TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.6, Jun
e
201
4, pp. 4740 ~ 4
7
4
6
DOI: 10.115
9
1
/telkomni
ka.
v
12i6.549
4
4740
Re
cei
v
ed
De
cem
ber 2
9
, 2013; Re
vi
sed
March 8, 201
4; Acce
pted
March 22, 20
14
Investigations of Wireless Sensor Networks for Indoor
Particulate Matter Monitoring
Heng Lu
o
1
*, Ai Huan
g Gu
o
2
, Jianping Chen
3
, Yu Ta
ng
4
, Weizho
ng Yu
5
, Yafei Ji
6
1,3,
4,5
Suzhou ke
y la
bor
ator
y
of mobil
e
net
w
o
rk
ing a
nd a
ppl
ie
d
technol
og
ies,
SuZ
hou U
n
iver
sit
y
of Scie
nce
and T
e
chno
log
y
, Chi
n
a
2
School of Elec
tronic an
d Infor
m
ation En
gi
ne
erin
g, T
ongji Universit
y
, Ch
in
a
*Corres
p
o
ndi
n
g
author; em
ail
:
luohe
ng
198
1
@
16
3.com
1
, tjgah
@mai
l.tong
ji.ed
u.cn
2
,
ala
n
jpc
h
e
n
@
y
aho
o.com
3
, usts
y
t
o
ng@s
i
na.c
o
m
4
, usts
w
z
yu
@sina.c
o
m
5
, ustsy
f
ji@ sohu.com
6
A
b
st
r
a
ct
Precise
me
asu
r
ement of i
ndo
or ma
ss co
nce
n
tration
of parti
culate
matter i
s
criticial
l
y i
m
p
o
rtant fo
r
the he
alth r
i
sk
eval
uatio
n si
nc
e mod
e
rn
peo
ple s
p
e
nd
mor
e
than
90
% of
their l
i
fe in
do
or
s. F
o
r the sake
of
accuracy, l
ong
-term
mo
nitori
ng syste
m
s sh
oul
d be
de
ploy
ed a
m
ong w
h
i
c
h w
i
reless s
e
nsor n
e
tw
ork is
a
soun
d sol
u
tio
n
.
How
e
ver, most of the w
i
reless s
ens
or n
e
tw
orks are b
a
ttery pow
ered
and th
us en
e
r
gy
ma
na
ge
me
nt s
c
he
me s
h
o
u
ld
be i
m
ple
m
ent
ed to pr
ol
ong
the lifeti
m
e of
the w
hole
net
w
o
rk. Meanw
hile
,
sample sites must be select
e
d
carefully to a
v
oid the resu
lts bias.
In
this pap
er,
the i
m
p
o
rtance
of s
a
mp
l
e
interva
l
as
w
e
l
l
as s
a
mpl
e
l
o
cations
is i
n
ve
stigated
the
o
r
e
tically
a
nd
pr
actically.
Res
u
lts show
that
by
ado
ptin
g efficic
ent p
o
w
e
r
ma
n
age
ment sc
he
me,
more
tha
n
67%
of ener
gy can be
save
d. F
i
nally,
metho
d
s
of sample i
n
ter
v
al confi
gurati
o
n as w
e
ll as
sa
mp
le sites se
le
ction are pr
op
o
s
ed.
Ke
y
w
ords
:
particulate
m
a
tter
pollutants,
wireless sensor network,
sample
interval, samp
le
l
o
ca
ti
o
n
,
ener
gy red
u
ction
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Extensive ep
idemiolo
gical
studi
es hav
e p
r
ov
ided
stron
g
evide
n
ce
of
a
s
so
ciation
s
betwe
en in
do
or p
a
rti
c
ulate
matters a
n
d
adverse
hu
man h
ealth,
esp
e
ci
ally for young
chil
dren,
becau
se a la
rge p
a
rt of lifetime, excee
d
ing
90%, of
mode
rn pe
o
p
le wa
s
spe
n
t indoo
rs. T
h
e
results in [1]
showed cl
ose relationshi
p
bet
ween
i
n
door particul
ate
matters and
the risks
for
respiratory d
i
sea
s
e
s
in y
oung
child
re
n. Other
stu
d
ies al
so
de
monst
r
ated t
he conne
ctio
n
betwe
en p
a
rt
iculate
matte
rs
and
pulm
onary i
n
jury
[2], neuro
d
e
gene
rative di
sorde
r
s [3]
a
nd
cardiova
scula
r
dise
ase [4].
