fff reflex v1

Transcription

fff reflex v1
ใบรับรองวิทยานิพนธ
บัณฑิตวิทยาลัย มหาวิทยาลัยเกษตรศาสตร
วิศวกรรมศาสตรมหาบัณฑิต (วิศวกรรมคอมพิวเตอร)
ปริญญา
วิศวกรรมคอมพิวเตอร
ภาควิชา
วิศวกรรมคอมพิวเตอร
สาขา
เรื่อง
การเขารหัสเครือขายเชิงเสนสําหรับปญหาการสื่อสารระหวางตนทางปลายทางแบบหลายคู
Linear Network Coding for the Multiple Source-Sink Pair Communication Problem
นามผูวิจัย นายมุนนิ ทร เอี่ยมโอภาส
ไดพิจารณาเห็นชอบโดย
อาจารยที่ปรึกษาวิทยานิพนธหลัก
(
ผูชวยศาสตราจารยจิตรทัศน ฝกเจริญผล, Ph.D.
)
(
ผูชวยศาสตราจารยชัยพร ใจแกว, Ph.D.
)
(
ผูชวยศาสตราจารยภุชงค อุทโยภาศ, Ph.D.
)
อาจารยที่ปรึกษาวิทยานิพนธรวม
หัวหนาภาควิชา
บัณฑิตวิทยาลัย มหาวิทยาลัยเกษตรศาสตรรับรองแลว
(
รองศาสตราจารยกัญจนา ธีระกุล, D.Agr.
คณบดีบัณฑิตวิทยาลัย
วันที่
เดือน
พ.ศ.
)
!"!#
Linear Network Coding for the Multiple Source-Sink Pair Communication Problem
>?
@A B@>C
DE ! @!F G @@#D"G@G @DE (G@@ )
.G. 2554
C% D#EF 2554: !"!# G@G @DE (G@
@ ) G@@ CG@@ NB
OF!: Q#G NN G RNQ!, Ph.D. 28 TDKD%D#UO# $%KFO(&FO
VW HDXDHCW $#YZ'&$[#O%VWWD
#YX EWOHO !"!#
VDX Iwama et al. VW&%YO
%D#D( k ( %D#%O& TCOW OD%D# k T
%D# &%&%YDXYX(B?N?N>?B
V Iwama et al. VW(HDWHWHEWH T
O
V | ℑ |3k EW U$%]^&W ℑ HDXEWH
V k 2|ℑ|2k EWXD
OD%D#D( k (
DWHW%D#'YXW& %O&O%&%Y%D#EW Iwama et al.
&W& (HDWHWODXO&%D
'&$%_Z`D%
Dilworth
/
&#
&#H%D# YZ%K&
/
Munin Eamopas 2011: Linear Network Coding for the Multiple Source-Sink Pair
Communication Problem. Master of Engineering (Computer Engineering), Major
Field: Computer Engineering, Department of Computer Engineering. Thesis Advisor:
Assistant Professor Jittat Fakcharoenphol, Ph.D. 28 pages.
Network coding is an another technique to improve efficiency of the network. This
thesis has the objectives to study and to improve the network coding, and focus on the multiple
source-sink pair communication problem
Recently, Iwama et al. present an algorithm for k source-sink pair network coding,
when k is some fixed constant. The running time of the algorithm depends on the upper bound
on number of vertices performing encoding operations. Iwama et al prove that ℑ 3k encoding
vertices are sufficient to obtain the network coding whose operated on field ℑ . In this work,
we improve the bound to k 2|ℑ|2k and this tighter bound, in turns, improves the running time of
the algorithm proposed by Iwama et al. The techniques used are elementary linear algebra and
Dilworthps theorem.
/
Studentps signature
Thesis Advisorps signature
/
Q#NADQ#G NN G RNQ! @BOF
"!Q#G N VN" @BOF@ BVOF"!"V
N @WO" X NVNBYZN[Q!@\???B
ADBXX >"!@!A@N]F^BAB"!!B
@#"!NA?@? X ADNBCG@@ B
@?"! XV A?BYAD?@?B?#"!VNV"!V
!VN#@@
@A B@>C
]FC@ 2554
(1)
"
"F
UC $
H
C '&$KD
C KD
G&'&$H
G&
H
C '&$'$
C
'$
'&$#
$
YZ'&$%
(1)
(2)
1
3
4
16
16
16
17
17
25
26
26
26
27
28
(2)
1
2
3
4
5
B@B"@! B@B
"@>?@
a>b? k+1 B@B Q!@
>b? p+1 %D#EWD(&%v
8
9
15
19
22
1
!"#$
%""&"'&(
Linear Network Coding for
the Multiple Source-Sink Pair Communication Problem
##
VNNA@@WV@#!V[B@B@ (Network Coding) [BBOBVN"!@@@WV
@#!B@Be@V!e@V\? >?\"!@#!B@B!
e@ @@B@ B V@ N@#! @ Y\@@W
@#!N!e@@@!Q!@\? NOV @B"[! (?N
"?V \) bOVC!?!
