T
E
L
KO
M
NI
K
A
,
V
ol
.
14,
N
o.
3,
S
ept
em
ber
20
16,
pp.
90
4
~
9
15
I
S
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed
A
b
y
D
IK
T
I,
D
e
c
r
e
e
N
o
:
58/
D
I
K
T
I
/
K
ep/
2013
D
O
I
:
10.
12928/
T
E
LK
O
M
N
I
K
A
.
v
1
4
i
3
.
3517
90
4
R
ec
ei
v
ed
F
ebr
u
ar
y
2
6
,
201
6
;
R
ev
i
s
e
d
J
un
e
8
,
201
6
;
A
c
c
ept
ed
J
un
e
2
9
,
2
01
6
To
w
a
r
ds
S
moot
h a
n
d
Hig
h
-
Q
u
a
lit
y
Bit
r
at
e A
d
ap
t
at
io
n
for
H
TTP
A
d
a
p
ti
v
e
S
tr
e
a
m
i
ng
L
i
h
o
n
g
G
e
n
g
1
,
L
i
a
n
g
P
a
n
2
,
Y
i
q
i
a
n
g
S
h
e
n
g
3
,
Z
h
i
c
h
u
a
n
G
u
o
*
4
1
,2
,3
,4
N
at
i
on
al
N
et
w
or
k
N
ew
M
e
di
a E
n
gi
n
eer
i
n
g R
e
s
ear
c
h C
en
t
er
,
I
n
s
t
i
t
ut
e
of
A
c
ous
t
i
c
s
,
C
hi
ne
s
e A
c
ade
m
y
of
S
c
i
en
c
e
s
,
B
ei
j
i
ng
1001
90,
C
h
i
na
1
U
ni
v
er
s
i
t
y
o
f
C
hi
n
es
e A
c
a
de
m
y
of
S
c
i
en
c
es
,
B
ei
j
i
n
g 10
019
0,
C
hi
n
a
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
m
ai
l
:
guoz
c
@
d
s
p.
a
c
.
c
n
A
b
st
r
act
A
l
t
hou
gh H
T
T
P
ad
apt
i
v
e
s
t
r
e
am
i
ng h
as
bee
n w
el
l
d
oc
um
e
nt
ed f
or
t
he
c
os
t
-
e
f
f
ec
t
iv
e d
e
liv
e
r
y
of
v
i
de
o s
t
r
eam
i
n
g,
i
t
i
s
s
t
i
l
l
a gr
eat
c
h
al
l
e
nge
t
o
pl
a
y
bac
k
v
i
deo s
m
oot
hl
y
w
i
t
h
hi
gh q
ual
i
t
y
under
t
h
e
f
l
uc
t
uat
i
ng ne
t
w
or
k
c
ond
i
t
i
o
ns
.
I
n t
hi
s
pa
per
,
w
e pr
o
pos
ed a
nov
el
b
i
t
r
at
e adap
t
at
i
on al
gor
i
t
hm
f
or
H
T
T
P
adapt
i
v
e
s
t
r
eam
i
n
g.
O
ur
al
g
or
i
t
hm
em
p
l
oy
ed t
w
o a
ppr
o
ac
he
s
f
or
t
hr
ou
ghpu
t
es
t
i
m
at
i
o
n a
n
d b
it
r
a
t
e
s
el
e
c
t
i
on,
w
h
i
c
h
w
as
e
v
a
l
uat
ed on
our
t
e
s
t
b
ed (
a
f
ul
l
y
f
u
n
c
t
i
o
nal
H
T
T
P
Li
v
e S
t
r
e
am
i
ng
s
y
s
t
em
)
ov
er
a
net
w
or
k
,
em
ul
at
e
d us
i
ng D
um
m
y
N
et
.
F
i
r
s
t
,
t
h
e t
hr
ou
ghp
ut
e
s
t
i
m
at
i
on m
et
hod,
b
as
e
d on t
he pr
edi
c
t
i
on of
t
he di
f
f
er
en
c
e b
et
w
een t
he e
s
t
i
m
at
ed an
d i
n
s
t
an
t
ane
ous
t
hr
oughp
ut
s
,
w
as
ob
s
er
v
ed t
o r
e
s
pon
d s
m
oot
h
l
y
t
o
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
on
s
and
r
api
d
l
y
t
o
l
ar
ge
f
l
u
c
t
u
at
i
o
ns
.
S
ec
on
d,
t
he
bi
t
r
at
e
s
el
e
c
t
i
on
al
gor
i
t
hm
,
ba
s
e
d
on
p
i
ec
ew
i
s
e
f
u
nc
t
i
on
s
t
o
d
ef
i
ne
t
h
e
v
ar
i
a
t
i
o
n
r
a
nge
of
t
he
c
ur
r
ent
bi
t
r
at
e,
w
as
f
ou
nd
t
o
r
e
s
ul
t
i
n
s
m
oot
he
r
c
han
ges
i
n
q
ual
i
t
y
w
i
t
h
a
h
i
gher
a
v
er
a
ge
qu
al
i
t
y
.
T
he
r
es
ul
t
s
o
f
our
e
x
per
i
m
ent
s
d
em
ons
t
r
a
t
ed
t
he
pr
os
p
ec
t
s
of
our
bi
t
r
a
t
e a
dapt
a
t
i
on
al
gor
i
t
hm
f
or
H
T
T
P
ad
apt
i
v
e s
t
r
eam
i
n
g.
Ke
y
w
o
rd
s
:
t
hr
ough
put
es
t
i
m
at
i
on
,
bi
t
r
at
e
s
e
l
ec
t
io
n
,
HL
S
C
o
p
y
r
i
g
h
t
©
20
16 U
n
i
ver
si
t
a
s A
h
mad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
eser
ved
.
1
.
I
n
tr
o
d
u
c
ti
o
n
W
i
t
h
t
he
di
f
f
us
i
on
of
new
v
i
de
o
-
ena
bl
ed
de
v
i
c
es
and
f
as
t
er
I
nt
er
ne
t
c
onnec
t
i
ons
,
v
i
d
eo
t
r
af
f
i
c
has
c
o
m
e t
o dom
i
nat
e I
nt
er
ne
t
t
r
af
f
i
c
.
S
ev
er
a
l
v
i
d
eo s
t
r
eam
i
ng pr
ot
oc
o
l
s
hav
e be
en
pr
opos
e
d f
or
t
he d
el
i
v
er
y
of
v
i
de
o c
ont
ent
.
T
r
adi
t
i
o
n
al
s
t
r
eam
i
ng pr
o
t
oc
ol
s
,
s
u
c
h as
r
eal
t
i
m
e
s
t
r
eam
i
ng pr
ot
oc
o
l
(
R
T
S
P
)
,
c
ont
r
ol
t
he
v
i
de
o t
r
a
ns
m
i
s
s
i
on r
at
e
di
r
ec
t
l
y
;
ho
w
ev
er
,
s
uc
h pr
ot
oc
ol
s
ar
e di
f
f
i
c
ul
t
t
o dep
l
o
y
bec
aus
e a s
pec
i
a
l
i
z
e
d s
t
r
ea
m
i
ng s
er
v
er
i
s
r
equi
r
ed
[
1
]
. B
y
c
o
n
tr
a
s
t,
h
y
p
er
t
ex
t
t
r
ans
f
er
pr
ot
oc
ol
(
H
T
T
P
[
2
]
)
adapt
i
v
e s
t
r
e
am
i
ng
has
bec
om
e a c
os
t
-
ef
f
ec
t
i
v
e and
popu
l
ar
o
pt
i
on
b
ec
aus
e
i
t
r
eus
es
t
he
ex
i
s
t
i
n
g
I
nt
er
net
i
nf
r
as
t
r
uc
t
ur
e,
pr
ov
i
d
es
ne
t
w
or
k
addr
es
s
t
r
ans
l
at
i
o
n (
N
A
T
)
f
r
i
endl
i
ne
s
s
and i
s
al
l
o
w
ed b
y
m
os
t
f
i
r
ew
al
l
s
[
3
]
.
C
ur
r
ent
l
y
,
on
e of
t
he
m
os
t
pr
om
i
nent
ap
pr
oac
h
es
i
s
H
T
T
P
Li
v
e
S
t
r
e
am
i
ng (
H
LS
)
,
as
pr
opos
e
d b
y
A
pp
l
e
I
nc
.
I
n ge
ner
a
l
,
a
n H
T
T
P
adapt
i
v
e s
t
r
eam
i
ng s
er
v
er
k
now
s
l
i
t
t
l
e
i
nf
or
m
at
i
on a
bout
t
he
c
l
i
en
t
,
i
t
i
s
t
h
e c
l
i
e
nt
’
s
r
es
p
ons
i
bi
l
i
t
y
t
o m
a
k
e dec
i
s
i
ons
r
e
gar
di
ng t
he s
el
ec
t
i
o
n o
f
appr
opr
i
at
e
al
t
er
n
at
i
v
es
t
o
m
ai
nt
a
i
n
a
g
ood
qu
al
i
t
y
of
ex
per
i
enc
e
(
Q
oE
[
4
]
)
.
T
o
pr
ov
i
d
e
t
he
h
i
ghes
t
pos
s
i
b
l
e
v
i
deo
qu
al
i
t
y
,
a
n a
da
pt
at
i
o
n a
l
gor
i
t
hm
m
us
t
appr
opr
i
at
el
y
es
t
i
m
at
e t
h
e a
v
a
i
l
ab
l
e t
hr
o
ugh
put
.
O
v
er
es
t
i
m
at
i
on
of
t
h
e t
hr
o
ughp
ut
m
a
y
l
e
ad
t
o
buf
f
er
f
r
ee
z
es
,
w
her
e
as
u
nder
es
t
i
m
at
i
on
of
t
he
t
hr
oug
hpu
t
m
a
y
l
ead
t
o
bu
f
f
er
ov
er
f
l
o
w
.
F
ur
t
her
m
or
e,
t
o
pr
ov
i
d
e us
er
s
w
i
t
h
a s
m
oot
her
v
i
deo
qua
l
i
t
y
,
a g
ood
bi
t
r
at
e s
e
l
ec
t
i
on
al
gor
i
t
hm
i
s
des
i
r
ab
l
e.
I
f
t
he bi
t
r
at
e
i
s
s
el
ec
t
ed
bas
ed o
nl
y
o
n t
he
es
t
i
m
at
ed
t
hr
oug
hpu
t
,
t
he
n
abr
upt
c
ha
ng
es
i
n qua
l
i
t
y
w
i
l
l
oc
c
ur
w
hen
t
he
a
v
a
i
l
a
bl
e
t
hr
oug
hput
dec
r
eas
es
or
i
nc
r
eas
es
dr
a
m
at
i
c
al
l
y
.
H
o
w
t
o
f
ul
f
i
l
l
bot
h
r
equ
i
r
em
ent
s
di
s
c
us
s
ed
abo
v
e
i
s
a
k
e
y
r
es
ear
c
h pr
obl
em
of
H
T
T
P
adapt
i
v
e s
t
r
eam
i
ng
[
5
]
.
R
es
ear
c
her
s
[
6
-
12
]
ha
v
e r
e
por
t
ed
v
ar
i
o
us
al
g
or
i
t
hm
s
f
or
t
hr
oughpu
t
e
s
t
i
m
at
i
on an
d b
i
t
r
at
e
s
el
ec
t
i
on.
R
egar
d
i
n
g t
hr
o
ug
hput
es
t
i
m
at
i
on,
i
n
[
6
]
, t
h
e
aut
h
or
s
us
ed
t
he
r
unn
i
n
g a
v
er
a
ge
of
t
he
m
eas
ur
ed
t
hr
oughp
ut
t
o
es
t
i
m
at
e
t
he
t
hr
ou
gh
put
.
H
o
w
e
v
er
,
t
h
i
s
m
et
hod
ex
hi
b
i
t
e
d
a
s
l
o
w
r
es
pons
e
t
o
l
ar
g
e
c
han
ges
i
n
t
hr
o
ugh
put
and
t
hus
ha
d
a
t
end
enc
y
t
o
s
uf
f
er
buf
f
er
under
f
l
o
w
.
I
n
[
7
]
,
t
he de
gr
ee
of
f
l
uc
t
uat
i
on of
t
he t
hr
ough
put
d
i
f
f
er
enc
e w
as
us
ed t
o d
y
n
am
i
c
al
l
y
c
ont
r
o
l
t
he
w
ei
g
ht
i
ng c
o
ef
f
i
c
i
ent
of
t
h
e m
et
hod pr
es
en
t
ed
i
n
[
8
]
f
or
t
hr
oughp
ut
es
t
i
m
at
i
on.
