QoC-based Optimization of End-to- End M-Health Data Delivery Services Ing Widya (UoT), Bert-Jan van Beijnum (UoT), Alfons Salden (TI)
QoC-based Optimization of End-to-EndM-Health Data Delivery Services
Ing Widya (UoT),
Bert-Jan van Beijnum (UoT),
Alfons Salden (TI)
Outline:– introduction
• mobile-healthcare case
– context flow graph
– computational model
• freshness, availability, costs QoC
– computational example
– conclusions
• resource configuration & alternatives
Introduction• M-Health Application
PNO
InternetBackbone
mHP
ISPADSL
BlueTooth
Zigbee
UMTS
WiFi
Zigbee
GPRS
m-Health PortalBody Area Network (BAN) Internet Access
Front-End
Front-End
MBUMBU
modem
Front-End
Front-End
sisi
sisi
Back-End
Back-End
Context Flow Graph (CFG)
Zigbee1
NFE12
BTooth1
= pre-select node
S1
S2
S3
S5
= context generator node
S4
wired link
wired links
NMBU3
NFE11
NFE21
= aggregating node
NMBU11
ADSL
BTooth2
WiFi
NGW1
NBE1 NBE2BE-
processingWiFi
GPRSUMTS
FE1-processing
NFE22FE2-
processing Zigbee2
NMBU22
NGW2
GW-processing
• optimal path to bring health-data to professional ?
CFG, QoC and QoS
• QoC based selection– Quality of Context (information)
– QoC freshness(/up-to-dateness), availability, “costs”;
node A node Bresource
(processing/communication)
Context Information
Context Information
QoCat_BQoCat_AQoS
– QoC impeded by QoS
• max-plus algebra:– arithmetic maximum (instead minimum)
• properties: commutative, associative, …–
Computational Model
• min-plus algebra– additive operation:
– multiplicative operation:
b} {a, minimum b a
b a b a
cbcacbaScb,a,
arithmetic domain
Computational Model (..)
• aggregation & concatenation elements– algebraic expression
).Fr (d )Fr (d
Fr
Fr
d0
0d II Fr
S22maxS11
S2
S1
2
1maxFE
}Fr {d maximum Fr Sii 2 1, i
FE
– QoC arithmetic expression
S1
S2
NFE11
d1
d2
Computational Model (..)
• concatenation & pre-select element– algebraic expression
FE12zb
FE12z
bMBU1
Fr ) d d (
Fr d
d II Fr
FE12iz} {b, i
MBU1 Fr } {d minimum Fr
– QoC arithmetic expression
Zigbee1
NFE12
BTooth1NMBU1
1
Computational Model (..)
• End-to-End Freshness algebraic expression
I
I
I
I
......)FrdFr(dd
......)FrdFr(dd
......)FrdFr(dd
......)FrdFr(dd
IIIIFr
FE21z2maxFE11z1wh
FE21z2maxFE11z1w
FE21z2maxFE11z1u
FE21z2maxFE111zg
BE2
,Frd FrdFrdFr S4wl4maxS2wl2maxS1wl1FE11
S5wl5FE21 FrdFr
Computational Model (..)
wired link
wired links
Zigbee1
NMBU11
NMBU3
NGW1
NBE1 NBE2BE-
processing
GPRS
NFE11
NFE12
S1
S2
S3
FE1-processing
NFE21
NFE22S5FE2-
processing Zigbee2
NMBU22
S4
NGW2
• (1,1) element
• Availability & costs: independently & similarly
Computational Example
• QoS values (illustration)
• QoC results– three independently calculated QoC matrices
– 4x4 matrices of ranked alternative E2E paths
• Weighing Metrics– weighted quadratic norms
GPRS UMTS WiFi WFADSL ZB1 BT1 ZB2 BT2
Delay 5 4 3 1 7 10 4 8
Avail. 0,990 0,985 0,98 0,999/0 0,96 0,97 0,96 0,97
Costs 418 0 260 0 0 0 0 0
Computational Example (..)
• QoC for a rehabilitation training scenario– weights: wFr = 1, w1-Av = 150, wCo = 0.02 (not normalized)
0
5
10
15
20
25
30
35
40
z1&
z2&
g
z1&
z2&
u
z1&
z2&
w
z1&
z2&
wh
z1&
b2
&g
z1&
b2
&u
z1&
b2
&w
z1&
b2
&w
h
b1
&z2
&g
b1
&z2
&u
b1
&z2
&w
b1
&z2
&w
h
b1
&b
2&
g
b1
&b
2&
u
b1
&b
2&
w
b1
&b
2&
wh
0
5
10
15
20
25
30
35
40
z1&
z2&
g
z1&
z2&
u
z1&
z2&
w
z1&
z2&
wh
z1&
b2
&g
z1&
b2
&u
z1&
b2
&w
z1&
b2
&w
h
b1
&z2
&g
b1
&z2
&u
b1
&z2
&w
b1
&z2
&w
h
b1
&b
2&
g
b1
&b
2&
u
b1
&b
2&
w
b1
&b
2&
wh
indoors outdoors
– indoors: path via ZigBee1, BTooth2 & WiFi + ADSL
– outdoors: path via ZigBee1, BTooth2 & UMTS
Conclusions
• QoC based selection of an optimal E2E transfer pathfor M-Health scenarios;
• Min-max-plus algebra for several QoC dimensions;• Future work:
– dynamic case
– use of (colored) Petri-Nets
– other workflow operations