mofi paper 2011 pmip
TRANSCRIPT
-
8/13/2019 Mofi Paper 2011 Pmip
1/6
PMIPv6-based Flow Mobility Simulation in NS-3
Hyon-Young Choi and Sung-Gi MinDepartment of Computer and Radio Communication Engineering,
Korea University,
Seoul, South Korea
Email: [email protected]
Youn-Hee HanSchool of Computer Science and Engineering,
Korea University of Technology and Education,
CheonAn, South Korea
AbstractProxy Mobile IPv6 (PMIPv6) is a network-basedmobility support protocol and it does not require MNs to beinvolved in the mobility support signaling. In flow mobilitysupport, although the virtual interface in the MN solved theproblems of vertical handover or flow mobility in heterogeneousnetwork, missing procedure of MN-derived flow handover makesflow mobility in PMIPv6 incomplete.
In this paper, enhanced flow mobility support is proposedfor actualizing and full-covering of the flow mobility supportin PMIPv6. Enhanced flow mobility support is based on a
virtual interface in the MN. Virtual interface makes all physicalinterfaces to hide from the network layer and above. FlowInterface Manager is placed at the virtual interface and FlowBinding Manager in the LMA is paired with Flow InterfaceManager. They manage the flow bindings and are used toselect proper access technology to send packets. Flow mobilityprocedure begins with three different triggering cases, which arecaused by new connection from the MN, the LMAs decision,and request from the MN, respectively. Simulation using NS-3network simulator is performed to provide clear and practicalprocedure of enhanced flow mobility support in PMIPv6.
Index TermsProxy Mobile IPv6; Flow Mobility; NS-3 Net-work Simulator
I. INTRODUCTION
The wireless network is increasing quickly with full growth
of wired network in communication market of recent times.
Equally, with the rapid growth in the number of mobile
subscribers and mobile nodes (MNs) such as cellular phone,
PDA (Personal Digital Assistant), and laptop computer, the
demand for the seamless mobility services such as VoIP (Voice
over IP), media streaming, is becoming one of the most
important issue in the mobility management. Moreover, such
mobile nodes often have more than one wireless interfaces.
As mobile devices can utilize various interfaces to access
the Internet, the demand of inter-technology handover (i.e.,
vertical handover) and flow mobility is becoming another
important issue in the mobility management as well.The network-based mobility management protocol called
Proxy Mobile IPv6 (PMIPv6) [1] is being actively standard-
ized by the IETF NetExt Working Group, and is starting to
attract much attention among the telecommunication com-
munity and Internet community. Unlike the various existing
protocols for IP mobility management such as MIPv6, which
are host-based approaches, a network-based approach such as
PMIPv6 has salient features and is expected to expedite the
real deployment of IP mobility management.
The fundamental foundation of PMIPv6 is based on MIPv6
[2] in the sense that it extends MIPv6 signaling and reuses
many concepts such as the home agent (HA) functionality.
However, PMIPv6 is designed to provide network-based mo-
bility management support to an MN in a topologically local-
ized domain. Therefore, an MN is exempted from participation
in any mobility-related signaling, and the proxy mobility agent
in the serving network performs mobility-related signaling on
behalf of the MN. Once an MN enters its PMIPv6 domain andperforms access authentication, the serving network ensures
that the MN is always on its home network and can obtain
its home address (HoA) on any access network. That is, the
serving network assigns a unique home network prefix to
each MN, and conceptually this prefix always follows the
MN wherever it moves within a PMIPv6 domain. From the
perspective of the MN, the entire PMIPv6 domain appears as
its home network. Accordingly, it is needless (or impossible)
to configure the care-of address (CoA) at the MN.
Despite the fact that network-based mobility management
is considered a better solution than host-based solution, the
progress of completing the detailed scheme is quite late. More-
over, there are some problems about flow mobility support.In this paper, we introduce enhanced flow mobility support.
Enhanced flow mobility support is based on the virtual inter-
face in the MN. Virtual interface makes all physical interfaces
to hide from the network layer and above. Flow Interface
Manager is placed at the virtual interface and Flow Binding
Manager in the LMA is paired with it. They manage the
flow bindings and are used to select proper access technology
to send the packet. Flow mobility procedure is divided into
three cases, which are caused by new connection from the
MN, the LMAs decision, and the MNs decision respectively.