Great effort
s, therefore,
have been
devoted to study t
he long-te
rm vari
ation of
con
c
e
n
tration
of particu
late matters
indoors
[5
]
[6], the relation
ship
betwe
en ind
o
o
r an
d outdo
or
PM polluta
nts [7
]
[8], spatial
variation of P
M
pollutants [
9
]
and factors influencing
v
a
riability in the
infiltration of PM pollutants and its comp
onent
s [10].
The re
sult
s obtained in mu
ch literatu
r
e, it is
observed, u
s
e long
-term
sample
s befo
r
e whi
c
h
a
wi
reless sen
s
o
r
netwo
rk sh
ou
ld
be
d
eploye
d
to avoid destroying origi
n
al room stru
cture
s
an
d reco
rd the d
ensity of particulate matter
pollutant
s wit
hout the ca
re
of human bei
ngs
while mai
n
taining a
c
cu
racy.
Ho
wever, m
o
st of wi
rele
ss sen
s
o
r
n
e
tworks
are
batt
e
ry po
we
red
due to th
e co
nstrai
nts
of sampl
e
lo
cation
s an
d
absen
ce of p
r
efixed po
we
r infrast
r
uctu
re. As a co
nseque
nce, en
ergy
con
s
e
r
vation
scheme
s
sh
o
u
ld be i
m
ple
m
ented to
pr
olong th
e lifetime of sen
s
or netwo
rk so that
more
sample
s
can
be
coll
ected
autom
a
t
ically. Mean
while,
sam
p
le
sit
e
s sele
ction is al
so
crucial
sin
c
e
ran
d
o
m
sa
mple
lo
cation
s m
a
y
lead to
up to
20% diffe
re
nce
amo
ng fi
nal me
asure
m
ent
res
u
lts
Error
!
Re
ference s
o
urce not found.
.
In this pa
pe
r, the proble
m
s of en
erg
y
redu
ction and sampl
e
sites sel
e
ct
ion
are
investigate
d
carefuuly foll
owe
d
by p
r
opo
sing
an
energy savin
g
sche
me b
a
sei
ng o
n
t
h
e
observation t
hat the difference betwe
e
n
two nei
gh
borin
g sam
p
l
i
ng re
sults i
s
marginal. T
h
e
pape
r is
org
anized a
s
fo
llows. Sectio
n 2 de
sc
ribl
es the
experiments tool
s and the
o
ret
i
cal
simulato
r. Th
e third p
a
rt
discu
s
ses th
e experi
m
ent
and
simulati
on re
sult
s an
d estimate
s t
he
energy red
u
ct
ion level by using e
nergy saving sc
hem
es. The final
se
ction con
c
l
ude
s this pa
p
e
r.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Inv
e
s
t
igations of Wireless
Sens
or Networks
fo
r Ind
o
o
r
Particulate Matter Monito
ring (He
ng Lu
o)
4741
2. Rese
arch
Metho
d
Practi
cal m
e
a
s
ureme
n
t an
d theo
retical
analysi
s
are
combi
ned
to i
n
vestigate
th
e issu
es
in particulate
matter wi
rele
ss
sen
s
o
r
net
works indo
ors.
2.1. Samplin
g Equipment
The dire
ct re
ading
monito
ring device, Dylos
Air Qu
ality
Monitors
(Mo
del DC1
700, with
external
dime
nsio
ns of 17.
