sibel sibel tombaz tombaz , pål frenger, fredrik athley, eliane...
Post on 11-Mar-2021
12 Views
Preview:
TRANSCRIPT
Sibel Sibel Sibel Sibel tombaztombaztombaztombaz, Pål Frenger, Fredrik Athley, Eliane Semaan, Claes Tidestav, Ander Furuskär
Ericsson research
Sibel.tombaz@ericsson.com
› Ultra-lean design and the high beamforming gain, enabling longer and more efficient sleep at BS.
› At expected traffic levels beyond 2020, 5G-NX can decrease the energy consumption by more than 50% while providing ~10 times more capacity.
Identify the Identify the achievable energy achievable energy savings with 5Gsavings with 5G--NX systems NX systems operating at operating at 15 GHz utilizing 15 GHz utilizing massive beamforming and massive beamforming and
ultraultra--lean design compared to lean design compared to
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 2
while providing ~10 times more capacity.
› Carrier aggregation with 5G-NX provides >> 100 Mbps user throughput with lower energy consumption (-35%) despite the comparably energy-inefficient LTE layer at expected traffic levels beyond 2020.
ultraultra--lean design compared to lean design compared to baseline LTE@2.6 in a dense baseline LTE@2.6 in a dense
urban (major Asian city) urban (major Asian city) scenario.scenario.
100% + 1% ≠ 101%
› 5G at a Glance› Scope of the Study› Definitions& Methodology
– Energy Performance– Cell DTX– Energy Performance Assessment Methodology
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 3
– Energy Performance Assessment Methodology
› Power Consumption Models– LTE Systems– 5G-NX Systems
› Simulation Setup and Results› Conclusions› Appendix
› In contrast to earlier generations, 5G wireless access is not a specific radio access technology; rather it is an overall wireless access solution addressing the various requirements of mobile communication.
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 4
› Two key features– Massive beamforming– Ultra-lean design
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 5
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 6
energy performance
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 7
Lower
Energy Consumption
Higher
Energy Efficiency
&
› Energy performance is defined as the daily averaged area power consumption, unit is kW/km2.› Daily traffic profile proposed by EARTH is used for evaluations.
( )241
(1 )BSN
t tactive i sleep iP P
tη η + − ∑ ∑
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 8
( )1 1
(1 )active i sleep it i
P Pt
EPA
η η= =
+ − =
∑ ∑
Cell DTX is a hardware feature enabling sleep mode operation at BS side during empty transmission time intervals (TTIs).
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 9
› Power consumption of a Macro BS 2x2 MIMO configuration– EARTH Model
0
0
( ) ( )TRX p radiated
tot TRX
N P P BS is active transmitting
P N P BS is inactive
∆ +=
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 10
0 modTRXN P BS is on sleep eδ
1
0.84
without cell DTX
with cell DTXδ
=
› Proposed power consumption model of large scale antenna systems with N RF branches– Based on [1,2,3]
0
sec 0
( )radiated
c
tot
PNP P BS is active transmitting
P N P BS is inactiveε
+ +
= ×
Notation Definition Value [W]
Radiated power per sector 40radiatedP
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 11
sec 0
0 modtotP N P BS is inactive
P BS is on sleep eδ= ×
Efficiency of the PAs 0,25
Baseline power consumption 260
Circuit (Hardware) power RF branch
1
Sleep power consumption
ε0P
cP
sleepP 0 ; (0,1)Pδ δ ∈
[1] Hong Yang; Marzetta, T.L., "Total energy efficiency of cellular large scale antenna system multiple access mobile networks," Online Conference on Green Communications (GreenCom), 2013 IEEE , vol., no., pp.27,32, 29-31 Oct. 2013[2] Qiao, Deli; Wu, Ye; Chen, Yan, "Massive MIMO architecture for 5G networks: Co-located, or distributed?," Wireless Communications Systems (ISWCS), 2014 11th International Symposium on , vol., no., pp.192,197, 26-29 Aug. 2014[3] Emil Björnson, Eduard Jorswieck, Merouane Debbah, and Björn Ottersten, Multi-Objective Signal Processing Optimization: The Way to Balance Conflicting Metrics in 5G Systems
1
0.29
w ithout cell D TX
w ith cell D T Xδ =
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 12
Capacity grid
� Center macro
� 7 sites/21 cells
� 200 m ISD
� ~45 m antenna height (rooftop)
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 13
Coverage grid
� Surrounding macro
� 28 sites/84 cells
� 400 m ISD
� ~30 m antenna height (rooftop)
Without cell DTX
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 14
Even at high traffic, cell DTX can bring savings due to high difference between day-night traffic.
