Content uploaded by Pascale Minet
Author content
All content in this area was uploaded by Pascale Minet on Jan 15, 2014
Content may be subject to copyright.
A Framework for Integrating Mobile-IP and OLSR
Ad-Hoc Networking for Future Wireless Mobile Systems
∗
Mounir Benzaid
1,2
, Pascale Minet
1
and Khaldoun Al Agha
1,2
1
INRIA, Domaine de Voluceau - B.P.105, 78153 Le Chesnay Cedex, FRANCE
mounir.benzaid, pascale.minet@inria.fr
2
LRI, B
ˆ
at 490 Universit
´
e Paris Sud, 91405 Orsay Cedex, FRANCE
alagha@lri.fr
Abstract—Trends in fourth generation (4G) wireless
networks are clearly identified by the full-IP concept
where all traffic (data, control, voice and video ser-
vices, etc.) will be transported in IP packets. Many
proposals are being made to enhance IP with the func-
tionalities necessary to manage the mobility of nodes,
so that networks can provide global seamless roaming
between heterogeneous wireless and wired networks.
The most widely known of these proposals is Mobile-
IP. Although Mobile IP offers a mechanism that al-
lows mobile nodes to maintain Internet connectivity
while moving through different networks, it causes a
huge amount of control traffic and disruption latency
during location update (especially, with wireless inter-
faces). To resolve this problem, mobility design is of-
ten divided into two parts: macro-mobility and micro-
mobility. In this paper, we propose an integrated
hierarchical architecture that extends micro-mobility
management to an ad-hoc network and connects the
ad-hoc network to the existing core network infras-
tructure and the Internet. The wireless multihop ac-
cess network of the proposed architecture is based en-
tirely on IP, by using Optimized Link State Routing
protocol (OLSR) and meets the requirements of future
full-IP wireless networks.
Keywords: mobile ad-hoc networks, mobility manage-
ment, macro-mobility, micro-mobility, Mobile-IP, OLSR.
I. INTRODUCTION
With recent technological advances in palm-top
computers and cellular phones, it is apparent that
wireless networks exist to facilitate mobility and
have generated a large movement to bring Internet
capabilities to these wireless mobile devices. Indeed
they enable a user to remotely access the Internet
anytime and anywhere (e.g. from university, visited
company, home, hotel, etc.). However, to forward
data to a mobile node, it is necessary to locate it and
follow it in its motion, which introduces complica-
tions in routing protocols. On the other hand, the
* this work has been partially financed by the french
RNRT (http://www.telecom.gouv.fr/rnrt/) project ARCADE
(http://www-rp.lip6.fr/arcade/)
introduction of GPRS in GSM infrastructure net-
works, the migration toward UMTS technologies,
the widespread use of the IEEE802.11 standard
and the development of Bluetooth wireless LAN
offer mobile wireless access for already supported
applications (e.g. multimedia applications with
data, voice and video traffic, etc.). Today, it is
crucial to offer to users of these mobile wireless
networks services that are as close as possible
to those already existing in wired networks, and
to guarantee service continuity and Quality of
Service (QoS). In addition, the convergence toward
full-IP networks in fourth generation (4G) wireless
networks enables flexibility to design the core IP
network independently from the access network
with many different access technologies.
The Mobile-IP [1] standard presents a funda-
mental solution for mobility management in the
Internet. With Mobile-IP, the mobile nodes are able
to send and receive data whatever their current point
of attachment to the Internet. Although Mobile-IP
is mostly used for nomadic computing with wired
connections in which a mobile node is unplugged
from one physical attachment point and plugged
into another. However, in the Mobile-IP protocol,
mobile nodes have to report every change of their
access points through foreign subnetworks to their
home network. Such a change during active data
transmission or reception is called a “handoff”. For
example, if we consider each wireless access point,
which covers only a small geographic area, as an IP
subnetwork (i.e. attachment point to the Internet),
then each handoff between cells would cause a huge
amount of control overhead. The registration pro-
cess is very long compared to the link-layer handoff
time, especially if the mobile node is far from its
home network, due to the high roundtrip time of the
Internet. In addition, in a micro-cellular and pico-
cellular network infrastructure environment (e.g.
UMTS, Bluetooth, HIPERLAN, etc.), Mobile-IP
would have to execute a costly registration process
at a high handoff rate.
To resolve this problem mobility management
is often divided into two parts: macro-mobility
and micro-mobility. Macro-mobility concerns the
management of mobile nodes moving on a large
scale, between different domains (or wide wireless
access networks), while micro-mobility covers the
management of mobile nodes moving at a local
scale, inside these domains (or a particular wireless
network). Macro-mobility is usually assumed to be
managed through Mobile IP and several protocols,
such as Cellular IP [2] and HAWAII [3], have been
proposed to manage the micro-mobility problem.
In situations (e.g., conference rooms, forests,
airports, subways and battle fields, etc.) where it is
either not possible or too expensive to deploy 2G
(GSM) and 3G (UMTS) cell-based mobile network
infrastructures, an infrastructureless mobile network
can be used. A network based on this idea is called
“Mobile Ad-hoc NETwork” (MANET) [4], [5]. The
advantage of an ad-hoc network is that it extends
mobility to wireless and autonomous domains,
where each node acts as a router and performs the
mobility’s functionalities. In addition to being able
to operate autonomously, an ad-hoc network can be
attached at some access point(s) to the fixed Internet.