HNP Update Request/Acknowledge message is defined for
LMA decision-based flow mobility. According to the concept
of network-based flow mobility, MN-derived flow handoverneeds the approval of the LMA.
I I . RELATEDW ORKS
There are several primitive solutions for supporting flow
mobility in PMIPv6 [4] [5] [6]. These approaches try to
support flow mobility without MNs involvement because
PMIPv6 is well-designed to support inter-technology han-
dover. However, performing inter-technology handover only in
the network side without MNs involvement has many practical
2011 Fifth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing
978-0-7695-4372-7/11 $26.00 2011 IEEE
DOI 10.1109/IMIS.2011.103
531
2011 Fifth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing
978-0-7695-4372-7/11 $26.00 2011 IEEE
DOI 10.1109/IMIS.2011.103
475
-
8/13/2019 Mofi Paper 2011 Pmip
2/6
issues [9]. Major issue is that the session cannot be sustained
because the configured address of each interface which is
created based on different interface identifier is not matched.
The issues of supporting inter-technology handover in
PMIPv6 mean that the MN should be modified. Moreover,
in flow mobility handover, to sustain the session of traffic, the
IP address must be the same even when the vertical handover
is performed. The MN with physical interfaces only cannot
afford it without any modification. Logical interface supports
[7] [8] [9] are proposed as a solution of network-based scheme
for multi-homing, vertical handover, and flow mobility.
There is a logical interface as well as physical interface
in the MN, and IP layer always exchanges data only with the
logical interface. Therefore, IP layer and above think that there
is only one interface even if the MN has many interfaces. In
the procedure of logical interface, it must have the dispatch
module used to decide which physical interface is chosen to
send data traffic. For the MN under PMIPv6 domain, as home
network prefix (HNP) is assigned to each physical interface,
HNP is the important metric for dispatch module in the logical
interface.PMIPv6 extension [10] is the major solution of IETF NetExt
WG. It assumes that MN has multiple physical interfaces and a
logical interface. IP layer only operates with logical interface.
Hence, sending packets from MNs IP layer are delivered to
the logical interface, and it forwards them to the network
through the appropriate physical interface. All packets that
arrived at physical interfaces are forwarded to logical interface
first, and then, they are delivered to the network layer.
The critical issues for supporting flow mobility are solved by
adapting virtual interface on the MN. However, the solution
of [10] has some issues. At first, there is no MN-initiated
flow mobility. Network-based flow mobility should support
flow handover from an interface to another interface based onthe networks decision. However, it is not the full support of
flow mobility without considering the flow handover request
of the MN, that is, the situation that the MN wants to move a
flow from the currently served interface to another interface.
It does not mean that the network approves the flow handover
whenever the MN wants but the MNs flow handover request
should support under the control of the network. Secondly,
it does not have detailed definition of the flow and the flow
management scheme in the LMA and the MAG.
III. PMIPV6- BASEDF LOW M OBILITY S CHEME
A. Architectural DesignTo manage and process flow handover, basic definitions of
flow, session, and traffic selector is needed.
A flow is defined as a series of packets matching a particular
flow description. To define a flow description, various param-
eters which are mentioned for defining a flow in [3] can be
used. Significant and frequently used parameters among these
are 5-tuple such as source address, destination address, source
port, destination port, and protocol. So, in this paper, these
5-tuple parameters are used for defining a flow.
Fig. 1. MN Stack Architecture
If a flow is represented with 5-tuple informationi, it implies
a direction because there is a source node for sending a traffic
and a destination node for receiving the traffic. So, a flow
can be categorized into inbound flow and outbound flow.
Inbound flow is sent from a CN and finally received at the
MN. Outbound flow is, on the contrary, sent from the MN to
a CN. What LMA manages is which MAG (i.e., which tunnel
interface) is used for an inbound flow. The LMA manages
inbound flow only because there are multiple paths through
MAGs for the inbound flows in the LMA whereas there is a
path for the outbound flow to the CN. As the same reason asthe LMA, the MN manages only outbound flow.
A session can be defined as a symmetrically congruent
pair of flows in case of bidirectional session. If a session is
unidirectional, a flow can be a session.
