handover parameter optimization in lte self- organizing ... 2100... · handover parameter...
TRANSCRIPT
FP7 ICT-SOCRATES
Handover parameter
optimization in LTE self-
organizing networks
TD (10)10068
COST 2100, 10th MCM
Athens, Greece
February 3rd – 5th
TUBS, Braunschweig, Germany
IBBT, Ghent, Belgium
VOD, Newbury, England
WWW.FP7-SOCRATES.EU
1. Introduction
2. Simulation scenario and LTE system-level simulator
3. Simulation metrics
4. Controllability and Observability studies
5. Performance of the non-optimised network
6. Handover optimisation SON algorithm
7. Simulation results
8. Conclusion
Outline
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
2/22
WWW.FP7-SOCRATES.EU
Problem
– Handover parameter optimisation is done manually
– high OPEX
– long optimisation intervals based on error reports
– Non-optimal handover performance
– handover failures
– ping-pong handovers
– call dropping
Handover parameter optimisation objective
– automate the optimisation
– adapt the handover parameters on a short-term scale
– optimise the handover performance
Approach
– analyse the system behaviour
– develop handover optimisation algorithm
Introduction
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
3/22
WWW.FP7-SOCRATES.EU
Assembling
Scenario Data
Processing Data
User locationsEnvironmentNetwork
Braunschweig
Scenario
Traffic Distri-
butionNetwork data OpenSteetMap
Generating
Source Data
Correlated User
Snapshots
Decorated User
Snapshots
Network
Information
LTE SimulatorSimulations
Realistic SOCRATES Scenario
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
4/22
WWW.FP7-SOCRATES.EU
Realistic SOCRATES Scenario
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
Computing the landuse information from openstreetmap.org
Landuse classes: Road, Building, Water, Street and Railway
5/22
WWW.FP7-SOCRATES.EU
Input data
– Realistic SOCRATES scenario
Power mask
– Soft frequency reuse
Call generation
– All users connected
Update RSRP/SINR
– Shadow fading maps
Handover procedure/algorithm
MATLAB LTE system-level simulator
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
Start Read input
data
Build Network Build Users
Yes
NoEnd of
Simulation? Set Power Mask
Call GenerationEnd
Update RSRP/SINR
HO procedureHO algorithm
Next step
6/22
WWW.FP7-SOCRATES.EU
Control parameters
– Hysteresis
– Time-to-Trigger
Assessment metrics
– Handover failure ratio
– Ping-Pong handover ratio
Simulation metrics
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
Control
parameterValues
Hysteresis
(0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5,
5.5, 6, 6.5, 7, 7.5, 8, 8.5, 9, 9.5, 10 )
in [dB]
Time-to-Trigger
(0 0.04 0.064 0.08 0.1 0.128 0.16
0.256 0.32 0.48 0.512 0.64 1.024
1.280 2.56 5.12) in [s]
succHOfailHO
failHO
HOFNN
NHPI
__
_
– Call dropping ratio
failHOnppHOppHO
ppHO
HPPNNN
NHPI
___
_
acceptedHO
droppedHO
DCN
NHPI
_
_
7/22
WWW.FP7-SOCRATES.EU
System metrics
– RSRP (Reference Signal Received Power)
– cell transmit power
– pathloss to the UE
– shadow fading with a standard deviation of 3dB
– SINR (Signal to Interference Noise Ratio)
– interfering cells
Simulation metrics
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
faduecuec LLPRSRP ,
N
n
ueRSRP
uecuec
n
RSRPSINR1
10
,
10,, 10log10
fadLueL
cP
N
8/22
WWW.FP7-SOCRATES.EU
Simulation parameter Value
Simulation time 200 [s]
Simulation step time 0.01 [s]
Simulation area (mobile users) 1.5 km * 1.5 km
Number of users 30
eNodeB transmit power 46 [dBm]
Number of considered cells in the scenario 76
Measured cells (N) 21
Considered interfering cells for SINR
calculations20
Critical ping-pong handover time (T_crit) 5 [s]
Handover execution time 0.25 [s]
SINR averaging window 0.1 [s]
Min. SINR threshold - 6.5 [dB]
Controllability and Observability studies
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
Objective
Analyse the system behaviour
and sensitivity
Find handover algorithm
approach
Simulation assumptions
All resources are used in all
cells (maximum interference)
Simulation approach
Perform system simulations for
all hysteresis and time-to-