VBV@#!" AD"! !@#!
@W@#!B N! \!!@@!Q!@\? bO
V! X DB V@W@#!\?>?\@ @#!B!Y "!@C
@#!
#"B[NA?@ GOF
"@! (multicast) B[B>? BO>? @#!\
>?!! X >?
@WBY\?[?B
NBY@BNBGOFBY[N@"!!A@V"NA?
VNBYVNVBOY !VNV
!!# V?! @B#>? k # B @#!WO @#!QNZNW@#!N" ! @W
\WO!B# \?W# 2
!"DXO w 2003 Lehman and Lehman VW(HOD%D#VHWH(
ODX T !" NP-Hard O w 2007 Wang
and Shroff \?#N@WDVB@B# 2 #\? " B?!\@@WV\?B@B#@ 2 #
NBYg N Iwama et al. (2008) ?!&%Y
O%D#D( k ( %D#%O& TCO
W OD%D# k T%D# &%&%YDXYX(B?N?
N>?BV Iwama et al. VW(HDWHWHEWH T
OV
| ℑ | 3k EW U$%]^&W ℑ HDXEWHV
k 2|ℑ|2 k EWXD
OD%D#D( k (
DWHW%D#'YXW& %O&O%&%Y%D#EW Iwama
et al. &W& (HDWHWODXO&%D
'&$%_Z`D%
Dilworth
3
%C"
1. g"!A!O@
"!!!#
4
%D
[OV@#!B@@@WV
@#!VVDBB@B\e@N!e@ >?BN!WO
YN!WO@"!YjBN[ V
EFG!H!"
ℑ b B@B@ k+k "! #D l · k >?@YBY@
no!? "!V?!BY[N
1. @B@ p? :
!Q!"! Q!#D @Vno!?N #CVno!?
2. @B@ !B :
@ a "! b V?XB#Vno!? a + b = b + a "! a · b = b · a
3. @B@ N?@# :
@ a, b "! c V?XB#Vno!? a + (b+c) = (a+b) + c "! a·(b·c) = (a·b)·c
4. #D@B@ N :
@ a, b "! c V?XB#Vno!? a·(b+c) = (a·b) + (a·c)
5. @B!FD 0 "! !FD#D 1 :
@ a V?XVno!? a + 0 = a "! a·1 = a
6. @B QQ :
@ a Vno!? @W QQ b Vno!?BV a + b = 0
7. @B QQ#D :
@ a Vno!? B\@ 0 @W QQ#D b Vno!?BV a·b = 1
no!?BV\?" no!?NN, no!?Nb
(Finite field) @WOno!?B@B@Vno!?@BNN? ?no!?
N?@B?!ZBA?B[\? 2 @B@ (0,1) B Binary field
5
!NBYN@Vno!?@VVA?N>? (Scalar)
H%D#D'
WDWD '&$ (Vector) H%D#D%XW'&$%%
V ]w&W $W $
'&$]w&W '&$D
W (Vector addition) '&$ (W& (Scalar
multiplication) $%X 2 H$
TV WWDX
1. D
^W :
%C|
x '&$ y O V x + y X(O V
2. D
&%D# :
x '&$ y OWO V , x + y = y + x
3. D
HW&W :
VW x y '&$ z OWO V , (x+y)+z = x+(y+z)
4. D&Z :
D 0 Ov
V %D#%O 0 + x = x x OW|O V
5. DGG :
x %D#(O V H$D
y %D#%O x + y = 0
6. (&D
^W :
%C a O]w&W '&$%C| x O V ax H$(O V EW ax DXD TG&
(& (scalar product)
7. (W&D
HW&W :
a '&$ b OW|O]w&W '&$ x O V VW a(bx) = (ab) x
8. (W&U$HVW#%D
:
a O]w&W '&$ x, y O V, a(x + y) = ax + ay
9. (W&U$HVW#%D& :
a,b O]w&W '&$ x O V, (a+b)x = ax + bx
10.D&Z( 1 (O]w&W :
%C x O V , 1x = x
6
O v1, v2, }, vn T
OW| O F(
G&O( a1v1+a2v2+} + aivi
# a1,a2,}, ai T& DG& (linear combination) v
G&
%C| TG& ! v1, v2, } , vn EWH$W Span(v1, v2, } , vn)
"#$% (Ordered Set) v
%D#DWKW$O
v
EWDK ≤ Wv
P # ≤ D
3 WDX
1. $% (reflexive) :
a ≤ a @ a [@V?XV P
2. @ a@@ (antisymmetric) :
W a ≤ b "! b ≤ a "! a = b @ a,b [@V?XV P
3. @ W? (transitive) :
W a ≤ b "! b ≤ c "! a ≤ c @ a,b,c [@V?XV P
#@V?X a "! b V P NBBB\? (comparable) W a ≤ b
b ≤ a \Y\@[NNBBB\@\? (incomparable) "!