Traffic selector is the information which is used for traffic
filtering to decide which MAG the traffic is forwarded to. The
definition of traffic selector is the same as that of flow. The
difference is that traffic selector can use a wildcard symbol (*,
asterisk) to match any value among 5-tuple in a flow.
The information of traffic selector and the priority list of
access technologies are managed as a flow binding.
Figure 1 depicts the stack architecture for flow mobility in
the MN. Virtual Interface layer is placed between networklayer and data link layer in the figure. It consists of Virtual
Interface and Flow Interface Manager. All packets from the
lower layer and from the upper layer are passed through the
Virtual Interface layer. Network layer and upper layers only
see Virtual Interface as a network interface. This can solve
many problems including address problem while performing
vertical handover or flow handover. Flow Interface Manager
makes a decision for choosing the physical interface for the
outbound flow from upper layers. Flow Interface Manager has
Flow Interface List for efficient physical interface choice.
Each entry of Flow Interface List contains flow ID, priority,
traffic selector, outbound interfaces, type, and lifetime fields.
Each entry is identified by flow ID. Traffic selector is used formatching the traffic with the entries. The matching order of
each entry is based on priority field. Outbound interface list
has the ordered list of physical interfaces and it is used for
selecting outgoing interface for the matched flow. It is noted
that outbound interface saved in the policy profile has only
access technology type. The information of physical interface
index matching with access technology type is filled after the
entry is inserted at Flow Interface List. In type field, available
values are static and dynamic. Static type means that the
532476
-
8/13/2019 Mofi Paper 2011 Pmip
3/6
Fig. 2. LMA Stack Architecture
entry is created by policy profile and dynamic type means
that the entry is created dynamically. Static-typed entry has
infinite lifetime and dynamic-typed entry has definite lifetime
for validity check.
Like the MN, the LMA also manages Flow Binding List in
Flow Binding Manager as shown in Figure 43. The difference
of Flow Binding Manager is that it manages flow binding for
inbound flow and flow binding list contains binding cache
entry matched with the access technology type.
B. Flow Handover Procedure
If the MN-initiated flow handover is considered, the starting
of flow handover can be made by following three triggering
cases:
1) Flow Mobility by a New Connection: If the MN is
connected to the network with new physical interface,
the flow mobility can be supported. In this case, the flow
binding information in the LMA and the MN is used for
the flow distribution.
2) Flow Mobility by LMAs Decision at a Time: the LMA
can get the link conditions such as the activation of the
wired/wireless link, congestion, the number of serving
nodes, or the number of flows. Therefore, the LMA can
start the flow handover by its own decision.
3) Flow Mobility by MNs Request and LMAs approval:
The MN may want to move a particular flow from served
interface to another interface. As the network-based flowmobility, the network should give a grant for the flow
handover. Hence, the MN must send a request to the
LMA for flow handover, and it can start flow handover
after receiving LMAs approval.
When a new physical interface is activated, there are two
actions that can be considered. One is that there is no flow
handover because all flows are using higher preferred inter-
faces. The other is that flow handover occurs. If flow handover
is expected for the new interface, the LMA assigns the same
home network prefix with the target flow. Otherwise, the LMA
assigns new home network prefix.
Figure 3 shows the flow handover procedure for the flow
mobility by a new connection. There are three traffic flowsover the interface 1. It is assumed that the interface 2 is
higher preferred interface for the flow Y. When 3G interface is
activated, flow Y will be moved into new interface. In binding
update process, because flow handover is expected for the
new interface, the LMA assigns HNP2 to the new interface.
Therefore, 3G interface got HNP2 as a home network prefix.
Figure 4 shows the case of flow mobility by LMAs de-
cision at a time. It also has two possible actions based on
whether the home network prefix of the target flow has been
Fig. 3. Flow Handover Procedure by New Connection
Fig. 4. Flow Handover Procedure by LMAs Decision at a Time
assigned to the target interface when the LMA decides to
move a flow to the other interface at one time. If the target
interface has the home network prefix of the target flow, the
flow handover is performed immediately. Otherwise, it needs
additional procedure to assign the home network prefix of the
target flow to the target interface. HNP Update Request (HUR)
and HNP Update Acknowledge (HUA) [11] are utilized for
this additional process. Then, the flow handover is performed.
