1. LNA
LNA
LNA
Uplink Optimization in Distributed Antenna Systems Presentation by Dennis McColl January 14th, 2014 IBTUF 2014 Austin, Texas
PUSCH
PUCCH

2. Discussion objectives…
Effect of excess uplink noise in LTE DAS
Noise sources, signal propagation, and Signal to Noise Ratio in the DAS uplink
Noise and path balance measurement setup examples
Share and discuss empirical results of optical degradation experiment

3. Why do we care about excess noise?
Unlike CDMA and EVDO, as noise increases, P0Targets do not change,
and the UE does not increase power. MCS is reduced to maintain SNR per bit.
UE Tx (P0-PUSCH) Uplink Target
Normal noise floor

4. 1 dB Additional Noise
UE Transmit Power is not affected by 1dB of additional noise
UE Tx (P0-PUSCH) Uplink Target
-Unchanged-
Noise floor +1dB

5. 3dB Additional Noise
UE Transmit Power is not affected by 3dB of additional noise
UE Tx (P0-PUSCH) Uplink Target
-Unchanged-
Noise floor +3dB

6. 10 dB Additional Noise
UE Transmit Power is not affected by 10dB of additional noise
UE Tx (P0-PUSCH) Uplink Target
-Unchanged-
Noise floor +10dB

7. long before the downlink throughput changes
Modulation Code Scheme selection will favor increased error correction and lower modulation rates as noise increases
Downlink Throughput
Increases in reverse link noise will dramatically affect uplink throughput
long before the downlink throughput changes
Uplink Throughput
UL = Upload
DL= Download
More uplink noise

8. DAS Noise and signal transport
Remote DAS Equipment
Splitter -
Combiner UL
LNA
e/o DL
Filters
Optical Fiber
Conditioning Tray
Head End DAS Equipment
ENodeB
LNA
LNA
o/e
Receiver

9. eNodeB RxOut with no conditioning. 35-40dB of excess noise.
eNodeB RxOut conditioning method 1
eNodeB RxOut conditioning method 2
PUSCH
PUCCH

10. Uplink signal flow through system components
Thermal Noise (kTB)
UE Signal
(No External Interference)
Electromagnetic energy generated by random molecular movement in the passive components of the systems that we are trying to optimize
Signal to Noise Ratio (SNR) is determined by the signal amplitude and thermal noise (kTB)
Noise Power = kTB k=Boltzmann's Constant = * 10^-23 (joules/kelvin)
T = Temperature in degrees absolute (Kelvin), B = Bandwidth in Hertz, 1 Watt = 1 Joule of energy per second
At room temperature (290 Kelvin)
kTB = -174dBm/Hz

11. Passive Network Uplink signal flow through system components
LNA
Ideal Signal
(No External Interference)
-Feed line attenuation
-Connectors
-Splitters
-Filters
-Passive Combining and distribution BEFORE the first LNA
Signal to Noise Ratio (SNR) is determined by the signal amplitude and thermal noise (kTB)
Low Noise Amplifiers (LNA’s) amplify signals
Variable attenuators allow us to …
Low Noise Amplifiers also generate massive amounts of noise
condition signals to maintain proper dynamic range,…
Signal is attenuated by the passive network
SNR degradation caused by an LNA is specified as its Noise Figure (NF, dB)
attenuate noise to achieve desired noise floors,…
Noise caused by kTB remains constant
Signal to Noise Ratio is degraded by passive attenuation only, not by increased noise
Ironically, NF doesn’t do a very good job of describing the noise generated by an LNA
and balance uplink and downlink path loss

12. Let’s take a closer look at an LNA…
(Example LNA specifications: Gain = 20dB, Noise Figure = 2dB)
Input to LNA
Signal = Terminated
Noise =kTB = -174dBm/Hz
50 Ω Load
kTB+
Gain+
Noise Figure
LNA Output
kTB + Gain+ NF
-174dBm/Hz +20dB+2dB
-152dBm/Hz
6.31* 10-16mW/Hz
kTB
LNA
Signal
It may be easier to think of this as the cost of doing business.
Every time signal flows through this LNA, a specific amount of noise is added which is independent of the signal present.