NBb ? (chain) W@A#Vb @WBB\? "!B
(antichain) @AX#Vb BB\@\? "!Bb ?BBB
\?B#Vb (Partially Ordered Set)
('"(&
#"@#!V"\?[ 3 CV @e@B 1. @#!"# (Unicast) # [@#!B@B!FD"
O O @B O \!O # [#"
@#!B\?\"!@BV
7
2. @#!"@! (Multicast) [@#!B@B!FD"O
! @B O \!! >?![@BNN
!A@ObO#V " \@V@Y@?V
3. @#!"? (Broadcast) [@#!B@B!FD"O
!?B@! " !"? Y@Y@?B#
V N@! B\!A@@BV
"(&%""&"&(
@#! !!#[@#!B@BN N! >? " ! N@B!B[## YX V@#!"
BY N @#!\!B ?BB[# @ N@\?[
@#!B@"O ON!\?
(Network Coding)
V@#! @ @#!NW##CV" (Packet) >?ADN#"
B" Y@Be@\BV?" \@@W#WO@#!B#CV" \? ADN@W?#@#!V" N!X" "!@#!!Y@@
?B\? bOVBBYNB[ NYADN@#!B\?N
\>?V" V@ "!!N@WW?\?W\?@#!B@B
[N@B
Ahlswede et al. (2000) \?"?V"@! YN@BV@#!
N@BV@ @V\@@W@#!\\? Z@C "!
@WV@#!V@BC@OY\?
8
N"? @#!B@V@B@!Q!@#!B>?!
N@W@#!\?@BC@OY
>?NVn"@B
GV"?!FD
s
a
a
b
a
b
v
t1
b
t2
1 B@B"@! OD
% s '&$
(&
a '&$ b V &%
t1 '&$ t2 WF%D# 1 EW'
&$#'%#
%D#DUO% 1
(& H$EW v (&UO %
(&
EW v H T
&(& a b V %D&$X #HU(&
VWDY#(& (&V &%%XEW v HY
(&X
DEW v U(& a '&$ b T(&OVW
EWO(&O a+b (&DX#V &%'& &%U' &(&&
T(&%D#
VW & t1 U(&%D#VW&#(& b VW O%
WD t2 U(&%D#VW(& a VW
9
s
a
b
a
b
a
v
a+b
a+b
b
.
a+b
t1
t2
2 B@B
!" (Linear Network Coding)
NBV "Y@B@A"!bb#
Li et al. (2003) NO\?"?@#!"@! Y@W
\?>?V@#!B[Q!@@#!B\? "!"?@B!O@
!>@BV\?" \@VBB?BA?V BVQ!@@#!VBYNB!"#!$%!&"'(#
!N@B"?YBBN\V\? (Li et al., 2003) " \@@BY BB?BVB Jaggi et al. (2005) NO\?B
@#!"@! CV!B[nA
@ >?B@#!V"!@ @!?
@BaVV@N \! V" !@
NA@@yBVVQ!@ bOQ!@NBOY@
BYN [ Q!@VAXB\e@?B V
A?@WOe@"!N@WW?\? N"!\@[ NV@ >?!V@y
10
@yBVBYNA@Nno!?N? ℑ >?@N[BQ!@N
N\@[ N"QQ?no!?BY @y@B
B[\\?@Q!@N@B>\@[ ! bO@BQ! N
YB "!!VY@?N#CV O(E ⋅ T ⋅ H 2 ) @
E
T
H
N@Y@?
Ne@
@@WV@#!"@! #%""&"&(
!!#Y[B
BVN >?@B!FD!"@! B@B!! " !!#Y" !!N B@#!N B[#
!YY
Vo 2003 Lehman and Lehman VW"?VZV@#!B\@V@! Y
"\@B @#!V#" "!VDBB
@B@#!V!FDB[@#! !!#Y \@
N?N# !N[ NP-Hard
Wang and Shroff (2007) \?NDVB@BB 2 #
!bO[N#A?B[\\? "!"?VDBBY@W
\?>?V""!\?CV!B[nA@ " B?!\@@WV\?B@BN# !@ 2 \?
@Vo 2008 Iwama et al. \?!O@V
B@BN# !@ 2 #CV!B[nA@ @?V
N# !Y[B k >?VBY Iwama et al. \?"?N[ V
11
>?B\@NO NBYOY#N# !"!?
np!?N?BVV >?N!B?VW?\
&K" Iwama et al.
VBYN @"!! WO ! O@ !N k # Iwama et al. "!N
>?B\?"?\
@@ V>? v \?@#!@[ x "! y @!? !"#!$"$
v
B"? σ (v ) [ "?WO>? v >??V
σ (v ) = (a, b ) @@#!BN \Y ax + by @ a "! b #Vno!?N? ℑ BV
V >?Y a "! b \@ 0 NB>? v BY[$
"#!$
(coding vertex)
σ @WOb >?Y@?B
@B @>? v V?X σ (v )
; &'()*(
+,)-./
*#
0(%
,!
