SLIDE 1
FP7 ICT-SOCRATES
Controllability for Self-Optimisation
- f Home eNodeBs
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Controllability for Self-Optimisation of Home eNodeBs Kristina - - PowerPoint PPT Presentation
Controllability for Self-Optimisation of Home eNodeBs Kristina - - PowerPoint PPT Presentation
FP7 ICT-SOCRATES Controllability for Self-Optimisation of Home eNodeBs Kristina Zetterberg, Ericsson AB Neil Scully, John Turk, Vodafone Ljupco Jorguseski, Adrian Pais, TNO Outline Introduction to Home eNodeBs (HeNBs) Related Work and
SLIDE 2
SLIDE 3
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Introduction to Home eNodeBs
LTE home base stations Create or extend coverage Improve capacity Typically within buildings, such as an office, a mall or a home Installed by customer Potentially large number Low transmit power Small coverage area Open or closed access
SLIDE 4
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Related Work and Scope
Feasability of HeNBs investigated in 3GPP and Femto Forum NGMN recognises the need of self-optimisation for HeNBs Self-optimisation discussed by H. Claussen et. al.
– Focus on open access HeNB power settings to optimise coverage to minimise
mobility signalling increase SOCRATES project Develops self-organisation methods to enhance the operations of LTE networks Two HeNB use cases considered
– Self-Optimisation of HeNB Interference and Coverage – Self-Optimisation of HeNB Handover
Controllability analysis
– How and to which extent different parameter settings affect performance – Evaluated using simulations of an LTE network with HeNBs deployed
SLIDE 5
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Self-Optimisation of HeNB Interference and Coverage
Optimise coverage area Minimise interference in the network Closed access HeNBs – open only for CSG users Same frequency as macro eNodeBs Main problem is dead zones Uplink and downlink HeNB power
varied to control trade-off
Only HeNB coverage (dead-zone) Only macro coverage Both macro and HeNB coverage
SLIDE 6
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Simulation Setup
Static Monte-Carlo simulator Hexagonal Layout, 7 sites with 21 cells
– Coverage Driven Scenario – 1732 meters s2s distance – Capacity Driven Scenario –
500 meters s2s distance Femto area with grid of houses
– 10 x 10 houses – HeNB density 10% – HeNB placement within house varies
One CSG user per HeNB house On average one non-CSG user per HeNB house Requested bitrate 0.25 Mbps UL, 1 Mbps DL Results collected from users within the HeNB houses
Femto area
SLIDE 7
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Controllability Study
Considered control parameters:
– Maximum DL Transmit Power
Varied from 0.2 mW to 20 mW in steps of 1 dB Reference signal power follows DL transmit power
– Maximum UL Transmit Power
Varied from 20 mW to 316 mW in steps of 1 dB Considered macro – HeNB distances; A CSG user is connected to the HeNB only if
RSRPHeNB > RSRPmacro
64 705 285 Macro-to-HeNB distance (m) 500 1732 1732 Site-to-site distance (m) Capacity Driven Scenario A Coverage Driven Scenario B Coverage Driven Scenario A
SLIDE 8
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Result Plots
X - axis show power difference in dB, compared to maximum setting Y - axis show ratio of users that can detect the reference signal, and have
non-zero uplink and downlink throughput, in the HeNB houses
The three different plots show the ratio of
– CSG users with RS, UL & DL coverage (from macro or HeNB) in the HeNB
houses
– CSG users with RS, UL & DL coverage from HeNB in the HeNB houses – Non-CSG users with RS, UL & DL coverage (from macro) in the HeNB houses
SLIDE 9
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Varying Downlink Power
Cov A: s2s 1732 m m2h 285 m Cov B: s2s 1732 m m2h 705 m Cap A: s2s 500 m m2h 64 m
SLIDE 10
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Varying Uplink Power
DL power 3.2 mW
Cov A: s2s 1732 m m2h 285 m Cov B: s2s 1732 m m2h 705 m Cap A: s2s 500 m m2h 64 m
SLIDE 11
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Throughput
For the given scenarios (with a maximum of one CSG user per home
eNodeB) throughput for HeNB connected UEs is equal to the requested throughput both in uplink and downlink
For the given scenarios, throughput for non-CSG users is not highly affected Seen effects are probably due to changed macro load
Cov A: s2s 1732 m m2h 285 m
SLIDE 12
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Conclusions and Further Work
Conclusions
– Dead-zones are the major problem when introducing closed access home
eNodeBs
– HeNB Maximum Transmit Power is a suitable parameter for controlling the
trade-off between HeNB coverage and the size of the dead-zone Possible Further Work
– Evaluate effects of adjusting the CSG user RSRP connect margin;
CSG user is connected to the HeNB only if RSRPHeNB > RSRPmacro – margin