In this work, we propose an integrated hier-
archical architecture that extends micro-mobility
management to an ad-hoc network and intercon-
nects the ad-hoc network into the existing networks
infrastructure and Internet. The wireless multihop
access network is based entirely on IP, uses the
Optimized Link State Routing protocol (OLSR) and
meets the requirements of future full-IP wireless
networks, such as providing high-rate video, voice
and data services to the cellular handset or handheld
Internet devices.
The remainder of this paper is organized as fol-
lows. In Section II, we describe different pro-
posals to manage mobility in the Internet; we fo-
cus on the definition of the macro/micro mobility
architecture and related work to combine infras-
tructure/infrastructureless networks with Mobile-IP.
Section III presents the proposed architecture for
managing mobility (intra/inter domain) in the Inter-
net and briefly describes the OLSR protocol. The
details of the integration of Mobile-IP and OLSR
are presented in section IV; also a short compari-
son with a GPRS cellular environment is depicted in
order to show similarities in mobility management.
II. SOLUTIONS FOR MOBILITY IN THE INTERNET
The mobile nodes in the Internet must be able
to reach their correspondents and to be reached by
them, while moving from one Internet attachment
point to another. The current version of the network
layer protocol (IPv4) is no longer sufficient to sup-
port these mobile nodes. Indeed, IP makes two im-
plicit assumptions. First, a point at which a node is
attached to the Internet remains fixed, and secondly
a node’s IP address identifies the network or subnet-
work to which it is attached. The protocol that man-
ages the mobility must be transparent to the trans-
port layer to guarantee continuity of service; it must
be compatible with IP; and for reasons of scalabil-
ity it must not require any changes to existing In-
ternet hosts and routers. To meet these requirements
“Mobile-IP” [1] was developed as a fundamental so-
lution for the management of mobility on the Inter-
net.
A. Mobile-IP
Routing of packets is based on the location
information contained in the node’s destination
IP address. This IP address includes two parts:
the network identifier and the host identifier. If a
mobile node moved to another network without
changing its IP address, it would no longer be
possible to correctly route packets to it. On the
other hand, if a node is configured with a new IP
address, it will lose its active upper layer protocol
connections, such as TCP sessions. In this context,
Mobile-IP protocol [1] has been developed within
the IETF [6] to handle mobility through different IP
networks. It is an enhancement of the Internet Pro-
tocol intended to enable nodes to maintain Internet
connectivity with the same IP address regardless
of their attachment IP subnetworks. Each mobile
node is assigned an IP home address belonging to
a particular network, known as its original or home
network. The home address is static: it remains
unchanged as the mobile node’s location varies, and
any packet addressed to it is routed to the home
network. When the mobile node is connected to the
home network, it behaves like any non-mobile node
and it may be reached through normal IP routing.
On the other hand, when the mobile node
moves to another network, it can no longer be
reached solely on the basis of its home address,
but must be assigned an address belonging to the
visited network, which is considered as a foreign
network for this node. This address is called the
care-of-address. The care-of-address makes it
possible to identify the actual location of the mobile
node and to route packets up to this new location.
The care-of-address changes whenever the mobile
node’s visited network changes.
The key feature of the Mobile-IP design is that
all required functionalities for processing and man-
aging mobility information are embedded in three
defined entities: the Mobile Node (MN), the Home
Agent (HA) and the Foreign Agent (FA).
• A MobileNode is a host or a router thatchanges
its attachment point from one network or sub-
network to another without changing its IP ad-
dress. The MN has only one home network.
• A Home Agent, which can be a host or a router,
is located on the home network. The task of the
HA is to maintain the current location informa-
tion for the mobile nodes when they are away
from home. Then, the HA tunnels packets to
their present location.
• A Foreign Agent is a host or a router on a vis-
ited or a foreign network that registers the en-
trance of mobile nodes, detunnels and forwards
packets to the visiting mobile node.
The care-of-address is the termination point of
a tunnel toward a MN. Mobile-IP can use two
different types of care-of-address: a “Foreign Agent
care-of-address” is an address proposed by the FA
with which the MN is registered; a “co-located
care-of-address” is an externally obtained local
address that the MN can dynamically acquire by
DHCP [7].
There are two mechanisms defined to update lo-
cation for an MN:
• Agent discovery and solicitation: Each MN
is associated to a unique home network as indi-
cated by its permanent home address. In order
to communicate with mobility agents (i.e. FA
and HA), a MN must discover the IP addresses
of these agents. These mobility agents peri-
odically advertise their presence via an Agent
Advertisement message (an extension added to
the ICMP Router Advertisement). A MN may
optionally solicit an Agent Advertisement mes-
sage from any locally attached mobility agent
through an Agent Solicitation message. Upon
receiving this Agent Advertisement the MN de-
termines whether it is currently connected to its
home network or to a foreign network, and also
detects when it has moved from one network to
another. In addition, this enables it to find the
care-of-address of the advertised FA.
• Registration: When a MN is away from home,
it uses a registration procedure to keep its HA
informed about its current location. To do so,
it contacts the closest FA and sends a registra-
tion request message directly (if it is used a co-
located care-of-address) or via this FA (if it is
used a FA-care-of-address) to its HA. When the
HA gets the request it knows where the MN is
located at present and to which care-of-address
it should relay the packets destined for the MN.
Then, it responds by a registration reply mes-
sage to grant or deny this mobility service. Af-
ter successful registration, the HA is responsi-
ble for intercepting packets arriving for the MN
on its home network, encapsulating and send-
ing these packets through a tunnel to the actual
mobile node’s FA.