1(/*#/-/(-+ !"%D#VWnaNAG"!# !N k # '&$D ‚
O>?%XW%D#DO] EW &%'
&$
VWƒ$(&%D#H
%%D# T(
%X '&$& σ
2.*/#3$* σ (O
%D# TV VW !"DX
Iwama et al. @@ nB\?@B!FD[ 2/1 Restricted Graph bO[nB@B
N?@OY@>?V?XB\@V "!!N@B# (indegree,outdegree)
[ (2,1) (1,2) \?Y "!@@ "! Y@?@B# (indegree,outdegree) [
(0,1) "!!@B# (indegree, outdegree) [ (1,0) Y@?
12
naNAGV?X@W"![ 2/1 Restricted Graph \?>?@N
>?"!@\V ">?B\@ \ YNND
>?N>?%D#D( (indegree, outdegree) T (2,1) %X WO#
EW v Y# T *#.,!"*- '&$DY# T *#.,!) (&%D#
V H$H
G&(&%D#VWH#%X '&$H$DEW%D#D( (indegree,
outdegree) T (1,2) 4(#$/- (Fork vertex)
&%YN V>?Y@? ?Y@BN>?
[N@!BVVZN@ @\? Iwama et al. NO\?"?
N >? B V Y \@ N [ @ N O >?@B BV >? @ N ? ! "! N@W"! V
N>?!?!>?Q!?@#\?
' &DX%VWEW &D#O#D
( & B [ 0 " NY N D@ \@ @B @Y # "!@ >?@B #
(indegree,outdegree) [ (1,1) bO"?@@?\? >?N@B!B
?BY
@@ V>? v @B σ (v ) = (a, b) !67'8#'9 α @WO
!B>?V[ (a + α , b) "!B !
8#' W α = −a V
?B !67'8" 9 β !BV>?[ (a, b + β )
"! !
8" W β = −b
NBYN"?@WV!BBY"!V@BN>?
!?!>?VQ!?@\? >? ?bB>?
O"!!BBB>??Q!B!B\N ?
>?N[ V@@?BY
!"#:;#(&7: L B"?@#!BV@ L @BV?
@B@#! >?@@ V si @#![ xi @#!BV@ L
13
k
N [Q!@@#! Y@? ∑ (ai xi ) @#!@ L
i =1
(a1 , a 2 , a3 ,…, a k )
=!>8$
v 9''8 867 i @WOB\! B i @
Y@?@#![ 0 "!V@#!BN>? v [ 1 B"? ei (v)
!=!>8$
v "? E (v) Q!N>? v
\!" ! E (v) = [e1 (v), e2 (v),…, ek (v)]
@@ >? v @B@#!@[ x "! y @b@ @!?
"!@BQ!N>? v \! B i [ ei !Bb\ a NV
@#!B\! B i @B!B\ a ⋅ x ⋅ ei V?B!B\b
NV@#!B\! B i @B!B\ b ⋅ y ⋅ ei
NZW@B>?B@BQ!\!@ 2 >? !B
>?Y >?V>?"@#!B@B!B\ a "!N>?
B@B!B\ − a N!B\ >?B@#!\!
?@#
MNO 1 $:6!"#!$%!&"'867?(#9
#
'?(#
':6$
"#!$8@$:
: 3k
ℑ $
>:6!"#!$%!&"'6A!;$B7867?(#9
#
':6$
"#!$8@$:
9:
3k
ℑ $
(D @@ V@B>? 2 >?B@B@#!Y[ u "! v "!@BQ!
\!@"![ (a1,b1) (a2,b2) >?>?"\@[Q#
Y (descendant) @N>?B Q!N>?"\NWO@b
>?B[ eL "!Q!N>?"\NWO@>?B[ eR N"[ 3 DB?BY
14
DB" a1eL ≠ 1 N ?bB>?" ?BYV@#!B
>?"!B\ -a1u "!@#!b>?B!B[ (1-a1eL)u @W
!BbB>?B\ a1/(1-a1eL) V@#!>?B!B\ €a1u \?
DBB b1eR ≠ 1 V?BDB" N ?B>?""!
!B\ b1/(1-b1eR) B>?B
DBA? a1eL = b1eR = 1 NY ?b"! ?B>?"
?BYNV>?"@#!!B\ -(a1u+b1v) "!@#!b"!
>?B[ €(b1/a1)v €(a1/b1)u " >?!Bb\ €a1 "!!B\
€b1 >?BN@B@#!B!B\ +(a1u+b1v) bON\!!BB>?"
N @#!"! Q![ k @ no!? ℑ ?Y@B
>?@ ℑ 3k N @B>?#B @\ "!@W ?
@B"?\ B"B@B>?\@ ℑ 3k \?
!O@ Iwama et al. YN@N!>?"!>?"
