Professional Documents
Culture Documents
US010080200B2
1 Min ,throughputi
cell edge CQE
converter
C QInoi:???? Sell - optimization
power controller
(SOPC )
Initial Power Transinit power
CQI Decision adjustment SOPC
Setting Controller Controller estiinator protection
(IPSC ) (CQDC ( TPAE) mechanism
(ANFIS 1 ) (ANFIS 2 ) ( ANFIS 3 )
Adjusled Tx. power
Initial Tx . power
Besi CQI Power allocator + Start
Tx, power
US 10 ,Page
080,2200 B2
OTHER PUBLICATIONS
Necmi et al., Channel estimation using an adaptive neuro fuzzy
inference system in the OFDM - IDMA system , Turkish Journal of
Electrical Engineering & Computer Sciences, pp . 352- 364 , Dec . 20 ,
2015 .*
* cited by examiner
U . S . Patent Sep . 18, 2018 Sheet 1 of 42 US 10 ,080,200 B2
1
.
FIG
-
-
-
!
loss
Path
| SINR
&
m|easured
UE
by
w NUE
.
,
1
-
u
For
DCecQisIon Contrler ((
)
CQIDC )
2
ANFIS
CQIBest
Iniu
.
lt
/Min.throughput celledgeCQI
convert CQImini
,u
sptaeoxtiwenargl Ofe
Bado
S
wimwenyewe
30 _ 2
.
FIG
*
*** ****
*
rCoavdeiraugse y
usor
of bRelsoucrke USTE
Of
Y ofrequiremnt
CeledgeCQI user
U . S . Patent Sep . 18, 2018 Sheet 3 of 42 US 10 ,080,200 B2
ljuomag 2080JO
starten
Butput
to www .
winstein
Westwem
ho +
+
. ..
ho 3
.
FIG
ho 4
*
**
*
*
wi
.. . ..
ECOT w
? 20 21-22
lpAvoaerstaghe Jo
Ki
?
Initial 931-and81 0
,
IX
.
bRelsoucrko u
user
of
U . S . Patent Sep .18,2018 Sheet 4 of 42 US 10 ,080 ,200 B2
r 2013
WANUGH
? . . .
4
.
FIG
rn
. ..
erver s ? metro
3195)
?????? ? ? ? ? ? r * Merry
?? armer.exere
?? ?
LILILLA
ti v ? ? ?
?? ?
ST
CR
EYE
? AIREmeasur nt
atent Sep .18, 2018 Sheet Sof 42 | US 10, 080 ,200 B2
Airtonaesvy
45
.
.
.….………………
.
…..…………
……………
.
?????
.
.
?
…
…
…
?
405
,
'
1
35
,
'
,
.
:
'
t
.
'
……
………
…
…
…
…
.… …. … … … … … …. ??
30
… :
?
-
-
-
-
--
--
.…
-
-
--
??
… …??
…
.
-
.… … …… …… … … … … … … … … …25
?
4????????
-?????“
Yy44444
.
5
.
FIG
. …
.
…
.
” . ]
dB
[
SNR
|
Assa-A .
titv .
.
…
*1520
4
.
-
??????
-
4444
-
???
- .
*
.
"
.
…
?????
"
"
??
id
,
"
.
"
,
"
:
-
-
?
.
,
.
"
"
"
"
-
…
.
…
.
…
:
.
|G … … … … … 5
.… . . ??
?? “
-
? qt?
.
. .
.
… :
.
…… … ……… :
.
.
?
…
.
…
.
)
600
0 IOS
LO
80 3…6000OLD3 CC12cal13CO314
11
.
CO
? ??.96
goo .
TO ZO
9
-
03 ?
DO ?
)
50
2
.
…
D3
90 .
…
…
CQI
15
10?? fr
.
iiiiiiii i
OL
.
titiiiiiii
go
BLER
U . S . Patent Sep . 18, 2018 Sheet 6 of 42 US 10 , 080, 200 B2
09
som
man
har
som
many
sourc
#
es
# 07
.
.
.
.
here,continuewithbroken efna
44 0€
.
ing
.
--
.
--
. OZ laplUNIS DID
'
9
.
.
. . .. . .. . . .. . .. . .. .. .. .. . . .. ... .. . . ... .. .. . 0
.
- 20100
1010 - D-106ED103 010 so10 90100-20100-8010 60100- -OL10 -
b1100
-EL100
21100 -DL100
-
. -
.
100
GL
Afgedeo
-
wafskS
!
g e og ? Û OL OL
)Mbps(Throughput
U . S . Patent Sep. 18 , 2018 Sheet 7 of 42 US 10 ,080,200 B2
296m
.
7
femtocel
*
*
suas
com
107
*
7
.
FIG
976m
.
20
?
:
U . S . Patent Sep . 18, 2018 Sheet 8 of 42 US 10 ,080,200 B2
20
19
.. . . ...
model
R
-
ITU modifeld 18
17
.measuremnt
. : .: . : .: . : .: . : .: . : .: . : .: . : . : . : .: .: . : . 1 . 1: . .: : : .: . : .: . : .: .: . : . U .: .: . : . : . : .: .: . :. : .: .: . : . : . : .: . : . : . : . : .: . : . : .: .: . : . : . : .: .: . : .: . : . . : . : . : .: .:
1 . - - .. : . : . : . : .:
..... .
. : . : . :. : . : . .. :. : . :. : . : . : . : . : .. : : : :: : : : : : : : : : : .. . . . . . . . . ..
16
-
ba
15
t. .I L 14
13
12
11 distance
)
m
(
10
8
.
FIG
9
- - -- - - - - - - - - - - -- - - - - - -- - - - -- - - - 7
.. . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. .. . . .. . . .. . . . . . .. .. . . .. 1 4
. .. 1
100
gutUReLET S
FILTEL ER G OL D E R Eudor 7
=
min
col
Ancol
min
10 90
=
80
.
2
A 70
*
.
4
*
.
.
*
.
*
#Ellen 60
X X BEATSBRYTEBRAX R*K ETR*AKTER .
*
.
L
50 ORB FIG
9
.
30
ot o ELDHOLD
DEDAHAAAA 0999 cossebehov
20
10
4
.
? 8 8 6 ) dBm7 ( Pini 8 ?
0
U . S . Patent Sep . 18, 2018 Sheet 10 of 42 US 10 ,080 ,200 B2
15
***#
**
* **
**
* * **
*
*
**
*2
2%
2%2
R R
)
m
(
R
)
a
(
10
.
FIG
*
**2
1931 *!
MENS 2222
A1
,
1 2385238
239 2438
thepeople
disSeq?o? o saibag
o
atent Sep. 18, 2018 Sheet 11 of 42 US 10 , 080, 200 B2
100
A23 90
80
9
82&
70
% A3 *
' 2
#
2
&
##
60
#
)
b
(
10
.