B. Micro-mobility approaches for wireless access
networks
With the Mobile-IP paradigm, micro-mobility
means the mobile node moves inside the same
access network without changing the register
mobility binding (i.e. association of a mobile node’s
home address with a care-of-address) on the home
network. On the other hand, macro-mobility means
movement between different networks, and requires
registering new care-of-address acquired in the
visited network.
Mobile-IP lacks smooth, fast and transparent
handoffs required for future full-IP wireless net-
works. Indeed, frequent handoffs inside a wireless
access network tend to generate a huge amount
of signaling overhead due to the required control
messages between a mobile node and home net-
work. Additionally, the need to acquire a new
care-of-address and tunnel establishment results in
latency and disruption to data traffic (e.g. packet
loss, jitter, etc.). Also, if a large number of mobile
nodes quickly migrate between foreign networks,
Mobile-IP does not provide a scalable solution to
support fast and seamless handoff control [2], [8].
Several proposals [2], [3], [9], [10] have been
made to deal with this micro-mobility problem in
wireless access networks. These protocols follow
the approach of hierarchical mobility support in
combination with Mobile-IP. They provide mobility
in a well defined area, e.g., an access network, and
allow mobility between different access networks
to be handled by Mobile-IP as a macro-mobility
solution. Two examples of protocols, Cellular-IP
and HAWAII, are presented in the following.
1) Cellular IP: The Cellular-IP [2] proposal
combines the smooth mobility support of cellular
telephony systems with the flexibility, robustness
and scalability of IP-based networks. The gateway
connecting a Cellular-IP network to the Internet
serves as the HA and FA within the wireless
access network. A Cellular-IP network contains
an arbitrary number of nodes, and Base Stations
(BSs) which have a wireless interface. The mobile
nodes are identified by their home addresse, and for
macro-mobility purposes they utilize the address of
the gateway as their care-of-address.
Cellular-IP routing is based on routes estab-
lished and updated by the mobile node during
its connection to the network. Each Cellular-IP
node maintains a routing cache that allows it to
forward packets from the gateway to the mobile
node and from the mobile node to the gateway. On
downlink, a beacon packet is periodically sent by
the gateway that floods the access network. This
mechanism allows each BS to record the interface
from which the beacon is received and uses it to
forward packets towards the gateway. On uplink,
in order to minimize control traffic, hop-by-hop
transmission of regular data packets from mobile
node to the gateway enables nodes on the path to
update their routing cache. On the other hand, the
routes can be established and refreshed by mobile
node’s transmission of route-update packets when it
connects to the network and each time it performs a
handoff.
The handoff is managed by two different mech-
anisms: hard handoff and semi-soft handoff. The
hard handoff provided no guarantees in terms of
packet loss. In semi-soft handoff, a mobile node
listening to beacons transmitted by two BSs can
establish new routing cache mappings before the
actual handoff takes a place. In this way, handoff
latency is reduced. Moreover, Cellular-IP presents
a support for passive connectivity with classical
paging mechanism: some BSs maintain paging
caches that are refreshed at regular intervals by
paging-update packets.
2) HAWAII: Handoff-Aware Wireless Ac-
cess Internet Infrastructure (HAWAII) [3] is a
domain-based approach for managing mobility.
As in Cellular-IP, HAWAII is responsible for
intra-domain mobility limited to an administrative
domain of an access network while inter-domain
mobility is handled by Mobile-IP. The gateway
into each domain is called a Domain Root Router
(DRR). Furthermore a HAWAII domain comprises
several routers and Base Stations (BSs) running the
HAWAII protocol, as well as mobile nodes. Each
mobile node is assumed to have a home domain, and
when entering a foreign HAWAII domain obtains a
co-located care-of-addressfrom the foreign domain.
A mobile node in the HAWAII domain exchanges
only Mobile-IP control messages with the BSs,
while the path setup within a HAWAII domain is re-
alized with three types of messages: power-up, up-
date and refresh. BSs periodically send Agent Ad-
vertisement messages. Whenever the host detects a
change of base station it issues a Mobile-IP regis-
tration request to the new BS. This registration is
used to trigger a HAWAII path setup scheme in-
side the domain. The BS then sends a HAWAII
path setup power-up or update message to the DRR
on a hop-by-hop basis. This has the effect of es-
tablishing a route for that mobile node to the DRR
and any intermediate routers on the path towards the
mobile node. In addition, aggregate hop-by-hop re-
fresh messages are periodically sent to maintain the
routes. Moreover, the choice of using a co-located
care-of-address and maintaining the mobile node
address unchanged within a domain simplifies QoS
per flow (e.g. RSVP).
C. MANET Capabilities
Solutions based on a cellular approach require
the deployment of an infrastructure. An alternative
solution consists of using a MANET network
that works without any pre-existing infrastructure
deployment. In fact, a MANET network [11] is
a collection of mobile nodes that communicate
using a wireless medium, forming an autonomous
network. There is no centralized access point or
pre-existing infrastructure. Such networks have
dynamic, random, sometimes rapidly changing
topologies, limited bandwidth, variable throughput
links, and limited power (e.g. battery operated
devices). When a node needs to communicate with
another node, it uses either a direct wireless link
or a multi-hop route to reach the destination. This
means that all the nodes must incorporate routing
capability to ensure that packets are delivered to the
designated destination.