A
l
t
hou
gh t
hi
s
m
et
hod s
how
ed s
up
er
i
or
i
t
y
i
n
det
ec
t
i
ng
l
ar
g
e c
han
ges
i
n t
hr
oug
hput
,
i
t
f
ai
l
e
d t
o s
m
oot
hl
y
es
t
i
m
at
e t
he t
hr
o
ugh
put
i
n
t
he c
as
e of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
a
nd t
h
us
i
nc
ur
r
ed r
edu
nd
ant
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
ow
ar
ds
S
moo
t
h a
nd
H
ig
h
-
Q
ual
i
t
y
B
i
t
r
at
e
A
d
apt
a
t
i
on f
or
H
T
T
P
A
da
pt
i
v
e
…
(
Li
hon
g G
eng
)
905
f
l
uc
t
uat
i
ons
i
n bi
t
r
at
e.
I
n
[
9
]
,
an ex
pon
ent
i
a
l
f
unc
t
i
on
w
as
ado
pt
e
d t
o d
y
n
am
i
c
al
l
y
c
ont
r
o
l
t
he
w
ei
g
ht
i
ng
c
o
ef
f
i
c
i
ent
of
t
he
m
et
hod
pr
es
ent
e
d
i
n
[
6
]
.
T
hi
s
m
et
hod
of
f
er
ed
s
m
oot
h
es
t
i
m
at
i
on
in
t
he
c
as
e
of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
at
a
r
el
at
i
v
e
l
y
h
i
gh
t
hr
oug
hpu
t
.
H
o
w
ev
er
,
at
l
o
w
t
hr
ou
ghp
ut
,
t
he es
t
i
m
at
ed t
hr
ou
ghp
ut
of
t
hi
s
m
et
hod
w
as
s
ens
i
t
i
v
e t
o s
uc
h f
l
uc
t
uat
i
o
ns
.
R
egar
d
i
n
g
bi
t
r
at
e
s
el
ec
t
i
o
n,
i
n
[
10
]
,
t
he
aut
hor
s
ac
hi
e
v
ed
t
he
s
m
al
l
es
t
c
han
ges
i
n
v
i
d
eo
qua
l
i
t
y
an
d t
he f
e
w
es
t
i
n
t
e
r
r
upt
i
o
ns
b
y
pr
es
er
v
i
ng t
he
m
i
ni
m
u
m
buf
f
er
l
engt
h
.
H
o
w
e
v
er
,
w
he
n
t
he m
i
s
m
at
c
h bet
w
een
t
h
e a
v
a
i
l
abl
e t
hr
ou
ghp
ut
an
d t
h
e
v
i
deo
b
i
t
r
at
e
w
as
s
i
gn
i
f
i
c
ant
,
t
h
i
s
al
g
or
i
t
hm
appe
ar
ed
t
o
s
uf
f
er
f
r
o
m
abr
upt
c
han
ges
i
n
bi
t
r
at
e.
A
dd
i
t
i
on
al
l
y
,
i
t
r
equ
i
r
ed t
oo
m
uc
h
t
i
m
e t
o adj
us
t
t
he qu
al
i
t
y
i
n ac
c
or
danc
e
w
i
t
h t
he a
v
a
i
l
a
bl
e t
hr
ou
ghp
ut
.
A
qu
i
c
k
boot
a
l
g
or
i
t
hm
w
as
pr
opos
ed t
o s
ol
v
e t
hi
s
t
y
p
e of
pr
obl
em
i
n
[
11
]
.
F
ur
t
her
m
or
e,
t
he aut
hor
s
us
ed a f
i
x
ed
-
i
nt
er
v
a
l
b
uf
f
er
m
odel
t
o k
ee
p t
he
b
i
t
r
at
e
unc
h
ang
ed
w
h
ene
v
er
t
he
buf
f
er
s
i
z
e
w
as
w
i
t
h
i
n
a pr
es
e
t
i
nt
er
v
a
l
.
A
l
t
hou
gh t
hi
s
m
et
hod r
e
duc
ed
t
he
num
ber
of
c
hanges
i
n b
i
t
r
at
e,
i
t
r
e
qui
r
e
d a
l
ar
g
e
buf
f
er
t
o c
ope w
i
t
h l
ong
-
t
e
r
m
v
ar
i
at
i
ons
i
n t
he a
v
a
i
l
a
bl
e t
hr
oug
hpu
t
.
I
n
[
12
]
,
t
h
e aut
h
or
s
onl
y
us
ed t
he buf
f
er
t
o ef
f
ec
t
i
v
el
y
r
e
duc
e
t
he r
eb
uf
f
er
r
at
e,
s
ugges
t
i
ng t
h
at
t
he c
ur
r
ent
buf
f
er
l
ev
el
w
as
s
uf
f
i
c
i
ent
f
or
an
adapt
at
i
o
n al
gor
i
t
hm
i
n
t
he s
t
ead
y
s
t
at
e.
I
n add
i
t
i
on
t
o
t
h
e
c
ur
r
ent
buf
f
er
,
w
e em
phas
i
z
e t
he i
m
por
t
anc
e of
t
he v
ar
i
at
i
o
n r
ang
e of
t
he c
ur
r
ent
bi
t
r
a
t
e i
n
an ada
pt
at
i
o
n
al
g
or
i
t
hm
.
I
n
t
hi
s
p
aper
,
w
e
pr
es
en
t
a
nov
el
bi
t
r
at
e
ad
apt
at
i
o
n
al
g
or
i
t
hm
f
or
H
T
T
P
adapt
i
v
e
s
t
r
eam
i
ng.
O
ur
bi
t
r
at
e a
d
apt
at
i
o
n al
gor
i
t
hm
i
nc
l
ude
s
a t
hr
ough
put
es
t
i
m
at
i
on
m
et
hod an
d a
bi
t
r
at
e
s
el
ec
t
i
on
a
l
gor
i
t
hm
.
T
he
c
ont
r
i
but
i
ons
of
t
he
pa
per
ar
e
t
hr
ee
-
f
o
ld
.
F
ir
s
t
,
w
i
t
h
t
he
i
nt
ent
of
r
eac
t
i
ng s
m
oot
hl
y
t
o s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
and q
ui
c
k
l
y
t
o l
ar
g
e f
l
uc
t
uat
i
ons
,
a no
v
e
l
t
hr
oug
hp
ut
es
t
i
m
at
i
on m
et
hod
i
s
pr
op
os
ed
b
as
ed
o
n
pr
e
di
c
t
i
ng
t
he
d
i
f
f
er
enc
e bet
w
e
en
t
he
es
t
i
m
at
ed
an
d
i
ns
t
an
t
an
eous
t
hr
o
ugh
put
s
.
S
ec
on
d,
t
o pr
o
v
i
d
e us
er
s
w
i
t
h a
hi
g
h an
d r
el
a
t
i
v
e
l
y
s
m
oot
h v
i
deo
qua
l
i
t
y
,
w
e
pr
op
os
e
an
i
n
n
ov
a
t
i
v
e
bi
t
r
at
e
s
el
ec
t
i
o
n a
l
gor
i
t
hm
bas
ed
o
n p
i
ec
e
w
i
s
e f
unc
t
i
o
ns
t
o
def
i
ne
t
he
v
ar
i
at
i
on
r
a
ng
e
of
t
he
c
ur
r
ent
bi
t
r
at
e.
F
i
na
l
l
y
,
t
o
v
er
i
f
y
t
h
e
p
er
f
or
m
anc
e
of
our
b
i
t
r
at
e
adap
t
at
i
on al
gor
i
t
hm
,
w
e r
e
por
t
t
h
e i
m
pl
em
ent
at
i
on of
a f
ul
l
y
f
unc
t
i
o
na
l
H
L
S
s
y
s
t
e
m
.
T
he paper
i
s
or
gani
z
e
d as
f
ol
l
o
w
s
.
I
n S
ec
t
i
on 2,
w
e f
i
r
s
t
pr
ov
i
de an o
v
er
v
i
e
w
of
H
LS
.
S
ec
t
i
on
3 des
c
r
i
b
es
our
bi
t
r
at
e
a
dapt
at
i
on
al
g
or
i
t
h
m
.
I
n S
ec
t
i
on
4,
t
he
ex
per
i
m
ent
s
ar
e
di
s
c
us
s
ed
i
n de
t
ai
l
.
F
i
na
l
l
y
,
c
onc
l
us
i
o
ns
an
d f
ut
ur
e
w
or
k
ar
e addr
es
s
ed i
n S
ec
t
i
o
n
5.
2.
O
v
er
v
i
ew
o
f H
L
S
H
LS
i
s
a
m
edi
a
s
t
r
eam
i
ng
pr
ot
oc
ol
b
as
ed
on
H
T
T
P
i
m
pl
e
m
ent
ed
b
y
A
p
pl
e
.
I
t
i
s
w
i
d
el
y
us
ed b
y
v
i
d
eo s
t
r
eam
i
ng
p
r
ov
i
der
s
f
or
i
t
s
eas
y
dep
l
o
y
m
e
nt
,
d
y
nam
i
c
al
ada
pt
a
bi
l
i
t
y
and s
t
r
on
g
pene
t
r
ab
i
l
i
t
y
[
13
]
.
C
onc
e
pt
ual
l
y
,
A
H
LS
s
y
s
t
em
has
t
hr
ee
p
ar
t
s
:
a
s
er
v
er
,
a
d
i
s
t
r
i
but
i
on
s
y
s
t
em
,
and
a
c
l
i
e
nt
.
T
he
s
er
v
er
i
s
r
es
pons
i
b
l
e
f
or
enc
od
i
n
g
t
h
e
i
npu
t
s
t
r
eam
s
as
MP
E
G
-
4
(
A
A
C
aud
i
o
and H
.
264
v
i
deo)
,
e
nc
aps
u
l
at
i
ng
t
hem
i
n MP
E
G
-
2 T
S
f
or
m
at
,
and
pr
epar
i
ng t
h
e
enc
aps
ul
at
e
d
m
edi
a
f
or
di
s
t
r
i
but
i
o
n.
T
he
di
s
t
r
i
b
ut
i
on
s
y
s
t
em
i
s
a
s
t
a
ndar
d
w
e
b
s
er
v
er
w
hi
c
h
i
s
r
es
pons
i
b
l
e
f
or
ac
c
ept
i
n
g a
nd r
es
pon
di
n
g c
l
i
e
nt
’
s
r
e
qu
es
t
s
.
T
he s
er
v
er
and
d
i
s
t
r
i
but
i
o
n s
y
s
t
em
ar
e i
nt
egr
at
ed
as
H
LS
s
er
v
er
i
n t
h
i
s
pa
per
.
T
he c
l
i
ent
i
s
r
es
po
ns
i
b
l
e
f
or
c
hoo
s
i
ng t
he
appr
o
pr
i
at
e m
edi
a
t
o
r
eques
t
,
do
w
nl
o
ad
i
ng
t
hem
,
and
t
he
n r
eas
s
em
bl
i
n
g t
h
em
t
o pr
es
ent
.
A
da
pt
at
i
o
n
i
s
an
i
m
por
t
ant
par
t
of
H
LS
[
14
]
.
O
n
t
he
H
LS
s
er
v
er
s
i
d
e,
t
he
or
i
gi
nal
v
i
de
o
c
ont
ent
i
s
e
nc
ode
d i
nt
o
m
ul
t
i
pl
e
al
t
er
nat
i
v
es
(
v
er
s
i
ons
)
at
di
f
f
er
ent
b
i
t
r
at
es
.
T
hen,
eac
h
al
t
er
n
at
i
v
e i
s
f
ur
t
her
par
t
i
t
i
o
ned i
nt
o a s
er
i
es
of
s
m
al
l
s
egm
ent
s
(
c
hun
k
s
)
w
i
t
h t
he
s
a
m
e du
r
at
i
o
n.
S
i
m
ul
t
an
eous
l
y
,
t
he c
h
ar
a
c
t
er
i
s
t
i
c
s
of
eac
h a
l
t
er
n
at
i
v
e
s
uc
h as
b
i
t
r
at
e
,
c
od
i
ng,
and r
es
o
l
ut
i
o
n
ar
e r
ec
or
de
d i
n t
h
e m
ani
f
es
t
f
i
l
e w
i
t
h t
he,
m
3u8 ex
t
ens
i
on.
A
l
l
t
he m
edi
a s
egm
ent
s
and
m
ani
f
es
t
f
i
l
es
ar
e
s
t
or
ed
i
n
t
he
di
s
t
r
i
but
i
on
s
y
s
t
em
.