FIG
A22 50 NRB
40
***
**
*1.*
**
2
ani res.
2
004 mars 30
9
2 2604
1 van 2*2
2
* 22,19822 22
293 20
2 433*247 28
1
,
A2 10
hogy a p ö
0
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 12 of 42 US 10 ,080 ,200 B2
19
14
A3
,
3 *
*
**19:
12
###
#####
#**
10
A3
,
2
-
-
* 8
CQliman )
c
(
10
.
FIG
thisit
6
*
*
4
*
*
A3
,
1 .HHHTrungGi?i?i
- *
*
*
*
una
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 13 of 42 US 10 ,080 ,200 B2
AS
14
*
"
*
***
3008
****
*8888
12
ansatte
***
# 10
2
8R
)
m
(
)
a
(
11
.
FIG
*
323
** 9 *
24
* 226 war tesz
10% 20%
24 e
22 CM
ook al ebeur o
membership of Degree
U.S.Patent
atent Sep. 18, 2018 Sheet 14 of 42 US 10 ,080,200 B2
#
* 99100
*
*
*
Wie:98 90
ge#
###
2
its *
## 99
w
#
#
**High 80
70
##
##
#
##
#
#
#
#
##
60
50
40 )
b
(
11
.
FIG
nRB
30
20
1968.a
.
te s
-
e
1
,
A2 *
*
10
s
time
4
c are
membership of Degree
U . S . Patent Sep . 18, 2018 Sheet 15 of 42 US 10 ,080 ,200 B2
14
.
14
##
**#
**
*
* 12
12
#
2
*
##
##
*
*
##
# 10
10
*
*
8
Colmin )
c
(
11
.
FIG
*
6
**?2237389
****
4
Ag
, and
***2236.****
in
fine*
*
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 16 of 42 US 10 ,080 ,200 B2
500
450
400
350
300
. . 0
L
5
.
0
5
.
3
) dBm( RMSE
U . S . Patent Sep . 18, 2018 Sheet 17 of 42 US 10 ,080 ,200 B2
. .
OL
.HAAT .
W
sin
i
.
.
99
. .
.
.
AAA .
"
09
0
penis
pada
seperti
.
.
istin
trinstine
.-020 0 SG
ya.mveongvione 13
.
FIG
** * ** iii vi
09 PL
)
dB
(
1309
IASI
.
1
.
bo St
1? 8 .
. soba
CQbest
.
.
itsi
.
??????
ri
"
!
n
dBm
40
-
PinjaPini
dBm
20
-
=
nom wgp
6
d
"
=
o
se
Ot
0€
U . S . Patent Sep . 18, 2018 Sheet 18 of 42 US 10 ,080,200 B2
As *
*
*
#
##
#
#
19
dit!
dan
22
70
# #
#
!
*** t
*
*!
W
65
60
*23* Ha!Ha
55
#
1989 #
#
#
2
,
1
) FIG
dB
(
PL )
a
(
14
.
at iti dei
50
45
40
35
posted
membershipof Degree
30
U . S . Patent Sep . 18, 2018 Sheet 19 of 42 US 10 ,080 ,200 B2
?????????????????????????????????? ???????
20
*
*
###
#
*
**
gitigödä*9* 10
* *
#*
*
*
**
*
*
*
0
*
**
*
**
*2
*2
##
10
-
#
*
*
#
*
#
20
-
.
2
Pini FIG
)
dBm
( )
b
(
14
.
30
-
* ***?2038193912222** CX 40
-
*
*** 26
*
22
336
722623
3592 50
-
298*.**
liste
:1999 #28 9.*X X X O
60
-
*
O
OO 70
-
*
membership of Degree
U . S . Patent Sep . 18, 2018 Sheet 20 of 42 US 10 ,080 ,200 B2
A3
,
3 100
*
90
#
## #
#
#
#
#
#
#
80
##
#
##
##
*#
**
** 70
*
60
)
c
(
14
.
FIG
50
SRB
itutän:jitite
40
9432767#2 8542
862tesat
387
**222** **Di
20
10
132sim
a
0
8
.
0
nö
a
membership of Degree
U . S . Patent Sep . 18, 2018 Sheet 21 of 42 US 10 ,080 ,200 B2
60
60
vier 199$
*
**29.999
19981997
443989
toDONARDO
A
#999
****#
19*
**
99
#
#
##
#
55
55
e
#
Wi #
#
#
#
#
#
##
50
50
) FIG
45 PL
dB
( )
a
(
15
.
45
weas
et
40
40
*
*
**
***
1934hesten*100
esi
*
35
35
vi
*
*
zaintWieis
thisthatus
417
1928att*aus
30
on membership oftomeDegreeda bagn o
U . S . Patent Sep . 18, 2018 Sheet 22 of 42 US 10 ,080 ,200 B2
* 0
2N
f*
*
:44999din
3
,
A2
A22 9
* #
*
*
*
#
** *
##
*# 10
-
#####
##
#
*
20
-
X ) FIG
30 Pini
-
40 dBm
( )
b
(
15
.
50
-
000
60
-
.
00
70
-
membershipof Degree
U . S . Patent Sep . 18 , 2018 Sheet 23 of 42 | - US 10 , 080 , 200 B2
eos meet
to
me
us 90
80
.
70
tttt
to
60
t
Ag2 50
nRB
)
c
(
15
.
FIG
I
H
THE
40
THE
E
F
A 30
PET
20
10
1
ad
ess the
?
membershipof Degree
U.S. Patent atent Sep. 18, 2018 Sheet 24 of 42 US 10 ,080 ,200 B2
500
450
400
350
300
eTrainorg 250 epoch FIG.16
200
150
100
50
0
4
.
0
RMSE
U . S . Patent Sep . 18, 2018 Sheet 25 of 42 US 10 , 080 ,200 B2
50
•
•
+
at
1
-
,
'
t
40
7
best
|
CQ
,
6
=
cQImin
e
10
best
CQ
,
6
=
CQImin
3
-
30
,
.
4
,
20
at
i
+
,
th
-
,
t
-
,
17
.
FIG
,
-
i
10 B
{
R
N
}
S
initi
an
th
5
-
41
111211 2 .
…
-
:
.
,
t
-
4
…
.
.
-
2
-
,
,
:
*
0
-
"
:
'
!
*
?
.
,
th
8
-
7
.
p
is
f
*
18
n
dr
-
10
-
-
-
'
,
ie
is
lt
20
-
( .
-
'
ie
ha
h
'
it
tr
i
.
:
8 8 8 8 8 9 8 8 8
) dB { P A
U . S . Patent Sep . 18, 2018 Sheet 26 of 42 US 10 ,080 ,200 B2
*
*
**
42N
14
14
An
3 !a
di
****
***
12
12
*43*
#*
10
)
a
(
18
.
