An Accurate Simulation Model for Mobile IPv6

Protocol
Eric Wu, Johnny Lai, Ahmet Sekercioglu
Department of Electrical and Computer Systems Engineering
Monash University, Australia

Abstract We present an overview of the Mobile IPv6

simulation model implemented by using OMNeT++ [1] simulation framework. We attempted to create a very accurate
model of the Mobile IPv6 protocol. The simulation model
consists of three key components: i) a class inheritance
approach for specifying a node to behave as a host or router,
ii) same approach for defining the Mobile IPv6 roles of mobile
node, correspondent node or home agent, and iii) conceptual
data structures necessary for storing binding information
and for responding to the received Mobile IPv6 signaling
messages.
An example network consisting of a mobile station moving
through a series of IPv6 subnets is used to demonstrate the
models capabilities. The handover behaviour is consistent
with the mathematical analysis and will be used in future
papers to investigate underlying issues with MIPv6.

I. I NTRODUCTION
Over the last two decades, arguably the main change
in telecommunication networks has been the deployment
of wireless access technologies. Wireless access has lead
to creation of a plethora of mobile devices with a wide
range of communication, computing, and storage capabilities. With wireless interfaces providing ease of use, these
devices have become increasingly popular.
The current Internet is based on an architecture created
in 1969 as the ARPANET. However, it does not support the
needed features and architectural structures for mobility.
Because of that, the existing general mobility support
solutions in the IP world have tried to hide the dynamic
change of IP addresses from the upper layers.
Mobility for IPv6, or Mobile IPv6 (MIPv6) [2] was developed to allow an IPv6 node to change points of network
attachment without disrupting applications or services. In
the MIPv6 specification, three network roles: mobile node,
correspondent node and home agent are essentially defined.
A mobile node is a node that can change its point of
attachment from one link to another, while still being
reachable via its home address. The mobile node maintains
two IPv6 addresses, home address and care-of address. The
home address is used as a permanent address of the node
and is also used as an identifier by the transport sessions.
The care-of address is essentially used as a current location
identifier of the mobile node by the home agent. It also
allows the correspondent node to send packets directly to
the mobile node. A correspondent node is a peer node with
which a mobile node is communicating. A home agent is

a router on the mobile nodes home link with which the

mobile node notifies its current care-of address. While the
mobile node is away from home, the home agent intercepts
packets on the home link destined to the mobile nodes
home address, encapsulates them, and tunnels them to the
mobile nodes registered care-of address.
Our research group focuses on developing simulation
systems for investigations into the performance and scalability of IPv6 and Mobile IPv6 protocols over the Internet.
To do our research, we have developed a comprehensive set
of models [3] for simulating fixed IPv6 networks. We have
also recently added an accurate model of IEEE802.11b
[4] to the simulation suite. Accurate modeling of MIPv6
protocol in the simulation will allow us to extend our
research into MIPv6 protocol performance analysis, such
as signaling and handover optimisations, network mobility,
and multicast mobility.
In the next section, the overall design of our model is
discussed, which includes our approach in obtaining an
accurate MIPv6 simulation model. The accuracy of the
model is assessed in the following section by comparing
performance results obtained from the mathematical analysis. A short section then covers additional support and
functionalities that may be added to the model. Finally,
the last section concludes with some remarks to review
the overall discussion that is presented.
II. D ESIGN

OF

T HE M OBILE IP V 6 M ODEL

The MIPv6 simulation model contains functionalities

such as handling of various IPv6 headers [5] and conceptual data structures to manage different aspects of the
protocol. The implementation is designed in a way such
that each module in the protocol stack handles a specific
type of IPv6 header or an ICMP message. Figure 1 depicts
the kernel of the protocol. The MIPv6 model interacts with
higher layer protocols via LocalDeliver and Send
modules. The IPv6 packet is sent to the network through
Output and received from PreRouting. The ICMPv6
Ping echo and reply messages are processed in the ICMP
module.
A. Host and Router
According to Neighbour Discovery in IPv6 [6], the
main difference between a host and router is the handling
of different control messages necessary for neighbour

provides the common behaviour for the different MIPv6

roles. The assignment of a node to a specific MIPv6 role
is done via XML configuration.

Fig. 3.

Class inheritance diagram of MIPv6 network role

C. Conceptual Data Structure

Fig. 1.

Kernel of MIPv6 protocol

discovery such as router advertisement, router solicitation,

neighbour advertisement and neighbour solicitation. Therefore, a class inheritance approach (shown in Figure 2) is
used to express the relationships between a router and host
as it also allows the transition between the two roles. Only
one instance of inherited class is created depending on
whether the node is a router or host at startup . The instance

Fig. 2.