@N>?Y@?B\@V "!! !BY[\\?@BA?
  N 2 
O    = O( N 2C )
 C  


(1)
@ N N>? "! C N>? !N!>?"!
>?""! N@>? ui "! vi @b"!@>?
pi "!@>? wi B@N>? pi @>? xi @>?
" qi "!@>? yi "! zi B@b"!N>? qi @!??
CB 3 "!b >? 2 b
OUT '&$ IN
15
vi
ui
xi
qi
pi
wi
yi
zi
3 "@>?@
OUT = {s1 , … , s k , w1 , … , wC , y1 , … , y C , z1 , … , z C }
(2)
IN = {t1 , … , t k , u1 , … , u C , v1 , … , vC , x1 , … , xC }
(3)
NY perfect matching b OUT "! IN "! vertex disjoint path
BVnB@N OUT \ IN >?\@Q>?"!>?"B!\ @B#"@ B[\\?\@ >? vertex disjoint path DXO&%Y
Fortune et al. (1980) bOV![A@ N " N[b>B! C Z\?
W@B vertec disjoint path BV\? !O@N!"!
NYV\?\@ bOBY[\?@BA?
2C
2
ℑ N" !>? @B\? ℑ #""!>?@B
Y@? C >?
!BV!O@@@AOY "! O(N ⋅ (3C + k )!⋅ ℑ ⋅ f ( N , C))
@ f ( N , C ) [!V vertex disjoint path @!O@ Fortune et al. (1980)
2C
2C
16
H'&%Q
H
1. AD"!#N]F^B
1.1 @A?O
1.2 ?"!!
2. ADV@#!"!OQ!
2.1 @ 2.2 a >?
2.3 A?>"@\@>bnnnp
%Q
1. S'&"&K##$%"
"'&&"&(
V"? Iwama et al. BNDB?N?N>?@g
B?N?B"OY "!#NB?N?B"OYBY>?V!BD "!]F^B Dilworth
17
T&'&%DH
T&
1. S'&"&K##$%"
"'&&"&(
NGOF!O@ Iwama et al. @W!?B?N?N
>?!\?>? ?"!!BB>? X >?BB
Iwama et al. "?\YND†VDBB>?Y@B@#!@
Y NO N DVDB B OY ! @# ! >? \@ N [ @Y@?
@W!?B?N?N>?N ℑ 3k [ k 2 ℑ 2k \? B?N?
B"OYBYV!V!O@ Iwama et al. !?! V#N
@B@@ @?BY
=!>8$
v 9'#(&7: L @WOB @ L @ Y@?@#![ 0 "!V>? v @B@#![ 1 B"? eL (v) @ L \@@W\WO\?N v NVQ!BY[ 0 "!V ei (v) "Q!N
>? v \@! t i @ 1 ≤ i ≤ k
!=!>8$
v "? E (v) Q!N>? v
\!" ! E (v) = [e1 (v), e2 (v),…, ek (v)]
!=!>8$
v 9:=#(&7: L Q!N>? v
!N ?@ L Nn B"? X L (v) "!V !=!>8$
v 867=#(&7: L Q! I L (v) = E (v) − X L (v)
18
ND@W@[b !?B@@@ ≺ ?BY >? u "! v V?XVb N! u ≺ v Z @@B@ v \WO\?N u
N!\? u [AAF v
MNO 1 D#$:6$
"#!$: k 2 ℑ k ?
C A# >#:6 k
C ′ ⊆ C 867:6"
k + 1 $!&:6 C ′ ⊆ C 867:6"
k ℑ + 1
]F^BBY[BZ\??N >?N]F^B Dilworth (1950)
#?BVBA?a>bNN>bBA?B@W"\?Nb !?
?Y\@@Ba>bB@B?WO k + 1 N @B>b? k ℑ k + 1 V?B
\@@B>bB?WO>bB@B?WO k ℑ k + 1 N@Ba>b? k + 1
MNO 2 D#$:6$
"#!$ : k 2 ℑ 2k A# >#:6"$
"#!$ C ′ 867 !=!>8$:& A> C ′ F867"
k + 1 $!&F867"
k
k ℑ +1
(D NB Q![ VC#@ k @ no!? ℑ ?Y Q!N" \?@A? ℑ k " @B>?@ k 2 ℑ 2k "!N @Bb >?B Q!@bO@B?@ k 2 ℑ k "!N
]FaBB 1 "? @Bb C ′ #VBY
!NBYN"?VDBB@Ba>b?@ k >b?@ k ℑ k bO
@B Q!@"! @W!BV>?!?!
\?>?@#!!\@!B"!