FIG
2
.
1
8
CQl
min
6
A1
,
1
& Š
membership of Degree
?
X 4
U . S . Patent Sep . 18, 2018 Sheet 27 of 42 US 10 ,080 ,200 B2
&
&
*
14
*
&
A2
,
3 *
12
10
Colbest )
b
(
18
.
FIG
*w ** 6
&
*
&
*
*
*
$
*
*
4
*
A21 **istnat
100am
*
*
*
*
*
XXX
X
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 28 of 42 US 10 ,080 ,200 B2
*
*
**
40
angers
A3
,
3 *
*
i
*
et
*
*
le 30
*
*
* 20
)
dB
(
SINR
A32 *
**
*
*
*
10 )
c
(
18
.
FIG
*
*
*
*
10
-
A3
,
1 *
*
**
**
*
*
*
*
**
20
-
TA
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 29 of 42 US 10 ,080,200 B2
14
14
12
2
,
1
10
10
lin
i
1905Hans
8
calmin )
a
(
19
.
FIG
2228****.*tex29622331
4
A1
,1
ó of
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 30 of 42 US 10 ,080 ,200 B2
-- - - - - - - - - - - - - - - - - - - -- -- - ----- -----
**
*
#*
#
#
14
#
*
**
*
#
#*
*
12
12
A22 10
10
dinwillthat##
8
CQlbest )
b
(
19
.
FIG
00o o *
*
*
*2*
wod0
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 31 of 42 US 10 ,080,200 B2
ON
40
#
3#
#
se*## * *
30
#
*
* # #
2
2.3882
** 20
#
*
#
#
##
#
##
#
10
)
dB
(
SINR )
c
(
19
.
FIG
286 10
-
2982.**98
A*
* 2954
2
9 tesa
membershipof Degree
U . S . Patent Sep . 18, 2018 Sheet 32 of 42 US 10 ,080 ,200 B2
500
450
*
400
*
*
350
300
eTrainorg *
250 epoch 20
.
FIG
200
min m
min min
150
100
thi
50
0
5
.
3
N 5
.
1 5
.
0
) dB( RMSE
U . S . Patent Sep . 18, 2018 Sheet 33 of 42 US 10,080,200 B2
+
=
1
)
1
+
i
(
wy
trip
L
.
10X
70€P
=
)
m
( 11
141
, for
APW
+
1
-
Pum
=
)
m
(
Pu NUE
.
,
1
=
u P>
Pmaxroiect Yes -
(
P
=
)
i
,
Phap
NUE
-
,
1
=
u
for >
AP
if
4
uli
,
Pump
=
)
1
+
i
(
x
.
Pemp APCOfor
0Pimpy
if
,
4
-
u
.
Pemp
=
)
1
+
i
( NUE
-
,
F1 Pmax
>
)
1
+
Preselili
< 21
.
FIG
Pouli
1
+
)
+
I
totul
Noo -
Yes
-
a
22
.
FIG
n
45
m
40
35
am an
30
20
- + - --
- - -
1. - - 1 . - 1. .. , .
.
Sage 15
(dB)SINRmeasurem nt FIG
.
23
a
(
)
10
L : 1 : . ... : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . t . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : .: . : . : . 5
5
-
10
-
) UE3)
IDCC
(
.. : . : . : . : . : . : . : .: . : .: . : .: . : . : . : .: .: . : . : .: .: . : . 1 . : .: .: . : . : .: .: . : . : .: .: . : . : .: .: . : . : .
) UE
IDCC
(
1
UE IDCC
(
2 15
-
)A
power
fixed
(
2
UE
- -
.
UE
fixed
(
3
power
)
.
-
fixed
(
1
UE
power
) 20
-
25
-
-
30
-
the sout? pa
333
U . S . Patent Sep . 18, 2018 Sheet 36 of 42 US 10 ,080 ,200 B2
50
45
VXXXXXXX 40
. : .: . : .: . : .: . :. : .: . : .: . : .: . : .: . .: . : .: .: . .: . : .: . : . 1 . 1 .1 .1 . 1 . 1 . 1 . . 1 . 1 . 1 .1 . 1 . 1 IS , ' , T , ' , , , N , : : . : .: .: . : . : . : . : . : . : . :2 1 : : _ 00
35
30
.. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . .. . . . . . . . . . .. . . .. . . . . . , , , , ,, ,
*
, , , , , ,, , , , , , , ,, ,, . . . . . . . . .. .. .. ........ .... .. .. 25
****
20
ESTATESKAL LOODDODODD00000000001 BBC
15
OGORA
omeCou 10
(dB)mSeaIsurNemRnt )
b
(
23
.
FIG
5
H 0
-
.
-
-
-
.
5
-
- . - . - -
10
-
) WE
power
fixed
(
1
UE )
power
fixed
(
2
)-
power
fixed
(
3
UE
-
A ) UE
IDCC
(
1
UE (
2
DCC
) UE)
DCC
(
3
15
20
-
-
-
-
-
-
-
-
@
-
- 25
30
-
50
.
45
40
.
35
.
30
.
...
25
.
00g *** 20
15
L. . : :
6
SSSSSSSSSSSSSSSS S
III
.
SSSSSSSS : : : :
.
:: : : : : : : :
** . . .. . 10
(dB)SINRmeasurem nt FIG
)
c
(
23
.
.
S 5
.
ogOOOOO .
.
.
L
0
5
-
.
:
.
.
10
-
UE
fixed
(
2
power
)
.
.
3n
)
p?xy
iemod
|
UE
IDCC
(
1
)UE )UE
IDCC
(
2 )
IDCC
(
3
.
.
-
A
UE
fixed
(
3
power
) 15
20
-
-
.
andrtima
.
-
7
.
.
0 O
-
0 0 0A
25
.
.
-
.
ja & 30
-
doo
CCDF
U . S . Patent Sep . 18, 2018 Sheet 38 of 42 US 10 ,080,200 B2
sz
.
UE
1
IDCC
(
)UE )UE
2
(
IDCC 3
(
DCC
) 22
*
power
fixed
(
2
UE
.
power
fixed
(
1
UE
-
-
A
UE
(
3
fixed
power
)
.
oo *
OC
*
A 44
*
. ---
*
"
.
826
GO OOGA
.
- - - - - - - - -
SL
.
FF
)(Mthrboupghsut FIG
)
a
(
24
.
SZL
.
:
6 .
OL
..........
.
700.00 ..
...
.,
06
.
.
.
*
62
9
.
-. - . .. * -. .
.
92
.
CCDF
U . S . Patent Sep . 18, 2018 Sheet 39 of 42 US 10 ,080 ,200 B2
25
26KXK
OK*
5
.