Figure 5 depicts the the MN-derived flow handover pro-
cedure. New messages, Flow Handover Indicate and Flow
Handover Acknowledge are defined for flow handover fromthe MN. When the MN decides flow handover, it sends Flow
Mobility Indicate message with traffic selector of the flow
to be moved through the target interface. The message is
received by MAG, and then it sends PBU message with traffic
selector option to the LMA. LMA approves the request and
sends PBA message back to the MAG, and the MAG sends
Flow Mobility Acknowledge to the MN. After receiving flow
handover approval from the LMA, the MN moves the flow to
requested interface.
533477
-
8/13/2019 Mofi Paper 2011 Pmip
4/6
Fig. 5. Flow Handover Procedure by MNs request and LMAs approval
Fig. 6. Simulation Topology
IV. SIMULATION ANDR ESULTS
NS-3 network simulator [12] [13] was used to perform sim-
ulation about flow mobility support. NS-3 is a new simulator
that is intended to eventually replace the aging NS-2. Even
though NS-2 still has a greater number of network models
included in the distribution, NS-3 has a good development
momentum, and is believed to have a better core architecture
and better suited to receive community contributions. NS-3
especially supports multiple interfaces fully in a node whereas
NS-2 cannot support purely.Figure 6 depicts the network topology with three physical
interfaces, WLAN, WiMax(WiBro), and 3G(PPP). Many CNs
are connected to the border router and can access the PMIPv6
domain network by passing through the LMA. The LMA
manages three MAGs. The wired link speed is all 100 Mbps,
and the link delay is 0.01 ms for all links. The link speeds of
WLAN, WiMax, and 3G are 54 Mbps, 75 Mbps, and 1 Mbps,
respectively. Interface queue size is 100 packets for wired
link, 400 packets and 5 seconds of lifetime for WLAN, 1024
TABLE ITRAFFICD ESCRIPTION
Flow Src. Dest. Proto. Rate Start Stop
UDP1 CN:56 MN:10 UDP 7 Mbps 2.0s 25.0s
UDP2 CN:57 MN:20 UDP 3 Mbps 2.0s 25.0s
UDP3 CN:58 MN:30 UDP 0.8 Mbps 12.0s 25.0s
UDP0.1 CN:59 MN:40 UDP 15 Mbps 2.0s 25.0s
UDP0.2 CN:60 MN:50 UDP 10 Mbps 14.0s 25.0s
TABLE IIINITIALF LOW B INDINGL IST INM N
Flow ID Priority Traffic Selector Binding IDs
3 15 (*, *, *, 30, UDP) 3G(-)WM(-)WL(-)
2 15 (*, *, *, 20, UDP) WM(-)3G(-)WL(-)
1 10 (*, *, *, *, UDP) WL(-)3G(-)WM(-)
packets including control packets for WiMax, and 100 packets
for PPP. In order not to change MNs default gateway after
receiving Router Advertisement, all MAGs MAC address on
edge network interface are unified to 00:00:AA:BB:CC:DD.
Therefore, the same link-local addresses for MAGs are shown
in the figure. The MN has virtual interface with Link-ID,00:00:11:22:33:44.
There are five UDP traffic flows from CNs to the MN as
shown in Table I. All traffics use HNP1 as destination address,
which is assigned to the first activated interface. To distinguish
each UDP, source port, and destination port are separated. In
order to check the flow handover operation, the start time of
UDPs is differentiated based on the scenario. The end time is
the same as the total simulation time. UDPs related to the flow
handover are UDP1, UDP2, and UDP3. UDP0.1 and UDP0.2
are used as background traffic.
Static Flow Binding list of the LMA and static Flow
Interface list of the MN are shown in Table II and Table
III, respectively. In static entries, Binding ID and Interface
ID contain only the priority among access technology types.
When an interface is activated or the PMIPv6 binding update
process is completed, the proper binding ID or interface index
is filled at the matched access technology type.
According to the lists, UDP2 is matched with Flow ID 2,
UDP3 is matched with Flow ID 3, and the other UDPs are
matched with Flow ID 1. Hence, the highest priority among
interfaces is WiMax for UDP2, 3G for UDP3, and WLAN for
the others.