19
4 a>b? k+1 B@B Q!@
MNO 3 ?$# C F"$
"#!$867:6 !=!>8$:& A>:6"
: k + 1 $!!"#!$%!&"' σ 867?(#$
"#!$ Q $
>:6!"#!$
%!&"' σ ′ 867?(#$
"#!$ Q-1 $
A>"#:;9''89
#$:& σ
(D V>?Va>b v0, v1, J, vk @B@#!b[ u0, u1, J, uk @!? ?CB
4 "!@B σ (vi ) = (ai , bi ) N@#!" !>?@NQ!@@#! k
NO@W ri BV
k
u 0 = ∑ ri u i
\? ?bB>? v0 NV
i =1
k
@#!B v0 !B\ Ku0a0 − ∑ ri u i a0 @!BbB>? vi
i =1
k
\ a0ri 1 ≤ i ≤ k "!Q!@@#!B!B\N>? v1 WO vk + ∑ riuia0
i=1
?Y@WV@B!BV σ′(v0) = (0,b0) "! σ′(v0 ) = (ai + a0ri ,b0 )
@ 1 ≤ i ≤ k >?BQ!@#!\!@ σ \?
DBB[>bYNOY"!?"a>b\@\?N!B
>?V>bN@BQ!@#!"! Q!>?XV>b?
>?VBBYN@NDNDBB>b@BB 2 >?
MNO 4 D#:6%;$
v1,v2 867
- v1 ≺ v2 A>
- :6"#:;"#8#'$:&%& u A>
- :6 !=!>8$:& 7%& E(v1 ) = E(v2 ) A>
20
- XL(v1 ) ≠ 0 :&7 L %&#(&7:"#8#'" v2
>:6!"#!$%!&"'
σ′(v2 ) = (0, B) :&7 σ (v2 ) = ( A, B) A> σ ′(w) = σ (w)
8QR w∈V −{v1 , v2 } 867"#:;9''89
#$:& σ
(D !Bb\ α B v1 N\?@#!BN!B\ α ⋅ u "!@#!
b>? v2 N@B[ u1 (1 + α ⋅ e L (v1 )) "!@ ?bB>? v2
@#!BN!B\ − A ⋅ u1 (1 + α ⋅ eL (v1 ) ) !V α = A (1 − A ⋅ eL (v1 ) ) N\?
α ⋅ u = A ⋅ u1 (1 + α ⋅ e L (v1 ) ) bO!Q!!B"!\\? >?@W α
\?@ A ⋅ e L (v1 ) ≠ 1 B[BYW A ⋅ e L (v1 ) = 1 N@N\?
X L (v1 ) = E (v1 ) − I L (v1 )
= E (v1 ) − E (v 2 ) ⋅ A ⋅ e L (v1 )
= 0
(4)
bO?"\B XL(v1 ) ≠ 0
NBYN"?V!B>?V>bY W Q!>?@BG?B !!B"! YN@B
G?B#
MNO 5 D#:6%;$
v1,v2 867
- v1 ≺ v2 A>
- !=!>8"8@%;:68S8
6' $!&"6'9
# E (v1 ) = δ ⋅ E (v2 )
$!67'8#'($!&8" )"$
v2 A> !=!>8?$:"$
v1 ,
E ′(v1 ) >':68S8
6' E (v 2 ) ';
(D @V L @b v2 I L (v1 ) = 0 !BbB v2 \@@BQ!
V?X E (v1 ) #"! ?YN"?#NVDBB I L (v1 ) ≠ 0 @@ V
σ (v 2 ) = (a 2 , b2 ) @ND Q!#
21
E (v1 ) = X L (v1 ) + I L (v1 ) = δ ⋅ E (v 2 )
(5)
I L (v1 ) = e L (v1 ) ⋅ a 2 ⋅ E (v 2 )
(6)
?Y XL(v1) N @BG?B E(v2) @!BbB>? v2 I L′ (v1 ) N@B?!B\" G?@# XL(v1) Y\@!B"!" V? NO
@WA\? E ′(v1 ) @B?!B\" @BG?B E (v2 ) # !BY@W#N\?V?B
N]F^BB 5 Y V@W!B"!>?V>b\?
>?B Q!@BG?@ W !Y@BG?B#N@B
!B V!? \N"?@W!B>bV@#!
!@B!B\ @B \?
MNO 6 D#:6"$
"#!$ C = {v1,J, vh+1 } 867
- v1 ≺ v2 ≺ J ≺ vh+1
- !=!>8" vh+1 ≠ 0 A>
- :6"#:;"#8#'$:&%& u A>
- vi ∈ C D#$ !=!>8 E (vi ) ≠ 0 !=!>8" vl+1 δ i δ i ⋅ E (vi ) = E (vh+1 )
- eL(i+1)(vi )ai+1 ≠ 0 $! 1 ≤ i ≤ k :&7?$# L(i) F#(&7:"#8#'"$
vi
A> σ (vi ) = (ai , bi ) $!8QR vi ∈ C A>
- !"#:;8" "$
vi %& γ (i ,1) y1 + γ (i , 2) y 2 + + γ (i ,i ) yi :&7
γ (i ,i ) ≠ 0 A> γ (i ,1) , , γ ( i ,i ) ∈ ℑ A> ! y1 , y 2 , , y i 8@$:
@W
$!!"#!$%!&"' σ A>%?