Different routing protocols are proposed in the
MANET working group of the IETF [11]. They
address the problem of unicast routing, while taking
into account the features of wireless, multi-hop,
mobile ad-hoc networks. Such protocols can be
divided into three categories: proactive, reactive and
hybrid. With reactive protocols, a node discovers
routes on-demand and maintains only active routes.
Thus, a route is discovered whenever a source
node needs to communicate with a destination node
for which a route is not available. This discovery
is based on pure flooding [12] in the network.
Examples of reactive protocols include AODV [13]
and DSR [14]. With proactive protocols, each
node continuously maintains the routes to all other
nodes in the network by the periodic exchange of
control messages. When a node needs to send a
packet to any other node in the network, the route
is immediately available. Examples of proactive
protocols include DSDV [15] (an adaptation of
Routing Information Protocol [16]), OLSR [17]
(an optimization of the Link State algorithm
OSPF [18]) and TBRPF [19]. Hybrid protocols,
such as ZRP [20], use a mixed approach of proactive
and reactive techniques.
Some proposals aim to facilitate connectivity of
stub ad-hoc networks to the Internet and routing in-
teroperability based on Mobile-IP is achieved. The
authors in [21] shows how to integrate an ad-hoc
routing protocol with Mobile-IP. Routing within the
ad-hoc network is achieved by routed, a modified
version of the RIP daemon, on each mobile node
within the network. The Foreign Agent participates
in the ad-hoc routing. The mobile nodes within
range of the Foreign Agent cooperate to forward
Agent Advertisements or Mobile-IP messages to
other nodes outside its range. Each mobile node
uses the Foreign Agent as its default router. A route
manager is used to coordinate the manipulation of
the IP routing table.
A proposal to integrate a reactive protocol,
DSR [14] with the Internet routing and Mobile-IP
is presented in [22]. An addressing architecture is
proposed, where all the nodes in an ad-hoc network
are assigned home addresses from a single IP
subnet. Nodes within range of the Foreign Agent
serve as gateways between the ad-hoc network and
the Internet. DSR is utilized for routing within the
ad hoc network, while standard IP routing applies
to the wired network. The gateway participates
in both DSR routing and Internet routing. In the
integration of Mobile-IP and DSR, Foreign Agents
(implemented on gateways) are responsible for
forwarding packets between the ad-hoc network and
the Internet.
The most recent work on this ad-hoc wired net-
works connectivity is Mobile-IP for Mobile Ad-Hoc
Networks (MIPMANET) [23]. In MIPMANET,
nodes in an ad-hoc network that need Internet access
register with the Foreign Agent and use their home
address for all communications. Mobile nodes tun-
nel all packets to their Foreign Agent, which decap-
sulates the packets and forwards them to the desti-
nation. This tunneling between mobile nodes and
the Foreign Agent is used to separate the ad-hoc net-
work from Mobile-IP. On the other hand, the pack-
ets from hosts on the Internet addressed to mobile
nodes are delivered by ordinary Mobile-IP mecha-
nisms to the Foreign Agent. Then, the AODV pro-
tocol is used to deliver packets to the mobile nodes
in the ad-hoc network. Nodes that do not require In-
ternet access have no knowledge of external routes.
Moreover, MIPMANET utilizes a mechanism called
“MIPMANET Cell Switching” (MMCS), that al-
lows a mobile node to determine when it should
register with another Foreign Agent. This solution
allows the coexistence of heterogeneous addresses
in the ad-hoc network, i.e. each node can use its
home address for communication inside the ad-hoc
network.
III. PROPOSED ARCHITECTURE FOR MOBILITY
MANAGEMENT
As mentioned above, although there have been a
number of proposals [21], [22], [23] for intercon-
necting an ad-hoc network with the fixed Internet,
none of them utilizes any of the emerging micro-
mobility solutions. They simply extend Mobile-
IP to serve the nodes further than a one hop dis-
tance from a base station. These schemes have the
disadvantage of concentrating traffic on the ad-hoc
nodes one-hop away from the gateway. The archi-
tecture that we propose for interconnecting ad-hoc
networks to the fixed Internet with a macro/micro-
mobility management follows the approach of hier-
archical mobility support, where Mobile-IP proto-
col is used to support macro-mobility (inter-domain)
and the OLSR protocol is used to support micro-
mobility (intra-domain). In this hierarchical arch-
tecture, the Base Stations introduce an intermediate
level that allows more flexibility in mobility man-
agement and higher bandwidth between the gateway
and the ad-hoc network.
A. Hierarchical mobility support
The proposed architecture is depicted in Figure 1.
An OLSR-IP access network constitutes an IP sub-
network and it is interconnected to the Internet via
an OLSR Gateway (OLSR-GW). The displacement
of a mobile node inside an OLSR-IP access network
is managed by the OLSR protocol. Mobility
between different OLSR-IP access networks or IP
subnetworks is managed by Mobile-IP.
An OLSR-IP access network consists of: (i) a
random topology of ad-hoc mobile nodes and (ii)
a fixed hierarchical structure connecting OLSR-GW
and OLSR Base Stations (OLSR-BS) by wired links.
For nodes having multiple interfaces like Base Sta-
tions (i.e. one wired and the other wireless), the
OLSR protocol is implemented on both interfaces.
A node inside the OLSR-IP access network uses its
IP home address to establish and maintain routes.
HA
: OLSR−N
: OLSR−MN
: OLSR−BS
: wired link
: wireless link
OLSR
GW
Internet with Mobile−IP
IPv4 Tunnel
Foreign network
Home network
Figure 1. An OLSR-IP Access Network Interconnected to a
Mobile-IP enabled Internet.