– Evaluate effects of using parts of the macro frequency band for the HeNB – Evaluate effects of adjusting the transmit power on parts of the frequency band – Algorithm development
SLIDE 13
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Self-Optimisation of HeNB Handover
Minimise dropped calls Maximise user throughput Open access HeNBs Indoor HeNBs, providing coverage also outdoor Macro – HeNB Handover HeNB – HeNB Handover
SLIDE 14
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Simulation Setup
Dynamic simulator Single 3-sector macro eNB
– Wrap-around – Site-to-site distance 500 meters
Row of houses with HeNBs 165 meters from macro eNB 15 meters between each HeNB User experience of a UE moving down
a street is modelled
Full buffer traffic Study considers
– Impact of UE speed – Impact of relative signal strengths
between eNB and HeNB
– Impact of macro network load 165 m
SLIDE 15
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Controllability Study
Considered control parameters:
– Hysteresis (HYST)
Set to the values 0, 3, 6, 9 and 12 dB
– Time to trigger (TTT)
Set to the values 0, 100, 320, 640 and 1280 ms Considered scenarios;
5 1 Macro cell load (UEs/sector) 280 165 50 Macro-to-HeNB distance (m) 100 30 3 UE speed (km/h)
SLIDE 16
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
SINR and Serving Cell
2 4 6 8 10 12 14
- 20
- 15
- 10
- 5
5 10 15 20 25 30 Time (s) Blue: D ata SIN R (dB) / R ed: Serving cell Hysteresis = 12 dB TTT = 640 ms UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector
30 40 UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector
Hysteresis = 12 dB TTT = 640 ms UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector 35
2 4 6 8 10 12 14
- 20
- 10
10 20 30 40 Time (s) Blue: Data SINR (dB) / Red: Serving cell Hysteresis = 0 dB TTT = 0 ms UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector
Low hysteresis and TTT
– Many handovers to HeNBs
High hysteresis and TTT
– UE stays on macro eNB most of the time
SLIDE 17
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Throughput
Low TTT and small hysteresis gives higher throughput More pronounced at higher UE speed Same trends in throughput are seen for higher HeNB-to-macro eNB distance At a lower distance the impact is not as large, as the UE stays connected
to the macro eNB
12 9 6 3 100 320 640 1280 1 2 3 4 5 Hysteresis (dB) UE speed = 100 km/h Separation distance = 165 m Load = 5 UEs/sector TTT (ms) Throughput (Mbps) 12 9 6 3 100 320 640 1280 5 10 15 Hysteresis (dB) UE speed = 3 km/h Separation distance = 165 m Load = 5 UEs/sector TTT (ms) Throughput (Mbps) 12 9 6 3 100 320 640 1280 5 10 15 Hysteresis (dB) UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector TTT (ms) Throughput (Mbps)
3 km/h 30 km/h 100 km/h
SLIDE 18
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Ping-pong Handover Ratio
Low hysteresis and TTT values gives high throughput, but could also
lead to ping-pong
12 9 6 3 100 320 640 1280 0.2 0.4 0.6 0.8 Hysteresis (dB) UE speed = 3 km/h Separation distance = 165 m Load = 5 UEs/sector TTT (ms) Ping-pong handover ratio 12 9 6 3 100 320 640 1280 0.1 0.2 0.3 0.4 Hysteresis (dB) UE speed = 30 km/h Separation distance = 165 m Load = 5 UEs/sector TTT (ms) Ping-pong handover ratio
3 km/h 30 km/h
SLIDE 19
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Improving Throughput by Avoiding Handover
12 9 6 3 100 320 640 1280 2 4 6 Hysteresis (dB) UE speed = 100 km/h Separation distance = 165 m Load = 0 UEs/sector TTT (ms) Throughput (Mbps) 12 9 6 3 100 320 640 1280 0.2 0.4 0.6 0.8 1 Hysteresis (dB) UE speed = 100 km/h Separation distance = 165 m Load = 0 UEs/sector TTT (ms) Femto ratio
Throughput does not always decrease as handover parameter values
are increased
Gain is only achieved for low macro cell load
Throughput Ratio of time connected to HeNB
SLIDE 20
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Conclusions and Further Work
- Conclusions
–
Handover settings have large impact on number of performed handovers and
- n UE troughput
–
Impact varies with distance, UE speed and macro load
–
Set HO parameters low, but high enough to avoid ping-pong
–
For rapidly moving UEs;
a) Set HO parameters low to handover promtly b) Set HO parameters low to avoid handover to HeNB
- Possible Further Work
–
Consider static UEs and ping-pong effects
–
Consider lower HeNB density and subsequent macro-HeNB-macro handovers
–
Study SINR and dropped calls together with throughput
–
Algorithm development
SLIDE 21
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB
Summary
Controllability studies performed for two use cases
HeNB interference and coverage
– Closed access – Uplink and downlink HeNB power evaluated – Dead-zones is the major problem – Downlink power is a suitable control parameter for the trade-off
HeNB Handover
– Open access – Time-to-trigger and hysteresis evaluated – HO parameters should be low, but high enough to avoid ping-pong – Rapidly moving UEs could gain from not handing over to HeNB
SLIDE 22
WWW.FP7-SOCRATES.EU Kristina Zetterberg, Ericsson AB