Table IV shows the simulation scenario. This simulation
scenario is set up to check the validity of the proposed
procedure. Total simulation time is 25 seconds. It is dividedinto three phases. The first phase is for the flow mobility
procedure by new connection, the second phase is by LMAs
TABLE IIIINITIALF LOW I NTERFACELIST INL MA
Fl ow ID Priorit y Traffic Sel ector Interface IDs
3 15 (*, *, 30, *, UDP) 3G(-)WM(-)WL(-)
2 15 (*, *, 20, *, UDP) WM(-)3G(-)WL(-)
1 10 (*, *, *, *, UDP) WL(-)3G(-)WM(-)
534478
-
8/13/2019 Mofi Paper 2011 Pmip
5/6
TABLE IVSIMULATION S CENARIO
Phase Time (s) Event
- 0.0 - Simulation starts
Phase1 0.0 - WLAN is activated(0.0s 10.0s) 2.0 - UDP1, UDP2, and UDP0.1 start to send
5.0 - WiMax is activated
Phase2 10.0 - 3G(PPP) is activated(10.0s 20.0s) 12.0 - UDP3 start s to send
14.0 - UDP0.2 starts to send15.0 - LMA moves UDP1 to WiMax
Phase320.0 - MN requests to move UDP3 to WLAN
(20.0S 25.0s)
- 25.0 - All UDP stop- Simulation stops
decision, and the third phase is by MNs request and LMAs
approval.
There are several detailed events for each phase and the
reactions for each event are easily expected. When UDP1,
UDP2, and UDP0.1 start to send at 2.0 seconds, these traffics
are delivered to the MN through MAG1 (WLAN) because
WLAN is the only activated interface at that time. Since UDP2has the highest priority for WiMax, when WiMax is activated
at 5.0 seconds, it performs flow handover from WLAN to
WiMax. It can be expected that WiMax gets the same HNP as
the WLAN because the LMA knows that UDP2 will perform
flow handover when new interface is activated. UDP1 and
UDP2 do not change anything when 3G is activated at 10.0
seconds. 3G interface gets new HNP which is different from
that of WLAN and WiMax because there is no influenced
flow for activating 3G interface. When UDP3 starts to send
at 12.0 seconds, it must use 3G interface which is the highest
priority for UDP3 and has been activated since 10.0 seconds.
At 14.0 seconds, UDP0.2 starts to send via WLAN. This traffic
causes the network congestion of WLAN and degradation ofsending rate of UDP1. Hence, by LMAs decision, the LMA
enforces flow handover of UDP1 to the WiMax by inserting
a new dynamic entry at 15.0 seconds. In similar way, the
MN requests a flow handover of UDP3 to the WLAN at 20.0
seconds with the assumption of running out of 3G data service.
Figure 7 depicts the throughput of the flows for three
phases in the MN. After WLAN is activated at the beginning
of the simulation, UDP1, UDP2, and UDP0.1 start to send
via WLAN at 2.0 seconds. Since the bandwidth of WLAN
is 54 Mbps, these traffic cannot serve with their full rate.
When WiMax is activated at 5.0 seconds, UDP2 performs flow
handover from WLAN to WiMax because WiMax has higher
priority for the UDP2. It can be noticed that the throughputof UDP2 increases with more than the serving rate just after
flow handover is performed. The interface queue of WLAN
is the main reason. When UDPs are using WLAN only, the
bandwidth of wireless link is smaller than the arriving rate.
Hence, the packets are queued in the interface queue. When
flow handover occurs, UDP2 packets in the interface queue are
still sending while UDP2 packets through WiMax arrived at
the MN. After this period, UDP2 serves with the full serving
rate. After UDP2 serves as its serving rate, the throughput of
UDP1 and UDP0.1 also increases.
In phase 2, as UDP3 starts at 12.0 seconds, there is a
throughput for the UDP3 in the MN. UDP1 and UDP0.1
which are served through WLAN show the degradation in
throughput from 14.0 seconds due to the start of UDP0.2. After
performing flow handover for UDP1 to the MAG2 based on
LMAs decision at 15.0 seconds, UDP1 is served as its full
rate as well as UDP2 is. The reason of the rapid increase of
throughput is that the MN receives UDP1 packets from both
WLAN and WiMax at the same time as explained situation in
phase 1. It ls also noted that there are oscillations in WiMax
interface because WiMax transfers series of packets at one
time and it has longer interval between transmissions.