The architecture is composed of several functional
entities:
• Home Agent (HA): a routerin the mobile node’s
home network.
• OLSR-GW: a router allowing an OLSR-IP ac-
cess network to be connected to the Internet
and implementing the role of a Foreign Agent
in order to manage the visitor mobile nodes. It
can also implement the role of a Home Agent if
the OLSR-IP access network is the home net-
work. Furthermore, OLSR-GW implements the
OLSR protocol to contribute to micro-mobility
management.
• OLSR-BS: a node having two interfaces: wire-
less and wired. In order to ensure routing
between the two parts of the architecture, it
implements the OLSR protocol on both inter-
faces.
• OLSR-MN: a mobile node in the ad-hoc net-
work having only a wireless interface. It im-
plements Mobile-IP’s registration procedures
which are trigged by a change of the OLSR-IP
access network. It also implements the OLSR
protocol, which makes it possible to build its
OLSR routing table to reach and maintain con-
nectivity with all nodes inside the OLSR-IP ac-
cess network. In addition, it uses the optimal
default route via the OLSR-GW to reach a host
outside the OLSR-IP access network.
• OLSR-N: a wired node that serves as a traffic
controller or supports micro-mobility manage-
ment functions.
The operation of this architecture is as follows.
When an OLSR-MN is away from its home network,
the Home Agent intercepts packets addressed to
it and sends these packets through a tunnel up to
the current mobile node’s attachment OLSR-IP
access network. The OLSR-MN would previously
have registered its OLSR-GW care-of-address with
its Home Agent. The Foreign Agent decapsulates
packets and forwards them to the visited OLSR-
MN according to the OLSR routing table. The
displacement of the OLSR-MN inside the visited
OLSR-IP access network is managed locally by
the OLSR protocol and does not require informing
or changing registered location information at the
level of the Home Agent. Briefly, the Home Agent
does not have to know the exact location of the
OLSR-MN, but in which OLSR-IP access network
the OLSR-MN is located. Micro-mobility inside the
visited network is dealt with locally by updating
the routing tables of each node in the network,
according to neighborhood and topology changes.
The advantages of this architecture are the follow-
ing:
• Increased bandwidth between OLSR-GW and
the mobile nodes: due to having Base Stations
connected by wired links to the OLSR-GW.
• Reduction in the amount of global location up-
dates by Mobile-IP: avoiding Mobile-IP regis-
tration procedures when executing handoffs be-
tween the Base Stations. This is because the
Base Stations are not directly connected to the
Internet.
• Shared traffic load between Base Stations: the
traffic to and from the outside of the OLSR-IP
access network is distributed over all the Base
Stations of the local network.
• Guaranteed use of the shortest route between
OLSR-GW and a mobile node: because the
OLSR protocol maintains the shortest routes
regarding the number of hops.
• Micro-mobility is handled in a transparent
manner for the Home Agent: when a mobile
node moves inside the OLSR-IP access net-
work, there is an automatic change of the Base
Station which receives the packets from the
OLSR-GW destined to a mobile node, without
location updates on the home network. We can
refer to this local change as a handoffat IP level
(IP-handoff) between a Base station’s cells.
• Preferential use of wired architecture to for-
ward packets inside the access network. The
wired paths are less vulnerable than the wire-
less ones between ad-hoc nodes, consequently
the Base Stations can reduce the number of
wireless hops and therefore the cost of the
route.
• Connections without Base Stations are possi-
ble: a mobile node can move out of the cov-
erage area of a Base Station, and can still be
reached using ad-hoc OLSR routing.
B. Optimized Link State Routing Protocol (OLSR)
OLSR [17], [24] is a proactive routing protocol,
providing the advantage of having routes immedi-
ately available in each node for all destinations in
the network. It is an optimization of a pure Link
State routing protocol. This optimization is based
on the concept of multipoint relays (MPRs) [25].
First, using multipoint relays reduces the size of
the control messages: rather than declaring all
links, a node declares only the set of links with
its neighbors that are its “multipoint relays”. The
use of multipoint relays also minimizes flooding of
control traffic. Indeed only MPRs forward control
messages. This technique significantly reduces the
number of retransmissions of broadcast control
messages [26]. OLSR is characterized by two types
of control messages: neighborhood and topology
messages, called respectively Hello messages and
Topology Control (TC) messages. Indeed OLSR
provides two main functionalities: Neighbor Dis-
covery and Topology Dissemination.
1) Neighbor Discovery: Each node must detect
the neighbor nodes with which it has a direct link.
Due to the uncertainties in radio propagation, a
link between neighboring nodes may enable the
transmission of data in either one or both directions
over the link. For this, each node periodically
broadcasts Hello messages, containing the list
of neighbors known to the node and their link
status. The link status can be either symmetric
(if communication is possible in both directions),
asymmetric (if communication is only possible in
one direction), mpr (if the link is symmetric and the
sender node of the Hello message has selected this
node as a multipoint relay), or lost (if the link has
been lost). The Hello messages are received by all
one-hop neighbors, but are not forwarded.