13. Signal flow through an LNA…
(Example LNA specifications: Gain = 20dB, Noise Figure = 2dB)
kTB+
Gain+
Noise Figure
50 Ω Load
kTB
LNA Output
Signal = -50dBm
Noise = -152dBm/Hz
6.31* 10-16mW/Hz
LNA
Signal
Signal+ Gain
-50
dBm
SNR
Input to LNA
Signal = -70dBm
Noise =kTB = -174dBm/Hz
-152
dBm/Hz
SNR of a 1Hz BW Signal
=-50dBm – (-152dBm)
=102dB
SNR of a 1Hz BW Signal
=-70dBm – (-174dBm)
=104dB

14. What happens when LNA’s are cascaded?
(Example LNA specifications: Gain = 20dB, Noise Figure = 2dB)
LNA
LNA
Signal = -30dBm
Noise = -132dBm/Hz +
(-152dBm/Hz) =
dBm/Hz
Signal = -50dBm
Noise = -152dBm/Hz
SNR (1Hz) = 102dB
Signal = -70dBm
Noise =kTB = -174dBm/Hz
SNR (1Hz) = 104dB
SNR (1Hz) =
SNR* = degraded by 0.04dB

15. The Effect of Reducing Noise Early
(Example LNA specifications: Gain = 20dB, Noise Figure = 2dB)
SNR (1Hz)=
102dB
SNR (1Hz)=
102dB
LNA
LNA
-22dB
Signal = -70dBm
Noise =kTB = -174dBm/Hz
SNR (1Hz) = 104dB
Signal = -52dBm
Noise = (-174dBm/Hz + 20dB +2dB)
+(-152dBm/Hz) = -149dBm/Hz
Attenuation added to bring excess noise down to kTB
SNR (1Hz) = 97dB
4.96dB of SNR is lost by attenuating noise out before the last LNA

16. Typical DAS Uplink Model Showing Cascaded Amplifiers
Remote DAS Equipment
Splitter -
Combiner UL
LNA
e/o DL
Filters
Attenuation here is designed to condition signal levels prior to the electrical to optical conversion.
***Use default/recommended settings!!!***
There is more to lose than to gain!!!
Optical Fiber
Conditioning Tray
Head End DAS Equipment
ENodeB
LNA
LNA
o/e
Receiver
Warning!!! There may be multiple places to attenuate/adjust signal levels in the head end DAS equipment. Use recommended settings, except for the last stage (conditioning stage).

17. 0dB 0dB 0dB Petco Park No Conditioning
(Unbalanced, default attenuation only)
0dB
0dB
0dB
Unknown Stable RFI
≈-53dBm
Ch. Pwr. Density
dBm/Hz
Ideal PSD for Rx noise floor in an Ericsson eNodeB is about
-156dBm/Hz. Typical levels are around -154dBm/Hz.

18. 40dB Total Attenuation 31dB 9dB 0dB
Petco Park Initial Deployment Settings
(Balanced RxOut Noise Floor Target between -151dBm to -153dBm)
40dB Total
Attenuation
31dB
9dB
0dB
Unknown Stable RFI
≈-93dBm
Ch. Pwr. Density
dBm/Hz
Ideal PSD for Rx noise floor in an Ericsson eNodeB is about
-156dBm/Hz. Typical levels are around -154dBm/Hz.

19. 36dB Total Attenuation 11dB 0dB 25dB Attenuation shifted to eNodeB
4dB less attenuation required to hit the same target
36dB Total
Attenuation
11dB
0dB
25dB
Unknown Stable RFI
≈-90dBm
Est. SINR Delta=
+3dB -0.4dB =2.6dB
Ch. Pwr. Density
dBm/Hz

20. 36dB Total Attenuation 0dB 11dB 25dB Attenuation shifted to eNodeB
4dB less attenuation required to hit the same target
36dB Total
Attenuation
0dB
11dB
25dB
Unknown Stable RFI
≈-90dBm
Est. SINR Delta=
+3dB -0.3dB =2.6dB
Ch. Pwr. Density
dBm/Hz