O
n
t
he
H
L
S
c
l
i
en
t
s
i
d
e,
ac
c
or
d
i
ng
t
o
t
h
e
s
t
at
us
of
t
he
t
er
m
i
nal
/
net
w
or
k
,
t
he m
os
t
appr
o
pr
i
at
e al
t
er
na
t
i
v
e i
s
dow
nl
oad
ed t
hr
oug
h t
h
e s
endi
ng
of
c
ons
ec
ut
i
v
e
H
T
T
P
r
eques
t
s
.
T
her
ef
or
e,
t
he
c
l
i
e
nt
e
v
e
nt
ua
l
l
y
get
s
t
h
e
w
ho
l
e
v
i
d
e
o
c
ons
i
s
t
i
n
g
of
t
he s
egm
ent
s
at
d
i
f
f
er
ent
bi
t
r
at
es
.
3
.
T
h
e
P
r
o
p
o
s
e
d
B
i
tr
a
te
A
d
a
p
ta
ti
o
n
A
l
g
o
r
i
th
m
I
n t
h
i
s
s
ec
t
i
on,
our
bi
t
r
a
t
e
a
dapt
at
i
on
al
gor
i
t
hm
,
w
h
i
c
h
c
ons
i
s
t
s
of
t
hr
ou
ghp
ut
es
t
i
m
at
i
on
and b
i
t
r
at
e
s
e
l
ec
t
i
on,
i
s
des
c
r
i
bed
i
n det
ai
l
.
A
s
s
t
at
ed a
bo
v
e,
m
ul
t
i
p
l
e a
l
t
e
r
nat
i
v
es
of
a
s
egm
ent
ed s
equ
enc
e
at
di
f
f
er
ent
bi
t
r
at
es
ar
e s
t
or
e
d i
n t
h
e s
er
v
er
.
E
ac
h s
e
gm
e
nt
c
ont
ai
ns
τ
s
ec
onds
of
pl
a
y
bac
k
.
T
h
e s
et
of
bi
t
r
at
es
f
or
m
di
f
f
er
ent
v
i
deo qua
l
i
t
i
es
i
s
deno
t
ed b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
04
–
91
5
906
{
}
12
,,
m
R
RR
R
=
,
w
her
e
i
R
is
t
h
e
i
t
h bi
t
r
at
e i
n
R
.
W
e as
s
u
m
e t
hat
ij
RR
<
if
ij
<
.
T
he
c
l
i
en
t
do
w
n
l
o
ads
t
he s
egm
ent
s
i
n c
hr
ono
l
o
gi
c
a
l
or
der
,
and al
l
do
w
nl
o
ads
ar
e no
n
-
pr
eem
pt
i
v
e
,
i
.
e.
,
t
he do
w
n
l
oa
d
of
s
egm
ent
i
c
annot
b
egi
n
unt
i
l
s
egm
ent
1
i
−
has
bee
n c
om
pl
et
el
y
do
w
n
l
oa
ded
.
A
f
t
er
eac
h
f
r
a
gm
ent
i
s
do
w
n
l
oa
ded
,
our
t
hr
oug
hpu
t
es
t
i
m
at
i
on
m
et
hod
i
s
i
n
v
ok
ed
t
o es
t
i
m
at
e t
he a
v
ai
l
ab
l
e
t
hr
ough
put
.
T
hen,
t
he
pr
opos
ed b
i
t
r
at
e a
dapt
at
i
on
al
gor
i
t
hm
w
i
l
l
r
u
n t
h
e
bi
t
r
at
e s
el
ec
t
i
o
n a
l
gor
i
t
hm
t
o s
el
ec
t
t
h
e m
os
t
appr
o
pr
i
at
e
b
i
t
r
at
e f
or
t
h
e n
ex
t
s
egm
ent
.
T
h
e
det
a
i
l
s
of
t
h
e a
l
gor
i
t
hm
ar
e des
c
r
i
bed
be
l
o
w
.
3
.
1
.
T
h
r
o
u
g
h
p
u
t E
s
ti
m
a
ti
o
n
M
e
th
o
d
T
hr
oughput
es
t
i
m
at
i
on
i
s
o
ne of
t
h
e
m
os
t
c
r
uc
i
al
c
on
c
er
ns
i
n
ada
pt
i
v
e
s
t
r
eam
i
ng
[
11
]
.
U
s
ual
l
y
,
t
he a
v
a
i
l
abl
e t
hr
o
ughp
ut
i
s
c
al
c
u
l
at
ed b
y
di
v
i
di
n
g t
he
dat
a s
i
z
e of
a
s
egm
ent
b
y
i
t
s
del
i
v
er
y
dur
at
i
on
,
as
d
eno
t
ed b
y
:
i
i
i
S
T
t
=
(
1
)
W
h
er
e
i
S
and
i
t
ar
e t
he s
i
z
e a
nd do
w
nl
o
ad t
i
m
e,
r
es
pec
t
i
v
e
l
y
,
of
t
he
i
t
h s
egm
ent
and
i
T
i
s
t
he
av
a
i
l
ab
l
e
t
hr
oug
hp
ut
f
or
t
h
e
i
t
h
s
egm
ent
.
B
e
l
o
w
,
w
e
c
al
l
i
T
t
he
i
ns
t
ant
ane
ous
t
hr
ou
ghpu
t
.
T
he
s
im
p
le
s
t
m
et
hod
t
o
es
t
i
m
at
e
t
he
av
ai
l
ab
l
e
t
hr
oug
h
put
i
s
m
er
el
y
t
o
us
e
t
h
e
i
ns
t
ant
ane
ous
t
hr
oug
hpu
t
.
T
hi
s
m
et
hod
y
i
el
ds
a s
t
a
bl
e
buf
f
er
l
ev
el
;
ho
w
e
v
er
,
t
he
v
i
de
o qu
al
i
t
y
f
l
uc
t
uat
es
.
A
s
m
oot
hi
n
g s
t
r
at
eg
y
w
as
a
dopt
ed
i
n
[
15
]
to
s
ol
v
e
t
h
e pr
ob
l
em
of
f
l
uc
t
uat
i
ons
.
H
o
w
e
v
er
,
t
hi
s
m
et
hod r
eac
t
s
s
l
o
w
l
y
t
o l
ar
ge f
l
uc
t
ua
t
i
ons
,
w
h
i
c
h
m
ay
r
es
ul
t
i
n
pl
a
y
b
ac
k
f
r
ee
z
es
.
I
n
[
9
]
,
a
d
y
n
am
i
c
w
e
i
ght
i
n
g c
oef
f
i
c
i
ent
w
as
app
l
i
ed
i
n t
he s
m
oot
hi
ng m
et
hod t
o s
ol
v
e
t
hi
s
pr
o
bl
em
.
A
l
t
hou
gh t
h
i
s
m
et
hod pos
s
es
s
es
t
he adv
ant
ages
of
bot
h m
et
hods
di
s
c
us
s
ed abo
v
e,
i
t
s
t
i
l
l
y
i
el
ds
f
l
uc
t
uat
i
ng es
t
i
m
at
es
w
h
en
t
he a
v
a
i
l
ab
l
e t
hr
oug
hpu
t
i
s
l
o
w
.
T
he goal
of
our
m
et
hod i
s
t
o ac
hi
ev
e a t
hr
o
ugh
put
e
s
t
i
m
at
i
on t
hat
1)
i
s
s
t
abl
e i
n t
he
c
as
e
of
s
hor
t
-
t
er
m
f
l
uc
t
ua
t
i
ons
a
nd
2)
r
e
ac
t
s
qui
c
k
l
y
t
o
l
ar
g
e
f
l
uc
t
uat
i
ons
.
T
he
un
der
l
y
i
ng
pr
i
nc
i
pl
e
of
our
m
et
hod i
s
a
s
f
ol
l
o
w
s
:
1
ee
i
ii
T
TD
+
=
+
(
2
)
W
h
er
e
1
e
i
T
+
i
s
t
he es
t
i
m
at
ed
t
hr
oug
hpu
t
f
or
t
he
(
1)
i
+
t
h s
egm
ent
and
e
i
D
i
s
t
he pr
edi
c
t
ed
di
f
f
er
enc
e v
al
ue.
A
pr
ec
i
s
e pr
edi
c
t
i
on of
e
i
D
y
i
e
l
ds
a pr
ec
i
s
e
1
e
i
T
+
.
e
i
D
i
s
def
i
ned b
y
(
3)
and (
4)
i
n
ou
r
m
et
hod:
e
ii
i
DT
T
=
−
(
3
)
*
e
ii
DD
δ
=
(
4
)
W
h
er
e
i
D
r
ef
l
ec
t
s
t
he de
gr
ee
of
f
l
uc
t
uat
i
on
of
t
he
t
hr
ou
ghpu
t
an
d
δ
i
s
a c
or
r
ec
t
i
o
n
t
er
m
t
hat
t
ak
es
v
al
ues
on t
he i
nt
er
v
al
(
)
0,
1
.
F
or
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
,
i
D
i
s
s
m
al
l
and
δ
s
houl
d
be
c
l
os
er
t
o
1
t
o
br
i
ng
1
e
i
T
+
c
l
os
er
t
o
e
i
T
,
r
es
ul
t
i
ng
i
n
a
s
m
oot
her
es
t
i
m
at
i
on.
F
or
l
ar
ge
f
l
uc
t
uat
i
ons
,
i
D
i
s
l
ar
ge a
nd a
δ
c
l
os
er
t
o 0 i
s
r
equ
i
r
ed t
o br
i
ng
1
e
i
T
+
c
l
o
s
er
t
o
i
T
t
o al
l
o
w
a m
or
e r
api
d
r
es
pons
e
t
o
l
ar
ge
f
l
uc
t
uat
i
o
ns
.
I
n
s
um
m
ar
y
,
t
he
v
al
ue
of
t
he
p
ar
am
et
er
δ
i
s
s
t
r
on
gl
y
r
e
l
at
ed
t
o
i
D
i
n
our
ana
l
y
s
i
s
.
T
he
c
ont
r
ol
f
unc
t
i
ons
r
e
l
at
i
ng
δ
and
i
D
ar
e
des
i
gn
ed
as
s
ho
w
n
i
n
(
5)
and
(6
):
i
nor
D
T
ρ
=
(
5
)
0
()
1
1
M
Ne
ρρ
δ
−
=
+
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
ow
ar
ds
S
moo
t
h a
nd
H
ig
h
-
Q
ual
i
t
y
B
i
t
r
at
e
A
d
apt
a
t
i
on f
or
H
T
T
P
A
da
pt
i
v
e
…
(
Li
hon
g G
eng
)
907
W
h
er
e
ρ
i
s
t
he
nor
m
al
i
z
at
i
on
v
a
l
u
e of
i
D
and
nor
T
is
a
nor
m
al
i
z
at
i
o
n f
ac
t
or
;
M
and
0
ρ
ar
e t
he
s
t
eepn
es
s
and
m
i
dpoi
nt
of
t
he
c
ont
r
o
l
f
unc
t
i
on
(
6)
,
r
es
pec
t
i
v
e
l
y
;
N
i
s
a
f
unc
t
i
o
n
of
t
he
buf
f
e
r
oc
c
upanc
y
.
I
n
our
m
et
hod,
w
e as
s
um
e t
hat
t
he f
l
uc
t
u
at
i
ons
ar
e s
hor
t
-
t
er
m
w
h
en
i
D
i
s
l
es
s
t
ha
n
10%
of
nor
T
(
i.
e
.
,
0.1
ρ
<
)
and t
he f
l
uc
t
uat
i
ons
ar
e l
ar
ge
w
h
en
i
D
i
s
l
ar
ger
t
h
an
20%
of
nor
T
(
i.
e
.
,
0.2
ρ
>
)
.
M
and
0
ρ
need t
o be s
et
pr
op
er
l
y
t
o m
a
k
e s
ur
e t
hat
a
ρ
s
m
al
l
er
t
han 0
.
1
w
i
l
l
y
i
e
l
d a
δ
c
l
os
er
t
o
1
,
i
n
ot
h
er
w
or
ds
,
our
m
et
hod
g
et
s
a
s
m
oot
h
es
t
i
m
at
i
on
f
or
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
;
b
y
c
o
nt
r
as
t
,
a
ρ
l
ar
ger
t
han
0.
2
w
i
l
l
obt
ai
n
a
δ
c
l
os
er
t
o
0,
s
ug
ges
t
i
ng
t
hat
o
ur
m
et
hod q
ui
c
k
l
y
r
es
ponds
t
o
l
ar
ge
f
l
uc
t
uat
i
o
ns
.
F
ur
t
her
m
or
e,
i
n
c
as
e
of
buf
f
er
dr
ai
ni
ng
-
up
,
N
w
i
l
l
do
m
i
nat
e
t
he
s
peed of
c
ont
r
ol
i
n (
6)
w
he
n t
he buf
f
er
l
ev
el
i
s
l
o
w
er
t
h
an t
he m
i
ni
m
u
m
t
hr
es
hol
d.