78mm*1
1
.43
mm*7.62mm
)
we
re
used in
this p
ape
r. It dep
end
s a
li
ght
scattering te
chniqu
e to determin
e
the d
ensity of
airb
orne m
a
tter
with 2 si
ze ra
nge
s sm
all (with
diamete
r
> 0.
5
μ
m) an
d large (dia
meter
> 2.5
μ
m). An
air sam
p
le is cont
in
uo
us
ly d
r
aw
n
in
to
th
e
instru
ment by
a pump. The
incomi
ng air
passe
s th
rou
gh a la
ser
be
am in a ph
otometer a
nd t
he
den
sity of particulate matte
r is displayed
and re
co
rd
ed
.
Figure 1. Sample Devi
ce
2.2. Samplin
g Locatio
ns
Five sa
mplin
g lo
cation
s in
the sch
ool
campu
s
a
r
e
chosen a
nd t
w
o are
sho
w
n i
n
Figu
re
2. The buil
d
i
ng was
con
s
t
r
ucte
d in 2
0
0
0
and the
r
e i
s
n’t any ai
r-condition
er
system in all fi
ve
locat
i
o
n
s.
(a)
Cla
s
sroo
m #1 (18m
*9
m)
(b)
Cla
s
sroo
m #2 (6m*
9m
)
Figure 2. Two
Sampling Lo
cation
s
2.3. Data
Col
l
ection
The inst
rume
nts we
re op
erated fo
r 47
days from 2
September
2013 to 18
Octob
e
r
2013. T
he
sa
mpling p
e
ri
od
s a
r
e 1
2
ho
urs from
8 am
to 10 p
m
. The
sam
p
ling i
n
terval i
s
1 mi
n
u
te
originally.
2.4. Theore
t
ical Model-Computa
t
iona
l Fluid D
y
namics (CF
D
)
Beside
s exp
e
r
imental m
e
a
s
ureme
n
ts, a
theor
eti
c
al
model, comp
utation fluid
dynamics
model, i
s
em
ployed. By solving a
colle
ction of
parti
al differe
ntial equatio
ns
n
u
meri
cally fo
r the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4740 – 4
746
4742
con
s
e
r
vation
of ma
ss,
ene
rgy a
nd tu
rbu
l
ence q
uantit
i
e
s,
CF
D mo
d
e
l is abl
e to
p
r
ovide t
he fiel
d
distrib
u
tion
s
of co
ntamina
n
ts. Despite
of so
me
un
certaintie
s in
t
he m
odel
the
CF
D m
odel
has
become mo
re
and more po
pular
with the
developme
n
t of computing
capa
city.
3. Results a
nd Analy
s
is
Both experim
ental re
sults
and theo
retical simulatio
n
results a
r
e an
alyzed in thi
s
se
ction.
3.1. Sample Interv
al
(1) Ex
pe
rime
nt result
s
For the sake of spa
c
e limitations, only two
sets of measurin
g sam
p
les a
r
e de
pi
cted in
Figure 3. Ho
wever, othe
r
sampl
e
s h
a
ve simila
r perf
o
rma
n
ce.
As seen
in Fi
gure
3(a), th
e ma
ss con
c
entrati
on
of P
M
2.5 de
crea
sed
slig
htly with time.
Ho
wever, a
n
unexpe
cted
swift growth
at the fi
nal stage i
s
ob
served
sin
c
e t
hat lectu
r
e
was
ende
d 3 min
u
tes a
hea
d of sched
uled
time. Like
wi
se, the d
e
n
s
i
t
y of PM2.5 in Figu
re 3
0
(b)
decli
ned
with time. Neverth
e
less, it decreased
mu
ch
quicke
r
than that in Figure 3(a
)
.
Mean
while, it is observed
that the mass
con
c
e
n
tra
t
ion in Figure 3
0
(b) i
s
hi
gher o
n
averag
e com
pare
d
to that
in 0(a
)
. Th
e
main
cau
s
e
is that the
wind
ow i
s
sh
ut down whe
n
temperature
decrea
s
e
s
an
d therefo
r
e th
e excha
ge rat
e
betwe
en in
door a
nd out
door ai
r is
sm
all.