Thus, Hello messages enable each node to dis-
cover its one-hop neighbors, as well as it two-hop
neighbors (the neighbors of its neighbors). Each
node m of the network independently selects its own
set of multipoint relays among its one-hop neigh-
bors. The multipoint relays of m cover, in terms
of radio range, all the two-hop neighbors. Figure
2 shows the MPRs selection by node m. One possi-
ble algorithm for selecting these MPRs is described
in [25]. Each node m maintains the set of its “mul-
tipoint relay selectors” (MPR selectors). This set
contains the nodes which have selected m as a mul-
tipoint relay. Node m only forwards broadcast mes-
sages received from one of its MPR selectors.
multipoint relays
of node m
m
Figure 2. Multipoint relays of node m.
2) Topology Dissemination: Each node of the
network maintains topological information about
the network obtained by means of TC messages.
Each node m selected as a multipoint relay, broad-
casts a TC message advertising its MPR selectors.
The TC messages are flooded to all nodes in the
network and take advantage of MPRs to reduce the
number of retransmissions.
The neighbor information and the topology infor-
mation are refreshed periodically, and they enable
each node to compute the routes to all known desti-
nations. These routes are computed with Dijkstra’s
shortest path algorithm [12]. Hence, they are opti-
mal as concerns the number of hops. Moreover, for
any route, any intermediate node on this route is a
multipoint relay of the next node. The routing table
is computed whenever there is a change in neighbor-
hood information or a change in topology informa-
tion.
IV. INTEGRATION OF MOBILE-IP AND OLSR
To achieve integration of Mobile-IP and OLSR
we propose the architecture depicted in Figure 3.
The ad-hoc routing protocol is implemented by
an OLSRd daemon, while the MOBILE-IP (MIPd)
is implemented by three specific daemons: HAd,
FAd and MNd, corresponding respectively to Home
Agent, Foreign Agent and Mobile Node function-
alities. Each node in the ad-hoc network needing
mobility services, such as a visiting OLSR-MN, or
a node participating in the management of mobil-
ity inside the access network, such as an OLSR-MN
and an OLSR-BS, runs the two daemons OLSRd and
MNd. We can also have nodes with only an OL-
SRd daemon. The OLSR-GW has two daemons OL-
SRd and FAd; also it can implement HAd if it plays
the role of the Home Agent. To ensure transparency,
the two daemons OLSRd and MIPd (i.e. HAd, FAd
or MNd) run independently without requiring any
change to their behavior.
HAd
INTERNET
Broadcast packets
(Agent Advertisement)
Wide Scope Broadcast packet
Wide Scope Broadcast packet
OLSRd OLSRd
MNd
OLSRd
FAd
MNd
OLSRd
Figure 3. Integration model.
The OLSR-GW periodically broadcasts Agent
advertisement messages targeting the limited
broadcast address “255.255.255.255”. So, these
Advertisements are only received by nodes within
the transmission range of GW-OLSR (i.e. one-hop
neighbors). In order to enable all nodes in the ac-
cess network to receive advertisements we propose
MPR-based flooding. In MPR-based flooding, a
node forwards a broadcast packet at most once,
and only if this packet has been received from
one of its MPR selectors. With this mechanism
an advertisement does not leave the GW-OLSR
node and it is redirected to the local interface
(i.e. loopback). Then, the OLSRd captures this
advertisement and encapsulates it in a new packet
called a “Wide Scope Broadcast packet” (WSB).
Hence, a broadcast packet reaches all the nodes in
the OLSR-IP access network and takes advantage
of MPRs, which significantly reduces the number
of retransmissions, in contrast to a purely classical
flooding. The messages in an ad-hoc network are
flooded with the multicast address “224.0.0.1”.
On the other hand, when an OLSRd receives the
WSB packet, it decapsulates the packet and sends
the original advertisement to the MNd on the local
interface, as if it had been sent directly by a Foreign
Agent.
In order to coordinate the manipulation of the IP
kernel routing table by the two independent dae-
mons MIPd and OLSRd, it is necessary to prevent
MNd from changing routes established by the OL-
SRd within the ad-hoc network, or alternatively to
anticipate the use of a route management tool as de-
scribed in [21]. Inside an OLSR-IP access network,
the mobility management contains two procedures:
an attachment procedure and a micro-mobility han-
dling procedure.
A. Attachment procedure
When a mobile node m powers up or moves to an
IP-OLSR access network, it triggers the attachment
procedure in order to have connectivity with its cor-
respondents. To do so, first it must establish con-
nection with the IP-OLSR access network, and then
perform registration with the home network. Briefly,
it follows these steps :
1) Neighbor discovery:
• m starts by sending an empty Hello mes-
sage announcing its IP home address in
order to establish links with its neighbors.
• a neighbor node receiving an empty Hello
from node m, replies with a Hello con-
taining the address of m. This allows m
to validate the link in both directions.
2) Multipoint relay selection: among the neigh-
bors’ replies, m selects its multipoint relays
(i.e. neighbors with a “symmetric” link, that
cover all two-hopneighbors). These MPRsare
declared in the next Hello (i.e, Hello contains
the IP addresses of those neighbors with the
link type “mpr”).
3) Topology dissemination: the declared nodes
will then broadcast TC messages to all the
nodes in the network declaring that they are
MPRs of m and thus m can only be reached
through them (i.e. an MPR is a last hop in the
route to m).
4) Routing table construction: all the nodes in
the network (including OLSR-GW) which re-
ceive TC messages, have a global vision of the
network topology, and thus calculate routes to
reach m.
5) Mobility detection:
• m may send Mobile-IP’s Agent Solici-
tation message to request a Mobile-IP’s
Agent Advertisement message.