21. A VSG with a -25dBm RSRP at any frequency allows
for channel and quality estimation on any frequency

22. VSG at remote end with baseline attenuation settings
VSG at remote end with baseline attenuation settings. LTE Analyzer on RxOut at head end.
DAS=31dB, Conditioning Tray=9dB, eNodeB=0dB
Remote VSG
LTE OTA Analyzer at Head End
31dB
9dB
0dB
Average RSRP over multiple measurement periods was 76.2dBm Average SINR over multiple measurement periods was 39.3dB
Channel Power Density = dBm/Hz

23. VSG at remote end with shifted attenuation settings
VSG at remote end with shifted attenuation settings. LTE Analyzer on RxOut at head end.
DAS=11dB, Conditioning Tray=0dB, eNodeB=0dB
Remote VSG
LTE OTA Analyzer at Head End
11dB
0dB
25dB
Average RSRP over multiple measurement periods was 72.3dBm
Average SINR over multiple measurement periods was 40.8dB
Channel Power Density = dBm/Hz
3.9dB Less Path Loss
1.5dB Better SINR

24. VSG at remote end with shifted attenuation settings
VSG at remote end with shifted attenuation settings. LTE Analyzer on RxOut at head end.
DAS=0dB, Conditioning Tray=11dB, eNodeB=25dB
Remote VSG
LTE OTA Analyzer at Head End
0dB
11dB
25dB
Average RSRP over multiple measurement periods was 72.3dBm
Average SINR over multiple measurement periods was 40.8dB
Channel Power Density = dBm/Hz
3.9dB Less Path Loss
1.5dB Better SINR

25. - Attenuate at the eNodeB first to maintain amplified noise levels through cascaded amplifiers thus protecting SNR. Leave some attenuation in the conditioners to maintain eNodeB receiver dynamic range.
Remote DAS Equipment
Splitter -
Combiner UL
LNA
e/o DL
Filters
Head End DAS Equipment
Conditioning Tray
ENodeB
LNA
LNA
o/e
Receiver

26. DAS Deployment Requirements
- Get the macro system out!! – Maintaining forward link dominance (SINR) will allow higher modulation code schemes to be utilized
- eNodeB Receive Channel Power Density should mimic macro levels- Set uplink attenuation values so that measured and reported values are the same as a typical macro eNodeB
- Maximize Uplink SNR – Attenuate appropriately. Leave noise in the system until the last possible stage keeping device dynamic range in mind
- Maintain Path Balance – Uplink path loss needs to match the UE measured downlink loss so that inter-cell interference is stable across the DAS and macro systems. If target receive levels are achieved and a path imbalance remains then P0 and PRACH target levels must be adjusted
- UE transmit behavior must mimic macro UE behavior - If UE’s transmit too low or too high at cell edges compared to adjacent cell UE’s, eNodeB receive levels will be affected and uplink SINR will not be stable

27. Example: Balanced UE Uplink in a macro system
Device transmit power increases as it moves towards cell edge
Equal RSRP Boundary
(Cell Edge)
Device transmit power increases as it moves towards cell edge
Inter-cell Interference is the same for both sectors given the same
Operational Path Loss and P0Nominal Targets

28. Two Scenarios where DAS uplink isn’t balanced, or where UE transmit power differs from macro behavior:
The DAS UL path loss is LESS than the perceived forward link loss
Once the UE completes the open loop process, its closed loop power control will reduce power to adapt to the actual uplink path loss
Increased inter-cell interference to DAS
DAS
eNodeB
Receiver
Device transmit power is lower than the same device on a macro system
Equal RSRP Boundary
(Cell Edge)
As the DAS UE reduces power, adjacent cell UE’s begin to interfere with the DAS receivers, degrading uplink SINR, throughput, and capacity

29. 2. The DAS UL path loss is MORE than the perceived forward link loss
Once the UE completes the open loop process, its closed loop power control increases power to adapt to the actual uplink path loss (if it can RACH)
Increased inter-cell interference to macro
DAS
eNodeB
Receiver
Device transmit power is higher than the same device on a macro system
Equal RSRP Boundary
(Cell Edge)
As the DAS UE increases power, adjacent cell receivers see more interference which degrades their UE uplink SINR, throughput, and capacity