22
UE
fixed
(
2
power UE
(
1
DCC
) UE)UE
2
(
DCC 3
DCC
(
)
-
UE
3
fixed
(
power
-
.
UE
fixed
(
1
power
)
- --
B mun
TOV 20
5
.
17
..
DOO
Byxor
.
..
.
: - . :. : . : . : . : . : . : . : .: . : . : . : . :. : . : .:. : . : . , . , . , . , , , , , , , , , , , , , , , , , ,' , ,' , ,' , N, ' , I , ' , N, ' , ' ' . - , , , ,, , , , ,, ,, , , , ,, , , : : : : : : : : : : : : :: : 15
.
Soo 5
.
12
(Mbps)throughput FIG
.
24
(
b
)
.
10
BOOK
danwanita
.
.. .
***
exe 5
.
7
.
. . . . : . : . : . : . : . : . : . : . . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . " . . : . : . : . : . : . : . : . :. . : . :. : . : . : . . :. . : . : . : .: . : . : . : . : . : . : . : . : . : . : . : . : . : . : . : . . : . : . : . : . : . : . : . : . : . : . : . : . : . : . '. 5
.
2
. ..
. . . . . .. .
.. . .
0 . . ....
non sono p
solo 4000 o
U . S . Patent Sep . 18, 2018 Sheet 40 of 42 US 10 ,080 ,200 B2
25
Det
)
power
fixed
(
1
UE
O
.
UE
fixed
(
2
power.VE
fixed
(
3
power
-
UE
1
DCC
(
)
.
UE
2
DCC
(
)
)
DCC
(
3
UE
A
S86019 5
.
22
20
*
*
o
K
...
com
M A N *
*
5
,
17
15
5
.
12
10
(Mbps)throughput FIG
)
c
(
24
.
ce ode
*
-
- 5
.
-- 7
......
.... ....
5
V
"
.
!
. ...
ie
and
maging
ntetrenanctioinoains
!
. 5
.
2
wa ring
.
-
. ..... 0
een
en op Ô og CCDF
elegante e pa o O
o
U . S . Patent Sep . 18, 2018 Sheet 41 of 42 US 10 ,080 ,200 B2
80
-
82
-
vi
)
power
fixed
(
Total
- )
power
fixed
(
2
UE
- )-
IDCC
(
Total
- UE
1
IDCC
(
)
www
UE
2
IDCC
(
)Am UE
3
IDCC
(
)
-
-
VE
fixed
(
1
power
) .
-
A
VE
(
3
fixed
power
) -
- *
*
*
*
84
-
-
86
-
.
.
*
.
*
*
more
(dBm)Interfer nce
-- --* -
.. 0- 100 88
90
92
- 25
.
FIG
-
94
-
$ ó 00
- Hosting . com
.
96
-
4.- .
.
.
.
.
.
98
-
-
60 BOL 100
-
) Mbps( throughput Average
U . S . Patent Sep . 18, 2018 Sheet 42 of 42 US 10 ,080 ,200 B2
)B
IDCC
(
1
VE
O )A
IDCC
(
2
UE
- VE
3
IDCC
(
)- 80
82
84
www
Total
-
IDCC
()
86
-
L
(dBm)Interfer nce
-
88
90
92
- 26
.
FIG
94
-
96
-
*
*
*
.
.
98
-
100
-
in ó seu ) dBm( power Tx Average
Hoe km
40
-
US 10 ,080 ,200 B2
INTELLIGENT DEPLOYMENT CASCADE global traffic distribution and interference levels. The algo
CONTROL DEVICE BASED ON AN rithm in a decentralized femtocell deployment has not been
FDD -OFDMA INDOOR SMALL CELL IN considered . A research report has proposed an adaptive
MULTI- USER AND INTERFERENCE neural fuzzy inference system (ANFIS )-assisted power con
ENVIRONMENTS 5 trol scheme for a multi- rate multimedia direct -sequence
code - division multiple -access (DS -CDMA ) system to pre
TECHNICAL FIELD cisely predict the channel variations and thus compensate for
the effect of signal fading in advance . The author in the
The invention presents an adaptive neural fuzzy inference above study also provides a procedure for determining the
system (ANFIS ) based intelligent deployment cascade con - 10 transmission rate based upon the output of the signal- to
trol ( IDCC ) device for frequency division duplexing ( FDD ) - interference - plus -noise ratio (SINR ) increment of the
orthogonal frequency division multiplexing access ANFIS power controlmechanisms at the sample period. The
(OFDMA) indoor small cell operated in the multi-user (MU ) fuzzy membership functions of ANFIS power control
and interference environments to self- optimize the MU mechanisms use seven Gaussian functions , so that there are
service reliability (SR ), throughput, minimum transmit 15 49 fuzzy inference rules. The ANFIS power controlmecha
power and interference for multimedia call services . The nisms use two input variables, including SINR error e (n ) and
proposed IDCC device is designed with a cascade architec SINR error change Ae(n ), to track the set point of target
ture , which mainly contains five units including a resource SINR . In the present technique , the target SINR value is set
allocator, a minimum throughput/ cell edge CQI converter , to a fix value of 1. 5 dB , let the power control process is not
an adaptive neural fuzzy inference system (ANFIS ) based 20 flexible enough . The input parameters of ANFIS power
initial transmit power setting controller (ITPSC ) in the first control mechanism totally depend on SINR control effi
cascade unit , an ANFIS based channel quality index (CRI) ciency . The power cannot be controlled by channel envi
decision controller (CQIDC ) in the second cascade unit and ronment. The technology has not considered the perfor
an ANFIS based self -optimization power controller (SOPC ) mance of multi -user (MU ) service reliability (SR ).
