Extended ZRP: a Routing Layer Based Outline
Service Discovery Protocol
for Mobile Ad Hoc Networks
Service Discovery in MANETs
Christopher N. Ververidis & George C. Polyzos E-ZRP: Routing Layer based Service Discovery
Simulation Results
Mobile Multimedia Laboratory Proactive part
Department of Computer Science
Athens University of Economics and Business Reactive part
Athens 10434, Greece Service Availability
{chris,polyzos}@aueb.gr Conclusions
http://mm.aueb.gr/
Tel.: +30 210 8203 650, Fax: +30 210 8203 686
polyzos@aueb.gr 2
Routing Layer Based Service Discovery:
Service Discovery in MANETs Motivation
Service Discovery in fixed networks If Service Discovery is implemented
Assumes reliable communication above the routing layer then
Mainly centralized approaches two message producing processes coexist:
(UDDI,Salutation, JINI, SLP, SDP) one for communicating service information among nodes
Service Discovery in MANETs one for communicating routing information among nodes
Needs to be distributed-decentralized hence a node is forced to perform the battery-draining operation
Needs to be scalable of receiving and transmitting packets multiple times
Needs to minimize energy consumption
(Allia, GSD, DEAPspace, Konark, SANDMAN) a Routing Layer based Service Discovery protocol: E-ZRP
SANDMAN and DEAPspace integration of routing with service discovery
power savings only by allowing nodes to go into ‘sleep’ mode cross-layer optimization
What if continuous connectivity is mandatory? an idea proposed by Koodli and Perkins
polyzos@aueb.gr 3 polyzos@aueb.gr 4
Review:
Zone Routing Protocol (ZRP) – Haas et al. Extended ZRP (E-ZRP)
combines reactive and proactive routing approaches
ZRP actually consists of 3 parts: Our goal
Neighbor Discovery Protocol (NDP) To provide an experimental assessment of energy savings
Intra-Zone Routing Protocol (IARP) obtained by implementing service discovery at the routing layer
responsible for proactively maintaining route records for nodes located inside
a node's routing zone (e.g. records for nodes located up to 2-hops away)
Inter-Zone Routing Protocol (IERP) Our approach
responsible for reactively creating routes for nodes located outside a node's
select interesting, appropriate MANET routing protocol
routing zone (e.g. records for nodes located further than 2-hops away)
exploit the capability of acquiring service information along with
routing information
we modified the Zone Routing Protocol
by piggybacking service information into routing messages
services are described using UUIDs (Unique Universal Identifiers), in
order to keep packet lengths of routing messages small
polyzos@aueb.gr 5 polyzos@aueb.gr 6
Simulation Results Simulation Results
Proactive part, Fixed topology Proactive part, Mobility
E-ZRP vs. a traditional Flooding application-layer service discovery protocol random waypoint model with the following parameters:
Flooding radius equals E-ZRP Zone radius Min. Speed = 0 (m/s)
a message in E-ZRP contains info about the sending node’s service and Pause Time = 30 s.
also about the services of its intra-zone neighbors Max. Speed: 0.5 m/s, 1 m/s, 2 m/s, 5 m/s, 7.5 m/s, 10 m/s and 12.5 m/s
a Flooding message contains info only about the sending node’s service
Avg. # of Services Discovered / Node Avg. Battery Consumption / Node
in order to be shorter Avg. # of Services Discovered / Node
12 0,9
6
same broadcast intervals 0,8
5,5 10 0,7
Average Energy Consum ption per Node
Energy (mWhr)
0,35 0,6
8
5
0,5
Services
0,3
6
Services
Energy (mWhr)
0,4
0,25 4,5
0,3
0,2 4
0,2 FLOODING
4
0,15 FLOODING
FLOODING E-ZRP
2 0,1
0,1 E-ZRP
FLOODING E-ZRP
3,5 0
0,05 E-ZRP 0
ec
ec
ec
/s
ec
/s /s
ec
ec
ec
0 3 /s /s /s /s
/s
/s
/s
/s
/s
/s
m
/s
m m m m m m
m
m
m
m
5 1 2 5 5 10 .5
m
m
m
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0. 7. 12
1
2
5
.5
5
5
10
0.
7.