Busy Hour
With cell DTX
› Advantages of 5G-NX:
1. In a given deployment, beamforming gain provides higher user data rate.
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 15
1. In a given deployment, beamforming gain provides higher user data rate.› Same traffic will be served by utilizing less amount of resources. � Longer time to sleep.
2. New system control plane enables BSs to sleep for longer consecutive time durations.› Deeper sleep for BSs� Lower sleep power consumption compared to LTE.
› Carrier aggregation with 5G-NX always increases the energy consumption.
› Almost 5 times increase in 5-percentile user throughput requires up to 38% increase in energy consumption.
› Most of the traffic is offloaded 5G-NX (>70%), taking advantage of
› Superior cell-edge performance by 5G-NX systems even at 15 GHz.› Beamforming gain compensates for the additional propagation loss.
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 16
› Most of the traffic is offloaded 5G-NX (>70%), taking advantage of efficient transmission.
LTE@2.6 layer is the dominant consumer, responsible of >60% of total energy consumption.
Power Split
LTE@2.6 ����17%,loaded5G-NX@15 ���� 13% loaded
� LTE can still serve the traffic with acceptable 5-percentile user throughput.
� 5G-NX enables up to 65% energy saving compared to LTE@2.6 while increasing the 5-percentile user throughput 7 times.
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 17
Traffic levels are so low to fully benefit from carrier aggregation in order to reduce the energy consumption.
� Both carrier aggregation solutions increases the energy consumption.
› Increased static power consumption due to the new system cannot be compensated by the reduction in dynamic power due to lower utilization.
› LTE@2.6 cannot handle the traffic. (Maximum ~500 Mbps/km2).
› 5G-NX alone can provide 70 Mbps 5-percentile throughput, with 64% energy saving.
› LTE@2.6+5G-NX@15: Serve the 5-percentile users consuming
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 18
› LTE@2.6+5G-NX@15: Serve the 5-percentile users consuming less dynamic power.
– Efficient transmission at NX layer and offloading benefits at LTE layer.
LTE@2.6 +5G-NX@15 can provide >> 100 Mbps user throughput with lower energy consumption, despite the comparably energy-inefficient LTE layer.
.
conclusions
› Standalone 5G-NX is the most energy efficient solution (-64% energy consumption for 1200 Mbps/km2)
› LTE@2.6 +5G-NX@15 can provide >> 100 Mbps user throughput with lower energy consumption (-35% for 1200 Mbps/km2).
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 19
› In order to achieve the potential energy savings, we have to fully utilize all the benefits obtained by 5G-NX (high BF gain, higher BW, ultra-lean design, etc) with cell DTX.
– Implementation is the key!
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 21
Energy Performance“Metrics”
I want the red one!
5
6
7
8
9
10
y
Inverted metric: y=1./x
Largeimprovement
Bit/Joule
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 22
1 2 3 4 5 6 7 8 9 10
1
2
3
4
5
xy
Large improvement of a “bad” system
Little improvement of index
improvement of index
Minor improvement of system that is already good
Energy Performance“Metrics”
6
7
8
9
10Linear metric: y=x
Largeimprovement
of index
I want the green one!