(a) Sampl
e
s
in 8:55 ~9:4
0 on 17 Septe
m
ber, 20
13 (23
Ԩ
~ 3
2
Ԩ
, Sunny)
(b) Sampl
e
s
in 8:55 ~9:4
0 on 8 Octo
ber,
2013 (2
0
Ԩ
~ 23
Ԩ
, Rainny)
Figure 3. Samples in a Pe
riod of 45 min
u
tes in
SuZh
ou Unive
r
sity of Science an
d Tech
nolo
g
y
(2) T
heo
retic
a
l simulatio
n
res
u
lts
CFD m
odel is used to sim
u
late the temporal
di
stributi
on of PM2.
5. The average
outdoo
r
den
sity of PM2.5 is confi
gure
d
to
75
μ
g/m
3
, which i
s
a
n
ave
r
ag
e
mass
co
nce
n
tration i
n
Ch
ina.
As see
n
in Fi
gure 4 the d
e
n
sity of PM2.5 varies
with time. However, the difference is not larg
e.
P
M
2.
5 t
e
m
p
o
r
a
l
di
s
t
r
i
bu
t
i
on
10
20
30
40
0
5
1
0
15
2
0
25
30
3
5
40
Ti
m
e
(
m
i
n
)
D
ens
i
t
y
(u
g/
m
3
)
P
M
2
.
5 t
e
m
p
or
a
l
d
i
s
t
r
i
bu
t
i
on
35
45
55
0
5
1
0
15
20
25
30
35
40
Ti
m
e
(
m
i
n
)
De
n
s
i
t
y
(
u
g
/
m
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Inv
e
s
t
igations of Wireless
Sens
or Networks
fo
r Ind
o
o
r
Particulate Matter Monito
ring (He
ng Lu
o)
4743
(a) 3 minute
s
after initial stage
(b) 5 minute
s
after initial stage
(c) 7 minute
s
after initial stage
Figure 4. Simulation Results by CF
D Mo
del
As see
n
in Fi
gure 4, the dif
f
eren
ce all three figure
s
is
margi
nal.
(3) Dis
c
u
s
sio
n
(a) Sam
p
ling
interval optim
ization
As con
c
lu
de
d from Figure 3, the con
c
entrati
on of
particul
a
te matters i
s
temperature
related. It is obse
r
ved, ho
wever, that the di
fference be
tween two n
e
i
ghbo
ring
sam
p
ling re
sult
s is
margi
nal. As
a con
s
e
que
n
c
e, the sa
mpl
i
ng interval
, it is sug
g
e
s
ted
,
should b
e
prolong
ed so that
much m
o
re e
nergy can be
saved.
Different
sam
p
ling inte
rval
s a
r
e em
ploy
ed to
test the
reliability of t
he propo
se
d
energy
con
s
e
r
vation
scheme
s
a
n
d
the re
sults
are
sho
w
n
in
Figure
5 wh
ere, for exa
m
ple, interval
_1
mean
s the sampling inte
rval equal
s 1
minute. As
d
epicte
d
, the differen
c
e b
e
t
ween
“interv
a
l_1”
and
“interval
_3”
are
ma
rginal. However, the
diffe
rence be
co
m
e
s o
b
viou
s
as the
interval
increa
se
s to 5 minutes. Th
erefo
r
e, the inte
rval, in this
c
a
s
e
, is
s
e
t to 3 minutes
.
Figure 5. Samples
with Di
fferent Interv
als in 8:55
~9
:40 on 17 O
c
tober, 20
13 (1
8
Ԩ
~ 22
Ԩ
,
Clou
dy)
P
M
2.
