Fast Handoffs with GPS Routing
For Mobile IP
Mustafa Ergen, Sinem Coleri
(ergen,csinem)@EECS.Berkeley.Edu
Department of EECS
University of California, Berkeley
Abstract
Mobile IP is designed to support uninterrupted
connectivity of mobile computers as they roam from
place to place. The general structure and the goals of
today’s mobile IP protocol are described. The previous
works to achieve fast handoff and to decrease the
control traffic is given. Then, our proposed system
using GPS (Global Positioning System) with its handoff
mechanisms and the changes that it brings to the
current system together with its improvements in
handoff and traffic are outlined. The simulation results
using network simulator (ns2) and C++ are presented.
1. Introduction
The Internet is expected to become a huge wired
network that consists of wireless attachment points at the
edges that connect the wireless mobile users inside their
coverage area to the wired part of the network.
The primary aim of Mobile IP is to support
mobility in addition to portability for computers without
charge. Portability, which means the operability of the
computer at any point of attachment except the time during
which it changes its point of attachment, is achieved today.
The aim of Mobile IP is to maintain these computers in
almost continuous reachability even when moving.
Wireless connections are expected to be
performed through radio or infrared attachments like the
cellular telephone technology. However, the limited
geographical coverage and the charging problems are not
yet solved for these systems. The coverage problem is only
solved by cellular telephone technology. However, this
brings the charging of connections. By ignoring the social
and economic concerns, we will work on the adaptation of
this cellular idea to Mobile IP.
As wireless computing becomes more prevalent,
the need for the achievement of real-time services through
wireless links increases. The need for real-time services
can only be solved through tiling the world into smaller
cells. These smaller cells bring also transceivers of lower
power for mobile nodes.
1
Although higher bandwidth, which is required for
real-time services, increases the quality of services
provided to users, small wireless cells cause frequent
handoffs. Frequent handoffs bring the frequent location
updates so an increase of the latency in the re-routing of
packets in handoffs. Therefore, although the wireless
sources improve in the future, high quality of service
cannot be achieved without making the handoffs
transparent to users.
Our system achieves fast handoffs by both using
hierarchical structure of the network [11] and GPS (Global
Positioning System) devices in the routers. According to
the hierarchical structure, the network is thought to be
composed of administrative or geographical domains. Each
domain has a hierarchical tree of foreign agents that
includes a domain foreign agent at the top. Using a GPS
device, each router knows its own position. With a special
advertisement messaging scheme, they make other routers
inside the domain learn their positions. They will then use
this position information to send the packets directed to
one foreign agent to the adjacent foreign agents in addition
to this foreign agent. Besides, we use this geographical
information of domain foreign agents to decide whether to
send the registration request to home agent or to the
previous domain foreign agent of the mobile host, which
brings local home agent functionality to domain foreign
agents if mobile host does not go away from this domain.
This paper addresses the course project of a
graduate course on High Speed Communication Networks
given at UC-Berkeley by Prof. Jean Walrand in the 20002001 academic year.
Different Mobile IP systems have been proposed
by several companies. In section 2, a brief overview of the
general structure of their proposed systems and the
fundamental differences between them are described. In
Section 3, the previous works on achieving fast handoff
and less traffic of advertisement messaging are introduced.
In section 4, the basic architecture of our system is
presented with the possible improvements that it brings to
the current structure. In Section 5, the performance results
of our simulations in Network Simulator (ns2) and C++
are presented. Section 6 highlights the areas for future
work. Section 7 concludes the paper.
2.Current Mobile IP Architecture
The basic structure of today’s Mobile IP has four
entities:
Mobile Host (MH): Host or router that is portable
with wireless network hardware while retaining a constant
IP address.
Home Agent (HA): A stationary router on a
mobile host home network that keeps the current location
information of this host and forwards the packets
accordingly.
Foreign Agent (FA): A router that has a
connection with the mobile host as it moves and that
delivers the packets coming to it by having both a radio
and wired Internet connection.
Corresponding Host (CH): Any host on the
Internet that sends packets to the mobile host.
HA
CH
Source
Location
Directory
(not enhanced by mobility binding) the routing is still via
HA. Also, there are proposals which address this problem.
The fundamental differences between most of the
proposed Mobile IP systems are in the following areas:
(i) How does a HA know where a MH is?
(ii) How can CHs send directly to the current FA
of the MH avoiding the triangular trip via HA?
The system objectives in judging these different
systems are performance as non-mobile protocols,
scalability and security. The first goal, achieving the
performance as non-mobile protocols, requires nontriangular routing since the packets go directly to the
destination address in non-mobile protocols. Additionally,
this performance issue requires fast handoffs when the user
moving from one cell to another since the user should not
understand that he is roaming from cell to cell. The second
goal, scalability, is necessary in order to avoid the
bottleneck of the MH location database as the number of
MHs increases. The third goal, security, is necessary in
order prevent the redirection of packets to malicious
eavesdroppers.