I
t
i
s
def
i
ned as
ex
pr
es
s
ed i
n (
7)
an
d (
8)
:
c
u
r
m
i
n
ma
x
mi
n
BB
BB
µ
−
=
−
(
7
)
1
0
100
0
N
µ
µ
µ
>
=
≤
(
8
)
W
h
er
e
c
u
r
B
i
s
t
he c
ur
r
ent
buf
f
er
l
ev
el
w
h
i
c
h i
s
m
eas
ur
ed i
n s
ec
onds
,
mi
n
B
and
ma
x
B
ar
e
t
he
m
i
ni
m
u
m
and m
ax
i
m
u
m
bu
f
f
er
t
hr
es
hol
ds
,
r
es
pec
t
i
v
e
l
y
,
and
µ
i
s
t
he b
uf
f
er
oc
c
upanc
y
.
3
.
2
.
B
i
tr
a
te
S
e
l
e
c
ti
o
n
A
l
g
o
r
i
th
m
I
n t
h
i
s
s
ec
t
i
on,
gi
v
e
n t
he
es
t
i
m
at
ed t
hr
o
ugh
put
as
d
es
c
r
i
bed
abo
v
e
,
w
e
pr
o
po
s
e an
i
nn
ov
at
i
v
e
bi
t
r
at
e
s
el
ec
t
i
o
n
al
gor
i
t
hm
t
hat
c
ons
i
der
s
t
he
v
ar
i
at
i
on
r
ang
e
of
t
h
e
c
ur
r
ent
b
i
t
r
at
e
t
o
ac
hi
e
v
e
a
h
i
g
h
and
r
e
l
at
i
v
e
l
y
s
m
oot
h
v
i
d
eo
q
ual
i
t
y
.
B
e
s
i
des
,
buf
f
er
under
f
l
o
w
and
ov
er
f
l
o
w
ar
e
c
ons
i
der
e
d i
n o
ur
al
gor
i
t
hm
.
T
o
ac
hi
e
v
e
our
g
oa
l
,
w
e
s
t
udi
ed
t
he
J
us
t
N
ot
i
c
ea
bl
e
D
i
f
f
er
enc
e
(
J
N
D
)
.
T
he
J
N
D
[
16
]
is
def
i
ned t
o des
c
r
i
be t
h
e s
m
al
l
es
t
per
c
ept
ua
l
di
f
f
er
enc
e/
c
hang
e bet
w
een t
w
o s
t
i
m
ul
i
(
e.
g.
,
t
w
o
v
er
s
i
o
ns
of
a
v
i
deo)
t
hat
c
o
ul
d
be
det
ec
t
ed
b
y
hum
an
per
c
ept
i
on;
t
h
i
s
w
or
k
c
onc
l
uded
t
h
at
i
f
t
h
e
v
i
deo
v
er
s
i
ons
w
er
e
eq
ua
l
l
y
s
p
ac
ed
b
y
3
J
N
D
uni
t
s
,
a s
ep
ar
at
i
o
n
t
h
at
y
i
el
d
ed no ob
v
i
ous
di
f
f
er
enc
e i
n
pr
ac
t
i
c
e
,
t
h
e t
y
p
i
c
al
num
ber
of
v
er
s
i
ons
w
as
4 t
o 7
.
O
n
t
he
bas
i
s
of
t
hi
s
c
onc
l
us
i
o
n,
t
he bi
t
r
at
es
of
t
he F
oot
b
a
l
l
an
d S
oc
c
er
s
eque
nc
es
w
er
e s
pac
ed b
y
1.
5 J
N
D
uni
t
s
i
n
[
17
]
,
r
es
ul
t
i
ng
i
n b
i
t
r
at
es
of
3000
,
149
5,
10
38,
773,
640
,
55
0,
42
7,
32
2,
2
60,
a
nd 2
22
K
bps
;
t
hi
s
l
i
s
t
of
bi
t
r
at
es
,
w
h
i
c
h
be
l
o
nge
d
t
o
f
as
t
-
m
ot
i
on
gr
oup
,
s
hou
l
d
t
hus
al
s
o
be
s
af
e
f
or
s
l
o
w
-
m
ot
i
on
gr
oup
.
B
as
ed
on t
hes
e c
onc
l
us
i
o
ns
,
t
w
o p
i
e
c
e
w
i
s
e f
unc
t
i
o
ns
ar
e des
i
gn
ed t
o
det
er
m
i
ne t
he s
af
e
v
ar
i
at
i
on r
a
nge
of
t
he c
ur
r
e
nt
b
i
t
r
at
e.
E
quat
i
o
ns
(
9)
an
d (
10)
pr
es
e
nt
t
h
es
e f
unc
t
i
o
ns
.
_
100
200
400
1400
c
ur
L
ow
T
h
L
ow
T
h
c
ur
M
i
dT
h
c
hg
up
M
i
dT
h
c
ur
H
i
gT
h
c
ur
H
i
gT
h
RR
R
RR
R
R
RR
RR
<
≤<
=
≤<
≥
(
9
)
_
100
(
,
200)
(
,
200)
(
,
400)
c
ur
L
ow
T
h
c
ur
L
ow
T
h
L
ow
T
h
c
ur
M
i
dT
h
c
hg
dow
n
c
ur
M
i
dT
h
M
i
dT
h
c
ur
H
i
gT
h
c
ur
H
i
gT
h
c
ur
H
i
gT
h
RR
mi
n
R
R
R
R
R
R
ma
x
R
R
R
R
R
ma
x
R
R
R
R
≤
−
<≤
=
−
<<
−≥
(
10
)
W
h
er
e
c
u
r
R
i
s
t
he c
ur
r
ent
bi
t
r
at
e;
Lo
w
Th
R
,
Mi
d
T
h
R
and
Hi
g
T
h
R
ar
e t
hr
ee bi
t
r
at
e t
hr
es
h
ol
ds
t
h
at
sa
t
i
sf
y
Lo
w
Th
M
i
d
Th
H
i
g
Th
R
RR
<<
;
t
he f
unc
t
i
ons
mi
n
and
ma
x
r
et
ur
n t
he m
i
ni
m
u
m
an
d
m
ax
i
m
u
m
v
a
l
ue,
r
es
pec
t
i
v
el
y
,
be
t
w
ee
n t
hei
r
t
w
o i
npu
t
s
;
and t
he r
et
ur
n v
al
u
es
,
_
c
hg
up
R
and
_
c
hg
dow
n
R
,
def
i
ne
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
04
–
91
5
908
t
he
s
af
e
v
ar
i
at
i
o
n
r
ange
of
t
he
c
ur
r
ent
bi
t
r
at
e.
I
f
t
he
av
ai
l
ab
l
e
t
hr
ou
gh
put
i
s
i
nc
r
ea
s
i
ng,
eq
uat
i
on
(
9)
i
s
us
ed
t
o
det
er
m
i
ne t
h
e v
ar
i
at
i
o
n r
an
ge,
i
.
e.
,
_
c
hg
up
R
.
B
y
c
ont
r
as
t
,
i
f
t
h
e a
v
a
i
l
ab
l
e t
hr
ough
put
i
s
dec
r
eas
i
ng
,
equ
at
i
on
(
10
)
i
s
us
ed
t
o
det
er
m
i
ne
_
c
hg
dow
n
R
.
W
e
c
onv
er
t
/
c
hg_up
c
hg_dow
n
R
R
i
nt
o
t
he
c
or
r
es
pondi
ng
bi
t
r
a
t
e
i
nd
ex
w
i
t
h
eq
uat
i
on (
11)
:
__
(
/
,)
c
hg
c
hg
up
c
hg
dow
n
c
ur
I
nde
x
f
R
R
R
=
(
11
)
W
h
er
e
f
i
s
a
c
on
v
er
s
i
on
f
u
nc
t
i
on
and
c
h
g
I
nde
x
i
s
t
h
e
s
af
e
v
ar
i
at
i
o
n
r
an
ge
of
t
he
bi
t
r
a
t
e
i
ndex
c
or
r
es
pondi
ng t
o
/
c
hg_up
c
hg_dow
n
R
R
.
B
as
ed o
n t
he s
af
e v
ar
i
at
i
on r
ang
e of
t
he c
ur
r
ent
bi
t
r
at
e as
det
er
m
i
ned abo
v
e,
t
he
pr
oc
edur
es
per
f
or
m
ed i
n
o
ur
al
gor
i
t
hm
ar
e des
c
r
i
b
ed
i
n a
l
gor
i
t
hm
1.
H
er
e
,
b
e
s
t
R
I
nde
x
is
in
i
t
i
a
li
z
e
d
as
t
he
b
i
t
r
at
e
i
ndex
w
i
t
h t
he h
i
g
hes
t
pos
s
i
b
l
e
v
a
l
u
e
t
hat
i
s
l
o
w
er
t
ha
n t
he c
ur
r
ent
es
t
i
m
at
ed
t
hr
oug
hpu
t
;
i
n ot
her
w
or
ds
,
i
t
s
at
i
s
f
i
es
:
{
}
|
b
es
t
e
R
I
nde
x
k
k
k
R
ma
x
R
R
T
R
R
=
<∈
，
(
12
)
l
as
t
R
I
nde
x
i
s
t
h
e b
i
t
r
at
e
i
nd
ex
of
t
he
l
as
t
s
egm
ent
.
D
i
f
I
nde
x
i
s
t
he
a
bs
ol
u
t
e
v
al
ue
of
or
i
D
w
hi
c
h
i
s
equ
al
t
o
t
h
e d
i
f
f
er
enc
e b
et
w
e
en
b
e
s
t
R
I
nde
x
a
nd
l
as
t
R
I
nde
x
.
mi
n
B
,
mi
d
B
and
ma
x
B
ar
e b
uf
f
e
r
t
hr
es
hol
ds
,
m
eas
ur
ed i
n
s
e
c
onds
,
w
h
i
c
h
sa
t
i
sf
y
mi
n
mi
d
m
a
x
B
B
B
<<
.
T
he i
nput
ar
gum
ent
s
of
our
al
g
or
i
t
hm
i
nc
l
ud
e t
h
e i
n
s
t
ant
an
eous
t
hr
oug
hpu
t
(
T
)
,
t
he
es
t
i
m
at
ed t
hr
ou
ghp
ut
(
e
T
)
,
t
he c
ur
r
ent
b
uf
f
er
l
ev
e
l
(
c
u
r
B
)
and t
he
b
i
t
r
at
e i
n
dex
of
t
he l
as
t
s
egm
ent
(
l
as
t
R
I
nde
x
)
.
T
he out
put
ar
g
um
ent
i
s
t
he b
i
t
r
at
e i
ndex
f
or
t
he
nex
t
s
egm
ent
(
n
e
x
t
R
I
nde
x
).
W
h
en
0
or
i
D
≥
,
an
e
qua
l
or
h
i
g
her
bi
t
r
at
e
c
an
be
r
e
ques
t
e
d
f
or
t
he
nex
t
s
egm
ent
.
T
he
c
ur
r
ent
buf
f
er
l
ev
e
l
i
s
c
o
ns
i
der
ed
i
n
t
hi
s
c
as
e.
I
f
c
u
r
m
i
d
BB
≤
,
i
t
i
s
not
w
i
s
e
t
o
i
nc
r
eas
e
t
he
b
i
t
r
at
e
i
m
m
edi
at
el
y
bec
aus
e
t
h
e
b
uf
f
er
l
ev
e
l
i
s
s
t
i
l
l
i
ns
uf
f
i
c
i
ent
.
T
hus
,
w
e
l
ea
v
e
t
h
e
bi
t
r
at
e
unc
han
g
ed
(
n
e
x
t
l
a
s
t
R
I
nde
x
R
I
nde
x
=
)
. If
c
u
r
m
i
d
BB
>
,
t
he
n i
t
i
s
s
af
e t
o i
nc
r
e
as
e t
he
bi
t
r
at
e.
A
s
des
c
r
i
be
d abo
v
e
,
i
t
i
s
nec
es
s
ar
y
t
o
ens
ur
e
t
ha
t
ev
er
y
c
ha
nge
i
n q
ual
i
t
y
r
em
ai
ns
w
i
t
hi
n
t
he
s
af
e v
ar
i
at
i
o
n r
ang
e of
t
he c
ur
r
ent
b
i
t
r
at
e.
T
hus
,
D
i
f
I
nde
x
i
s
c
om
par
ed w
i
t
h
c
h
g
I
nde
x
. If
c
h
g
D
i
f
I
nde
x
I
nde
x
≥
,
i
t
m
ean
s
t
hat
b
e
s
t
R
I
nde
x
i
s
m
u
c
h hi
gher
t
ha
n
l
as
t
R
I
nde
x
.