• m receives either an Agent Advertisement
from OLSR-GW in response to its Agent
Solicitation or a periodic AgentAdvertise-
ment.
• therefore, m can detect if it has moved
away from its home network (if the net-
work prefix of m’s home IP address dif-
fers from prefix network of OLSR-GW’s
IP address) or has moved to another ac-
cess network (if the network prefix of the
advertised care-of-address differs from
current network prefix of the registered
OLSR-GW).
6) Registration:
• if m is away from its home network or
has moved to another access network, it
sends a Mobile-IP’s registration request
message to its Home Agent via the OLSR-
GW.
• m receives Mobile-IP’s registration reply
message from its Home Agent, via the
OLSR-GW.
Once m has been registered, it is attached to the
OLSR-IP access network and can send and receive
data packets to/from their correspondent nodes in
the IP-OLSR access network or on the Internet.
B. Micro-mobility handling procedure
As mentioned above, inside the OLSR-IP access
network, mobility is naturally handled by the OLSR
protocol. It is implemented by all the nodes within
the access network. The operation scheme as ex-
ecuted by a node while moving within the access
network can be summed up as follows:
• The node periodically broadcasts Hello Mes-
sages to the one-hop neighborhood.
• It selects its multipoint relays whenever its one-
hop or two-hop neighborhood changes.
• It periodicallybroadcasts TC messages overthe
whole network, to announce the set of its mul-
tipoint relays selectors.
• It establishes and keeps topological informa-
tion updated, which gives a global view of the
network (all multipoint relays of the nodes in
the network).
• It constructs the routing table from the neigh-
borhood and topology information: routes to
all nodes in the network.
C. Comparison with the GPRS
The objectiveof our architecture is to define a uni-
versal mobility model that provides global seamless
roaming between heterogeneous wireless and wired
IP-based networks. Nowadays, the most widespread
architecture in cellular networks that really pro-
vides this mobility service is obviously GSM/GPRS
2G (and in the future UMTS 3G). Briefly, the
GSM/GPRS network works as follows:
• Base Stations selection: when a mobile node
powers up in a cellular environment, it tunes its
radio interface and listens to beacons transmit-
ted by Base Stations. It then selects the Base
station with the strongest signal.
• Location updating/Attachment: a mobile
node must register with the MSC (Mobile ser-
vices Switching Center)/VLR (Visitor Location
Register) and the connected SGSN (Serving
GPRS Support Node) to indicate its current lo-
cation. The MSC/VLR and SGSN record the
location information, and then send this in-
formation to the subscriber’s HLR (Home Lo-
cation Register). Although a mobile node is
uniquely identified by the International Mo-
bile Subscriber Identity (IMSI), it has a tempo-
rary alias called Temporary Mobile Subscriber
Identity (TMSI), assigned by the VLR or SGSN
which manage the Location Area or Routing
Area. This TMSI changes whenever a mobile
node moves to a new Location Area. Once
these procedures have been executed a mobile
node is attached to the GSM/GPRS network.
• Handoff: while communicating, a mobile
node or the MSC/SGSN can initiate a handoff
between cells, based on signal strength mea-
surements.
A comparison between the GSM/GPRS architec-
ture and our Mobile-IP/OLSR architecture shows
similarities in mobility management. The care-of-
address assigned by the Foreign Agent whenever a
mobile node changes access network plays the same
role as the TMSI in a GSM/GPRS environment, and
the IP home address is the equivalent of the IMSI.
The following table summarizes equivalent tasks for
both architectures.
Table I. Comparison between GSM/GPRS
architecture and Mobile-IP/OLSR architecture.
GSM/GPRS Mobile-IP/OLSR
Base stations selection Multipoint relays selection
SGSN attachment/
Mobile-IP registration
Location updating
Handoff
Neighbor discovery/
Micro-mobility
VLR registration Topology dissemination
V. CONCLUSION
4G wireless networks, in addition to incorpo-
rating all the services currently provided by 3G
wireless networks, will integrate full-IP based
core and access networks. In this way, universal
mobility management will be possible through
heterogeneous wireless and wired networks on the
Internet. In this paper, we have discussed some
limitations of existing mobility solutions on the
Internet (e.g. Mobile-IP). We have proposed an
integrated architecture that manages this universal
mobility both for large scale macro-mobility and lo-
cal scale micro-mobility. The proposed architecture
extends a wireless access network’s micro-mobility
management to an ad-hoc access network, connects
an ad-hoc network to the Internet and is based on a
hierarchy of OLSR-IP access networks: where the
Mobile-IP standard is used for macro-mobility man-
agement between access networks and the OLSR
protocol is used for micro-mobility management
within the access network.
We have proposed a mechanism to integrate
Mobile-IP with the OLSR protocol in this architec-
ture. A comparison with the GSM/GPRS cellular
infrastructure shows some similarities in mobility
management. In parallel, work [27] has been car-
ried out on the new OLSR extension, denoted Fast-
OLSR. The Fast-OLSR proposes integrating fast
mobility routing in the OLSR protocol in order to
account for fast micro-mobility within an OLSR-IP
access network. While the OLSR protocol currently
supports best effort traffic only, in further work we
will study how to integrate Quality of Service (e.g.
in terms of bandwidth) to mobile nodes, using an
admission control mechanism.
REFERENCES
[1] C. Perkins, “IP Mobility Support for IPv4,” RFC
3220, IETF: The Internet Engineering Task Force,
http://www.ietf.org, January 2002.