3. Previous Work on Fast Handoffs
Encapsulation
FA
MH
Decapsulation
H
Figure 1: General Structure of Mobile IP
The earliest mobile IP protocols have proposed
the following scheme: When a CH wants to send a packet
to a MH, it assigns the destination address of the packet as
the constant IP address of the MH. Then the packet goes to
the HA of the MH. The HA has the location information
for the MH as another IP address, which is named care-ofaddress. At this stage, the HA tunnels the packet by
encapsulating it with the destination address as the care-ofaddress of the MH. The packet is received at the tunnel
end point, which is either the FA or the MH and delivered
to MH. When the MH moves, it informs the HA of the
current location.
The basic problem with the above system is the
triangle routing. Datagrams with destination address of
MH have to first go to the HA even if CH and MH are near
each other and both away from HA. This problem is solved
with mobility binding, which is a functionality of the CH to
have an entry for MH. When a CH wants to send a
datagram to MH for the first time, the normal procedure of
packet forwarding is implemented. Then either HA or MH
informs the CH of the current care-of-address of MH so
that the following packets are forwarded directly to this
care-of-address. When the MH moves from one FA to
another FA, again either the MH or the HA informs the
CH of the current location so that CH will send the packets
directly to MH avoiding triangle routing. For ordinary CH
2
The problem of excessive mobility management
traffic and time has been addressed in various works
during the past years. The main problem in the system
developed to solve the triangle routing problem is that it
requires a lot of signaling between the MH, HA and CHs
and a lot of time to reroute packets even if MH moves
inside a very small area.
One of the proposed solutions is the hierarchical
mobility
scheme
proposed
by
Caceres
and
Padmanabhan[6]. In this system, the network consists of
domains, each of which includes a domain foreign agent
keeping track of all the regional movements of the MHs
inside the domain. The disadvantage is that the updating
procedure of the mobile node entry in the routing table of
the domain foreign agent takes time, which may cause the
loss of MH packets coming in between. To solve this
problem, they propose the buffering of the packets coming
to old FA. This can decrease the number of lost packets
due to handoff. However, in the real-time case, learning
the new place of the MH and sending them to the new
location can take so much time that it is better to ignore
them instead of causing additional overhead in the
network.
Another alternative proposed by Charles
Perkins[11] is to form a hierarchy of foreign agents. Every
foreign agent is supposed to know its ancestors in the tree.
Then the registrations due to the handoff from one FA to
another will only be up to the lowest common ancestor in
the tree. This prevents the trip of the registration messages
to the domain foreign agent each time the MH moves.
However, all FAs in the route from the domain foreign
agent to the MH’s FA have to keep the MH location
information and this may cause bottleneck as the number
of MHs increases. In addition, when their lowest common
ancestors is very high up the tree, the handoff can again
take a lot of time.
The proposed scheme by Seshan et al. [7]is to
assign each MH a multi-cast address. When a MH registers
a FA, this FA identifies the handoff targets and makes
them join to the multi-cast group. These handoff targets
buffer the last packets to transmit them in case of a hand–
off. This avoids the wasteful trip to HA to make it aware of
the new position. However, multicast address conflicts
may occur if one packet passes through another network
where the same multi-cast address is used.
The solution brought to the multi-cast conflict
problem by C. L. Tan et al. [5] is to assign a multi-cast
address to each MH by domain foreign agent inside the
domain. They propose dynamic virtual macro-cells (DVM)
for each FA containing the FAs around it and assign a
multicast address to each DVM in addition to each mobile
in the domain. They propose the allocation of a range of
multi-cast address to each domain to eliminate the multicast address conflict completely. However, this will
require a continuous messaging between all routers to
learn about the new multi-cast addresses allocated by
them. Moreover, in the last proposals, an intelligent
neighborhood discovery mechanism is needed to eliminate
the manual configuration of handoff targets and DVM
members. Also, an inter-domain handoff mechanism needs
to be described for these systems since they ignore the
handoffs between different domains by using hierarchical
structure.
4. Proposed System
Our proposed system aiming at improving
handoff for real-time applications assumes that the
network contains the domains, which have a hierarchical
structure of FAs, and that each router knows the positions
of all routers inside its domain. This can be achieved by a
special advertisement messaging and GPS (Global
Positioning System) devices at each router although
today’s routers do not have such a capability. Then the
packets coming to MH will first go to the domain foreign
agent. At this time, domain foreign agent and the other
routers will cooperate with each other by using location
information in order to achieve fast handoffs. Our system
reaches smooth handoff goal even in inter-domain
movements by using location information.
4.1 Structure of the Network
Home Network
Internet
The Internet is composed of administrative or
geographical domains as shown in Figure 2.
The structure of each domain is given in Figure 3.
Every domain has a hierarchy of foreign agents that
includes a domain foreign agent at the top. Every FA
knows its ancestors and its children in the tree.
Gateway
DFA
FA1
FA3
FA2
FA4
FA5
Figure 3: Structure of the domain
4.2 Handoff Mechanisms
4.2.1 Intra-Domain Handoffs
Our system uses the hierarchical structure in order
to eliminate the wasteful trip to HA for each movement of
the MH inside the domain. Each domain has one domain
foreign agent (DFA), which is the ancestor of all other
routers inside this domain.