I
f
w
e c
hoos
e
b
e
s
t
R
I
nde
x
t
o
f
ol
l
o
w
t
h
e c
ur
r
ent
av
a
i
l
ab
l
e t
hr
o
ugh
put
,
a
n ob
v
i
ous
c
hang
e i
n q
ua
l
i
t
y
w
i
l
l
be not
i
c
ed b
y
t
he
us
er
.
T
hus
,
w
e c
ho
os
e
l
as
t
c
hg
R
I
nde
x
I
nde
x
+
as
t
he bi
t
r
at
e i
n
dex
f
or
t
he nex
t
s
egm
ent
,
w
hi
c
h pr
o
v
i
des
a r
el
at
i
v
el
y
r
a
pi
d
r
es
pons
e
t
o
t
h
e a
v
a
i
l
abl
e
t
hr
ou
gh
p
ut
w
h
i
l
e
m
ai
nt
a
i
ni
n
g
a
us
er
-
f
r
i
endl
y
q
ua
l
i
t
y
c
ha
nge.
I
f
<
c
h
g
D
i
f
I
nde
x
I
nde
x
,
t
hen
b
e
s
t
R
I
nde
x
c
an be c
hos
en s
af
el
y
.
H
o
w
ev
er
,
w
hen t
h
e
buf
f
er
l
ev
e
l
i
s
hi
g
her
t
han
t
he
m
ax
i
m
u
m
buf
f
er
t
hr
es
hol
d
and
b
es
t
R
I
nde
x
R
i
s
l
es
s
t
han
t
he
es
t
i
m
at
ed
t
hr
ou
ghpu
t
,
a
s
c
he
m
e t
o av
oi
d
buf
f
er
ov
er
f
l
o
w
i
s
r
eq
ui
r
e
d.
W
e i
nc
r
eas
e
b
e
s
t
R
I
nde
x
b
y
1 t
o pr
e
v
en
t
buf
f
e
r
ov
er
f
l
o
w
.
A
f
t
er
t
he app
l
i
c
at
i
on of
t
hi
s
ov
er
f
l
o
w
-
c
ont
r
ol
m
eas
ur
e,
D
i
f
I
nde
x
w
i
l
l
be l
es
s
t
han or
equa
l
t
o
c
h
g
I
nde
x
,
w
hi
c
h
i
s
s
t
i
l
l
a
s
af
e qu
al
i
t
y
c
han
ge.
W
h
en
0
or
i
D
<
,
t
he
c
ur
r
ent
av
ai
l
ab
l
e
t
hr
o
ug
hput
i
s
i
ns
uf
f
i
c
i
ent
t
o m
ai
nt
ai
n t
he
pr
e
v
i
o
us
v
i
deo
qu
al
i
t
y
.
B
i
t
r
at
e
r
ed
uc
t
i
on
i
s
i
ne
v
i
t
ab
l
e.
T
he
c
ur
r
ent
buf
f
er
l
e
v
el
s
hou
l
d
be
c
ar
ef
ul
l
y
c
ons
i
der
e
d i
n t
hi
s
c
as
e.
I
f
c
u
r
m
i
n
BB
≤
,
t
hen t
he buf
f
er
c
an be eas
i
l
y
em
pt
i
ed i
f
t
he s
el
ec
t
ed
bi
t
r
at
e
i
s
hi
g
her
t
h
an t
h
e a
v
a
i
l
abl
e
t
hr
oug
hpu
t
.
W
e i
nc
r
e
m
ent
al
l
y
dec
r
eas
e
b
e
s
t
R
I
nde
x
b
y
1 i
n a l
oop
unt
i
l
b
es
t
R
I
nde
x
R
i
s
l
es
s
t
ha
n t
he c
ur
r
en
t
a
v
ai
l
ab
l
e t
hr
ou
ghp
ut
.
T
hi
s
s
t
r
at
eg
y
w
i
l
l
c
a
us
e t
he
buf
f
er
l
ev
e
l
t
o
i
nc
r
eas
e
dur
i
ng
t
he
do
w
n
l
o
ad
d
ur
at
i
on
of
t
he
nex
t
s
eg
m
ent
t
o
pr
ev
ent
pl
a
y
b
ac
k
f
r
ee
z
es
,
bu
t
a
n
uns
af
e
c
hange
i
n
qua
l
i
t
y
m
a
y
oc
c
ur
.
I
f
t
he
buf
f
er
l
ev
el
s
at
i
s
f
i
es
B
mi
n
c
u
r
ma
x
B
B
<≤
,
t
he
s
i
t
uat
i
on
i
s
r
el
at
i
v
el
y
s
af
e and a
s
om
ew
hat
c
ons
er
v
at
i
v
e
s
c
hem
e
i
s
adopt
ed.
F
i
r
s
t
,
w
e
det
e
r
m
i
ne
t
he
b
es
t
pos
s
i
bl
e
v
i
d
eo
q
ual
i
t
y
(
l
as
t
R
I
nde
x
k
−
)
b
y
dec
r
eas
i
n
g
bi
t
r
at
e
l
e
v
e
l
i
n
a
l
oo
p
t
o
ens
ur
e
t
h
at
at
l
e
a
s
t
a m
i
ni
m
u
m
buf
f
er
i
s
pr
es
er
v
ed
.
T
hen,
a
bi
t
r
at
e
l
e
v
el
t
hat
i
s
no
hi
gh
er
t
h
an
l
as
t
R
I
nde
x
k
−
is
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
ow
ar
ds
S
moo
t
h a
nd
H
ig
h
-
Q
ual
i
t
y
B
i
t
r
at
e
A
d
apt
a
t
i
on f
or
H
T
T
P
A
da
pt
i
v
e
…
(
Li
hon
g G
eng
)
909
adop
t
ed.
I
f
D
i
f
I
nde
x
i
s
l
es
s
t
han or
equa
l
t
o
c
h
g
I
nde
x
,
i
t
r
e
v
e
al
s
t
h
at
t
h
e
av
a
i
l
ab
l
e t
hr
oug
hput
has
not
c
ha
nge
d
c
ons
i
d
er
ab
l
y
.
I
t
i
s
unn
ec
es
s
ar
y
t
o s
w
i
t
c
h
t
o
b
e
s
t
R
I
nde
x
t
o a
dapt
t
o
t
h
e dec
r
e
a
s
i
ng
t
hr
oug
hpu
t
.
T
her
ef
or
e,
n
e
x
t
R
I
nde
x
i
s
s
et
t
o t
h
e m
i
ni
m
u
m
v
al
u
e be
t
w
ee
n
1
b
e
s
t
R
I
nde
x
+
and
l
as
t
R
I
nde
x
k
−
.
If
D
i
f
I
nde
x
i
s
l
ar
ger
t
h
an
c
h
g
I
nde
x
,
t
he
n
a l
ar
ge
t
hr
oug
hput
f
l
uc
t
uat
i
on
has
oc
c
ur
r
ed.
T
o adapt
t
o t
he
dec
r
eas
i
n
g t
hr
ou
ghp
ut
qu
i
c
k
l
y
and s
af
el
y
,
t
he m
ax
i
m
um
c
hange i
n
i
nd
ex
c
or
r
es
pond
i
ng t
o t
h
e s
af
e v
ar
i
at
i
o
n r
ang
e,
c
h
g
I
nde
x
,
i
s
c
hos
en.
I
n ac
c
or
da
nc
e w
i
t
h t
he
l
i
m
i
t
s
i
m
pos
ed
ab
ov
e,
n
e
x
t
R
I
nde
x
i
s
s
et
e
qua
l
t
o
t
he
m
i
ni
m
u
m
v
al
u
e
b
et
w
e
en
l
as
t
c
hg
R
I
nde
x
I
nde
x
−
and
l
as
t
R
I
nde
x
k
−
.
F
in
a
ll
y
,
if
c
ur
m
ax
BB
>
,
t
he
n
a
m
or
e
c
ons
er
v
at
i
v
e
s
c
hem
e
i
s
adop
t
ed
t
o
r
ed
uc
e
t
he
buf
f
er
.
I
f
c
h
g
D
i
f
I
nde
x
I
nde
x
≤
,
t
he
n
n
e
x
t
R
I
nde
x
i
s
s
i
m
pl
y
s
et
eq
ua
l
t
o t
he
l
as
t
bi
t
r
at
e
l
as
t
R
I
nde
x
t
o
ens
ur
e
a s
m
oot
h
v
i
de
o qua
l
i
t
y
.
I
f
c
h
g
D
i
f
I
nde
x
I
nde
x
>
,
t
he
n
n
e
x
t
R
I
nde
x
i
s
d
ec
r
em
ent
ed as
1
l
as
t
R
I
nde
x
−
,
j
us
t
i
n c
as
e an a
br
upt
c
h
a
nge i
n b
i
t
r
at
e
oc
c
ur
s
w
h
en
t
he buf
f
er
l
ev
el
s
w
i
t
c
hes
f
r
o
m
t
he c
ur
r
ent
buf
f
er
s
t
at
e t
o t
he s
t
at
e
<
mi
n
c
u
r
ma
x
B
B
B
≤
.
I
t
s
hou
l
d
b
e n
ot
e
d t
h
at
n
e
x
t
R
I
nde
x
m
u
s
t
l
i
e
i
n
a r
eas
o
na
bl
e
r
ange
of
[
]
1,
m
i
n our
al
g
or
i
t
hm
.
4
.
E
xp
er
i
m
en
t
s an
d
D
i
scu
ssi
o
n
I
n
t
h
i
s
s
ec
t
i
on,
w
e
g
i
v
e
a
n
ov
er
v
i
e
w
of
t
he
ex
per
i
m
ent
al
m
et
hodo
l
og
y
a
nd
e
v
a
l
u
a
t
e
our
bi
t
r
at
e
ad
apt
at
i
on
a
l
g
or
i
t
h
m
on
our
H
L
S
s
y
s
t
em
.
A
f
t
er
r
epor
t
i
ng
t
he
r
es
u
l
t
s
of
t
he
ex
per
i
m
ent
s
,
w
e pr
es
e
nt
a s
i
m
pl
e di
s
c
us
s
i
on of
our
al
gor
i
t
hm
.
4
.
1.
E
x
p
e
r
i
m
e
n
t S
e
tti
n
g
T
he s
t
r
uc
t
ur
e
of
our
H
LS
s
y
s
t
em
i
s
depi
c
t
ed
i
n
F
i
g
ur
e
1 a
nd
c
ons
i
s
t
s
of
t
hr
ee
c
o
m
ponent
s
,
i
.
e.
,
a H
L
S
s
e
r
v
er
,
a H
LS
c
l
i
e
nt
a
nd
w
i
r
e
d l
oc
a
l
ar
e
a net
w
or
k
(
LA
N
)
.
O
n t
he s
er
v
er
s
i
de,
t
he or
i
gi
na
l
v
i
deo
w
as
enc
oded i
n C
B
R
m
ode t
o pr
oduc
e 20 a
v
a
i
l
ab
l
e bi
t
r
a
t
e v
e
r
si
o
n
s f
r
o
m
100 t
o 20
00 K
bps
w
i
t
h a
s
t
ep of
100
K
bps
.
I
n ad
di
t
i
on,
eac
h
v
er
s
i
o
n
w
as
c
hoppe
d i
nt
o
s
egm
ent
s
of
t
he
s
am
e
dur
at
i
o
n
of
5
s
ec
onds
(
i
.
e.
,
=5
τ
)
.
A
l
l
s
egm
ent
s
and
m
ani
f
es
t
f
i
l
es
w
er
e
s
t
or
ed on
t
he
v
er
s
i
on
2.
4.
9 A
p
ac
he H
T
T
P
s
er
v
er
t
h
at
i
s
i
nt
e
gr
at
e
d i
nt
o
Mac
Mi
n
i
1
0.
10
.
2.
Mor
eo
v
er
,
t
he
s
er
v
er
’
s
T
i
m
eout
w
as
s
e
t
t
o
60
s
f
or
al
i
v
e c
on
nec
t
i
ons
.
O
n
t
h
e
c
l
i
e
nt
s
i
de,
t
he
c
l
i
en
t
w
as
i
m
pl
em
ent
ed on
t
h
e A
ndr
o
i
d 4.
4.
2
p
l
at
f
or
m
.
A
l
l
ada
pt
at
i
o
n
a
l
gor
i
t
hm
s
w
er
e
i
m
pl
em
ent
ed on
t
he c
l
i
ent
.
T
hr
e
e s
egm
ent
s
w
i
t
h t
he
f
our
t
h bi
t
r
at
e
i
nd
ex
w
er
e
b
uf
f
er
ed bef
or
e
t
he s
t
ar
t
of
v
i
d
eo pl
a
y
b
ac
k
.