[2] A.T. Campell, J.Gomez, A.Valko, “An overview of Cellular
IP,” IEEE Wireless communications and Networking Con-
ference, vol. 2, pp. 606–610, September 1999. New Or-
leans.
[3] R.Ramjee, T.L. Porta, S.Thuel, K.Varadhan, S.Y.Wang,
“HAWAII: a Domain-based Approach for supporting Mo-
bility in Wide-area Wireless Networks,” IEEE interna-
tional Conference on Network Protocols, November 1999.
Toronto.
[4] C. Perkins, Ad Hoc Networking. Addison Wesley, 2000.
[5] C.K. Toh, Ad Hoc Mobile Wireless Networks protocols and
Systems. Prentice HALL PTR, 2002.
[6] IP Routingfor Wireless/Mobile Hosts (mobileip),mobileip-
charter. http://www.ietf.org/html.charters/mobileip-
charter.html, 2002.
[7] R. Droms, “Dynamic Host Configuration Protocol,” RFC
2131, IETF, http://www.ietf.org, March 1997.
[8] R.Caceres, V.N. Padmanabhan, “Fast and Scalable Hand-
offs for Wireless Internetworks,” ACM International Con-
ference on Mobile Computing and Networking, pp. 56–66,
November 1996. NY, USA.
[9] S. Das, A. Misra, P.Agrawal,
“TeleMIP:Telecommunication-Enhanced Mobile IP
Architecture for Fast Intra-Domain Mobility,” IEEE
Personal Communications, vol. 7, pp. 50–58, August 2000.
[10] A. O’Neill, G. Tsirtsis, S.Corson, “Edge Mobility Architec-
ture,” draft-oneill-ema-01.txt, IETF, March 2000.
[11] Mobile Ad-hoc Networks (MANET), manet-charter.
http://www.ietf.org/html.charters/manet-charter.html,
2002.
[12] A. S.Tanenbaum, Computer Networks. Prentice Hall, 1996.
[13] C. Perkins, E. Royer, S. Das, “Ad hoc On-Demand
Distance Vector (AODV) Routing,” draft-ietf-manet-aodv-
08.txt, IETF, March 2001.
[14] J. Broch, D. Johnson, D. Maltz, “The Dynamic Source
Routing Protocol for Mobile Ad Hoc Networks,” draft-ietf-
manet-dsr-01.txt, IETF, December 1998.
[15] C. Perkins, P. Bhagwat, “Highly Dynamic Destination-
Sequenced Distance-Vector Routing (DSDV) for Mobile
Computers,” in ACM SIGCOMM’94, (London, UK), Au-
gust 1994.
[16] J. Malkin, “Routing Information Protocol (RIP) V2,” RFC
1388, IETF, January 1993.
[17] P. Jacquet, P. Muhlethaler, A. Qayyum, A. Laouiti, L. Vi-
ennot, T. Clausen, “Optimized link state routing protocol,”
draft-ietf-manet-olsr-06.txt, IETF, September 2001.
[18] J. Moy, “Open Shortest Path First (OSPF) V2,” RFC 2328,
IETF, January 1998.
[19] R. G. Ogier, F. L. Templin, B. Bellur, M. G. Lewis, “Topol-
ogy broadcast based on reverse-path forwarding (TBRPF),”
draft-ietf-manet-tbrpf-05.txt, IETF: The Internet Engineer-
ing Task Force, March 2002.
[20] Z. J. Haas, M. R. Pearlman, P. Samar, “The Zone Routing
Protocol (ZRP) for Ad Hoc Networks,” draft-ietf-manet-
zone-zrp-02.txt, IETF, July 2000.
[21] C. Perkins, H.Lei, “Ad hoc Networking with Mobile IP,”
Second European Personal Mobile Communication Confer-
ence, pp. 197–202, October 1997.
[22] J. Broch, D.A. Maltz, D.B. Johnson, “Supporting Hiear-
archy and Heterogenous Interfaces in Multi-HOP Wireless
Ad Hoc Networks,” Workshop on Mobile Computing, IEEE
International Symposium on Parallel Architectures, algo-
rithms and Networks, pp. 75–85, June 1999. Australia.
[23] U. Jonsson, F.Alriksson, T.Larsson, P.Johansson, G.Q.
Maguire, “MIPMANET - Mobile IP for Mobile Ad Hoc
Networks,” IEEE/ACM Workshop on Mobile and Ad hoc
Networking and Computing, pp. 75–85, August 2000.
Boston MA USA.
[24] P. Jacquet, P. Muhletaler, A. Qayyum, A. Laouiti, T.
Clausen, L. Viennot, “Optimized Link State Routing Pro-
tocol,” in IEEE INMIC, (Pakistan), December 2001.
[25] A. Qayyum, L. Viennot, A. Laouiti, “Multipoint Re-
laying: An Efficient Technique for flooding in Mo-
bile Wireless Networks,” Tech. Rep. RR-3898, INRIA,
http://www.inria.fr, February 2000.
[26] A. Qayyum, A. Laouiti, L. Viennot, “Multipoint relay-
ing technique for flooding broadcast messages in mobile
wireless networks,” in HICSS: HAWAII INTERNATIONAL
CONFERENCE ON SYSTEM SCIENCES, (Hawai, USA),
January 2002.
[27] M. Benzaid, P.Minet, K.A. Agha, “Integrating fast mobility
in the OLSR routing protocol,” Fourth IEEE Conference in
Mobile and Wireless Communications Networks, Septem-
ber 2002.