In the MIPv6 specification, the conceptual data structures are described as follows:
Binding Cache: A cache of bindings1 for other nodes.
This cache is maintained by home agents and correspondent nodes. Each entry in the cache contains information
about the home address that is visible for the upper layer
protocols and the care-of address that corresponds to the
home address.
Binding Update List: This list is maintained by each
mobile node. The list has an item for every binding that the
mobile node has or is trying to establish with another node.
Both correspondent and home registrations are included in
this list. Entries from the list are deleted as the lifetime of
the binding expires.
Home Agents List: Home agents need to know which
other home agents are on the same link. This information
is stored in the Home Agents List. The list is used
for informing mobile nodes during dynamic home agent
address discovery.
Figure 4 shows the design of MIPv6 conceptual data
structures. The base class MIPv6CDS contains the binding

Class inheritance diagram of neighbour discovery

is created in Neighbour Discovery, a simple module

that resides in ICMP module. This approach provides
extensibility when adding MIPv6 extensions, particularly
the signaling and handover optimisations by allowing new
subclasses to reuse the capabilities of the base class as well
as override certain functions to test different algorithms
accordingly.
B. Mobile Node, Correspondent Node and Home Agent
The mobile node, correspondent node and home agent
process the messages differently. This is done in the
Mobility module depicted in 1. The Mobility module
also implements a class inheritance approach as shown
in Figure 3. The base class MIPv6MobilityState

Fig. 4.

Design of MIPv6 conceptual data structures

cache. The mobile node interface contains binding update

list and the home agent contains the home agent list. The
instance of specific interface is instantiated in Mobility
module. As the node is assigned to a particular MIPv6
role, the specific class instance is created accordingly.
III. M OBILE IP V 6 N ETWORK S IMULATIONS
This section demonstrates the capabilities and correctness of the model by comparing the handover result with
1 binding is the association between the home address and the care-of
address

mathematical analysis based on the protocol described in

the MIPv6 specification. The configuration of the test network is seen in Figure 5. Each router connects to an access

Fig. 6.

Fig. 5.

One simulation sample for showing disruption of ICMP stream

A test scenario for the MIPv6 simulation model

point forming a wireless subnet. A mobile node roams

across different subnets and returns to its home network
eventually. A correspondent node connecting to one of the
routers acts as server for the mobile node. The mobile
node sends an ICMP Ping echo to the correspondent node
and the correspondent node acknowledges with an ICMP
Ping response. This demonstrates connectivity between the
mobile node and correspondent node while the mobile
node changes points of attachments. Table I shows a list
of parameters configured in this experiment.
Simulation Model
Mobility
Ping Application
MIPv6

Parameter
MN MoveSpeed
PingInterval
MaxRtrAdvInterval**
MinRtrAdvInterval**
AllowedMissedRtrAdv
RetransTimer
DupAddrDetectTransmits
** parameters configured for router only

Value
3m/s
0.05s
1s
0.25s
1
1s
1

TABLE I
L IST OF PARAMETERS IN MIP V 6 TEST NETWORK

Figure 6 illustrates disruption of the ICMP stream and

Figure 7 shows the complete results of 10 simulation runs
generating 30 handovers in total. The average handover
latency in Figure 6 is 2332ms.
The MIPv6 handover latency is caused by three major
components: movement detection, duplicate address detection, and a delay interval for a router to reply with a router
advertisement upon a receipt of a router solicitation.
The router periodically sends an unsolicited multicast
router advertisement at least once every RtrAdvInterval
seconds, the actual interval is randomly computed to
be between MaxRtrAdvInterval and MinRtrAdvInterval.
If the mobile node does not receive a router advertisement within RtrAdvInterval seconds, AllowedMissedR-

Fig. 7.

Handover latencies

trAdv decrements by one. When AllowedMissedRtrAdv

reaches zero, the movement detection is triggered. Hence,
the mobile node starts performing handover operation.
Let denote as the actual router advertisement interval
between zero and RtrAdvInterval. The delay caused by
the movement detection can be expressed by the following
equation:

  movement detection

   !   #"$
%'&)(+* "
,.-   
&  ''/ ,  0 &211  3. 