The communication between the MH and FA is
achieved through the wireless network hardware of the
MH and one of the radio interfaces of the FA. Each FA
broadcasts a beacon packet with a beacon period. When a
MH wants to register to a FA, it registers with the DFA
and sends the address of the DFA to its HA. Therefore,
when a MH moves inside this domain, it does not need to
inform HA of its current FA.
Switching from one FA to another can still be a
problem inside the domain for real-time services since this
kind of services is not tolerable to delay. To solve this
problem, we propose to send each datagram directed to a
specific FA to all the FAs adjacent to this FA since the
MH is expected to move to one of these FAs if it changes
FA. While the current FA of the MH is forwarding packets
to the MH, the adjacent FAs buffer the last packets in
order to forward these packets to the MH if a handoff
occurs.
Our system involves two ways of achieving
smooth handoff by sending the packets to all the FAs
adjacent to the FA of the MH without needing multi-cast
address.
First way of Fast Intra-Domain Handoff:
According to the first way, the following
exchange of messages takes place:
Foreign Network
Figure 2: Structure of the Internet
3

DFA takes the encapsulated packet coming from HA
or CH and decapsulates it.
 DFA looks at the destination address and finds from
its visitor list the care-of-address of MH.
 DFA finds the adjacent FAs of the MH’s FA from the
location-IP address table. Then from its routing table,
it decides to which of its branches it should send these
packets in order to cover all of the adjacent FAs, and
sends these packets to these branches by encapsulating
them.
 When the routers at the end of these branches take the
packets, they decapsulate the packet, check the
location-IP address to find the adjacent FAs of the
MH’s FA and decides to which of its branches it
should send these packets, and send them with
encapsulation.
By applying this method at each of the routers taking
the encapsulated packets, we achieve the multi-cast
forwarding of datagrams without allocating any multi-cast
address.
the new FA and the registration time to DFA. Then it
immediately starts to receive packets from the new FA.
The disadvantage of the first method is that it sends the
packets to all of the FAs adjacent to the current FA all the
time. This increases the traffic in the domain even if the
MH does not move at all.
The advantage of the second way is that it does
not create traffic if the MH does not move. It only creates
extra traffic when FA does not know the new location of
the MH. However, if the number of links between the old
and new FAs is very high, the forwarding of packets from
the old to new FA can take time and can increase the realtime delay.
The second way of intra-domain handoffs can
also be improved with appropriate additional hardware.
For instance, MH may order FA to begin to send the data
to the adjacent FAs when it starts to receive weak signal
from FA. Moreover, MH can send the location direction of
its movement in order that its packets come to FAs in that
direction.
Second way of Fast Intra-Domain Handoff:
According to the second way, the following
exchange of messages occurs:
 DFA takes the encapsulated packet coming from HA
or CH and decapsulates it.
 DFA looks at the destination address and finds from
its visitor list the FA that the MH is connected to and
sends the packet to that FA.
 The FA taking the encapsulated packet decapsulates
the packet and checks whether the destination address,
the constant IP address of the MH, is inside its visitor
list.
 FA decides whether to send the packet over the radio
link, send the packets to adjacent FAs or ignore them.
 If MH is inside its visitor list, it sends the packet
via radio link.
 If MH is not inside its visitor list but inside its
cache list, this means that MH has just left this
FA. If the destination address of this cache list
entry is “all ones”, this means that FA does not
know the new care-of-address of the MH. Since
MH has moved to one of its adjacent routers, it
finds its adjacent routers from the location-IP
address table and sends the coming packets to its
adjacent FAs. If the destination address is not “all
ones”, this means that FA knows the new care-ofaddress of the MH and send the incoming packet
only to this address.
 If MH is neither inside its visitor list not inside its
cache list, it buffers the packet.
Performance of Intra-Domain Handoff Mechanisms:
Our system aims at providing real-time services,
which means providing minimum delay for each packet
and minimum number of lost packets.
Delay
We can separate the time needed to complete the
handoff into two parts: the discovery time and the handoff
latency.
The discovery time is the time from roaming out
of the old FA’s cell until hearing beacon from the new FA.
In the worst case, this discovery time is equal to the
beacon period. The worst case situation occurs when the
MH leaves just immediately after the reception of the
beacon packet.
The handoff latency is the time from sending a
registration request to new FA until it accepts the request.
This means that this time includes the travel time of the
registration request message from MH to DFA and the
travel time of the registration reply message from DFA to
MH.
According to the first way of intra-domain
handoffs, the time necessary for the MH to take the first
packet is given by the sum of the discovery time and
handoff latency time.
Comparison of Two Ways of Intra-Domain Handoffs:
The advantage of the first way is that it provides
the minimum delay for the reception of the next packet
from the new FA. The MH only waits for the discovery of
4
Handoff time=discovery time+handoff latency
According to the second way of intra-domain
handoff, the handoff time depends on how much time is
necessary to go packets from old FA to new FA. This
depends on the traffic of the links that the packets pass by
and on the number and length of the links between them.