in the third cascade unit . The SOPC consists of three parts, 25
namely the transmit power adjustment estimator ( TPAE ), SUMMARY
transmission power assignment and self-optimization power
controller protection mechanism . In view of the disadvantages of prior art , the primary
object of the present invention is to present an adaptive
BACKGROUND 30 neural fuzzy inference system (ANFIS ) based intelligent
deployment cascade control (IDCC ) device for frequency
Currently, the macrocells are deployed by operators . division duplexing (FDD )-orthogonal frequency division
Since the deployment of femtocells can be in orders of multiplexing access (OFDMA ) indoor small cell operated in
magnitude more numerous than traditional cellular deploy - the multi- user (MU ) and interference environments . The
ments and a network operator may not be able to control the 35 proposed IDCC device is designed with a cascade architec
femtocells directly . The femtocells are self-deployed by ture , which mainly contains five units including a resource
users rather than operators. Therefore , the femtocell base allocator, a minimum throughput/ cell edge CQI converter,
station ' s (BS ) self-optimization deployment control soft- an adaptive neural fuzzy inference system (ANFIS ) based
ware must have the characteristics of easy operation to make initial transmit power setting controller (ITPSC ) in the first
the BS with the least human action to satisfy the required 40 cascade unit, an ANFIS based channel quality index (CRI)
performance , which are stated hereinafter. The user just decision controller (CQIDC ) in the second cascade unit and
needs to plug - and- play and the BS of the femtocell can an ANFIS based self -optimization power controller ( SOPC )
automatically configure the system parameters in the MU in the third cascade unit. The SOPC consists of three parts ,
and interference indoor environments. In addition , the self- namely the transmit power adjustment estimator (TPAE ),
optimization control software deployed the eNode B ( NB ) 45 transmission power assignment and self - optimization power
of femtocell in an interference environment can self- opti- controller protection mechanism . In the experimental
mization control the transmit power of the BS to save example , it assumes that the number of indoor users is three ,
energy , reduce co - channel interference for the adjacent cell, the system bandwidth of the femtocell is set as 20 MHz and
and meet the requirement of service reliability . User input the required minimum throughputs for each of users are 2 . 76
settings include service reliability , the cell edge throughput 50 Mbps, 7 .44 Mbps, and 14 . 13 Mbps, respectively .
corresponding to the cell edge CQI and cell radius to match The principal object of the present invention is that it can
the size of the room coverage . A research report forecasts the autonomously control the assignments of the resource block ,
global small cell market to grow from $ 690 .0 million in initial power, the best channel quality index (CRI) and the
2014 to $ 4 . 8 billion by 2019 , at a Compound Annual Growth minimum transmit power, so that the indoor small cell can
Rate of 41. 7 % . Thus, the short distance femtocell technol- 55 produce the optimum throughput, minimum transmit power
ogy in the future development of next- generation wireless and interference for multimedia services . The resource allo
communication networks and applications will play a very cator sets the average number of resource blocks for each
important role. indoor user according to the number of users and system
A previous study has proposed a coverage adaptation bandwidth in the small cell. The minimum throughput/cell
approach for femtocell deployment in order to minimize the 60 edge CQI converter sets the minimum (cell edge ) CQI for
increase of core network mobility signaling. The informa- each indoor user in accordance with the minimum ( cell
tion on mobility events of passing and indoor users are used edge ) throughput requirement. Here the cell edge CQI
to optimize the femtocell coverage . An approach based on corresponds to the SINR threshold for the BLER of the
genetic algorithm was presented in to automatically opti- transceiver equal to 10 - 1 . The present invention uses cas
mize the coverage of a group of femtocells in an enterprise 65 cade ANFIS architecture to adapt the initial transmit power
environment. The algorithm is able to dynamically update setting to the requested throughput at the cell edge , coverage
the pilot powers of the femtocells as per the time varying radius and the allocated number of resource blocks ; to adapt
US 10 ,080 ,200 B2
the best CQI to the initial transmit power setting and average FIG . 14 (b ) shows the initial transmit power initialization
path loss (PL ) measured by user equipment (UE ) and the (Pini ) membership functions of CQIDC unit .
allocated number of resource blocks; to adapt the transmit FIG . 14 ( c ) shows the initial number of resource block
power adjustment estimator ( TPAE ) in SOPC unit to the (nRB ) membership functions of CQIDC unit .
requested CQI at the cell edge , the best CQI and measured 5 FIG . 15 (a ) shows the learned average path loss (PL )
average SINR . The present IDCC device is design to self- membership functions of COIDC unit.
optimize the signal- to - interference -plus -noise (SINR ) and FIG . 15 (b ) showsthe learned transmit power initialization
throughput service reliabilities of the indoor small cell in the (P . .) membership functions of COIDC unit .
multi- user (MU) and interference environments , while FIG . 15 ( c ) shows the learned number of resource block
maintaining the blocking error rate (BLER ) less than 10 - 1 10 (nRBR ) membership functions of CQIDC unit.
and minimizing the transmit power and interference power FIG . 16 shows the rootmean square error of CQIDC unit.
to achieve the aims of energy saving and interference FIG . 17 shows a set of training data for the ANFIS based
reducing TPAE in the SOPC unit: The optimum power adjustment
Other aspects and advantages of the present invention will
ntion 15
become apparent from the following detailed description , 15 ent corresponding to CQImin = 6 , the best CQI= 7 , 10 , and differ
con average measured
taken in conjunction with the accompanying drawings , SINR .
illustrating by way of example the principles of the present FIG . 18 (a ) shows the initial cell edge CQI (CQImin )
invention . membership functions of the ANFIS based TPAE in the
SOPC unit .
BRIEF DESCRIPTION OF THE DRAWINGS 20 FIG . 18 (b ) shows the initial best CQI (CQIbest ) member
ship functions of the ANFIS based TPAE in the SOPC unit .
The present disclosure willbecomemore fully understood FIG . 18 ( c ) shows the initial measured average SINR
from the detailed description given herein below and the (SINR ) membership functions of the ANFIS based TPAE in
accompanying drawings which are given by way of illus - the SOPC unit.
tration only, and thus are not limitative of the present 25 FIG . 19 (a ) shows the learned cell edge CQI (CQImin )
disclosure and wherein : membership functions of the ANFIS based TPAE in the
FIG . 1 is a block diagram of intelligent deployment SOPC unit
cascade control (IDCC ) device for FDD -OFDMA indoor FIG . 19(b ) shows the learned best CQI (CQIbest) mem
small cell.
FIG . 2 shows the architecture of an ANFIS based initial 30 FIG . functions
bership
19 ( c )
of power increment estimator unit.
shows the learned measured average SINR
transmit power setting controller (ITPSC ) unit. (SINR ) membership functions of the ANFIS based TPAE in
FIG . 3 shows the architecture of an ANFIS based channel the SOPC unit.
quality index (CRI) decision controller (CQIDC ) unit. FIG . 20 illustrates the root mean square error of the
FIG . 4 shows the architecture of an ANFIS based transmit
power adjustment estimator in SOPC unit. ANFIS based TPAE in the SOPC unit.
FIG . 5 illustrates the BLER performance of DL OFDM FIG . 21 shows the flow chart of power allocation algo
transceiver in IOA channel for CQI= 1 , 2 . . . , 15 . rithm .
FIG . 6 illustrates average throughput and system capacity FIG . 22 shows the simulation flow chart of IDCC .
of DL OFDM transceiver in IOA channel for CQI= 1 , FIG . 23 (a ) shows the CCDF of SINR measurement for
2 . . . , 15 . 40 different UEs using fixed power and the IDCC in interfer
FIG . 7 shows the scenario of laboratory with femtocell e nce free environment.
eNB and measurement points . FIG . 23 (b ) shows the CCDF of SINR measurement for
FIG . 8 shows the modified path loss model obtained from different UEs using fixed power and the IDCC in interfer
the field trial in the laboratory . ence power = - 90 dbm environment.
