Link Adaptation and Carriers Detection Errors inMultibeam Satellite Systems with Linear Precoding
Anxo Tato, Stefano Andrenacci, Symeon Chatzinotas, CarlosMosquera
atlanTTic Research Center, University of VigoSnT, University of Luxembourg
September 11, 2018
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 1 / 18
Introduction
High Throughput Satellite (HTS) at Ka-band• Multibeam satellite + Linear Precoding + Link Adaptation
Full Frequency reuse, 245 beams
Imperfect Channel State Information at the Transmitter (CSIT)• Carriers detection errors = Nullification
Unicast
Random interbeam scheduling
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 2 / 18
Introduction
Actual channel
CSIT:Estimated channel
available at GW
Precoding matrix
SINR
MODCOD
LUT
MODCOD
Actualprecoded
SINR
Estimatedprecoded
SINR
¿Allows quasi-error-freetransmission with
this MODCOD?
_ +
SINR absolute error
LUTLookup Table
SelectedModulation andCoding Scheme
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 3 / 18
System model
Signal model:y = Hx + n = HWs + n
Channel model:ESA’s 245 beams radiation pattern
H: Imperfect CSIT due to...
NullificationGaussian estimation errors
Linear Precoding: MMSE with Sum Power Constraint (SPC)
W = η · HH
(HHH +
1
snrIN
)−1
(1)
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 4 / 18
Nullification effect
Actual channel
CSIT:Estimated channel
available at GW
#1 #2 #3
Actual channel
CSIT:Estimated channel
available at GW
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 5 / 18
Nullification description
Estimation of CSI coefficients:
Asynchronous systems → I/C ≈ −15 dBSynchronous systems: → I/N ≈ −15 dBReal system:
CSI estimation performanceArchitecture of the receiver
for CSI detection and estimation
-10 -5 0 5 10 15 20C/N (dB)
0
5
10
15
20
25
30
C/I
(dB)
Carrier is nullified
Carrier is estimated
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 6 / 18
Example of nullification
Carrier to Noise in a subset of 9 beams
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 7 / 18
Number of estimated channel coefficients
Total number of coefficients per channel vector = 245DVB-S2X standard allows to report up to 32 coefficientsNumber of estimated coefficients with nullification: 1-15
50 100 150 200 250 300
20
40
60
80
100
120
140
160
180
200
Estimated coefficients
2
4
6
8
10
12
14
Results with real nullification
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 8 / 18
SINR absolute error due to nullification
Actual channel
CSIT:Estimated channel
available at GW
Precoding matrix
SINR
MODCOD
LUT
MODCOD
Actualprecoded
SINR
Estimatedprecoded
SINR
¿Allows quasi-error-freetransmission with
this MODCOD?
_ +
SINR absolute error
LUTLookup Table
SelectedModulation andCoding Scheme
SINR calculated by the GW
ˆsinrk =|h⊥k wk|2∑
j 6=k |h⊥k wj |2 +N0
Actual user SINR
sinrk =|h⊥k wk|2∑
j 6=k |h⊥k wj |2 +N0
SINR absolute error in dB
ek = 10 log10ˆsinrk − 10 log10 sinrk
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 9 / 18
System parameters
Parameter Value
Satellite orbit GEODownlink frequency Ka-band (20 GHz)Number of beams 245Color scheme Full frequency reuseFading No fading
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 10 / 18
Interbeam scheduling
Unicast: only one user served per beam in each frame
Interbeam scheduling: Users selecting user randomly among...
central positions of the beam within a radius of 2/6, 3/6 or 4/6 ofthe total beam radiusall beam positions
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 11 / 18
Maximum of the SINR absolute error (global results)
Synchronous nullification
Maximum error over the whole satellite coverage [SYNC]
-10 dB -15 dB -20 dB -25 dB
I/N null. threshold (dB)
0
0.5
1
1.5
2
2.5
3
SIN
R a
bso
lute
err
or
(dB
)
Users within a radius of 2/6
Users within a radius of 3/6
Users within a radius of 4/6
All beam positions
Comparison differentnullifications
Maximum error over the whole satellite coverage
ASYNC I/C = -15 dB REAL SYNC I/N = -15 dB
Nullification type
0
1
2
3
4
5
6
7
SIN
R a
bsolu
te e
rror
(dB
)
Users within a radius of 2/6
Users within a radius of 3/6
Users within a radius of 4/6
All beam positions
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 12 / 18
Complementary CDF of the error
Margin required to guarantee a given Frame Error Rate (FER) inall the coverage
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
SINR absolute error (dB)
10 -6
10 -5
10 -4
10 -3
10 -2
10 -1
10 0
CC
DF
: P
(err
or
> x
)
Error over the whole satellite coverage
Users within a radius of 2/6
Users within a radius of 3/6
Users within a radius of 4/6
All beam positions
High capacity losses due to the required large margins!
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 13 / 18
Geographical distribution of the error
Map of the SINR maximum error per beam over 1,000
realizations with MMSE-SPC precoder, real nullification
and scheduling users from inner circle of radius 2/6
Maximum error per position
when in the rest of the beams a
random user is scheduled from
the 3x3 central users
Best solution: independent margin per user to avoid performanceloss of worst-case margin.
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 14 / 18
Countermeasure: link adaptation with adaptive margin
Actual channel
CSIT:Estimated channel
available at GW
Precoding matrix
SINR
MODCOD
LUT
MODCOD
Actualprecoded
SINR
Estimatedprecoded
SINR
¿Allows quasi-error-freetransmission with
this MODCOD?
_ +
SINR absolute error
LUTLookup Table
SelectedModulation andCoding Scheme
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 15 / 18
Countermeasure: link adaptation with adaptive margin
Add to the SINR calculated by the gateway an adaptive margin,independent for each user, updated with its ACK/NAK exchange
Note: these are not used for retransmission purposes
Margin
SINRLUT
DVB-S2XModCods
SelectedModCod
Calculated by the gateway
Updated withACK/NAK exchange
ACK: Margin �Margin + ∆ACKNAK: Margin �Margin - ∆NAK
∆ACK∆NAK
=p0
1 − p0, p0 = Target FER (Typically 1E-5)
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 16 / 18
Simulation results
Scenario:
Continuous transmission of frames to 10 users located in 10non-neighbour beams scattered over all Europe.Target FER: p0 = 10−3, 10−4, 10−5
Results:Without adaptive margin
FER � Target value p0
With adaptive marging
Experimental FER of users is within 90 − 110% of the target FERAvoiding the performance loss of a global fixed margin.
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 17 / 18
Thanks for your attention!
The research was done during a stay at the Interdisciplinary Centre forSecurity, Reliability and Trust (SnT), University of Luxembourg,
supported by the project SatNEx-IV, co-funded by the European SpaceAgency (ESA).
Anxo Tato (atlanTTic, UVigo) ASMS/SPSC 2018 Berlin September 11, 2018 18 / 18
Top Related