E
ac
h s
ubs
eq
uent
r
e
ques
t
w
as
s
ent
af
t
er
t
he l
as
t
s
eg
m
ent
had been
c
o
m
pl
et
e
l
y
r
ec
e
i
v
ed.
P
ar
t
i
c
ul
ar
l
y
,
w
h
en t
h
e buf
f
er
l
ev
el
w
ou
l
d b
e l
ar
g
er
t
han t
h
e t
a
r
get
buf
f
er
,
an
i
dl
e de
l
a
y
bef
or
e t
h
e s
end
i
n
g of
t
he nex
t
r
equ
es
t
w
as
s
et
t
o ac
c
ount
f
or
a l
i
m
i
t
ed buf
f
er
c
apac
i
t
y
.
I
n our
ex
per
i
m
ent
s
,
t
h
e t
ar
get
b
uf
f
er
w
as
s
et
t
o
7 s
e
gm
ent
dur
at
i
ons
.
W
i
r
ed LA
N
i
s
b
ui
l
t
b
y
a
T
P
-
LI
N
K
r
out
er
an
d t
her
e
ar
e no ot
h
er
dev
i
c
es
ex
c
ept
a H
LS
s
er
v
er
and a H
L
S
c
l
i
e
nt
i
n t
h
i
s
L
AN
.
Wi
r
e
d
L
A
N
HL
S
S
e
r
v
er
HL
S
C
l
i
en
t
D
um
m
y
N
e
t
F
ig
ur
e
1
.
T
es
t
bed
or
gan
i
z
at
i
on f
or
ex
per
i
m
ent
s
F
or
a f
ai
r
c
o
m
par
i
s
on,
di
f
f
er
ent
al
g
or
i
t
hm
s
s
houl
d be e
v
a
l
uat
ed un
der
t
he s
am
e n
et
w
or
k
c
ondi
t
i
o
ns
.
T
he
D
um
m
y
N
et
net
w
or
k
e
m
ul
at
or
[
18
]
w
as
us
ed
t
o
c
on
t
r
ol
t
he
a
v
a
i
l
ab
l
e
b
and
w
i
dt
h.
T
hi
s
e
m
ul
at
or
i
s
eas
i
l
y
c
o
nf
i
gur
e
d
on
W
i
ndow
s
.
T
o
c
o
nt
r
ol
t
h
e
av
ai
l
ab
l
e
ba
nd
w
i
dt
h
on
t
he
c
l
i
ent
s
i
de
i
n a
s
i
m
pl
e m
anner
,
bot
h
D
um
m
y
N
et
and
our
c
l
i
e
nt
w
er
e
i
ns
t
al
l
ed
on
a
W
i
ndow
s
7
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
04
–
91
5
910
P
r
of
es
s
i
ona
l
des
k
t
op
w
i
t
h
a 3.
1
0 G
H
z
I
nt
el
C
or
e
i
3
C
P
U
a
nd
4 G
B
of
R
A
M.
I
n ad
di
t
i
on
,
an
A
ndr
o
i
d
v
i
r
t
ua
l
m
ac
hi
ne
w
a
s
need
ed t
o r
un
our
c
l
i
en
t
o
n
W
i
ndow
s
.
4
.
2
.
T
h
r
o
u
g
h
p
u
t E
s
ti
m
a
ti
o
n
E
x
p
e
r
i
m
e
n
t
B
ef
or
e our
bi
t
r
at
e ada
pt
at
i
on a
l
gor
i
t
hm
w
as
i
nv
es
t
i
g
at
ed,
o
ur
t
hr
oug
hput
es
t
i
m
at
i
on
m
et
hod w
as
e
v
a
l
ua
t
ed s
e
par
at
e
l
y
i
n t
hi
s
ex
per
i
m
ent
.
T
he t
w
o t
hr
oug
hput
es
t
i
m
at
i
on
m
et
hods
pr
opos
e
d
i
n
[
7
]
an
d
[
9
]
w
er
e
i
m
pl
em
ent
ed
t
o
dem
ons
t
r
at
e
t
he
ef
f
ec
t
i
v
en
es
s
of
ou
r
m
et
hod.
F
or
s
i
m
pl
i
c
i
t
y
,
t
he
m
et
hods
of
[
7
]
an
d
[
9
]
ar
e c
a
l
l
e
d t
h
e D
F
I
m
et
hod
a
nd
t
he
T
hang
m
et
hod,
A
l
gor
i
t
hm
1
B
i
t
r
at
e s
el
e
c
t
i
on
al
g
or
i
t
h
m
I
nput
:
,,
,
l
as
e
c
u
r
t
R
I
nde
x
TT
B
O
u
tp
u
t:
n
e
x
t
R
I
nde
x
1:
b
e
s
t
or
i
l
as
t
R
I
nde
x
R
I
nde
x
D
−
=
2:
n
e
x
t
b
e
s
t
R
I
nde
x
R
I
nde
x
=
3:
i
f
0
or
i
D
≥
th
e
n
4:
i
f
c
u
r
m
i
d
BB
≤
th
e
n
5:
n
e
x
t
l
a
s
t
R
I
nde
x
R
I
nde
x
=
6:
el
se
7:
i
f
c
h
g
D
i
f
I
nde
x
I
nde
x
≥
th
e
n
8:
n
e
x
t
l
a
s
t
c
h
g
R
I
nde
x
R
I
nde
x
I
nde
x
=
+
9:
el
se
10:
i
f
&&
b
es
t
e
c
ur
m
ax
R
I
nde
x
B
R
T
B
≥<
th
e
n
11:
1
n
e
x
t
b
e
s
t
R
I
nde
x
R
I
nde
x
=
+
12:
e
n
d i
f
13:
e
n
d i
f
14:
e
n
d i
f
15:
el
se
16:
i
f
c
u
r
m
i
n
BB
≤
th
e
n
17:
w
h
i
le
b
es
t
R
I
nde
x
RT
>
th
e
n
18:
1
b
e
s
t
b
e
s
t
R
I
nde
x
R
I
nde
x
=
−
19:
e
n
d w
hi
l
e
20:
n
e
x
t
b
e
s
t
R
I
nde
x
R
I
nde
x
=
21:
el
se
i
f
c
ur
m
ax
BB
≤
th
e
n
22:
0
k
=
23:
w
h
i
le
()
(
/
1)
*
la
s
t
R
I
nde
x
k
c
ur
m
i
n
R
T
BB
τ
−
−
>−
th
e
n
24:
1
kk
=
+
25:
e
n
d w
hi
l
e
26:
i
f
c
h
g
D
i
f
I
nde
x
I
nde
x
≤
th
e
n
27:
(
)
1
,
n
e
x
t
l
a
s
t
b
e
s
t
R
I
nde
x
m
i
n
R
I
nde
x
k
R
I
nde
x
−
+
=
28:
el
se
29:
(,
)
ne
x
t
l
as
t
l
as
t
c
hg
R
I
nde
x
m
i
n
R
I
nde
x
k
R
I
nde
x
I
nde
x
=
−−
30:
e
n
d i
f
31:
el
se
32:
i
f
c
h
g
D
i
f
I
nde
x
I
nde
x
≤
th
e
n
33:
n
e
x
t
l
a
s
t
R
I
nde
x
R
I
nde
x
=
34:
el
se
35:
1
n
e
x
t
l
a
s
t
R
I
nde
x
R
I
nde
x
−
=
36:
e
n
d i
f
37:
e
n
d i
f
38:
e
nd i
f
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
ow
ar
ds
S
moo
t
h a
nd
H
ig
h
-
Q
ual
i
t
y
B
i
t
r
at
e
A
d
apt
a
t
i
on f
or
H
T
T
P
A
da
pt
i
v
e
…
(
Li
hon
g G
eng
)
911
r
es
pec
t
i
v
el
y
.
C
om
pl
ex
n
et
w
or
k
c
ondi
t
i
ons
w
i
t
h
bot
h
l
ar
g
e
f
l
uc
t
u
at
i
ons
and
s
h
or
t
-
t
e
rm
f
l
uc
t
uat
i
ons
w
er
e
em
ul
at
ed
us
i
ng D
um
m
y
N
et
.
I
n
our
m
et
hod,
i
n
or
der
t
o
obt
a
i
n
a
s
m
oot
h
es
t
i
m
at
i
on
w
he
n
0.1
ρ
<
and
an
a
ggr
es
s
i
v
e
es
t
i
m
at
i
on w
hen
0.2
ρ
>
,
M
an
d
0
ρ
w
er
e s
et
t
o 2
1
and
0.
1
67,
r
es
pec
t
i
v
e
l
y
.
I
n
pr
ac
t
i
c
e
,
w
e
r
egar
d
t
he
f
l
uc
t
u
at
i
ons
w
hi
c
h ar
e
l
es
s
t
h
an
one
s
t
ep
of
bi
t
r
at
es
(
100
K
bps
i
n o
ur
ex
per
i
m
ent
s
)
as
t
he
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
.
T
her
ef
or
e,
nor
T
w
as
s
et
t
o
10
00
K
b
ps
ac
c
or
di
n
g
t
o
t
h
e
as
s
u
m
pt
i
on
bel
o
w
eq
uat
i
on (
6)
i
n s
ec
t
i
on
3.
1.
mi
n
B
and
ma
x
B
w
er
e s
et
t
o 1 an
d 4 s
egm
ent
dur
a
t
i
ons
,
r
es
pec
t
i
v
el
y
.
T
he
ex
per
i
m
ent
al
s
et
t
i
n
gs
f
or
t
he
D
F
I
met
ho
d
and
t
h
e
T
hang
m
et
h
od
w
er
e
s
et
as
des
c
r
i
bed
i
n
[
7
]
and
[
9
]
.
S
p
ec
i
f
i
c
al
l
y
,
i
n
t
he
D
F
I
m
et
ho
d,
c
w
as
s
et
t
o
0.
1
6
7
f
or
f
ai
r
nes
s
and
ε
w
as
s
et
t
o
0.
0
5 t
o
obt
ai
n
t
h
e opt
i
m
al
r
es
ul
t
s
.
(
a)
E
s
t
i
m
at
ed t
hr
o
ugh
put
(
b)
A
da
pt
ed
b
i
t
r
at
e
(
c
)
B
uf
f
er
l
ev
el
F
i
gur
e
2
.
C
om
par
i
s
on of
di
f
f
er
ent
t
hr
ou
ghp
ut
es
t
i
m
at
i
on m
et
hods
F
ig
ur
e
2
(a
),
F
i
g
ur
e
2(
b)
an
d
F
i
g
ur
e
2(
c
)
c
om
par
e
t
he
r
es
ul
t
s
i
n
t
er
m
s
o
f
t
he
es
t
i
m
at
ed
t
hr
oug
hpu
t
,
ada
pt
e
d bi
t
r
a
t
e
and buf
f
er
l
ev
e
l
,
r
es
pec
t
i
v
el
y
.
N
ot
e t
hat
t
h
e ban
d
w
i
dt
h c
ur
v
e i
n e
ac
h
of
t
he f
ol
l
o
w
i
n
g f
i
gur
es
r
epr
es
ent
s
t
he t
he
or
et
i
c
a
l
c
a
pa
c
i
t
y
of
t
he
l
i
nk
(
c
ont
r
ol
l
ed
b
y
D
um
m
y
N
et
)
;
t
he es
t
i
m
at
ed t
hr
o
ugh
put
s
ar
e t
h
e r
es
ul
t
s
of
t
he
di
f
f
er
ent
t
hr
o
ug
hput
es
t
i
m
at
i
on
m
et
ho
ds
;
t
he
adap
t
ed
b
i
t
r
at
e
i
s
t
he
h
i
g
hes
t
v
a
l
ue
t
h
at
i
s
l
o
w
er
t
han
t
he
es
t
i
m
at
ed t
hr
o
ugh
put
,
s
h
o
w
n
as
equa
t
i
o
n (
12)
;
an
d t
h
e b
uf
f
er
l
ev
e
l
,
w
hi
c
h
i
s
s
ho
w
n
on t
h
e r
i
ght
v
er
t
i
c
al
ax
i
s
i
n F
i
g
ur
e
2
(c
),
r
epr
es
ent
s
t
h
e c
ur
r
ent
buf
f
er
s
t
at
e
i
n s
ec
on
ds
.
T
he f
ol
l
o
w
i
ng
b
eh
av
i
or
s
c
an b
e obs
er
v
ed
f
r
o
m
t
hos
e f
i
gur
es
.
I
n t
he c
as
e
of
l
ar
ge
f
l
uc
t
uat
i
ons
(
e.
g.