R γ 1′ , γ 2′ , γ l′ ' ∈ ℑ >:6!"#!$%!&"' σ ′ 867
σ ′(vi ) = (ai′ , bi′ ) $!8QR vi ∈ C − {v h +1 } A> σ ′(v h +1 ) = (a h′ +1 , bh +1 ) A> σ ′( w) = σ ( w)
$! w∈V / C 8678?$#"#:;"#"'8 ti 67'9 (γ 1′ y1 + γ 2′ y2 + + γ l′' yl ) ⋅ ei (vh+1 )
22
(D N#N]F^BBY?#N>?A (induction) h
B@#!j h = 1 ND#>? v1 "! v2 W E (v1 ) ≠ 0 N!B
\B B v1 V@#!B! i !B\ γ 1′ y1 ⋅ ei (v2 ) # δ 1 ⋅ E (v1 ) = E (v2 ) ?Y
!B>?V B =
γ 1′
δ 1 y1
N\?!B@#! @B VDBB E (v1 ) = 0 N!BV E (v1 ) ≠ 0 >?@BB
!Bb\ + eL (v1 ) B>? v1 !BBY\@@BQ!V?X@#!B!N
Q![ 0 "!\?
σ ′(v1 ) = (a1 + e L 2 (v1 ) , b1 )
(7)
HY@ ?bB v2 NV Q! v1 !B[
E ′(v1 ) = X L ( 2 ) (v1 ) = −e L ( 2 ) (v1 ) ⋅ a 2 ⋅ E (v 2 )
(8)
"!@#!! B i !B\ − a 2 ⋅ u ⋅ ei (v2 ) NY@!BbB v1 \
− e L ( 2) (v1 ) @#!! B i N!B\ + a 2 ⋅ u ⋅ ei (v 2 ) bON! ?b
B v2 "!D σ ′′(v1 ) = σ (v1 ) "!\? E (v1 ) ≠ 0 bO@W!B
\??B"?\ "!
5 >b? p+1 %D#EWD(&%v
23
YA Y@@ j]F^B[N@ h = p N"?[NDB h = p+1
ND>? vp "! vp+1 V>b? p+1 ?CB 5 W E (v p ) = 0 N!B >?B?BY@#!jV
E (v p ) = −e L ( p +1) (v p ) ⋅ a p +1 ⋅ E (v p +1 ) ≠ 0
?V (γ 1′
y1 + γ 2′ y 2 + + γ ′p' y p ) ⋅ ei (v p +1 )
γ ′p
B i B >?!B\
δ p yp
[!B"!@#!B! N\?@#!B! B i !B\
@ φ [ BOY
!B@#!B B
(φ + γ ′p' y p ) ⋅ ei (v p +1 )
(9)
y1 , y 2 ,..., y p −1
Y?YN!
(γ 1′ y1 + γ 2′ y 2 + + γ ′p'−1 y p −1 − φ ) ⋅ ei (v p +1 ) = (γ 1′ y1 + γ 2′ y 2 + + γ ′p'−1 y p −1 − φ ) ⋅ δ p ⋅ ei (v p )
= (γ 1′′y1 + γ 2′′ y2 + + γ ′p′'−1 y p −1 ) ⋅ ei (v p )
(10)
bON@@ jB]F^B[NB h = p "?@W!BBYN>? v1, J,
vp-1 \? "!#N]F^BBY[N
N]F^BBYV@W ?B>?V>b"!!B
@?!B"!@#!B?N ?BY\?
MNO 7 D#$:6!"#!$%!&"' σ 867?(#$
"#!$ Q $
A>:6 C "
$
"#!$867:6 !=!>8$:& B7:6"
'#' k ℑ k + 1 A# >:6!
"#!$%!&"' σ ′ 867:6!"#:;9''8$:& σ '?(#$
"#!$#' Q $
(D N @#!@@B\?@BA? ℑ k #" ?YN @Bb
C ′ ∈ C k + 1 B"? v1 , v 2 , , v k +1 @ v1 ≺ v 2 ≺ ≺ v k +1 "!
V Li [@b>? vi @B#>? vi,vj B@B X L (vi ) ≠ 0 @W!?
j
>?\?>?]F^BB 4 NDDBB XLj(vi) = 0 V yi [ @#!
24
@>? vi Nb C ′′ ∈ C ′ B @\V]FaBB 6 >?