5 t
e
m
por
al
di
s
t
r
i
but
i
o
n
40
44
48
52
0
5
10
15
2
0
25
30
35
40
45
Ti
m
e
(
m
i
n
)
D
ens
i
t
y
(ug/
m
3
)
i
n
t
e
r
v
al
_1
i
n
t
e
r
v
al
_3
i
n
t
e
r
v
al
_5
i
n
t
e
r
v
al
_9
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4740 – 4
746
4744
(b)Tra
nsmi
ssi
on interval op
timization
Figure 6
d
e
scribe
s
th
e p
o
wer
co
nsumpti
on of
the
wh
o
l
e PM
se
nsor network. A
s
sho
w
n,
transmissio
n
energy cost
cou
n
ts fo
r a
bout 4
2
% of
the total e
n
ergy
con
s
um
ption. The
r
ef
ore,
more e
nergy may be save
d by extendin
g
the sampli
n
g
interval.
Figure 6. Power
Con
s
u
m
ption of DC17
00 (PM2.5 m
onitor) and
CC25
30 (TX d
e
vice)
(c) Total ene
rgy redu
ction
The total red
u
ce
d ene
rgy is estim
a
ted via:
=+
__
s
a
m
p
l
e
_
PP
P
re
duc
e
d
all
r
e
duc
ed
red
u
c
e
d
T
X
(1)
Whe
r
e
P
reduced_all
is the
total powe
r
redu
ction,
P
r
educed
_sa
m
ple
and
P
reduced_
TX
denote p
o
w
er
redu
ction by
sampli
ng inte
rval increa
se
as we
ll as
less
trans
m
is
s
i
on times
res
pec
tively.
Table 1. Esti
mated Energ
y
Cost for 3
minutes
sampling pow
er
cost (
m
w)
T
X
pow
e
r
cost (mw)
T
o
tal pow
e
r
cost (
m
w
)
interval_1
1.4 × 3
1 × 3
7.2
interval_3
1.4 × 1
1 × 1
2.4
Table
1 item
ize
s
the
po
wer
con
s
u
m
pti
on of
sam
p
li
ng a
nd tran
smissi
on
part
usi
n
g
Eqation (1).
As seen, th
e
total po
we
r co
nsu
m
pt
io
n of inte
rval
_1 tripl
e
s th
at of interval
_3
,
demon
stratin
g
the efficien
cy of the prop
ose
d
ene
rgy saving me
ch
anism the
o
retically.
3.2. Samplin
g Sites
Beside
s sam
p
le interval, sample lo
catio
n
s may also
have gre
a
t impact on the
accu
ra
cy
of the monitoring re
sults.
(1) Ex
pe
rime
nt result
s
For the sake of spa
c
e limitations, only two
sets of measurin
g sam
p
les a
r
e de
pi
cted in
Figure 7. Ho
wever, othe
r
sampl
e
s h
a
ve simila
r perf
o
rma
n
ce.
As sh
own, the mass
con
c
entration
of PM2.5 in
ba
ck door i
s
hi
ghe
r than that in the front
door si
nce t
he ba
ck d
o
o
r
was al
way
s
o
pen
befo
r
e
cla
ss.
On
e mo
re findi
ng i
s
that t
h
e
con
c
e
n
tration
of PM was tempe
r
ature related an
d
it wa
s high
er in
cold day
s. Last but not th
e
least, the
diff
eren
ce
of
co
nce
n
tration
b
e
twee
n
the
front
do
or
an
d back doo
r sa
mples wa
s
la
rge
at the initial
sampling
sta
g
e
after
whi
c
h
those tw
o lin
es
overla
ppe
d with
time u
n
til rea
c
hi
ng t
h
e
same value finally.