We define the time necessary for packets to go from the
old FA to new FA as re-routing time. Handoff time will be
given by the following formula in the second case:
If rerouting time>(discovery time+handoff
latency)
Handoff time=rerouting time
else
Handoff time=discovery time+handoff latency
Packet Loss
If there were no handoff mechanism described
above and all packets going to the old FA were lost, the
number of lost packets would have been found by the
following formula:
number of packets lost = (discovery time +
handoff latency)/(packet inter-arrival time)
In the worst case, when discovery time is equal to
beacon period, this gives the following formula:
number of packets lost = (beacon period +
handoff latency)/(packet inter-arrival time)
The packet loss of our system implementations
depends on the size of buffers kept for each MH.
According to the first way of intra-domain
handling, where each adjacent FA takes all packets coming
to a FA, the number of lost packets are given as follows:
Registration time = (discovery time + handoff
latency)
If buffer size>(registration time/packet interarrival time)
packet loss=0
else
packet loss=(registration time/packet interarrival time)-buffer size
According to the second way of intra-domain
handoffs, where the FA sends to its adjacent FAs only if it
does not know the new location of the MH, the minimum
number of saved packets is given as follows:
Minimum number of saved packets=
(Registration time-re-routing time)/inter-arrival time
4.2.2 Inter-Domain Handoffs
The inter-domain handoffs will be needed more
and more as the battery-power of laptop computers
increases and the world becomes tiled into small wireless
cells. Therefore, our system tries to achieve fast handoffs
between the domains by using geographical information in
contrast to the other proposed mechanisms.
As the world is tiled into small cells, the
probability that a user will pass from one domain to
another increases. Suppose a user powers up his laptop
computer in Berkeley campus domain. Then he decides to
go to Los Angeles. He will work during the journey in his
car. He wants to stay connected to Internet during his
entire journey. Since today’s laptop computers have
enough electrical storage to last an entire day, he will be
5
able to stay connected to Internet with a proper protocol
and tiling of the places that he passes through into small
cells. Our system tries to make all handoffs transparent to
the user during his journey, which requires fast interdomain handoffs.
Our system gives local home agent functionality
to the DFA that the user is connected first as long as the
user roams inside a specific range of DFAs. This range is
found through a global angle and the distance between the
first DFA that the user has connected and home agent.
The procedure of changing domains without interdomain fast handoffs upon reception of the registration
request from a MH is as follows:
 The MH sending the registration request is not inside
the visitor list of the DFA receiving its packet.
 DFA sends registration request to the HA of the MH,
takes registration reply from HA, updates its visitor
list with the MH entry and sends the registration reply
to MH.
The procedure of changing domains with inter-domain
fast handoffs upon reception of the registration request
from a MH is described as follows (This is the procedure
of our system and we assume that the registration request
coming from the MH includes the location and IP address
of its old local home agent and the IP address and location
of its HA):
 The old local home agent DFA entry in the
registration request is not the same as the address of
the DFA receiving the packet.
 DFA decides whether to become the new local home
agent with the global tube algorithm.
 If it decides to become the new local home agent,
it sends the registration request to HA and takes
the registration reply from HA.
 If it does not decide to become the new local
home agent, it sends the registration request to the
old local home agent and receives the registration
reply from it.
 Then it sends the registration reply to MH by
encapsulating it with the care-of-address of the MH,
which includes the new local home agent information.
Global tube algorithm:
The main aim of the global tube algorithm is to
eliminate the unnecessarily long handoff times and the
traffic created to inform HA when MH changes domains.
Our system aims to use the DFA of the previously
connected domain as the local home agent and send
registration requests to the DFA of this domain as long as
the DFAs of the following domains lie inside an R radius.
This R radius is determined by the global tube in Figure 4.
LIFETIME
TYPE
New
DFA
Old
DFA
HOME ADDRESS
d2
d1

TARGETED DOMAIN FOREIGN AGENTS ADDRESS
R
CARE OF ADDRESS
HA
IDENTIFICATION
LOCAL HOME AGENT IP ADDRESS
LOCAL HOME AGENT GEOGRAPHICAL LOCATTON
Figure 4: Global Tube
HOME AGENT GEOGRAPHICAL LOCATION
The need for the global tube arises from the
necessity of an R radius depending on the location of the
DFA treated as a local home agent. If the local HA DFA is
near the HA, it is better to send the registration request to
HA instead of this DFA even if the difference between the
two DFAs is small. On the other hand, if the previous DFA
is far from the HA, it is better to send the registration
request to DFA instead of HA for the same distance
between the two DFAs. Therefore, the R radius must
decrease as the local HA DFA becomes nearer to the HA.
The global tube algorithm is as follows:
 Find the distance between previous DFA and HA, d 1.
 Find the R distance from the global  angle and the
distance d1 as follows:
R = d1 * tan()
 Find the distance between the local HA DFA and the
current DFA, d2.