@N C ′′ = {y1 } "!V y1′ = y1 NYbY?BY
1. V j [?BA?B vj \@#V C ′′ W yj TG&
y1′ , y 2′ , , y ′j −1 A?bY
2. "?@B y ′′ B[ y1′ , y 2′ ,, y ′j −1 '&$%O
y j = γ ( j ,1) y1′ + γ ( j , 2) y ′2 + + γ ( j , j −1) y ′j −1 + y ′′
γ ( j ,1) , γ ( j , 2) , , γ ( j , j −1) ∈ F # vj V C ′′ "!V
y ′j = y ′′
! 1
N @#!!BY[ V k @ @WbY\?@BA? k !NYN @#!BOY k B@B#@
NZb D = C ′′ + {v j } @\" !V]F^BB 6 ?N
#"! "!VB eL (i +1) (vi ) ⋅ ai +1 ≠ 0 Y[N X L (i +1) (vi ) = 0 "!NB
E (vi ) = E (vi +1 ) V\? e L ( i +1) (vi ) ⋅ ai +1 = 1
@@Bb B ]F^BB 6 "! @W ?B>? vj V>?
BY\@[>?"!V@#!B!!B\
(γ 1′ y1 + γ 2′ y 2 + + γ ′j '−1 y j −1 ) ⋅ ei (v j )
(11)
bON]F^BB 6 @W!B>? v1, v2 ,J, vj-1 B@B
!B"!B@#!B!! ?BY\?
N]F^BB 3 "!]F^BB 7 @WAN>?\?Y
VDB>b"!a>b"! V]F^BN!WON>?Y@?V
MNO 2 $:6!"#!$%!&"'867?(#9
#
'?(#$
"#!$: k 2 ℑ 2k $
>:6!"#!$%!&"'6A$B7867?(#
':6$
"#!$';9: k 2 ℑ 2k >?
25
(D @B>?@ k 2 ℑ 2k N]F^BB 2 N\?@B>ba>b
bO @\V ]F^BB 3 ]F^BB 7 ?Y@W
V@B@B>?\@ k 2 ℑ 2k \?
%DH
BB?N?N>?VNBY[g N
Iwama et al. >?@W!?>? VDBB@B@#! OY @#! B
@B?BY\? N?@B @B@#!@Y \Z @
BBY>?N[ @B Q!B@# NDVDBB
Q!\@@?N@WB?NN>?B"
BY\?B
26
'&'
NBYB!?B?N?>?N?@B[ ℑ 3k V!
2k
k 2 ℑ >? >?V!BD "!]F^B Dilworth V#N@W# B ? N ? V@ bO B ?N ? B " OY BY V ! O @ V D B@B !!#V!V!?!\?
'
VNBYB"B?N?N>?V@Y \@\?
A!O@ "!O@V@>?OWO!FD? N@W
g!O@V@BC"!V!V!BY\?B
27
'&"""
Ahlswede, R., N. Cai, S. R. Li, and R. W. Yeung. 2000. Network information flow. IEEE
Transactions on Information Theory
Dilworth, R. P. 1950. A Decomposition Theorem for Partially Ordered Sets. Annals of
Mathematics. Vol. 51. No 1
Fortune, S., J. Hopcroft, and J. Wyllie. 1980. The directed subgraph homeo-morphism problem.
Theoret. Comput. Sci.
Iwama, K., H. Nishimura, M. Paterson, R. Raymond, and S. Yamashita. 2008. Polynomial-time
construction of linear network coding. ICALP 2008.
Jaggi, S., P. Sanders, P. A. Chou, M. Effros, S. Egner, and L. Tolhuizen. 2005. Polynomial time
algorithms for multicast network code construction. IEEE Transactions on
Information Theory. Vol. 51. No 6
Lehman, A. R. and E. Lehman. 2003. Complexity classification of network information flow
problems. In Proc. 41st Annual Allerton Conference on Communication, Control,
and Computing.
Li, S-Y. R., R. W. Yeung, and N. Cai. 2003. Linear network coding. IEEE Transactions on
Information Theory
Wang, C.C. and N. B. Sheoff. 2007. Beyond the butterfly a graph-theoretic characterization of the
feasibility of network coding with two simple unicast sessions. IEEE International
Symposium on Information Theory
28
%KN'&#"
?B
WB?
GOF
"NNA
WBNNA
Q!?B?/Q!
AGOFB\?
@A B@>C
16 @F 2528
C\ N?A@
G.. (G@\nne) @!F G \?A>DE GOF CG@@ @!F G (.G. 2551)

Similar documents

4.0.3 spy sweeper

4.0.3 spy sweeper ใด คอมพิวเตอร์ สามารถเก็บได้ เช่น ข้อมูลสํามโนครัวประชากรของประชาชนทัว ประเทศ สามารถเก็บ ข้อมูลรายละเอียดทั"งหมดของประชากรทั"งประเทศหรื อทัว โลกได้ ตั"งแต่ ชื อ-สกุล วันเดือนปี เกิด และ สถานที ...

More information

Online Group Singing System by Background Subtraction

Online Group Singing System by Background Subtraction รู ปที@ 3 แผนภาพแสดงวิธีการทํางานของอัลกอลิธึมทํา Background Subtraction ................................................. 6 รู ปที@ 4 ภาพพื=นหลังที@ไม่มีผใู ้ ช้ .....................................

More information