W/km2
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 23
1 2 3 4 5 6 7 8 9 10
1
2
3
4
5
6
x
y
Large improvement of system
Little improvement of index
Minor improvement of system
0 50 100 150 2000
0.05
0.1
0.15
0.2
0.25
Time [ms]
PA
load
[%
]
System Signature Index (SSI): 0.4 ms every 100 ms
Access Information Table (AIT):3.6 ms every 1-10 second(s)
LTE-NX:PA on time ratio ≈ 0.5%DTX duration ≈ 99.6 ms
100 x less
500 x longer
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 240 1 2 3 4 5 6 7 8 9 100
0.05
0.1
0.15
0.2
0.25
Time [ms]
PA lo
ad [%]
0 1 2 3 4 5 6 7 8 9 100
0. 05
0.1
0. 15
0 .2
0. 25
Time [ms]
PA
load
[%
]
LTE:PA on time ratio ≈ 52%DTX duration ≈ 0.2 ms
1
0.29
w ithout cell D T X
w ith cell D T Xδ
=
Lte-nx@15 GHz
10
12
14
16
Are
a P
ower
Con
sum
ptip
n [k
W/k
m2]
LTE@2.6
Low Load - Without Cell DTX
Low Load - With Cell DTXHigh Load - Without Cell DTX
High Laod - With Cell DTX
8
10
12
14
16
Are
a P
ower
Con
sum
ptip
n [k
W/k
m2]
LTE-Nx@15
Low Load - Without Cell DTX
Low Load- With Cell DTXHigh Load - Without Cell DTX
High Load - With Cell DTX
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 25
0 5 10 15 20 254
6
8
Time
Are
a P
ower
Con
sum
ptip
n [k
W/k
m2]
0 5 10 15 20 25
2
4
6
Time
Are
a P
ower
Con
sum
ptip
n [k
W/k
m2]
If energy performance is included into the design of LTE-Nx from the start, ultra-lean design will enable significant energy savings with cell DTX.
› Wrap-around:– No-wrap around. – Simulate larger system, discard statistics from edge cells
› Path Loss Model
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 26
– New frequency dependent model is used.
– Height dependent probability that the building is of type “new”
– More information at FRA 160 90 percent presentation– https://ericoll.internal.ericsson.com/sites/FRA/Documents/Activities%202014/160%20-%205G-
LTE%20with%20BF%20capabilies%20and%20densification/90PercSeminar_final_v8.pptx
LTE @2.6 GHz/LTE @15 GHz/Lte-Nx @15 GHzLTE @2.6 + LTE@15 GHz/ LTE@2.6+LTE-Nx@15 GHz
Parameter ValueBandwidth 20 MHz/100 MHz/100 MHz /40+100 MHz /40+100 MHz
UE Transmit power 23 dBm
BS Transmit power 46 dBm
Number of UE receive antennas 2
UE max antenna gain -8 dBi/3dBi/3 dBi**/ -8 dBi + 3 dBi*/ -8 dBi + 3 dBi*
Energy Performance of 5G-NX Wireless Access | © Ericsson AB 2015 | 2014-06-05 | Page 27
UE max antenna gain -8 dBi/3dBi/3 dBi**/ -8 dBi + 3 dBi*/ -8 dBi + 3 dBi*
BS Antenna 742215_fitted*/742215_fitted/Antenna array/742215_fitted*/742215_fitted*+Antenna array***
Indoor traffic 80%
Duplex mode TDD Configuration 5 (1 8 1)
Noise Figure UE 9 dB
Noise Figure BS 2.3 dB
Beam-forming at BS No/No/UE-specific BF (GoB)/No/UE-specific BF (GoB)
• Max Gain = 18, Horizontal Half Power Beamwidth = 65, Front-To-Back Ratio = 25, Vertical Half Power Beamwidth = 6.5, Side Lobe Level = -17, Max Total Attenuation = 30
** More information on UE antenna gain can be found in the appendix.
***Antenna array with (5x20)
top related