FIG . 9 shows a set of training data for ITPSC unit : The 45 FIG . 23 (c ) shows the CCDF of SINR measurement for
minimum transmit power corresponding to coverage different UEs using fixed power and the IDCC in interfer
radius = 5 m , the cell edge CQI= 3 , 7, 10 and number of ence power= - 80 dbm environment.
resource block = 1 to 100 . FIG . 24 (a ) shows the CCDF of throughput for different
FIG . 10 ( a ) shows the initial coverage radius ( R ) mem - UEs using fixed power and the IDCC in interference free
bership functions of ITPSC unit . 50 environment.
FIG . 10 (b ) shows the initial number of resource block FIG . 24 (b ) shows the CCDF of throughput for different
(nRB ) membership functions of ITPSC unit. UEs using fixed power and the IDCC in interference
FIG . 10 (c ) shows the initial cell edge CQI (CQImin ) power = - 90 dbm environment.
membership functions of ITPSC unit . FIG . 24 ( c ) shows the CCDF of throughput for different
FIG . 11 (a ) shows the learned coverage radius (R ) mem - 55 UEs using fixed power and the IDCC in interference
bership functions of ITPSC unit . power= - 80 dbm environment.
FIG . 11 (b ) shows the learned number of resource block FIG . 25 shows the average throughput of DL transceiver
(nRB ) membership functions of ITPSC unit . for different UEs using fixed power and the IDCC in the MU
FIG . 11 ( c ) shows the learned cell edge CQI (CQImin ) interference environment.
membership functions of ITPSC unit. 60 FIG . 26 shows the average transmit power of DL OFDM
FIG . 12 shows the rootmean square error of ITPSC unit . transceiver for different UEs using the IDCC in the MU
FIG . 13 shows a set of training data for CQIDC unit: The interference environment.
best CQI corresponding to the initial transmit power of
ITPSC = - 40 dBm , - 20 dBm , 0 dBm , average path loss = 30 DETAILED DESCRIPTION
dB to 70 dB and number of resource block = 1 to 100 . 65
FIG . 14 ( a ) shows the initial average path loss (PL ) For your esteemed members of reviewing committee to
membership functions of CQIDC unit. further understand and recognize the fulfilled functions and
US 10 ,080 ,200 B2
structural characteristics of the invention , several preferable number of resource blocks (NRBW), the output parameter for
embodiments cooperating with detailed description are pre the uth user is the best CQI (CQIbestu ). Under interference
sented as the follows. free environments, the CQIDC unit adapts the best CQI to
The invention presents an adaptive neural fuzzy inference the changing PL , , Pini , and nRB , The Gaussian shape
system (ANFIS ) based intelligent deployment cascade con - 5 membership function of each input parameter is divided into
trol (IDCC ) device for FDD -OFDMA indoor small cell three levels . There are 27 fuzzy inference rules. The CQIDC
operated in the multi-user (MU ) and interference environ unit defines the average path loss ( PL , ) in 30 dB ~ 40 dB for
ments to self-optimize the MU service reliability (SR ), L , it defines average path loss (PLU) in 41 dB ~ 60 dB for M ,
average throughput, transmit power and interference for it defines average path loss (PLU ) in 61 dB ~ 70 dB for H ; it
multimedia call services .
The principal structure of the present invention is an 10 51defines initial transmit power setting (Pi ) in - 75 dBm ~
dBm for L , it defines initial transmit power setting ( Pi )
ANFIS based IDCC device as shown in FIG . 1 ; which in –50 dBm - - 4 dBm for M , it defines the initial transmit
ini,
There are 27 fuzzy inference rules . The ITPSC unit defines defines CQImin , in 11 - 15 for H ; it defines the CQIhesty, in
the Ry of less than 5 meters as low (L ), it defines 5 1 ~ 5 for L , it defines CQIbestu in 6 - 10 for M , it defines the
meters ~ 10 meters as medium (M ), it defines 11 meters - 15 COIbest ,, in 11 - 15 for H ; it defines the SINR , less than - 25
DESTU
meters as high (H ); it defines the nRB , of 1 RB - 25 RBs as 55 dB ~ - 5 dB for L , it defines the SINR ,, in - 4 dB - 25 dB for
low (L ), it defines 26 RBs~ 74 RBsasmedium (M ), it defines M , it defines the SINR , in 26 dB ~ 45 dB for H . The
75 RBs~ 100 RBs as high (H ); it defines CQImin ., in 1 - 5 for
7717 7 . signal- to - interference -plus-noise ratio (SINR ) is estimated
L , it defines CQImin117 , „0, in 6 - 10 for M , it defines CQImin
min ., U, in through the reference signal received power (RSRP ) mea
11 ~ 15 for H ; the output for the uch user is the initial transmit sured from the user equipment (UE ) and sends back to the
power setting (Piniu ), which satisfies the uth user requests 60 IDCC device in e Node B ( NB ). The estimated path loss
under interference free environments . (PL ) is obtained by subtracting RSRP from the transmit
The architecture diagram of ANFIS based COIDC unit is reference signal.
shown in FIG . 3 , which contains five layers , a total of three ( A ) The Architecture of the ANFIS Controller:
inputs and one output. There are three input parameters for The ANFIS based TPAE in the SOPC unit is used as an
the uc user including average path loss (PL ) between 65 example to describe the framework of the ANFIS controller :
orthogonal frequency division multiplexing (OFDM ) trans - Layer 1 : The generalized bell shape membership func
mitter and receiver, initial transmit power setting (Piniu ) and tions are defined as:
US 10 ,080 ,200 B2
the BLER and throughput of the LTE downlink (DL ) trans
Ajin (Xj,m ) = 2 (1) ceiver for indoor small cell are simulated to generate the
1+1time@j,n simplesite
xj,m - Cin |2bj,n training data for the ANFIS ITPSC . The system parameters
are shown in Table 1 and fundamental parameters of the
for n = 1, 2, 3 and j = 1, 2 , 3 5 transceiver are shown in Table 2 . In the simulation of the
present embodiment, for the different channel quality index
(COI), the BLER of 1x1 SISO - OFDM transceiver is simu
where x ; m is the m , input and the premise parameters aina lated where the system bandwidth is 20 MHz, the indoor
bin , C ; n pertaining to the node outputs are updated according office A ( IOA ) channel is selected as channelmodel, the least
to given training data and the gradient descent approach . 10 square (LS ) channel estimation and minimum mean square
Layer 2 : The output of node i, denoted by 02.i, is the error (MMSE ) equalizer are used , and the user speed is
product of all the incoming signals for the ish rule. It is given assumed to be 10 km /hr. The 1000 sub frames are applied for
by : the simulations. The results are shown in FIG . 5 , which is
Wi,m = 02,;= 41,p(* 1.m )XA2,4(x2,m )x43, (X3.m ) 15 used as a training data to define the SINR threshold for
BLER = 10 - 1 under different CQI, as shown in Table 3 . In
for i= 1,2,27;p = 1,2,3;q = 1,2,3;r= 1,2,3 (2) each COI (corresponding to each pair of modulation and
Layer 3 : The output of node i, denoted by 03.1, is called code rate mode ), the difference between the SINR measure
3 . 77
the normalized firing strength and calculated as: ment value and the SINR threshold value is used to control
the size of the transmit power adjustment AP , which must
compliance with the provisions of Table 3 , where the inter
Wi.m (3) ference power is not considered . The physical resource
03,i = Wim = = - , for i = 1 ~ 27 blocks ( PRBs) of Table 3 is generated according to the
Wi.m system bandwidth of 20 MHz and three indoor users . When
the system bandwidth is 20 MHz, the corresponding total
number of physical resource blocks (PRBs ) is 100 PRBs .
