestrian and indoor tests

9. Performance Measures
Complex channel measurement: H = [ H ij] for the ith transmit and jth receive antenna
Capacity (instantaneous and averaged over 1 second):
C = log2(det[1 + H†H]) =  log2(1 + /4i)
where  is the signal-to-noise ratio and
i is the ith eigenvalue of H†H
To eliminate the effect of shadow fading, the capacity is normalized to the average capacity with a single antenna:
Cn =  log2(1 + /4i) / (1/16)  log2(1 + Hij)

10. Performance Measures (cont.)
For the multibeam antenna, we normalize the capacity by the average capacity of the strongest beam:
Cn =  log2(1 + /4i) / (1/4)  log2(1 + Hijmax)
Correlation (averaged over 1 second):
Transmit signal correlation
i1,i2 = | [H†H] i1,i2 / ([H†H] i1,i1 [H†H] i2,i2 )1/2 |
Receive signal correlation
j1,j2 = | [HH†] j1,j2 / ([HH†]j1,j1 [HH†] j2,j2 )1/2 |

11. Effect of Time Averaging on Capacity
Simulation results with independent- Rayleigh-fading equal-power channels
Distribution of capacity does not vary significantly with averaging
100 fades during 1 second average at 30 mph
Spatial averaging reduces the effect of fading
Capacity for pedestrians is similar to mobile users

12. Effect of Time Averaging on Correlation
Simulation results with independent- Rayleigh-fading equal-power channels
Distribution of correlation does vary significantly with averaging
Correlation decreases with:
Speed
Terminal antenna rotation
Low signal level
Relative correlation only should be considered to identify antennas causing capacity reduction

13. MIMO Field Test Results
Amplitudes of 16 channels between the 4 transmit and 4 receive antennas
1 second average
Channel powers are approximately equal for dual-polarized transmit and receive antennas

14. Field Test Results
Dual-polarized, spatially-separated base station and terminal antennas
Instantaneous normalized capacity
Spatial averaging reduces variations due to Rayleigh fading
Capacity increase is close to 4 times that of a single antenna
50% and 90% of the signal correlations are less than 0.23 and 0.47, respectively

15. Correlation distribution - 1 second average
Dual-polarized, spatially-separated base station and terminal antennas
Tx antennas: 30% decrease in correlation for spatially-separated antennas
Rx antennas: 60% decrease in correlation for cross-pol, spatially- separated antennas

16. MIMO Field Test Results
Measured capacity distribution is close to the ideal for 4 transmit and 4 receive antennas

17. Field Test Results
Dual-polarized multibeam base station antenna (2 center beams) and dual-polarized terminal antennas
Instantaneous normalized capacity
Capacity increase is close to 2 times that of a single antenna
50% and 90% of the Rx signal correlations are less than 0.61 and 0.87, respectively

18. Correlation distribution - 1 second average
Dual-polarized multibeam base station antenna (2 center beams) and dual-polarized terminal antennas
Correlations between co-pol beams are high

19. Field Test Results
Measured capacity distribution is close to the ideal for 4 transmit and 4 receive antennas with dual-polarized, spatially-separated base station and terminal antennas (cases 1-4)
Capacity gain is 2 times a single antenna with the dual-polarized multibeam antenna (case 8)
Capacity gain is 1.4 times a single antenna with the orthogonally-polarized multibeam antenna (case 7), and slightly lower with vertically-polarized multibeam antenna (cases 5,6)

20. Conclusions
Conducted the first field tests to characterize the mobile MIMO radio channel in a typical cellular environment
With 4 transmit and 4 receive antennas close to 4 times the capacity of a single antenna can be supported
Dual-polarized spatially-separated base station and terminal antennas
Multibeam antenna has lower capacity - only twice the gain
Capacity distribution is close to the ideal and is nearly independent of terminal speed
Capacity for pedestrians is similar to mobile users
Correlation results can be used to compare antenna diversity performance
Field test data and results are valuable inputs to design and deployment of mobile MIMO systems
Future work: Wideband channel measurements

21. MIMO Field Test Results
Pedestrian Tests
Mobile Tests
Amplitudes of 16 channels between the 4 transmit and 4 receive antennas
No averaging
Channel characteristics vary for pedestrian and mobile users
Capacity for pedestrians is similar to mobile users - spatial averaging reduces the effect of fading

22. MIMO Field Test Results
Instantaneous normalized capacity
Spatial averaging reduces variations due to fading
Potential capacity increase is close to 4 times that of a single antenna