12
Node population Node Population Maxim um Speed Maxim um Speed
polyzos@aueb.gr 7 polyzos@aueb.gr 8
Extended Comparison
Proactive Part Extended Comparison, Proactive part—cont.
optimal configuration for application-based 250 Nodes, 1000 s simulation time
service discovery scheme (restricted zone Low Mobility: min. speed 0 m/s, max. speed 0,5 m/s and pause time 30 s
updates)
often High Mobility: min. speed 0 m/s, max. speed 12,5 m/s and pause time 30 s
service discoverability is equal to or better than 6
E-ZRP
that achieved by a routing layer based approach G H dotted lines: E-ZRP with a broadcast interval of 10 s
10 s F
5 E Application
no mobility, 250 nodes, 1000 s simulation time D
layer
Flooding broadcast interval Service deletion interval
Avg. Energy Consumption per node and
vertical and horizontal blue dotted lines: E-ZRP 4 C (flooding) Avg. Num. of Services per node A 200 s 800 s
with a broadcast interval of 10 s
Services
16 B 160 s 640 s
Flooding broadcast interval x 4 = Service 3 B 14 C 80 s 320 s
deletion interval A C D D 40 s 160 s
2 seldom 12 E F G H
messages are shorter for Flooding than for B E 20 s 80 s
10
ZRP/E-ZRP
Services
F 15 s 60 s
1 A
8
Flooding performs better than E-ZRP in terms F
G H G 10 s 40 s
6 DE
of service discoverability for broadcast intervals 0
C application layer based service discovery reaches
0,1 0,2 0,3 0,4 B
higher than 40 s, but Energy consumption is 4 A 12.5m/sec its optimal performance in terms of energy
0.5m/sec
increased by > 30% Energy (m Whr) 2 consumption when the broadcast interval is > 160 s,
longer intervals fewer messages transmitted 0 saving 3% more power but discovering 43% less
0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 services for low mobility cases and 22% less
nodes receive less services information Energy (m Whr)
services for high mobility cases.
polyzos@aueb.gr 9 polyzos@aueb.gr 10
900
800
IERP
Reactive Part 700
600
FLOODING
Service Availability
Delay (ms)
500
400
Flooding imposes significant delays 300
200
for discovering out of zone services 100 Service Availability Duration (SAD)
IERP: node needs 10 .. 50 ms 0
3 4 5 6 7 8 9 10 decreases when speed increases
Flooding: node needs 200 .. 800 ms # of Hops to Service Provider
Each point on the diagram is an average obtained over 20 service
discovery requests between different node pairs @ the same distance Average Transaction Duration (ATD)
for a node, for any service
Since IERP uses the mechanism of bordercasting, it can
efficiently and quickly “scan” distant areas of the network Tradeoff between
to find the requested service average SAD
Flooding takes a long time to “scan” the network since it number of discovered services
relies on hop-by-hop broadcasting
polyzos@aueb.gr 11 polyzos@aueb.gr 12
Service Availability Results Conclusions
Average SAD actually decreases when speed increases
E-ZRP leads to significantly smaller energy consumption (approximately 50%
high mobility (max. speed = 14 m/s): highest # services discovered less), but also, in certain cases, it achieves higher service discoverability
high ATD: the discovery protocol would perform better in a low mobility setting
low ATD: a high mobility setting would be ideal for the discovery protocol ‘Favoring’ the application layer based service discovery protocol with larger
In high density cases, the average SAD is decreased flooding intervals (in order to become more economic in terms of energy
despite the existence of multiple paths and providers
consumption (savings of 3%)), had a detrimental effect in service
discoverability, reducing it by 22% or more, compared to the proposed routing
because of higher contention layer based approach
The total number of services discovered is higher in denser environments
350
Our experiments for out-of-zone services revealed that E-ZRP consumes 5%
100
90
700
300
more energy than Flooding but achieves one order of magnitude smaller
delay for discovering services
Avg. SAD (seconds)
Number of Services
Number of Services
80 650 250
70
600 200
60
50 150
40
550
100
We introduced a new metric, Service Availability Duration (SAD) for
500
30
20 50 measuring the “quality” of discovered services
450
10
0 400
0 examined the implications of network density and node mobility on the
0 5 10 15 0 5 10 15
0 200 400 600
SAD in Seconds
800 1000
availability of services discovered with E-ZRP
Speed (m/s) Speed (m/s) 30 Nodes. Speed Range (0-14m/s). E-ZRP's Zone Radius = 3
15 Nodes. Speed Range (0-14m/s). E-ZRP's Zone Radius = 3
polyzos@aueb.gr 13 polyzos@aueb.gr 14
Thanks!
George C. Polyzos
Joint work with my Ph.D. student
Christopher N. Ververidis
Email: chris@aueb.gr or polyzos@aueb.gr
Mobile Multimedia Laboratory
Department of Computer Science
Athens University of Economics and Business
http://mm.aueb.gr/
polyzos@aueb.gr 15