 If d2>d1
The current DFA is the new local home agent
else
The previous DFA stays as local home agent
OTHER EXTENSIONS……
Figure 5: Registration Request Format
As can be seen from Figure-5, the only change in
the packet header is the association of old local home
agent DFA IP, old local home agent DFA location and HA
location to the extension of the Mobile IP message format.
The encapsulated format of this message is
expected to go from the FA that the MH is connected to
the DFA, and then from the DFA to either HA or the
previous DFA that the MH has visited.
The registration reply format that our system uses
is given in Figure 6.
TYPE
CODE
LIFETIME
HOME ADDRESS
4.3 Changes in Control Message of Mobile IP
TARGETED DOMAIN FOREIGN AGENTS ADDRESS
IDENTIFICATION
4.3.1 Changes in Registration Messages of
Mobile IP
LOCAL HOME AGENT IP ADDRESS
LOCAL HOME AGENT GEOGRAPHICAL LOCATTON
Our system uses the same Mobile Agent
discovery mechanism as Mobile IP. On the other hand, it
requires using the extension parts of the registration
messages of Mobile IP due to the use of domain foreign
agent concept and location information.
The registration request and reply messages are
used to inform the home agent of the current care-ofaddress of the MH and tells the HA how long it wants to
use this care-of-address.
The registration request format that our system
uses is given in Figure 5.
6
HOME AGENT GEOGRAPHICAL LOCATION
Figure 6: Registration Reply Format
As can be seen from Figure-6, the only change in
the packet header is the association of new local home
agent DFA IP, new local home agent DFA location and
HA location to the extension of the Mobile IP message
format. This registration reply message extensions is used
to inform the MH of the new local home agent
specifications.
4.3.2 Location Advertisement Messages
Our system assumes that each FA contains a GPS
device so knows its current location. Location
Advertisement Messages are used in order to send this
location information to all routers inside the same domain.
Each router is expected to have a location-FA
table, the entries of which include the IP address and the
location of each foreign agent care-of-address inside its
domain.
Each foreign agent advertises its position to its
neighbors. When one router gets the position of a foreign
agent, it has to make a decision whether to send this packet
to its neighbors and update its table. We know that each
router in Internet has a routing table and make routing
decisions according to this routing table. In order to
prevent creating loops and too much traffic with each
advertisement messages, every router will decide whether
to process a location advertisement message if it comes
from the router, which is the next router on the shortest
path going to the source of this location packet. The
processing of the location advertisement message includes
sending of the packet to all the neighbors of this router
except the router that the packet is coming from and
updating the table according to this location information.
This location advertisement messaging provides
each router to know all the FAs around each FA. This
avoids the manual configuration of handoff targets of each
FA.
assumes that each FA in the network covers only one cell
and acts as a base station.
4.4.1.2 Domain Foreign Agent Tasks
Processing of registration request message:



4.4 Entities of Our System
4.4.1 Domain Foreign Agent
4.4.1.1 Domain Foreign Agent Tables
Domain Foreign Agent (DFA) has two types of
tables: visitor list and location-FA table.
Visitor list keeps all the MHs visiting the domain
at this time. It includes the constant IP address of each of
these MHs and their current care-of-address. This current
care-of-address can either be the care-of-address of a FA
inside the domain that MH is connected to at that time or
the address of the DFA of another domain if MH has
moved to another domain while giving to this DFA a local
home agent functionality.
Location-FA Table is the table that keeps the
location information of each FA inside the domain
together with its care-of-address. This will be used in the
forwarding of MH packets to multiple FAs to provide fast
handoffs.
.
There may be various structures different from
our scheme. FA functionality and base station functionality
might be separated. Each FA may have more than one base
station so can cover more than one cell. This indicates that
Location-FA table can vary depending upon the network.
Although the structures vary, idea is the same. Our system
7
Upon reception of a registration request message,
DFA checks its visitor list to see whether MH is new
in the domain or was already roaming inside the
domain.
If the MH is in its visitor list, this means that this is an
intra-domain handoff. Therefore, the DFA only
updates the MH entry of its visitor list with the new
care-of-address.
If MH is not in the visitor list, this means that this is
an inter-domain handoff. It will decide whether to
register to HA or to the old local home agent.
 If the old local home agent entry of the
registration request is home agent of the MH
then DFA becomes the new local home agent of
this MH since this means that MH has just
powered up his computer.
 If the old local home agent entry of the
registration request is not home agent in interdomain handoff, then it will decide whether to
become the new local home agent or to connect
to the send registration request to the old one.
 If it decides to become the new local home
agent, it sends registration request to HA. HA
sends registration reply. Then the DFA
produces a new visitor list entry for this MH
and sends a registration reply to the MH.
 If it decides to know the old local home agent
DFA as the new local home agent, then it
only sends registration request to this DFA.
This old DFA updates its visitor list entry
with this MH and the current DFA. Then
sends a registration reply to the new DFA.
The new DFA produces a new visitor list
entry for this MH and sends a registration
reply to the FA of the MH.
Sending packets to MHs:

When a DFA takes a packet with destination address
as one of the MHs in its visitor list, then it directs
these packets to the corresponding care-of-address in
its visitor list.