,
6
0
-
80 s
and 22
0
-
24
0 s
)
,
t
he es
t
i
m
at
ed t
hr
o
ug
hput
s
of
al
l
m
et
hods
r
es
pon
d
qui
c
k
l
y
as
a
r
es
ul
t
of
t
he
d
y
n
am
i
c
c
ont
r
ol
s
t
r
at
eg
i
es
.
F
or
t
hi
s
r
eas
on,
t
h
e
y
a
l
l
h
a
v
e
r
eas
o
nab
l
y
s
af
e
buf
f
er
s
(
hi
gher
t
h
an
8
s
)
.
H
o
w
e
v
er
,
i
n
t
he
c
as
e
of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
(
e.
g.
,
11
5
-
190
s
and
2
57
-
3
32
s
)
,
t
h
e
t
hr
o
ughp
ut
es
t
i
m
at
ed
b
y
t
he
D
F
I
m
et
hod
v
ar
i
es
f
r
equent
l
y
w
i
t
h
t
h
e
f
l
uc
t
uat
i
ng ba
nd
w
i
dt
h,
w
h
i
c
h c
aus
es
f
l
uc
t
uat
i
o
ns
i
n t
he ad
apt
e
d b
i
t
r
at
es
.
T
he
T
hang m
et
hod
obt
a
i
ns
s
m
oot
h es
t
i
m
at
es
at
hi
gh
ba
nd
w
i
dt
hs
(
e.
g.
,
2
57
-
33
2
s
)
but
f
l
uc
t
uat
i
ng
e
s
t
i
m
at
es
w
hen
t
he
ban
d
w
i
dt
h
i
s
l
o
w
(
e
.
g
.
,
11
5
-
190
s
)
.
B
y
c
o
nt
r
as
t
,
t
he
pr
op
os
ed m
et
hod
o
bt
ai
ns
s
m
oot
h
es
t
i
m
at
es
f
or
al
l
s
h
or
t
-
t
er
m
f
l
uc
t
uat
i
ons
b
y
v
i
r
t
ue
of
t
h
e
ap
pr
opr
i
at
e
des
i
gn
of
t
he
nor
m
al
i
z
at
i
on
f
ac
t
or
nor
T
.
T
her
ef
or
e,
t
he
pr
op
os
ed
m
et
hod
ac
hi
e
v
es
t
h
e
s
m
oot
hes
t
adap
t
ed
b
i
t
r
at
e
.
I
n
s
hor
t
,
0
50
100
150
200
250
300
350
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
T
i
me
(s
)
B
i
t
r
at
e (
K
bps
)
B
an
dw
i
dt
h
D
F
I
M
et
h
od
T
h
an
g M
et
h
od
P
r
opos
ed M
et
h
od
0
50
100
150
200
250
300
350
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
T
i
me
(s
)
B
i
t
r
at
e (
K
bps
)
B
an
dw
i
dt
h
D
F
I
M
et
h
od
T
h
an
g M
et
h
od
P
r
opos
ed M
et
h
od
0
50
100
150
200
250
300
350
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
B
i
t
r
at
e (
K
bps
)
0
50
100
150
200
250
300
350
0
5
10
15
20
25
30
35
40
45
50
55
60
T
i
me
(s
)
B
uf
f
er
l
ev
el
(
s
)
B
an
dw
i
dt
h
D
F
I
M
et
h
od
T
h
an
g M
et
h
od
P
r
opos
ed M
et
h
od
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
04
–
91
5
912
t
he
pr
o
pos
ed
m
et
hod
ac
hi
e
v
es
a
s
m
oot
h
r
es
pons
e
t
o
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
o
ns
and
a
f
as
t
r
es
pons
e t
o
l
ar
ge f
l
uc
t
u
at
i
o
ns
.
4
.
3
.
B
i
tr
a
te
A
d
a
p
ta
ti
o
n
A
l
g
o
r
i
th
m
E
x
p
e
r
i
m
e
n
t
T
he s
ec
ond ex
per
i
m
ent
w
as
c
onduc
t
ed t
o i
n
v
es
t
i
gat
e t
he per
f
or
m
anc
e of
our
bi
t
r
at
e
adap
t
at
i
on
al
gor
i
t
hm
,
i
.
e.
,
t
he b
i
t
r
at
e
s
el
ec
t
i
o
n a
l
gor
i
t
h
m
gi
v
en t
h
e t
hr
oug
hput
es
t
i
m
at
i
on m
et
hod
i
n
v
es
t
i
ga
t
e
d ab
ov
e.
A
s
de
s
c
r
i
bed abo
v
e,
our
b
i
t
r
at
e
s
el
ec
t
i
o
n al
gor
i
t
hm
al
r
ead
y
i
nc
l
u
des
a
s
t
r
at
eg
y
f
or
pr
e
v
e
nt
i
ng
buf
f
er
under
f
l
o
w
.
T
her
ef
or
e,
t
o bes
t
i
l
l
us
t
r
at
e
t
he
per
f
or
m
anc
e of
our
bi
t
r
at
e s
el
ec
t
i
o
n a
l
g
or
i
t
hm
,
t
he p
ar
am
et
er
N
w
as
s
et
t
o 1
f
or
t
hi
s
ex
per
i
m
ent
.
T
he o
t
h
er
s
et
t
i
ngs
w
er
e t
h
e s
am
e as
i
n
t
h
e pr
ev
i
ous
ex
per
i
m
ent
.
F
ur
t
h
er
m
or
e,
i
n our
bi
t
r
a
t
e s
e
l
ec
t
i
o
n a
l
gor
i
t
hm
,
w
e
s
et
mi
n
B
,
mi
d
B
and
ma
x
B
t
o
1.
5,
2
and
6
s
egm
ent
dur
at
i
o
ns
,
r
es
pec
t
i
v
el
y
.
T
he
bi
t
r
at
e
t
hr
es
h
ol
ds
Lo
w
Th
R
,
Mi
d
T
h
R
an
d
Hi
g
T
h
R
w
er
e
s
et
t
o
70
0,
1
00
0 an
d 15
00
K
b
ps
,
r
es
pec
t
i
v
el
y
.
A
no
t
her
b
i
t
r
at
e ad
apt
at
i
o
n al
gor
i
t
hm
w
i
t
h
goo
d p
er
f
or
m
anc
e,
Q
AAD
[
10
]
,
w
hi
c
h
es
t
i
m
at
es
t
hr
ough
put
bas
e
d on s
am
pl
es
of
t
he
do
w
nl
o
ad t
hr
oug
hpu
t
,
w
as
i
m
pl
em
ent
ed f
or
c
o
m
par
i
s
on.
T
he
ex
per
i
m
e
nt
al
s
et
t
i
ngs
w
er
e t
h
e s
am
e as
i
n
[
10
]
.
I
n par
t
i
c
ul
ar
,
t
he pr
edef
i
n
ed
m
ar
gi
nal
b
uf
f
er
l
engt
h
µ
w
as
s
et
t
o
5 s
egm
ent
d
ur
at
i
ons
and
t
he
m
i
ni
m
u
m
buf
f
er
l
engt
h
δ
wa
s
s
et
t
o 1.
5 s
egm
ent
dur
at
i
on
s
f
or
t
hi
s
ex
per
i
m
ent
.
C
om
pl
ex
net
w
or
k
c
ondi
t
i
o
ns
w
i
t
h bot
h l
ar
ge f
l
uc
t
u
a
t
i
ons
an
d s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
w
er
e
em
ul
at
e
d
us
i
ng
D
um
m
y
N
et
.
T
he
r
es
u
l
t
s
i
n
t
er
m
s
of
t
he
a
dapt
ed
bi
t
r
a
t
es
a
nd
b
u
f
f
e
r
le
v
e
ls
ar
e c
om
par
ed i
n F
i
gur
e
3.
T
he a
dap
t
ed
b
i
t
r
at
es
a
r
e t
he
o
ut
pu
t
s
of
t
he
b
i
t
r
at
e
ada
pt
at
i
o
n
al
g
or
i
t
hm
s
.
F
ro
m
F
i
g
u
re
3(
a)
,
w
e c
a
n s
ee t
hat
w
he
n
ban
d
w
i
dt
h
i
s
s
har
pl
y
d
ec
r
eas
i
n
g (
e
.
g.
,
1
86
-
20
6
s
and 3
01
-
32
1 s
)
,
t
he
Q
A
A
D
a
l
gor
i
t
hm
r
es
ul
t
i
n l
ar
g
e c
han
ges
i
n bi
t
r
at
e (
e
.
g.
,
214
-
2
31 s
an
d
342
-
3
58 s
)
bec
a
us
e i
t
i
ni
t
i
al
l
y
at
t
em
pt
s
t
o c
han
ge
q
ual
i
t
y
w
i
t
h t
he s
m
al
l
es
t
p
o
s
s
i
bl
e s
t
e
p i
n
bi
t
r
at
e.
B
y
c
ont
r
as
t
,
t
he
pr
o
pos
ed a
l
g
or
i
t
hm
at
t
em
p
t
s
t
o ens
ur
e t
hat
e
ac
h qu
al
i
t
y
c
hange
i
s
s
af
e.
T
hi
s
m
eas
ur
e
ac
hi
ev
es
s
m
oot
h
er
c
hang
es
i
n
qu
al
i
t
y
.
Mor
eo
v
er
,
t
h
e
pr
op
os
ed
a
l
gor
i
t
hm
has
a
f
as
t
er
r
eac
t
i
on
t
i
m
e i
n
r
ea
c
hi
ng
t
he
o
pt
i
m
al
b
i
t
r
at
e (
e.
g.
,
0
-
48
s
,
2
66
-
3
23 s
a
nd 3
64
-
4
19 s
)
.
T
hus
,
t
he pr
opos
ed
al
gor
i
t
h
m
y
i
el
ds
a hi
gher
a
dapt
ed
bi
t
r
at
e.
F
r
om
F
i
g
ur
e
3(
b)
,
i
t
i
s
ev
i
de
nt
t
ha
t
t
he pr
o
pos
e
d a
l
gor
i
t
hm
ac
hi
e
v
es
a s
af
e b
uf
f
er
.
(
a)
A
da
pt
ed
b
i
t
r
at
e
(
b)
B
uf
f
er
l
ev
e
l
F
ig
ur
e
3
.
C
om
par
i
s
on of
di
f
f
er
ent
bi
t
r
a
t
e a
dap
t
at
i
on
al
g
or
i
t
hm
s
T
o
m
or
e c
l
ear
l
y
i
l
l
us
t
r
a
t
e
t
he a
dv
ant
ages
of
t
he pr
opos
ed
al
gor
i
t
hm
,
f
i
v
e ad
di
t
i
on
al
ex
per
i
m
ent
s
w
er
e
c
on
duc
t
ed.
W
e ev
al
u
at
e
d
t
h
e p
er
f
or
m
anc
e
bas
ed
on
t
he
a
v
e
r
age
v
a
l
ues
of
t
he
f
i
v
e
s
et
s
of
r
es
ul
t
s
.
S
ev
er
a
l
s
t
at
i
s
t
i
c
s
r
egar
d
i
ng
t
h
e
t
es
t
ed
al
gor
i
t
hm
s
ar
e
pr
o
v
i
ded
i
n
T
abl
e
1.
T
he bi
t
r
at
e
v
a
l
u
es
ar
e s
ho
w
n
i
n t
h
e f
i
r
s
t
t
hr
ee
r
o
w
s
i
n uni
t
s
of
K
bps
.
T
he ne
x
t
t
hr
ee r
o
w
s
c
onc
er
n
t
he
b
uf
f
er
v
al
ues
,
and
t
h
e
r
es
ul
t
s
i
n
t
er
m
s
o
f
qua
l
i
t
y
c
han
ges
ar
e
s
ho
w
n
i
n
t
he
l
as
t
t
w
o
r
ow
s
.
T
he
“
Max
i
m
u
m
c
hange
i
n
b
i
t
r
at
e”
i
n
t
he
t
hi
r
d
r
o
w
i
s
t
h
e
l
ar
g
es
t
bi
t
r
at
e
di
f
f
er
enc
e
bet
w
een
an
y
t
w
o c
ons
ec
ut
i
v
e s
egm
ent
s
.
T
he s
t
andar
d
de
v
i
a
t
i
on (
S
T
D
)
i
n t
he s
ec
o
nd a
nd f
i
f
t
h r
ow
s
i
s
us
ed t
o qu
ant
i
f
y
t
h
e am
oun
t
of
v
ar
i
at
i
on
of
t
he
bi
t
r
at
e
a
nd b
uf
f
er
.