In phase 3, it performs MNs flow handover request at 20.0
seconds, which is moving UDP3 from 3G interface to WLAN
interface. In WLAN interface, there are two background traffic
UDP1 and UDP2 while congested. As a result, the throughput
of UDP3 slightly decreases after flow handover is completed.
V. CONCLUSION
PMIPv6 is a network-based mobility support protocol and
it does not require MNs to be involved in the mobility support
signaling. Even though the virtual interface in the MN solved
the problems of vertical handover or flow mobility in hetero-
geneous network, definite procedure of flow mobility has not
been provided yet, and missing procedure of MN derived flow
handover makes flow mobility in PMIPv6 incomplete.
In this paper, enhanced flow mobility support with MN
virtual interface have been presented for actualizing and full-
covering of the flow mobility support in PMIPv6. Enhanced
flow mobility support is based on the virtual interface in the
MN. Virtual interface makes all physical interfaces to hide
from the network layer and above. Flow Interface Manager is
placed at the virtual interface and Flow Binding Manager inthe LMA is paired with it. They manage the flow bindings and
are used to select proper access technology to send the packet.
Flow mobility procedure is divided into three triggering cases,
which are caused by new connection from the MN, the
LMAs decision, and the MNs decision respectively. NS-3
simulation showed that the enhanced flow mobility support can
perform well for possible flow mobility scenarios including
MN derived mobility support.
ACKNOWLEDGEMENT
This work was supported by the IT R&D program of
MKE/KEIT [10035245: Study on Architecture of Future Inter-
net to Support Mobile Environments and Network Diversity].REFERENCES
[1] S. Gundavelli, K. Leung, V. Devarapalli, K. Chowdhury, and B. Patil,Proxy Mobile IPv6, IETF RFC 5213, August 2008.
[2] D. B. Johnson and C. E. Perkins and J. Arkko, Mobility Support inIPv6, IETF RFC 3775, June 2004.
[3] G. Tsirtsis, Traffic Selectors for Flow Bindings, draft-ietf-mext-binary-ts-04, February 2010.
[4] M. Hui, G. Chen, and H. Deng, Service Flow Identifier in Proxy MobileIPv6, draft-hui-netext-service-flow-identifier-03, July 2010.
[5] F. Xia, and B. Sarikaya, Flow Binding in Proxy Mobile IPv6, draft-xia-netext-flow-binding-02, June 2010.
535479
-
8/13/2019 Mofi Paper 2011 Pmip
6/6
Fig. 7. UDP Throughput Trace at the MN
[6] R. Koodli and K. Chowdhury, Flow Handover for Proxy Mobile IPv6,draft-koodli-netext-flow-handover-01, October 2009.
[7] H. Yokota, S. Gundavelli, T. Tran, Y. Hong and K. Leung, VirtualInterface Support for IP Hosts, draft-yokota-netlmm-pmipv6-mn-itho-support-03, March 2010.
[8] T. Melia, Ed., Logicla Interface Support for multi-mode IP Hosts,draft-melia-netext-logial-interface-support-01, July 2010.
[9] S. Krishnan, H. Yokota, T. Melia, and C. Bernardos, Issues withNetwork based Inter-Technology Handovers, draft-krishnan-netext-intertech-ps-02, July 2009.
[10] CJ. Bernardos, Ed., Proxy Mobile IPv6 Extensions to Support FlowMobility, draft-bernardos-netext-pmipv6-flowmob-02, Feburary 2011.
[11] T. Tran, Y. Hong, and Y. Han, Flow Mobility Support in PMIPv6,draft-trung-netext-flow-mobility-support-01, October 2010.
[12] NS-3 Network Simulator, http://www.nsnam.org[13] H. Choi, et. al, Implementation and Evaluation of Proxy Mobile IPv6
in NS-3 Network Simulator, In Proceedings of the IEEE InternationalConference on Ubiquitous Information Technology & Applications2010, December 2010.
536480