4.4.2 (Normal) Foreign Agent
4.4.2.1(Normal) Foreign Agent Tables
(Normal) Foreign Agents (FA) has three types of
tables: visitor list, cache list and location-FA table.
Visitor list contains the MHs connected to this FA
at that time. It includes the constant IP address of all these
MHs. It updates this table according to the responses
coming from MHs to each beacon packet that it sends.
Cache list keeps all the MHs recently visited this
FA. When FA does not take a response for a MH in the
visitor list, it puts this entry to the cache list with a specific
lifetime. If it does not know the new care-of-address of this
MH, it puts an “all ones” entry as the new location of the
MH. If it learns the new place of the MH, it puts the new
care-of-address of the MH to the table. This list is used for
the second way of intra-domain handoff.
Location-FA table keeps the location information
and the care-of-address of each FA inside the domain.
4.4.2.2 (Normal) Foreign Agent Tasks
Processing of registration request message:



FA sends beacons periodically.
If it sees that a MH is newly connected to itself, it will
send the registration request to its DFA and update its
visitor list entry with this new MH entry upon the
reception of the registration reply from the DFA.
If it sees that some MHs are disconnected from itself,
it puts these entries to the cache list with a specific
lifetime.
Sending packets to MHs:
(for the first way of intra-domain hand-off)
 When a FA takes an encapsulated datagram sent to its
advertised care-of-address, it compares the inner
destination address to its entries of its visitor list.
 If this inner destination address matches the address of
one of the entries of the visitor list, it forwards the
decapsulated datagram to the MH.
 If this inner destination address does not match any
entry of visitor and cache list, it buffers the coming
packets in case there is a handoff.
(for the second way of intra-domain hand-off)
 When a FA takes an encapsulated datagram sent to its
advertised care-of-address, it compares the inner
destination address to its entries of its visitor list.
 If this inner destination address matches the address of
one of the entries of the visitor list, it forwards the
decapsulated datagram to the MH.
 If this inner destination address matches the address of
one of the entries of the cache list, and the
corresponding new location address is an “all ones”
address, then it finds the adjacent FA addresses from
its location-IP address table and sends the
decapsulated packet by encapsulating it with the
destination address of these FAs.
 If this inner destination address matches the address of
one of the entries of the cache list, and the
corresponding new location address is not an “all
ones”, then it sends the decapsulated packet by
encapsulating it with this destination address.
8

If this inner destination address does not match any
entry of visitor and cache list, it buffers the coming
packets in case there is a handoff.
4.5 Security
Mobile computing environment is very different
from the ordinary computing environment in terms of the
security. Since the links are wireless, the traffic can easily
be disrupted by malicious users. Our system provides the
security by using the Mobile IP features.
The protection against these malicious users is
obtained by inserting a time stamp or randomly generated
number into the identification field of the registration
request and reply messages. The home agent and the
mobile node have to agree on time stamp or random
number values. There are three kinds of authentication
extensions named as mobile-home, mobile-foreign and
foreign-home
authentication
extensions.
Security
Parameter Index (SPI) in these authentication extensions
defines the security context of authenticator computing.
The first connection of the MH to the Internet is
through the HA. HA and MH share keys that are
configured manually.
The manual configuration of keys between MH
and HA is not enough in our system since there are local
domain registrations. The problem is solved as it is solved
in the hierarchical Mobile IP structure. The home agent
will pick a registration key for use by MH and DFA. If the
DFA and HA share a security association, the FA wants
the registration key using this association and takes the
result back as a result of the registration reply. The HA
informs MH of this value using the usual key between
them. If the FA and HA do not have security association
but have public key, they use the public key with the
registration.
Our system sometimes requires sending a
registration request from one DFA to another DFA. We
assume that these DFAs share a security association or
public key in order to learn the registration key.
The procedure of sending packets to the adjacent
FAs is also secure in our system. When an adjacent FA
receives packets, it decapsulates and check its visitor list to
find appropriate MH whose home address is the same with
the decapsulated packet`s original source address. If it
finds, this means that secure registration is achieved so it
sends the packets. If it does not, it buffers them and does
not relay the packets unless the MH finishes its registration
procedure with the appropriate DFA.
5. Simulation Results
The performance of our algorithm can be
measured by examining the decrease in packet loss and
delay during a handoff.
The simulation scenario is split into two: IntraDomain Handoff and Inter-Domain Handoff. The
simulations are performed with Network Simulator (ns2)
and C++. We have measured packet loss and handoff time
in different wired link delays, wireless link delays and
beacon periods.
The handoff time is calculated by measuring the
round trip time of a 64 byte packet and by using the
beacon period. The packet loss is found by using the
handoff time calculated in ns2 and the network topology
formed in C++.
5.1. Simulation of Intra-Domain Handoff
The aim of the Intra-Domain Handoff simulation
is to measure how much time is required for a MH to
complete handoff, how many packets are saved by our
system compared to the current Mobile IP structure and
what is the inter-arrival time of the packets just before,
during and just after the handoff. Inter-arrival times,
handoff delay and packet loss is important for real time
systems.
