Scalability of FMIPv6 and HMIPv6

Youngjune GwonJames KempfAlper Yegin

Ravi JainDoCoMo Communications Labs USA

Objective

• Determine signaling scalability of HMIPv6, FMIPv6, and combined HMIP and FMIP (HFMIPv6).

• Compare signaling scalability against standard MIPv6 (SMIP).

• Use a piecewise simulation to assess.– Removes need to implement the protocols in a

simulator.– Reduces amount of compute time needed to perform

simulation.

Piecewise Simulation Procedure

• Simulate mobility traces for 100K mobile nodes.– Custom developed mobility simulator used.

• Measure per handover signaling costs and latencies on actual implementations of the protocols.– SMIP implementation is MIPL.– FMIP implementation from DoCoMo (03 draft).– HMIP implementation from Monash.– HFMIP integration performed by DoCoMo.

• Straightforward, could be further optimized.• Not draft-jung-mobileip-fastho-hmipv6-01.txt.

• Simulate aggregate signaling cost using mobility traces, traffic model, and per handover measurements.

Mobility Model

• Mobility model from ETSI (i.e. 3GPP) Technical Report 101 112 v3.2.0 (Release 98), ETSI, April 1998 used.

• 100K users simulated.• Two levels of mobility:

– Pedestrian mobility suitable for WLAN.

– Vehicle mobility suitable for WAN.

Initial position

Final destination

Direction angle and speedupdated each time after traveling decorrelation length with probability of 0.2

1 – TurnProb

TurnProb/2

TurnProb/2

1 – TurnProb

TurnProb/2

TurnProb/2

WAN Mobility Model

WLAN Mobility Model

Wireless Access Network Model

• 100 x 100 km planar area.

• Two wireless networks:– WAN: 1 km radius cells.

– WLAN: 100 m radius cells.

• Optimal packing of wireless cells into hexagonal geometry.

• Single access point per cell.

Overlapping width

Cell coverage is based on circular radius r

r

Wired Backhaul Model• Star topology.• Access routers connected

to multiple access points.– All cells under one access

routers are in same subnet.• Aggregation routers

connected to access routers.

• HMIP MAP above aggregation router (when appropriate).

• Measured 10, 20, and 50 ARs per MAP or Access Network.– Results only presented here

for 20.

Core Netw ork (IP)

G ateway R outer(M A P )

A ggregationR outer

A ccessR outer

A ccessR outer

A ccessR outer

A ccessR outer

A ccessR outer

A ccessR outer

A ggregationR outer

A ccessR outer

A ccessR outer

A ccessR outer

A ccessR outer

InternetA ccess

N etwork 2

A ccessN etwork 3

A ccessN etwork 4

Traffic Models• Two models:

– Real time Voice over IP.– Web traffic.

• Voice:– Poisson arrival process.– Mean call duration 120 seconds.– Markov process for transition between talking and silence states.

• Data:– Poisson arrival process.– Time between sessions is Pareto.

• Refs:– Voice: ETSI Technical Report TR 101 112 v3.2.0 (Release 98),

ETSI, April 1998.– Data: Shankaranarayanan, N., et al., “Performance of a Shared

Packet Wireless Network with Interactive Data Users,” Mobile Networks and Applications (MONET), Vol. 8, pp. 279 – 293, June 2003.

Results: Number of Handovers Per Hour

0 5 10 15 20 25 30 35 40 45 500.5

1

1.5

2

2.5

3

3.5

4

4.5x 10

6 Total IP handoffs

Tota

l IP

handoff

s p

er

hour

Number of cells (base stations) per AR

Results: Handover Signaling Load

0 5 10 15 20 25 30 35 40 45 500

1

2

3

4

5x 10

9 Total IP signaling (20 ARs per MAP/AN)

Am

ount

in o

cte

ts (

per

hour)

Number of cells (base stations) per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

0 5 10 15 20 25 30 35 40 45 500

0.5

1

1.5

2

2.5x 10

9 Total IP signaling over air (20 ARs per MAP/AN)

Am

ount

in o

cte

ts (

per

hour)

Number of cells (base stations) per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

Results: Mean IP Blackout Duration

10 15 20 25 30 35 40 45 500

200

400

600

800

1000

1200

1400Average IP layer blackout duration per handoff

IP layer

bla

ckout

dura

tion (

msec)

Number of ARs per MAP/AN

SMIP FMIP HMIP FMIP/HMIP Hybrid

Results: Handover Packet Loss

0 5 10 15 20 25 30 35 40 45 500

1

2

3

4

5x 10

7 Total Lost Voice Packets (20 ARs per AN/MAP)

AMR/

RTP/

UDP/

IPv6

pac

kets

Number of APs per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

0 5 10 15 20 25 30 35 40 45 500

100

200

300

400

500Total Lost Voice Packets per User (20 ARs per AN/MAP)

AMR/

RTP/

UDP/

IPv6

pac

kets

Number of APs per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

Voice Packets

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10x 10

8 Total Lost Data (20 ARs per AN/MAP)

Data

pay

load

in b

ytes

Number of APs per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

0 5 10 15 20 25 30 35 40 45 500

2000

4000

6000

8000

10000Total Lost Data per User (20 ARs per AN/MAP)

Data

pay

load

in b

ytes

Number of APs per AR

SMIP FMIP HMIP FMIP/HMIP Hybrid

Data Packets

Results: Traffic Tunnel Overhead

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8x 10

6 FMIP Tunneled Packets Handled per AR

Total

tunn

eled p

acke

ts

Number of APs per AR

10 ARs per AN20 ARs per AN50 ARs per AN

0 5 10 15 20 25 30 35 40 45 500

5

10

15x 10

8 HMIP Tunneled Packets Handled per MAP

Total

tunn

eled p

acke

ts

Number of APs per AR

10 ARs per MAP20 ARs per MAP50 ARs per MAP

Total Tunneled Packets

0 5 10 15 20 25 30 35 40 45 5010

20

30

40

50

60

70

80

Percent Tunneled Packets

Per

cent

(P

er A

R)

Number of APs per AR

90

100

FMIP

HMIP

Tunneled vs. UntunneledPackets

Conclusions• More APs per AR results in decreased signaling

load at IP level.– No surprise here.

• HMIP has lower handover signaling cost.• FMIP has lower handover blackout time and lower

handover packet loss.– But more APs per AR reduces HMIP blackout time

and packet loss to slightly more than FMIP.

• FMIP has much less traffic tunnel overhead.• Bottom line: FMIP should be simplified to reduce amount of

over the air signaling associated with IP handover.