Po
w
e
r
c
o
s
t
58
%
42
%
DC
170
0
C
C
2530
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Inv
e
s
t
igations of Wireless
Sens
or Networks
fo
r Ind
o
o
r
Particulate Matter Monito
ring (He
ng Lu
o)
4745
(a) Sam
p
le
s in 8:55 ~ 9:40
on 17 Septe
m
ber, 20
13 (23
℃
~ 3
2
℃
, Sunny)
(b) Sam
p
le
s in 8:55 ~9:4
0 on 16 O
c
tobe
r, 2013 (13
℃
~
20
℃
, Cl
ou
dy)
Figure 7. PM2.5 Sample
s
(2) T
heo
retic
a
l simulatio
n
res
u
lts
Mass con
c
en
tration of ind
oor p
a
rticulat
e ma
tter in two site
s a
r
e
evaluated. O
ne site i
s
sele
cted
ne
ar the ve
ntilatio
n
site
while
th
e othe
r
i
s
lo
cated in
the
ro
om. The
resul
t
s a
r
e
sh
own
in
Figure 8. A
s
observed, th
e
differen
c
e
in
two
sites be
comes smalle
r with time
bef
ore
overl
appi
ng
with eac
h
other.
Figure 8. PM2.5 Spatial Di
stributio
n
4. Conclusio
n
Wirel
e
ss
se
nso
r
net
work is a
pro
m
ising
altern
ative for ind
oor p
a
rticula
t
e matter
pollutant
s monitorin
g
. Ho
wever, bot
h
the sampl
e
interval co
nfiguratio
n a
nd sa
mple
sites
sele
ction may
lead to the re
sults bi
as.
An energy saving schem
e is prop
ose
d
in this pap
er to redu
ce
the powe
r
co
st for the
sen
s
o
r
n
e
two
r
k fo
r p
a
rti
c
ul
ate matter m
onitorin
g
. Th
eoreti
c
al
anal
ysis
sh
ows t
hat 66%
ene
rgy
may be save
d by this method.
Two lo
catio
n
s
(the front
door
and b
a
ck
doo
r)
we
re sele
cted t
o
deploy the
sen
s
o
r
netwo
rks fo
r
particulate m
a
tters m
onito
ring in thi
s
p
a
per
so th
at the effects
of random
sel
e
ct
ion
P
M
2.
5 s
p
a
t
i
c
al
d
i
s
t
r
i
bu
ti
on
20
35
50
0
5
10
15
20
25
30
35
4
0
4
5
Ti
m
e
(
m
i
n
)
D
ens
i
t
y
(
ug/
m
3
)
b
a
c
k
_d
oo
r
f
r
o
n
t
_
do
or
P
M
2.
5 s
pati
a
l
di
s
t
r
i
buti
o
n
120
140
160
0
5
10
15
20
25
30
35
40
45
Ti
m
e
(
m
i
n
)
De
n
s
i
t
y (
u
g
/
m3
)
bac
k
_door
f
r
ont
_door
P
M
2.5
s
p
ati
a
l
di
s
t
r
i
bu
ti
on
20
40
60
80
0
5
10
1
5
20
2
5
30
35
4
0
45
Ti
m
e
(
m
i
n
)
D
ens
i
t
y
(
ug/
m
3
)
si
t
e
_
1
s
i
t
e_2
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4740 – 4
746
4746
of sampl
e
po
int can b
e
tested. Extensi
v
e exper
ime
n
t
s sh
ow that
different sam
p
le point
s m
a
y
result in 50% differen
c
e at most.
Ackn
o
w
l
e
dg
ements
The p
r
oj
ect
wa
s fun
d
e
d
by the
State
Key Labo
rato
ry
of Advance
d
Opti
cal
Comm
uni
cati
on Systems
Networks a
n
d
Suzhou
S
c
ie
nce a
nd Te
ch
nology Fun
d
(SZS20130
4).
Referen
ces
[1]
F
r
anck U, Herb
arth O, Röder S, et al. Respir
ator
y
e
ffects of indo
or particl
e
s
in
youn
g chi
l
d
ren ar
e size
dep
en
dent.
Sci
ence of the T
o
t
a
l Envir
o
n
m
e
n
t
. 2011; 40
9(9): 162
1-16
31.
[2]
Künzli N, Mudw
a
y
IS, Götschi T
,
et al. Comparis
on
of o
x
i
d
ative pr
op
ertie
s
,
light absor
b
ance, and
total
and
el
ement
al
mass conc
entr
a
tion
of am
bie
n
t
PM2. 5 c
o
ll
ected
at 20
Eu
rope
an s
i
tes.