30. Discussion objectives…
Effect of excess uplink noise in LTE DAS
Noise sources, signal propagation, and Signal to Noise Ratio in the DAS uplink
Noise and path balance measurement setup examples

31. Over the air Path balance measurement setup example
ENodeB
DAS
RxOut
LTE Analyzer at Uplink Frequency
Single Antenna
Downlink path loss is the difference between the transmitted Reference Signal and the received RSRP
Uplink path loss is the difference between the Uplink VSG RSRP and the received RSRP minus the eNodeB Uplink Gain
Circulator
Uplink VSG
LTE Analyzer at Downlink Frequency

32. Noise Floor Measurement Setup Example
Using the ‘minimum hold’ trace option will allow for noise floor measurements to be made under ANY load at any time of the day.
The longer a MIN HOLD measurement is left to settle, the lower it will go.
Be sure to be consistent when comparing measurements.
Questions: Call Dennis McColl
Anritsu Analyzer Set Up Instructions for Rx Noise Power Measurement.
Shift/Mode → Interference Analysis
Select Amplitude:
Reference level = -50dBm
LNA ‘ON’ (Improves overall NF by 1dB)
Input Attenuation = 0dB
Select BW
Set RBW = 30kHz (Could be narrower or wider based on need, but choose 30kHz for a standardized measurement.)
Set VBW = to RBW/10
Select Measurements:
Select Spectrum (If not already selected)
Select Spectrum (To see measurements)
Select Channel Power
Set Center Freq = 782MHz
Set Ch Pwr Width = 4MHz
Set Span 26MHz
Select Shift/Trace
Select Trace A Operations
Select Min Hold (or whatever your choice)
*The longer it sits the lower it will get so be consistent when comparing output.
(Best Channel Power Density for an Ericsson eNodeB measured at the RxOut port is about -156dBm/Hz)

33. Discussion objectives…
Effect of excess uplink noise in LTE DAS
Noise sources, signal propagation, and Signal to Noise Ratio in the DAS uplink
Noise and path balance measurement setup examples
Share and discuss empirical results of optical degradation experiment

34. Optical Path Degradation Experiment
Remote DAS Equipment
Splitter -
Combiner UL
LNA
e/o DL
Filters
Variable Optical Attenuation
Remote DAS Equipment
Conditioning Tray
ENodeB
LNA
LNA
o/e
Receiver

35. Optical Path Degradation Measurements
Optical Attenuation
RSRP
RSRQ
SINR
Channel Power
EVM RMS%
Not Inserted
-76.9
-9.7
38.3
-78.5
4.78
-79.0
-9.6
32.0
-80.7
6.50 1
34.4
-81.9
7.14 2
-81.3
31.9
-83.6
8.57 3
-84.4
31.1
-85.5
10.03 4
-86.0
27.4
-87.4
12.10 5
-88.3
-10.0
27.6
-89.5
15.53 6
-90.3
-10.3
22.7
-91.4
18.55 7
-92.6
-9.9
24.2
-94.5
25.90 10
-97.5
-11.0
15.9
-98.6
50.00

36. Key take aways… Understand the attenuation system in your eNodeB’s
Attenuate as late as possible in the cascaded LNA chain to achieve a nominal noise floor
Adjust P0 and PRACH Targets to finish the balancing process
Clean your optics, consider logging quality metrics over time

37. Thank you!

38. Macro system impact on DAS performance
If allowed, the macro system downlink presence will significantly impair the DAS.
Even at low levels macro involvement is serious.
Serving sector downlink (RSRP) remains constant
Downlink throughput degrades almost immediately as the adjacent
cell degrades SINR
Downlink throughput reduced 49.01%
CQI=10
Remember to ask, “And what’s the theoretical SINR in the middle of a sector split???”.
Throughput
An adjacent sector is gradually allowed to overlap
Theoretical C/I at every sector split!

39. Reference – Wikipedia, Channel Capacity for Complex Constellations
(bits / symbol)
Reference – Wikipedia, Channel Capacity
for Complex Constellations