T
he “
N
um
ber
of
uns
af
e qu
al
i
t
y
0
50
100
150
200
250
300
350
400
450
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
T
i
me
(s
)
B
i
t
r
at
e (
K
bps
)
B
an
dw
i
dt
h
Q
AAD
M
e
t
h
o
d
P
r
opos
ed M
et
h
od
0
50
100
150
200
250
300
350
400
450
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
B
i
t
r
at
e (
K
bps
)
0
50
100
150
200
250
300
350
400
450
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
T
i
me
(s
)
B
uf
f
er
l
ev
el
(
s
)
B
an
dw
i
dt
h
Q
AAD
M
e
t
h
o
d
P
r
opos
ed M
et
h
od
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
ow
ar
ds
S
moo
t
h a
nd
H
ig
h
-
Q
ual
i
t
y
B
i
t
r
at
e
A
d
apt
a
t
i
on f
or
H
T
T
P
A
da
pt
i
v
e
…
(
Li
hon
g G
eng
)
913
c
hanges
”
r
ep
or
t
ed
i
n
t
he
l
as
t
r
o
w
r
ef
er
s
t
o c
hang
es
i
n q
ua
l
i
t
y
t
hat
ar
e out
s
i
d
e of
t
he s
af
e
v
ar
i
at
i
on r
a
nge
of
t
he c
ur
r
e
nt
b
i
t
r
at
e.
T
abl
e 1.
S
ta
ti
s
ti
c
s
o
f
A
v
er
a
ge V
al
ues
o
f
F
i
v
e
E
x
per
i
m
ent
s
S
ta
t
i
s
t
i
c
s
Q
AAD
P
r
opos
ed al
gor
i
t
h
m
I
m
pr
ov
e
m
ent
r
a
t
e
A
v
er
age bi
t
r
at
e (
K
bps
)
1122
1232
9.
8%
S
T
D
o
f b
i
tr
a
te
s
(
K
b
/
s
)
532
450
15.
4%
M
a
x
i
m
u
m
c
hange
i
n
bi
t
r
a
t
e (
K
bp
s
)
937
408
56.
5%
A
v
er
age buf
f
er
l
ev
el
(
s
)
29.
8
21.
7
-
27.
2%
S
T
D
o
f
buf
f
er
l
ev
el
(
s
)
7.
78
7.
01
9.
9%
N
um
ber
o
f
i
n
t
er
r
upt
i
ons
0
0
0%
N
um
ber
o
f
qual
i
t
y
c
hanges
52.
6
31.
4
40.
3%
N
um
ber
o
f
un
s
af
e
qual
i
t
y
c
hanges
2.
6
2
23.
1%
F
ro
m
t
he
s
t
at
i
s
t
i
c
s
,
w
e c
a
n s
ee t
ha
t
t
he Q
A
A
D
al
g
or
i
t
hm
has
a hi
g
her
buf
f
er
l
ev
el
bec
aus
e
of
i
t
s
c
ons
er
v
at
i
v
e
s
t
r
at
eg
y
an
d
t
hat
ne
i
t
h
er
al
gor
i
t
hm
s
u
f
f
er
s
an
y
i
nt
er
r
u
pt
i
o
ns
.
I
n
f
ac
t
,
a hi
gh
er
buf
f
er
w
i
t
h no i
nt
er
r
upt
i
ons
does
not
pr
o
v
i
d
e a bet
t
er
Q
oE
f
or
t
he us
er
.
B
ec
aus
e t
he
us
er
w
i
l
l
no
t
n
ot
i
c
e
a
c
han
g
e i
n t
he
buf
f
er
l
ev
el
un
l
es
s
t
he
buf
f
er
i
s
ex
ha
us
t
ed.
A
s
ex
pec
t
ed,
t
h
e
pr
opos
e
d a
l
gor
i
t
hm
r
es
ul
t
s
i
n f
e
w
er
(
uns
af
e)
qu
al
i
t
y
c
ha
nges
,
s
m
al
l
er
m
ax
i
m
u
m
c
hange
i
n
bi
t
r
at
e
and
l
o
w
er
S
T
D
of
bi
t
r
at
es
,
w
hi
c
h
i
m
pl
i
es
t
hat
o
ur
bi
t
r
at
e
a
dapt
at
i
on
a
l
gor
i
t
hm
c
an
obt
ai
n
s
m
oot
her
v
i
deo
qua
l
i
t
y
.
F
ur
t
her
m
or
e,
t
he pr
opos
e
d al
g
or
i
t
hm
ac
hi
ev
es
a hi
gher
a
v
er
ag
e b
i
t
r
at
e.
A
l
l
of
t
h
es
e r
es
u
l
t
s
c
onf
i
r
m
t
hat
t
he
pr
op
os
ed
al
gor
i
t
hm
ac
hi
ev
es
a c
o
ns
i
d
er
ab
l
e
a
dv
a
nt
a
ge
ov
er
t
he Q
A
A
D
al
gor
i
t
hm
.
4
.
4
.
D
i
scu
s
si
o
n
A
s
m
ent
i
one
d
a
bo
v
e,
a t
hr
ough
put
es
t
i
m
at
i
on
m
et
hod s
hou
l
d
be
s
t
abl
e
i
n
t
he
c
as
e
of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
w
hi
l
e a
l
s
o r
eac
t
i
n
g q
ui
c
k
l
y
t
o
l
a
r
ge f
l
uc
t
ua
t
i
o
ns
.
T
he m
et
ho
d pr
o
pos
ed
i
n
[
6
]
ac
h
i
e
v
es
s
m
oot
h es
t
i
m
at
i
on f
or
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
b
ut
f
ai
l
s
t
o
c
o
pe
w
i
t
h
l
ar
ge
f
l
uc
t
uat
i
ons
.
T
he D
F
I
m
et
hod an
d t
he T
han
g m
et
hod
r
es
pond
q
ui
c
k
l
y
t
o
l
ar
ge f
l
uc
t
uat
i
ons
b
ut
y
i
e
ld
f
lu
c
t
uat
i
ng
es
t
i
m
at
es
i
n
t
he
c
as
e
of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
o
ns
.
T
he
r
es
ul
t
s
of
t
he
f
i
r
s
t
ex
per
i
m
ent
s
ho
w
t
hat
o
ur
t
hr
ough
put
es
t
i
m
at
i
on
m
et
ho
d
beh
av
es
w
el
l
f
or
bot
h
s
hor
t
-
t
e
rm
f
l
uc
t
uat
i
ons
an
d l
ar
ge f
l
uc
t
u
at
i
o
ns
.
I
n our
t
hr
oug
hpu
t
es
t
i
m
at
i
on m
et
hod,
nor
T
i
s
dec
i
d
ed b
y
t
h
e s
t
ep of
b
i
t
r
at
es
.
A
l
l
t
he
s
t
eps
ar
e
equ
al
i
n
o
ur
ex
p
er
i
m
ent
s
.
W
hen
t
he
s
t
eps
ar
e
n
ot
e
qua
l
,
t
h
i
s
s
i
t
u
at
i
o
n
ne
eds
f
ur
t
her
r
es
ear
c
h;
f
or
ex
am
pl
e,
t
h
e s
m
al
l
es
t
s
t
ep c
a
n b
e c
hos
en t
o d
ec
i
d
e
nor
T
.
B
es
i
de
s
,
f
r
o
m
t
he
obs
er
v
a
t
i
ons
,
w
e no
t
e t
ha
t
t
he nor
m
al
i
z
at
i
on f
ac
t
or
nor
T
pl
a
y
s
a
n i
m
por
t
ant
r
ol
e i
n t
h
e
es
t
i
m
at
i
on
of
s
hor
t
-
t
er
m
f
l
uc
t
uat
i
ons
and t
he p
ar
am
et
er
M
pl
a
y
s
an i
m
por
t
ant
r
ol
e i
n t
he e
s
t
i
m
at
i
on o
f
l
ar
ge f
l
uc
t
ua
t
i
o
ns
.
W
i
t
h
f
ur
t
her
r
es
ear
c
h,
a d
y
nam
i
c
s
c
hem
e
c
an be des
i
gn
ed f
or
det
er
m
i
ni
ng
nor
T
and
M
bas
e
d on
t
he
c
har
ac
t
er
i
s
t
i
c
s
of
di
f
f
er
ent
net
w
or
k
s
.
T
he al
gor
i
t
hm
pr
opos
ed i
n
[
19
]
at
t
em
pt
s
t
o
m
ai
nt
ai
n a
s
t
abl
e b
i
t
r
at
e t
o ens
ur
e a s
m
oot
h
bi
t
r
at
e.
H
o
w
e
v
er
,
t
hi
s
s
t
r
at
eg
y
c
aus
es
e
v
e
nt
ua
l
abr
up
t
c
hang
es
i
n
b
i
t
r
at
e
i
n
t
h
e
c
as
e of
l
ar
ge,
r
api
d d
ec
r
eas
es
i
n t
h
e av
a
i
l
a
bl
e t
hr
ou
ghp
ut
.
W
hen
t
he av
a
i
l
ab
l
e t
hr
o
ugh
put
i
s
i
nc
r
eas
i
ng
,
t
he
Q
A
A
D
a
l
gor
i
t
hm
c
hanges
t
he qua
l
i
t
y
w
i
t
h t
h
e s
m
al
l
es
t
pos
s
i
bl
e s
t
ep t
o ac
h
i
e
v
e a
s
m
oot
h bi
t
r
at
e.
H
o
w
e
v
er
,
t
hi
s
s
t
r
at
eg
y
do
e
s
not
t
ak
e
f
ul
l
ad
v
a
nt
ag
e
o
f
t
he
a
v
a
i
l
ab
l
e ban
d
w
i
dt
h.
T
he
pr
opos
e
d
bi
t
r
at
e ada
pt
at
i
o
n al
g
or
i
t
h
m
c
hanges
t
he qua
l
i
t
y
w
i
t
hi
n t
h
e s
af
e v
ar
i
at
i
o
n r
ang
e of
t
he c
ur
r
ent
bi
t
r
at
e.
T
hi
s
s
t
r
at
eg
y
av
oi
d
s
abr
upt
c
h
an
ge
i
n
bi
t
r
at
e
and
get
s
a
hi
gh
ban
d
w
i
dt
h
ut
i
l
i
z
at
i
on
.
T
he
r
es
ul
t
s
of
t
he
s
ec
ond
ex
per
i
m
ent
i
nd
i
c
at
e
t
ha
t
our
al
go
r
i
t
hm
c
an
get
h
i
g
her
a
v
er
a
ge
q
ua
l
i
t
y
a
nd
s
m
oot
her
v
i
d
eo
qu
al
i
t
y
.
I
t
s
houl
d b
e
not
e
d t
hat
t
h
e t
hr
es
hol
ds
us
e
d t
o d
ef
i
ne
our
p
i
ec
e
w
i
s
e
f
unc
t
i
on
ar
e
s
ui
t
abl
e
f
or
s
l
ow
-
m
ot
i
on
[
16
]
v
i
deos
an
d
s
om
e
f
as
t
-
m
ot
i
on
v
i
de
os
.
T
h
us
,
f
or
v
i
deos
w
i
t
h f
as
t
er
m
ot
i
on,
t
hes
e
t
h
r
es
hol
ds
s
h
oul
d b
e r
es
et
b
as
ed o
n t
h
e J
N
D
.
I
n m
os
t
bi
t
r
at
e
ad
apt
at
i
on
al
gor
i
t
hm
s
,
m
di
f
f
er
ent
v
i
d
eo q
ua
l
i
t
i
es
m
us
t
i
ne
v
i
t
ab
l
y
b
e
t
r
av
er
s
ed
t
o
s
el
ec
t
t
h
e
bes
t
bi
t
r
at
e
(
b
es
t
R
I
nde
x
R
)
.
T
hi
s
i
s
al
s
o
t
r
ue
of
our
al
gor
i
t
hm
.
T
her
ef
o
r
e,
our
a
lg
o
r
it
h
m
’s
t
im
e
c
o
m
p
le
x
it
y
is
()
Om
,
i
den
t
i
c
a
l
t
o
t
hat
of
t
he
Q
A
A
D
al
gor
i
t
hm
.
I
n t
hi
s
p
aper
,
w
e f
oc
us
ed p
ur
el
y
on
i
m
pr
ov
i
ng t
he p
er
f
or
m
anc
e of
our
bi
t
r
at
e
ada
pt
at
i
on
al
g
or
i
t
hm
f
or
a s
i
ngl
e c
l
i
e
n
t
.
H
o
w
e
v
er
,
t
he c
as
e
of
m
ul
t
i
pl
e c
l
i
ent
s
i
s
bot
h m
or
e pr
ac
t
i
c
al
and
m
or
e c
hal
l
e
ng
i
ng
[
20
]
.
T
he ques
t
i
on of
ho
w
t
o ac
h
i
e
v
e a ba
l
a
nc
e bet
w
een
ef
f
i
c
i
enc
y
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.