Layer 4 : Every node in the fourth layer is an adaptive The resource assignmentmethod of this invention is the
node with a node function : orthogonal frequency division multiplexing access
04.1= wimfim = winx ( x1 mn + BMX2,m +Y43,m +W ;); (OFDMA ) for the frequency division duplexing (FDD )
mode of indoor small cell operated in the multi- user (MU )
for i= 1 ~ 27 (4) 30 environments. The eNB of the indoor office will perform the
resource allocation for each UE with 33 RBs at each time
where 04.j is the node output, fim is a crisp output in the instant
consequence , and the dig Big Yie W ; are the consequent system . capacity
For practical implementation considerations, the
of the downlink (DL ) OFDM transceiver
parameters ofnode i. The 27 fuzzy inference rules of fim are os formula is modified as [ 10 ]:
constructed as follows:
R1: if (* 1,m is Au ) and (x2.m is A21) and (X3,m is A31)
then (output is fim ); nRBH
C (bps ) = BW
BW ..BW
BW __ eff
eff..nRBtotal
* . n .log ( 1 + SINR /SINR _ eff)
R2: if (* 1,m is Au1) and (x2.m is A21) and (X3,m is A32)
then (output is f2,m ); 40
R3: if (* 1,m is A11) and (x2.mn is 421) and (x3.mn is 433 ) where nRBjotai is the total number ofRBs and nRB , is the
then (output is f3 ,m ); number of RBs assigned for the uin user; BW and BW _ eff
are system bandwidth and effective system bandwidth ,
R26: if (* 1.m is A13) and (x2.mn is A23) and (x3,m is respectively . The parametery is a correction factor. SINR
A32) then (output is f26,m ); 45 and SINR _ eff are signal to interference plus noise power
ratio and effective signal to interference plus noise power
ratio , respectively . In this invention , the simulation param
R2 : if (x1.7 is A13) and (x2m is A23) and (x3.,m is eters of DL SISO OFDM transceiver is given in Table 1 ,
433) then (output is f27 ,m ) (5) where BW = 20 MHz, BW _ eff= 0 . 83 , n = 0 .43 and SIN
The above 27 fuzzy inference rules are used for deter - 50 R _ eff = 2.51199 (4 dB ). The average throughput of DL trans
mining the assigned data rate to achieve optimization objec ceiver in IOA channel for CQI = 1 , 2 . . . , 15 is shown in FIG .
tive . 6 , where the system capacity is denoted by dotted curve and
Layer 5 : The single node in the fifth layer is a fixed node the simulated average throughput for CQI= 1, 2 . . . , 15 are
labeled 2 , which computes the overall output Og as the denoted by solid curve. It is observed that the simulated
summation of all incoming signals . 55 average throughput approximates to the Shannon capacity
bound for LTE specifications. Thus , the throughputs corre
sponding to SINR threshold for the different COIs are
27
calculated with (7) and listed in Table 3 . As can be seen from
Cm=0;= Šmiem kfiem table , the input throughput requirement settings for the uth
60 user in the range of 1.99 – 2 .76 Mbps , 6 . 76 ~ 7 . 44 Mbps,
12 . 99 – 14 . 13 Mbps correspond to the cell edge CQI
(B ) The Minimum Throughput/ Cell Edge CQI Conver- (CQImin,w ) in 3 , 7 , and 10 , respectively .
sion Unit: (C ) Initial Transmit Power Setting Controller (ITPSC )
In order to satisfy the user requirements of indoor small Unit :
cell in throughput and blocking error rate (BLER ) of less 65 In order to control the initial transmit power of small cell
than 10 -1, the relationship between the throughput and SINR eNB for satisfying the requirements of the uth user, the
threshold for the different CQImust be obtained . Therefore , BLER performance of the LTE downlink (DL ) transceiver is
US 10 ,080 ,200 B2
10
simulated to generate the training data for the ITPSC . This the ITU - R indoor path loss model [ 12 ], the path loss
invention considers multi - user system reliability (SR ) between a femtocell eNB and an UE separated by a distance
requirements of indoor small cell in fading environments . d (m ) in a given cell is
The received signal strength P , at the UE is log-normally
distributed . The coverage probability of P , greater than the 5 PL (d)= 20 log100 )+ 10N log10(d )+ LAn) -28 (dB ) ( 15 )
receiver sensitivity Pr.min from the femtocell to a UE at the where the carrier frequency f (MHz) is set as 2350 MHz
distance d is:
with 20 MHzbandwidth in the experiment. N is the path loss
exponent, where the nominal value in the indoor office is set
roo
Pw , (d ) = p [ P , > Pr,min ] = I1WO p (Pr)dP , (8 ) 10 asasfloors
33 [[ 12
12 ].: Ln
L & n )) ((dB
AB )) isis the penetration loss between the
the pen
, where n is the number of floors . The penetration loss
is not considered in the simulations.
1 Pr.min – K + 10Nlog - In addition , the standard deviation Ow of the received
= 3 - NIzerf| shadow fading signal power in the indoor office environment
Vzow 15 is set as 10 dB .
Experiment Measurements in the Laboratory :
where R is the coverage radius, K is the average signal For the purpose of determining the path loss exponent N
strength (dBm ) at the cell edge, K -Pr.min (dB ) is the fade and the standard deviation Ow of the received shadow fading
margin (FM ) at the cell edge ( d = R ) which is used to signal in the indoor office environments , the power mea
guarantee the reliability at the cell edge . O is the standard 20 surement of small cell eNB (ITRI- SC -CUT3 ) is performed
deviation of received signal strength (dB ) and N is the path inFIG the. 7 ,laboratory
where the
. The scenario of laboratory is shown in
femtocell eNB andmeasurement positions
loss exponent.