En
vi
ronm
e
n
t
a
l
hea
lth pers
pec
tives
. 2006; 1
1
4
(5): 684-
69
0.
[3]
Peters A, Ver
o
nesi
B, Ca
lder
ón-Garci
due
ña
s L,
et al.
T
r
anslocati
on an
d potenti
a
l ne
uro
l
ogic
a
l
effect
s
of fine an
d ultr
afine p
a
rt
icles
a critical u
pdat
e.
Part Fibre Toxicol
. 2
0
0
6
; 3(13): 1-13.
[4]
Miller KA, Sis
c
ovick DS, Sh
epp
ard L, et a
l
. Lon
g-term e
x
p
o
sur
e
to air
poll
u
tio
n
an
d
incid
enc
e of
cardiovascular events
in
w
o
m
en.
New
Engl
a
nd Jour
nal
of Medici
ne
. 20
07
; 356: 447-
458.
[5]
MacNei
ll M,
Wallac
e
L, Ke
arn
e
y
J, et
al. Fac
t
ors
influ
enci
n
g
varia
b
ilit
y i
n
th
e infiltr
a
tion
of PM2.5 mas
s
and its comp
on
ents.
Atm
o
sphe
ri
c En
vi
ro
nm
en
t
. 2012; 61:5
1
8
-53
2
.
[6]
Mollo
y SB, C
hen
g M, Galb
all
y
IE, et al.
Indoor
air q
u
a
lit
y in t
y
p
i
ca
l
temperate z
o
ne Austra
lia
n
d
w
el
li
ngs.
Atmospheric Envir
o
nm
ent
. 2012;
54: 400-
40
7.
[7]
Lóp
ez-Ap
a
rici
o
S, Smolík J,
Mašková L, et
al. Re
l
a
tions
hi
p of ind
oor a
n
d
outd
oor a
i
r p
o
llut
ants in
a
natura
l
l
y
venti
l
a
ted histor
ical buil
d
i
ng
e
n
vel
o
pe.
Buil
din
g
an
d Enviro
n
m
ent.
2011; 46: 14
6
0
-14
68.
[8]
Molle
R, Mazo
ué S, Géh
i
n É
,
et al. Indo
or-
out
do
or re
latio
n
shi
p
s of a
i
rb
orne
particl
es
and
nitro
g
e
n
dio
x
i
de i
n
sid
e
Parisia
n
bus
es
.
Atm
o
spheric Environm
ent.
2
013; 69: 2
40-2
48.
[9]
Eeftens M,
T
s
ai MY, Ampe C, et al. Spatia
l vari
ati
on of P
M
2. 5, PM10,
PM2. 5 absor
b
ance a
nd PM
coarse
conc
en
trations
bet
w
e
en
and
w
i
t
h
in
20 E
u
rop
e
a
n
s
t
ud
y ar
eas
an
d the
rel
a
tio
n
s
h
ip
w
i
th
NO2-
Results of the
ESCAPE project.
Atmospher
i
c
Environ
m
ent.
2012; 6
2
: 303-
317.
[10]
MacNei
ll
M, W
a
llac
e
L K
earn
e
y
J, et
al. F
a
c
t
ors
infl
uenc
in
g
vari
abi
lit
y i
n
th
e i
n
filtratio
n
of
PM2.5 ma
s
s
and its comp
on
ents.
Atm
o
sphe
ri
c En
vi
ro
nm
en
t.
2012; 61: 5
18-5
32.
[11]
Hui PS, W
ong
LT
, Mui KW.
Evalu
a
tion of
professi
ona
l choic
e
of sampl
i
ng loc
a
tio
n
s for indo
or air
qua
lit
y
assess
ment.
Build
ing and
e
n
viro
n
m
e
n
t
. 2007; 42(
8): 2900-
29
07.
Evaluation Warning : The document was created with Spire.PDF for Python.