(1)

Note that if AllowedMissedRtrAdv is zero the movement detection is triggered as soon as one router advertisement is missed. Due to the high probability of packet
loss in the wireless network environment, the mobile
node may not receive unsolicited router advertisements.
Consequently, the MIPv6 handover process is triggered
while the mobile node is still in the same subnet. To
avoid this, AllowedMissedRtrAdv is generally set to one
or higher.
Another dominating component to the handover latency

is the delay during the duplicate address detection operation. A node must perform a duplicate address detection
when it attaches to a new link and reconfigures a new
care-of address. During the duplicate address detection,
the node has to wait for at least RetransTimer seconds, to
check if any neighbouring node responds with a solicited
neighbour advertisement. If the mobile node does recieve
one, then a neighbouring node already has the same IP
address. The duplicate address detection delay can be
expressed by following equation:

'3 duplicate address detection

    
1 &   $
(2)
,-   
1
 & 1



 3 
  
    

The third component is a random delay interval,

RA DELAY TIME, computed within the range of 0 to
500ms. A router does not immediately reply with a solicited router advertisement upon receipt of a router solicitation. It waits for this random delay time and then sends
the router advertisement. Let be the actual delay interval.
This avoids an outburst of solicited router advertisements
if there are multiple routers on the same link.
Therefore, the theoretical MIPv6 handover latency can
be expressed as follows:

'3 

'3 movement detection (

'3 duplicate address detection (
     
   30

 

(3)

By adapting the values of the parameters used in the

experiment (shown in Table I) into Equation 3, we can
obtain a possible range of values for theoretical handover
delays.
The minimum value can be:

'3  %! &  "

 (
* (

*1

where,
RetransTimer = 1 second
DupAddrDetectTransmits = 1
For the minimum possible value of the handover delay, the
router advertisement is received instantly, hence assuming
the router advertisement interval is extremely small.
The maximum value of the delay can be:

'3   %"  "

* ( *(
* (
 #&%
$
1

' #&%
(

where,
AllowedMissedRtrAdv = 1
RetransTimer = 1 second

DupAddrDetectTransmits = 1
RA DELAY = 0.5 seconds
For the maximume possible value of the handover delay,
we assuming that first router advertisement is missed and
the second one is received at the longest time, which is
near MaxRtrAdvInterval.
Therefore, the theoretical handover delay values range
from 1 seconds to 3.5 seconds. In the experiment, the
average handover delay of 2.123 seconds falls within the
range provided by the mathematical analysis.
IV. F UTURE A DDITIONS
Most of the MIPv6 functionalities have been implemented in the simulation model. However, there is a
number of attributes, which is still under consideration and
may be implemented in the future. It includes:
) Return Routability Procedure
) Full Proxy Neighbour Discovery
) Improve robustness of handover process when the
mobile node returns home
) Dynamic Home Agent Address Discovery
) Router renumbering at home
V. C ONCLUSION
In this paper, we present an overview of the MIPv6
simulation model implemented under OMNet++ simulation framework. Accurate modeling of MIPv6 protocol
in the simulation will allow us to extend our research
into MIPv6 protocol performance analysis, such as signaling and handover optimisations, network mobility, and
multicast mobility. The example network in the paper
demonstrates the capabilities and verifies the correctness
of the model. The handover performance obtained from
the example network produces a delay we theoretically
expected.
R EFERENCES
[1] A. Varga, OMNeT++ Discrete Event Simulation System User Manual, 2nd ed., Technical University of Budapest, Dept. of Telecommunications.
[2] D. Johnson, C. Perkins, and J. Arkko, Mobility Support
in IPv6, RFC 3775, June 2004. [Online]. Available: http:
//www.ietf.org/rfc/rfc3775.txt
[3] J. Lai, E. Wu, A. Varga, Y. A. Sekercioglu, and G. K. Egan,
A Simulation Suite for Accurate Modeling of IPv6 Protocols, in
Proceedings of the 2nd International OMNeT++ Workshop, Berlin,
Germany, Jan 2002, pp. 222.
[4] S. Woon, E. Wu, and Y. A. Sekercioglu, A Simulation Model of
IEEE 802.11b for Performance Analysis of Wireless LAN Protocols, in Australian Telecommunications, Networks and Applications
Conference (ATNAC), Melbourne, Australia, December 2003.
[5] S. Deering and R. Hinden, Internet Protocol, Version 6 (IPv6)
specification, RFC 2460, December 1998. [Online]. Available:
http://www.ietf.org/rfc/rfc2460.txt
[6] T. Narten, E. Nordmark, and W. Simpson, Neighbor Discovery
for IP Version 6 (IPv6), RFC 2461, December 1998. [Online].
Available: http://www.ietf.org/rfc/rfc2461.txt