The first part of the handoff time, detection time,
simply depends on beacon period and agent solicitation.
Maximum detection time can be taken beacon period
unless MH sends agent solicitation.[5]
The second part of the handoff occurs after MH
detects the FA. It includes sending of a registration request
message to DFA and waiting for the registration reply
message.
According to this setup, the round-trip time for a
typical 1Mbps bandwidth and 4ms delay wireless link and
1ms delay wired link is found to be 19ms in Figure 8.
Registration Time with Variable Wireless Link
22
20
Round Trip Time
18
5.1.1 Simulation Scenario
DFA
1.2MB wireless
1MB wireless link
2MB wireless link
16
14
12
FA
FA
10
1
2
3
4
5
6
Wireless Delay
Figure8: Wireless Delay
M
H
The simulation results for different wired link
delays and a 1Mbps bandwidth and 4ms delay wireless
link are given in Figure 9. The handoff time is expected to
be approximately these round-trip times plus the beacon
period of the FAs.
Figure.7 Intra- Domain Handoff Scenario
Our simulation scenario is shown in Figure 7. In our
simulations, wired and wireless links are taken to be
10Mbps duplex and 1Mbps duplex links respectively. In
order to be able to examine the system for a real-time data
flow, the source of packets coming to MH is chosen to be a
packet audio source. This brings 20ms as packet interarrival time and 200 bytes as average packet size.
We measured :
 Registration time
 Packet sending time from old FA to new FA
 Calculated number of packets buffered in the new FA
 Maximum required beacon period according to
different buffer size
 Packet inter arrival time
by the help of this scheme.
Registration Period with variable Wired Line
40
35
30
Round Trip Time
M
H
25
20
4ms Wless Delay,10 MB Wired R, 1MB
Wless R
15
10
5
0
0
1
2
3
4
5
6
7
Wired Delay
Figure9: Wired Delay
5.1.2 Handoff Delay Performance
5.1.3 Optimum Beacon Period
The aim of this simulation is to measure the
handoff time of the MH roaming inside a domain. The
handoff time can be divided into two: FA detection time,
and registration requests and reply time between new FA
of the MH and DFA.
The buffer size and beacon period can be seen as
variables for the system to satisfy some specifications.
Therefore, the simulation to find the maximum beacon
period that FAs can use with different buffer sizes is
performed. The result is that as the size of the buffer
increases, the necessary beacon period increases for no
9
As can be seen from Figure 11, the maximum
time interval between the arrival of packets is 127ms.
The second way of the intra-domain handoff of
our system also brings improvement to bandwidth usage.
According to the first way of intra-domain handoff, when
MH is connected to a FA, all of the adjacent Fas take one
copy of each packet but there may be situations when MH
does not make any movement and remain attached to the
same FA. In that case, bandwidth in other cells is also
occupied.
For our second intra-domain handoff proposal,
the measurement of the time necessary for old FA to send
packet to a new FA and then the calculation of the number
of the packets that will be buffered when MH involves in
handoff is performed. Detection time and traffic may vary
and the number of buffered packets cannot be a constant
value. With a detection range between 50ms and 200ms,
the number of buffered packet changes in the range from 2
to 9 packets.
Figure10: Maximum Required Beacon Period
packet loss, as expected. In addition, the observation is that
as the wired link delay increases, the necessary beacon
period is almost constant but decreases slightly. The reason
for this is that as the wired link delay increases, the
necessary round trip time increases as much as the packet
arrival-times. However, since there is a constant discovery
time, the number of packets coming to the new FA at that
time slightly decreases.
Sending Time a packet from Old FA to New FA
14
12
Sending Time
10
5.1.4 Inter-Arrival Time Performance
A good inter-arrival time performance is
necessary for real time services. It has been demonstrated
that human ear can tolerate approximately 200 ms delay at
maximum.
The inter-arrival time performance simulation is
performed with a 100ms beacon period and a buffer size of
10 in 1Mbps bandwidth and 4 ms wireless delay and
10Mbps bandwidth and 4ms wired delay.
8
10MB Wired Rate
6
4
2
0
0
1
2
3
4
5
6
Figure12: Sending Time From old FA to new FA
Buffered Packets
12
50ms Beacon Period
100ms Beacon Period
150ms Beacon Period
200ms Beacon Period
10
Number of Packets
8
6
4
2
0
1
2
3
4
5
Wired Delay
Figure13: Number of Buffered Packets
Figure11: Time Interval Between Packets
10
7
Wired Delay
6
5.2. Simulation of Inter-Domain Handoff
Registration Request+Replay Time
80
70
60
50
Time to Register
In this part of the simulation, the comparison of
the round trip time of the current hierarchical structure and
our system with simulation scenario of MH movement in
different domains is performed.
We assume that the distance between DFAs can
be used to judge their physical layer distance. The reason
for this is that there is a strong relation between the
geographical distance and delay for macro movements
although we cannot say that IP addresses are distributed
geographically in micro movements.