trigger value combination
(handover operating point)
9/22
WWW.FP7-SOCRATES.EU
C & O: Handover failures
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
0
2
4
6
8
10
0
0.1
0.250.5
21
5
0
0.02
0.04
Hysteresis [dB]
Handover Failures
Time-to-Trigger [s]
Han
do
ver
failu
re r
ati
o
10/22
WWW.FP7-SOCRATES.EU
C & O: Ping-Pong handovers
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
0
2
4
6
8
10
0
0.1
0.25
0.5
1
2
5
0
0.2
0.4
0.6
0.8
Hysteresis [dB]
Ping-Pong Handovers
Time-to-Trigger [s]
Pin
g-P
on
g h
an
do
ver
rati
o
11/22
WWW.FP7-SOCRATES.EU
C & O: Call dropping
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
0
2
4
6
8
10
0
0.1
0.25
0.5
1
2
5
0
0.2
0.4
0.6
0.8
Hysteresis [dB]
Call drops
Time-to-Trigger [s]
Call d
rop
pin
g r
ati
o
12/22
WWW.FP7-SOCRATES.EU
Handover performance weighting function
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
HP = w1 HPIHOF + w2 HPIHPP + w3 HPIDC
– wx is the weight of the individual HPI
– HPIHOF is the handover failure performance indicator
– HPIHPP is the ping-pong handover performance indicator
– HPIDC is the dropped calls performance indicator
Weighting parameter Value
w1 0.5, 0.6, …, 2.0
w2 0.5, 0.6, …, 2.0
w3 0.5, 0.6, …, 2.0
4096 valid weighting parameter combinations have been considered
If (HP<0.05) => “meaningful” handover parameter operating point
13/22
WWW.FP7-SOCRATES.EU
Handover performance
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
0
2
4
6
8
10
0
0.1
0.25
0.5
1
2
5
0
0.5
1
Hysteresis [dB]
Handover Performance (weights = [1 0.5 2])
Time-to-Trigger [s]
No
rmali
sed
su
m o
f w
eig
hte
d H
O f
ailu
re r
ate
,p
ing
-po
ng
HO
rate
an
d c
all d
rop
pin
g r
ate
14/22
WWW.FP7-SOCRATES.EU
0
1
2
3
4
5
6
7
8
9
10
0
0.1
0.25
0.5
1
2
5
Operating Points (Threshold: 5%)
Hysteresis [dB]Time-to-Trigger [s]
“meaningful” handover operating points
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
15/22
WWW.FP7-SOCRATES.EU
Simulation parameter Value
Simulation time 1000 [s]
Simulation step time 0.01 [s]
Simulation area (mobile users) 1.5 km * 1.5 km
Number of users 50
eNodeB transmit power 46 [dBm]
Operating points
(Hysteresis, Time-to-Trigger)
(4, 0.48), (6, 0.32), (8, 0.1), (9, 0.08)
in [dB, s]
Number of considered cells in the scenario 78
Measured cells (N) 21
Considered interfering cells for SINR
calculations20
Handover performance averaging window 60 [s]
Critical ping-pong handover time (T_crit) 5 [s]
Handover execution time 0.25 [s]
SINR averaging window 0.1 [s]
Min. SINR threshold - 6.5 [dB]
Simulation parameters for the performance analysis
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
16/22
WWW.FP7-SOCRATES.EU
Performance of the non-optimised network
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
100 200 300 400 500 600 700 800 900 10000
5
10
15
20
25
Time [s]
Rati
o [
%]
Handover Performance for the operating point (4, 0.48)
Handover failure
Ping-Pong handover
Call dropping
17/22
WWW.FP7-SOCRATES.EU
Performance of the non-optimised network
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
100 200 300 400 500 600 700 800 900 10000
0.5
1
1.5
2
2.5
3
3.5
4
Time [s]
Han
do
ver
failu
re r
ati
o [
%]
Handover failure performance
Operating point (4, 0.48)
Operating point (6, 0.32)
Operating point (8, 0.1)
Operating point (9, 0.08)
100 200 300 400 500 600 700 800 900 10000
5
10
15
20
25
Time [s]
Pin
g-P
ong
hand
over
rat
io [%
]
Ping-Pong handover performance
Operating point (4, 0.48)
Operating point (6, 0.32)
Operating point (8, 0.1)
Operating point (9, 0.08)
100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
Time [s]
Cal
l dro
ppin
g ra
tio [%
]
Call dropping performance
Operating point (4, 0.48)
Operating point (6, 0.32)
Operating point (8, 0.1)
Operating point (9, 0.08)
Comparison of the network
performance for four different
operating points
(4 dB Hys, 0.48 s TTT)
(6 dB Hys, 0.32 s TTT)
(8 dB Hys, 0.1 s TTT)
(9 dB Hys, 0.08 s TTT)
18/22
WWW.FP7-SOCRATES.EU
Handover optimisation SON algorithm
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
HO SON algortihm
Next cell
1)
Update HPIs
HPIs <
threshold?