The percentage of the UE in a cell of radius R for P , are illustrated . The length and width of the laboratory are
greater than the receiver sensitivity Pr.min is defined as the os 20located
.98 meter and 7 . 30 meter , respectively . The black stars
in the laboratory represent 38 measurement posi
service reliability (SR ), which is given as : tions, which are distributed at distance of 1 - 19 meters from
the eNB . 1000 RSRPs are measured at each measurement
position .
SR= }{1+ erflp)+exgf 200 +1)1-eri 194 )} UE (Samsung Galaxy Note Edge SM -N915G ) used with
drive test tool report RSRPs for different distances between
where transmitter and receiver in the laboratory and calculate their
FM 10Nlogio e standard deviation . Then the path loss model of the labora
p = 50, 9 = 20w tory can be obtained by modifying the ITU -R indoor office
path loss model. FIG . 8 shows the modified path loss model
obtained from the field trial in the laboratory, where circle
The minimum transmit power of the ITPSC is evaluated solid curve is the ITU -R indoor office path loss model ; the
by link budget formula for the different SR , coverage radius triangular dotted curve is the measured path loss model; the
(R ) of indoor office, and the cell edge CQI (CQIminu ) square solid curve is the modified path loss model of the
requested by the u , user. The minimum transmit power in laboratory . Therefore , the formula of the ITU path loss
dBm of the ITPSC is given by : w40 model
m is modified as
Pini uG= Prmin
,+L ,
,u(CQImin,u)+L;-G ,+PL (R .)+FM (SR ) (11) PL (d = 20 log100 + 28 log10 (d )- 36 (dB ) ( 16 )
where Prmin, (CQIminu) is the receiver sensitivity of the where the standard deviation Ow of the received shadow
cell edge CQI (CQIming) for the uth user. Lydenotes the cable 45 fading
2 . 8 ,
signal and the path loss exponent N are 4 .27 dB and
respectively . Finally , by substituting O and N into
loss in dB . G , and G , are the antenna gains in dBi of the ( 9 )( 10 ), the fade margin FM for 90 % service reliability is
femtocell and the UE , respectively . PL (RM) denotes the calculated as 2 . 14 dB .
maximum path loss between a femtocell and the uth user at The training data of the ITPSC is generated from the
the cell edge . L , in dB is the body loss of the UE . FM (SR ) simulation results of the transceiver BLER , as shown in
denotes fade margin in dB corresponding to the SR se set by
by 505 Table
Tah 3 . Integrating Table 3 with equations ( 11), ( 12 ), ( 13 ),
the uth user. The receiver sensitivity of the given cell edge ( 14 ) and ( 16 ), the minimum transmit power is calculated for
CQI (CQIminu ) for the uth user is obtained by: the service reliability of 90 % , different coverage radius ( 2 .5 ,
Prmin,u(CQImin,u)= PN. +SNR (CQ!minu) (12 ) 5 , 7 .5 , 10 , 12. 5 and 15 meters ), different number of resource
block ( 1 – 100 ) and cell edge CQI (1 - 15 ) . FIG . 9 shows a set
where SNR .1 (CQImin
min , „ ) denotes the SNR threshold of the 55 of training data of the ITPSC for coverage radius of 5 meter.
receiver for different CQImin ., , which is generated from the cell edge CQI (3 , 7, 10 ) and different number of resource
performance simulations using the transceiver specification block ( 1 - 100 ). For example , when the number of resource
listed in Table 2 . The receiver noise power Py , in dBm for block is 50 , the minimum transmit power corresponding to
the uth user is given as: different cell edge CQI ( 3 , 7 , 10 ) is ( - 35 . 33 dBm , - 27 . 33
PN, = NF (dB )+ (- 174)+ 10 log10 (BWr. )(dBm ) 60 dBm , - 18 . 33 dBm ).
The function of the ITPSC is to set the initial minimum
where NF is the noise FIG . of the UE receiver and BWr. transmit power of the femtocell eNB , which satisfies the
is the receiver bandwidth . requirements of the different RBs (nRB ), coverage radius
( R ) of indoor office , and the cell edge COI ( COImin
min , u )
BW , , = 15 kHzx12xnRB ,
Y 21
(14 ) 65 requested by the u , user in the interference free environ
where nRB , is the allocated RBs of the uch user. The SNR ments. Each input uses three generalized bell shape mem
thresholds for BLER = 0 . 1 are summarized in Table 3 . Using bership functions (MFs ), which are defined as:
US 10 ,080 , 200 B2
11
(17) CO' best,4 = (21)
ajn umplacing
Ajin (Xj,m ) = 2 xj,m - Cin |2bj,n '
1 +( Pom
for n = 1, 2, 3 and j = 1, 2 , 3 5
( CQIi, if SNR:n ( CQL_ i) < SNR , < SNR:n (CQI i + 1), i = 1 ~ 14
{ CQL_ 15 , if SNR« > SNR:h (CQI_ 15 )
(CQI_ 1, if SNR, < SNR:h(CQI_ 1)
where xj,m is the min input and the premise parameters aj,ne The training data of the CQIDC is generated from the
bin , Cin pertaining to the node outputs are updated according 10 simulation results of the transceiver BLER . as shown in
to given training data and the steepest descent approach . The Table 3 . Integrating Table 3 with equations ( 19 ), (20 ) and
27 fuzzy inference rules are constructed in Table 4 . A (21), the best CQI is calculated for different measured
minimum transmit power optimization problem of the average path loss (30 dB ~ 70 dB ) , resource block ( 1 - 100 )
ANFIS -ITPSC for the uth user is formally formulated as and initial minimum transmit power (- 75 dBm ~ 20 dBm ).
follows: 15 FIG . 13 shows a set of CQIDC training data of the best CQI
corresponding to different average path loss (30 dB ~ 70 dB ),
when the resource block is 50 , the minimum transmit power
Optimize Pini = f (x ), f « ) is the objective function ; (18 ) is ( - 40 dBm , - 20 dBm , 0 dBm ). It is observed that the best
subject to : CQI will decrease as the path loss becomes larger, and at the
20 same average path loss, the best CQI increases as the initial
TE {Ry ,nRBy, CQImin,u) transmit power becomes larger.
Om < Ry < 15 m The function of the CQIDC is to determine the best COI
of the femtocell at the ucn user 's location of indoor office ,
1 SnRB, 5 100 which satisfies the receiver performance of BLERs0. 1 in the
15 CQlmin,u = 15 25 interference free environments . Each input uses three Gauss
ian MFs, which are defined as
nUE
Pinige €{ s 20 dBm }
30
30 Ain (Xim ) = e
tiim203,bojama
n , for n = 1, 2 , 3 and j = 1, 2, 3
( 22 )