Domain Foreign Agent
Home Agent
40
30
20
10
0
5.2.1 Simulation Scenario
1
2
3
4
5
6
Distance From HA/DFA
We measured the round trip time between HA and
new DFA and between new DFA and old DFA. We
modeled the distance by increasing number of nodes and
link delays between HA and new DFA.
HA
Figure14: Comparison of Registration Time HA/DFA
As it can be seen from the Figure 14, in the global
tube of our system, our structure gives better results and
eliminates the handoff latency because of the far home
networks. Our method can be simply described as
distribution of home agent functionality locally.
6 Future Work
DFA
F
A
DFA
F
A
M
H
F
A
M
H
DFA
F
A
F
A
Our work can be investigated further in various
ways. Neighborhood discovery mechanism can be made
adaptive to different domain structures and different cell
size. Movement detection of the MH by the FA can be
investigated in order to decrease bandwidth usage.
Geographical distribution can be studied within the domain
and this geographical structure can be adapted to the adhoc networks and rooftop networks.  angle can be made
adaptive to the environment, i.e. country  , continental ,
city . In addition, 3-D volumes can be considered in
global tube calculations and procedures can be adjusted
accordingly. Finally, security concern between domains
can be investigated.
F
A
Figure13: Simulation Scenario
5.2.2 Handoff Delay Performance
As the frequency of the movement of MH
increases, it will change its domain frequently. Current
researches focuses on shielding MH movement within the
domain from HA. However, when people want to use their
devices without interruption when moving, MH will
request more handoffs. Domain change according to
previous structure requires MH to register HA and this
cause delay if the HA is away.
As described in , our system does not require to
inform HA as long as the new FA is inside the global tube.
The comparison of the registration times by our
system and the previous systems for mobile nodes whose
distances to home agents vary is performed with the ns
simulator.
11
7 Summary
In this paper, we have presented an algorithm for
acquiring fast handoffs and less control traffic in wireless
networks, which uses the hierarchical structure of the
network, which implies that internet is a network of
domains, and location information of the foreign agents
inside every domain of the network. We have shown the
necessary required changes in registration messages and
the format of location advertisement messages. We have
achieved an intelligent neighborhood discovery
mechanism eliminating the manual configuration of
adjacent FAs by these location advertisement messages.
Moreover, we have obtained fast handoffs inside the
domain by sending the packets to multiple foreign agents
without needing any multi-cast address allocation. We
have demonstrated that our scheme meets the delay
requirements of one of the most demanding real-time
applications, audio applications. Furthermore, we have
achieved faster handoffs between the domains in contrast
to all other systems ignoring inter-domain handoffs. By
using local home agent functionality inside the global tube,
we avoided the need to register home agent for each
domain change.
The popularity of real-time applications and
wireless networks has demonstrated the need for the
support of these applications over the wireless Mobile IP.
The requirement of bandwidth by the real-time
applications will be provided by tiling the world into
smaller cells, which brings frequent hand-off problem.
Since real-time applications are not delay tolerant, the fast
handoff described in this paper is required to achieve good
quality service in the future.
8 References
[1] C. Perkins. “Mobile IP Design Principles and
Practices” Addison-Wesley Wireless Communication
Series 1997.
[2] T. Blackwell, K. Chan, K. Chang, T. Charuhas, J.
Gwertzman, B. Karp, H. T. Kung, W. David Li, D. Lin, R.
Morris, R. Polansky, D. Tang, C. Young, J. Zao.”Secure
Short-Cut Routing for Mobile IP” Summer USENIX, June
1994.
[3] R. Jain, A. Puri and R. Sengupta. “Geographical
Routing Using Partial Information for Wireless Ad Hoc
Networks”
Technical Memorandum no. UCB/ERL M99/69.
[4] B. Karp and H. T. Kung , “GPSR: Greedy Perimeter
Stateless Routing for Wireless Networks” Proc. of
ACM/IEEE MobiCom 2000.
[5] C. L. Tan, S. Pink and K. M. Lye. “A Fast Handoff
Scheme for Wireless Networks” Proc. of ACM/IEEE
WoW-MoM 1999.
[6] R. Caceres and V. N. Padmanabhan. “Fast and scalable
handoffs for wireless internetworks.” Proc. of ACM/IEEE
MobiCom 1996.
[7] S. Seshan, H. Balakrishan, and R. H. Katz. “Handoffs
in cellular wireles networks: The daedalus implementation
and experience” Kluwer International Journal on Wireless
Personal Communications, January 1997.
[8] K. Malki and H. Soliman. “Fast Handoffs in Mobile
Ipv4” draft-elmaki-mobileip-fast-handoffs-03.txt,
September, 2000.
12
[9] A. Yegin, M. Parthasarathy, C. Williams. “Mobile Ipv6
Neighborhood Routing for Fast Handoff ” draft-yeginmobileip-nrouting-00.txt, September, 2000.
[10] R. Katz. “Adaption and Mobility in Wireless
Information Systems” IEEE Personal Communication,
First Quarter,1994.
[11] C. Perkins “Mobile-IP Local Registration with
Hierarchical Foreign Agents” draft-perkins-mobileiphierfa-00.txt, February,1996.