Transcript Presentation

TIMIP
Micro-Mobility Performance Evaluation
of a Terminal Independent Mobile
Architecture
TIMIP - Paper 84
Pedro Estrela, Teresa Vazão, Mário Serafim Nunes
IST/INESC-ID, Portugal
HET-NETs ‘04, 28 July 2004
[email protected]
1
Outline



Introduction

Proposed Framework Model (simplified)

TIMIP Protocol / Architecture
Simulation Studies

NS2 Simulation Scenario

Experiments Evaluation
Conclusions
2
Proposed Framework Model

Proposed Framework Model (simplified)

Classifies existing protocols by their operations type and their efficiency

Considers both Handover and Stable Operations Efficiency

Protocol Efficiency = Handover Efficiency + Resource Optimization Efficiency

More details in the Paper (including protocol classification)
Phase
Operations
Differentiating point
Proposed Models
Detection
Mobile movement
detection
Layer 2 interactions/
dependence?
Passive, Reactive,
Predictive, Active
Registration
Network routing tables
update
Location and number of
updated nodes?
Inter-Domain,
Intra-Domain, Cluster
Data packets Routing
Shortest paths used?
Non-Optimized, Optimized
State Maintenance
Maintenance Cost?
Explicit, InBand
Execution
3
TIMIP/sMIP – Architecture

TIMIP – Terminal Independent Mobility for IP

Micro-Mobility proposal with surrogate behaviour and high efficiency

Terminal Independence - support for any IP terminal


Detection and registration are both performed / generated at network side only
Efficiency




Detection – Reactive Model – Highest possible in network side
Registration – Cluster Model – Update Messages follow the shortest paths between the APs
Execution – InBand State Maintenance – IP data traffic of the mobiles refresh routing paths
Execution – Optimized Routing – Packets follow shortest paths inside the network
…
…
…
LT
ANG/
sFA
AR
…
AP
Core
Network
…
LT
AR
LT
AP
…
ANG/
sFA
AR
…
TIMIP Domain 1
…
TIMIP Domain 2
4
NS2 Simulation Studies

NS2 2.26 Simulator featuring




InterDomain
Generator
GW
10 Mbit/s, 40 ms delay
10 Mbit/s, 20 ms delay
10 Mbit/s, 10 ms delay
IntraDomain
Generator
Simulation Scenario




TIMIP protocol simulation
CIMS v1.0 mobility protocols:
HAWAII (MSF) / Hierarchal MIP/
Cellular IP (hard handoff)
Modified 802.11 infra-structured
behaviour with multiple channels
(hard handoffs forced)
One Mobile Node / Multiple APs
Increasing hierarchal link delays
Intra/Inter domain CBR Traffic
generation
Old AP
New AP
Mobil e Receiv er
Objective: Evaluate Efficiency


Handover Efficiency by protocol
Execution Efficiency by protocol
5
Experiment A – Registration
Test probes:



CBR intra domain probes to MN, moving from node 9 to node 10
METRIC: Hard Handoff Lost probes measure Handover Latency
Handover Latency – Intra Domain traffic results





TIMIP – Time needed to reach crosshover node 4 (close to new AP)
HAWAII MSF – Time needed to reach Old AP (node 9) + OOO
Cellular IP – Time needed to reach CrossHover node 1 (close to GW)
Hierarchical MIP – Time needed to reach the domain’s GW (node 0)
Receive vs Generation Time during handover by protocol
Intra-Domain
255
Sequence Numbers (n)
245
235
225
GENERATOR
TIMIP
HAWAII
CIP
HMIP
215
205
195
185
175
165
0.95
1
1.05
1.1
1.15
HANDOVER
Time (s)
1.2
1.25
1.3
6
Experiment A – Registration

Test probes:



Handover Latency – Inter Domain traffic results



CBR inter domain probes to MN, moving from node 9 to node 10
METRIC: Hard Handoff Lost probes measure Handover Latency
TIMIP / HAWAII / Cellular IP – Time needed to reach CrossHover node 1
Hierarchical MIP – Time needed to reach the domain’s GW (node 0)
This experiment is also capable to evaluate execution phase efficiency (in the paper)
Receive vs Generation Time during handover by protocol
Inter-Domain
250
Sequence Numbers (n)
240
230
GENERATOR
TIMIP
HAWAII
CIP
HMIP
220
210
200
190
180
170
1.95
2
2.05
2.1
2.15
HANDOVER
2.2
2.25
2.3
Time (s)
7
Experiment B – Execution

Test probes:



CBR intra domain Traffic to MN, moving through all APs
METRIC: Sum of packets forwarded in all nodes per time interval measure Resource Optimization
Resource Optimization – Intra Domain traffic results




TIMIP – decreasing utilization, subject to location inside network
HAWAII – similar values (+ handover instants visible)
HMIP – all packets are forced to pass through the GW
Cellular IP – similar to HMIP + packets completely exit the domain
TIMIP
Average Packets Forwarded in the Network - Intra Domain
HAWAII
HMIP
45
CIP
Average Packets Forwarded / 50 ms
40
35
30
25
20
15
10
5
0
0
2
AP 8
4
AP 9
6
Time
AP 10
8
AP 11
10
8
Experiment B – Execution
Test probes:



CBR inter domain probes to MN, moving through all APs
METRIC: Sum of packets forwarded in all nodes per time interval measure Resource Optimization
Resource Optimization – Inter Domain traffic results



All protocols - Similar utilization, in all locations
HAWAII - Out-of-Order effect also visible in the graph (= increased network utilisation)
TIMIP
Average Packets Forwarded in the Network - Inter Domain
HAWAII
HMIP
45
CIP
Average Packets Forwarded / 50 ms
40
35
30
25
20
15
10
5
0
0
2
AP 8
4
AP 9
6
Time
AP 10
8
AP 11
10
9
Conclusions

Conclusions

TIMIP: micro-mobility protocol with Terminal Independence and Efficiency

TIMIP’s Efficiency – classified via proposed Framework



Detection – Reactive Model (closest: MIP fast-handovers Triggers)
Registration – Cluster Model (closest: HAWAII)
Execution



In-Band State Maintenance
Optimized Routing
(closest: Cellular IP)
(closest: HAWAII)
Future Work




Multicast
QoS Support
IPv6 Networks Support + IPv4 Terminal Independence
Non Hierarchical Networks
10
Questions?
Questions?
Thank You!
11