Yes No
Increase bad
performance time
Increase good
performance time
2)
4)
Reset good
performance time
Reset bad
performance time
Good
perform-
ance?
Yes No
Decrease HPI
thresholds Bad
perform-
ance?
No
Reset good
performance time
Change handover
operating point
Yes
Reset bad
performance time
11)
8)
3)
5)
6)
7)
9)
10)
12)
13)
Optimisation criteria for HPIs
Handover
Performance
Indicator
Hysteresis
Time-
to-
Trigger
Optimisation
Handover
failure ratio
< 5 dB ↑ TTT
5 dB – 7 dB ↑ TTT & ↑ HYS
> 7 dB ↑ HYS
Ping-Pong
handover
ratio
< 2.5 dB ↑ TTT
2.5 dB – 5.5 dB ↑ TTT & ↑ HYS
> 5.5 dB ↑ HYS
Call dropping
ratio
> 6 dB > 0.6 s ↓ TTT & ↓ HYS
<= 6 dB > 0.6 s ↓ TTT
> 7.5 dB <= 0.6 s ↓ TTT & ↓ HYS
3.5 dB – 6.5
dB
<= 0.6 s ↑ HYS
< 3.5 dB <= 0.6 s ↑ TTT & ↑ HYS
Optimisation actions are added up
Hys and TTT are only changed by one step at a time
The new operating point has to belong to the set of “meaningful operating points”
19/22
WWW.FP7-SOCRATES.EU
Handover optimisation simulation results
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
7
8
9
10
Time [s]
Rat
io [
%]
Handover performance for the operating point (6, 0.32)
Handover failure
Ping-Pong handover
Call dropping
20/22
WWW.FP7-SOCRATES.EU
Handover optimisation simulation results
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
7
8
Time [s]
Rati
o [
%]
Handover performance (Optimisation)
Handover failure
Ping-Pong handover
Call dropping
21/22
WWW.FP7-SOCRATES.EU
The system behaviour to different handover operating points has been
analysed
Handover performance can be optimised using the proposed algorithm
Handover operating points are chosen for every cell individually
The overall network performance is increased and the handover failure ratio
and ping-pong ratio drop to zero in the shown case
Next steps
– Run the algorithm in a larger scenario
– Improve the SINR calculation (scheduling)
– Introduce background traffic (implication on system throughput)
– User specific handover parameters
Conclusion
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
22/22
Thank you very
much for
your attention
FP7 ICT-SOCRATES
WWW.FP7-SOCRATES.EU
Handover procedure I
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
HO procedure
Next active user
HO
command
send?
YesNo
Find the best server
YesConnected
to the best
server?
1)
2)
3)
4)
NoSave best server
as HO candidate
5)
8)
If new cell is best
server set back HO
crit. time
6)
Increase handover
criteria time
7)
Yes
HO
criteria time
> TTT?
No
Send Handover
command
24/22
WWW.FP7-SOCRATES.EU
Handover procedure II
Dipl.-Ing. Thomas Jansen, TU Braunschweig, Institut für Nachrichtentechnik
Yes
No
13)
Increase HO
duration time
10)HO
duration > HO
execution?
11)
Handover complete
12)
Save successful
HandoverUpdate UE History
14)
Next active user
1)
HO failure
occured?
17)
Save Handover
failure
18)
Hand back
successful?
19)Yes
No
Yes
No
Reconnect to
Source eNode B
21)
Save call drop
during Handover
20)
Ping-Pong
HO detected?
15)
No
Yes
Save Ping-Pong
Handover
16)
The handover procedure is executed in every simulation time step
Handover procedure is independent of the handover algorithm
25/22