Title: Anton Kozlov Mobile IP: Security Issues
1Anton Kozlov Mobile IP Security Issues
- Survey of security chapters from the book by
- James D. Solomon, Mobile IP The Internet
Unplugged, Prentice Hall, 1998 - Applied Crypto and e-Security Lab
- http//www.cs.bu.edu/groups/aces/
- Boston University 2000
2Current State of Mobile Computing
- Mobile computers are one of the fastest growing
segments of the PC market - Short-range wireless networks (Bluetooth)
available from IBM, Toshiba, Dell, HP - High-speed (11 Mbps) wireless LAN products are
now easily and cheaply available (IEEE 802.11a,
IEEE 802.11b) - Low speed (currently 128 Kbps) Metropolitan Area
Wireless Network services are available in some
cities and spreading (Metricoms Ricochet)
3Mobile Computers Characteristics
- May change point of network connection frequently
- May be in use as point of network connection
changes - Usually have less powerful CPU, less memory and
disk space - Less secure physically
- Limited battery power
4Wireless Networks Characteristics
- Generally lower bandwidth
- Higher latency and variability
- Higher error rate
- More susceptible to interference and eavesdropping
5Outline of the Tutorial
- Part 1 The Need for Mobile IP
- Part 2 Mobile IP Overview (for IPv4)
- Part 3 Security Issues
- A Simple Mobile IP Application (Private Network
without Internet connection) - A More Complicated Application Internet-Wide
Mobility
6Part 1 The Need for Mobile IP
- Problems
- Terminology
- What Happens When a Node Changes Link?
- Cant We Solve This Problem with Host- Specific
Routes? - Why Not Just Change the Nodes IP Address?
- Cant We Just Solve the Problem at the Link
Layer? - What If We Only Need Nomadicity?
7Mobile IP solves the following problems
- If node moves from one link to another without
changing its IP address, it will be unable to
receive packets at the new link - If a node changes its IP address when it moves,
it will have to terminate and restart any ongoing
communications each time it moves - Mobil IP solves these problems in secure, robust,
and medium-independent manner whose scaling
properties make it applicable throughout the
entire Internet
8The Need for Mobile IP
- Terminology
- A home link is the link on which a specific node
should be located that is the link, which has
been assigned the same network-prefix as the
nodes IP address - A foreign link is any link other than a nodes
home link that is, any link whose
network-prefix differs from that of the nodes IP
address - Host-specific route is a routing-table with
Prefix-Length of 32 bits, it will provide a match
for exactly one IP Destination Address namely,
the address specified in the Target field - Mobility is the ability of a node to change its
point of attachment from one link to another
while maintaining all existing communications and
using the same IP address at its new link
9What Happens When a Node Changes Link?
10Cant We Solve the Mobility Problem with
Host-Specific Routes?
- How Might Host-Specific Routes Solve the Problem?
- If it Solves the Problem, Is This Solution a Good
One?
11Is This Solution a Good One?
- How Many Mobile Nodes We Can Expect?
- How Many Routes Are Required for Each Mobile
Node? - How Fast Will a Node Change Links?
- Is This Solution Robust?
- Is It Secure?
12ConclusionHost Specific Routes is an Unworkable
Solution to Node Mobility in the Internet
- Minimally, host-specific routes must be
propagated to all nodes along the path between a
mobile nodes home link and its foreign link - Some (in the worst case all) of these routes must
be updated every time the node moves from one
link to another - We expect millions of nodes to be operating
13Host-Specific routing has severe scaling,
robustness, and security problems
- Unless host-specific routes are propagated to a
much larger set of routers than minimal set
described in the first item above, then the
Internet mobility to route around isolated node
and link failures is negated by host-specific
routing - Serious security implications would require
authentication, and complicated key management
protocol to address
14Why Not Just Change the Nodes IP Address?
- Can Connections Survive a Changing IP Address?
- No, because all open TCP connections will be
terminated - How Do We Find a Node Whose IP Address Keeps
Changing? - Only if a mobile node itself initiates
communication, a huge overhead to keep entries in
DNS updated, address returned by a name server is
subject to change at any moment - Cant we just solve the problem at the Link
Layer? (Cellular Digital Packet Data - CDPD
(11Kbps), IEEE 802.11) - Provides node mobility only in the context of a
single type of medium and within a limited
geographic area
15What If We Only Need Nomadicity?
- A nomadic node is one which must terminate all
existing communications before changing its
point-of-attachment, but then can initiate new
connections with a new IP address once it reaches
its new location. - If all communications are initiated by the user
of a mobile node, and the user does not mind
shutting down his applications and restarting
then at a new location, then nomadicity is indeed
sufficient
16Why Mobility Is Preferable to Nomadicity?
- Many applications have configuration data bases
which depend on IP addresses, as opposed to host
names - In the future Servers and not just Clients might
need to become mobile (Clients know their Servers
only by their IP addresses) - Some license application vendors provide
network-licensing systems which restrict access
to only those nodes possessing specific ranges of
IP addresses - Some security mechanisms provide access
privileges to nodes based upon their IP
addresses. Mobile nodes employing Mobile IP allow
such mechanisms to work in the presence of node
mobility - Limited availability of IPv4 addresses, need for
specific address assignment mechanisms
17Summary
- A node that changes from one link to another is
incapable of communicating at the new location
unless it changes its IP address - Host-specific routing is not workable solution in
the context of the global Internet - Changing a nodes IP address is undesirable
- The difference between mobile and nomadic
computing (impossible for other node to know at
what address a nomadic computer can be reached at
any given moment)
18Summary (cont.)
- All link-layer solutions share limitations in
their geographic applicability and the media over
which they can run. - Even in those instances where a node requires
only nomadicity, the more subtle advantages
offered by Mobile IP mobility can make network
administration much easier.
19Part 2 Mobile IP Overview (for IPv4)
- Is Mobile IP an Official Standard?
- What Is the Scope of the Mobile IP Solution?
- What Are the Requirements for Mobile IP?
- What Assumption Does Mobile IP Make?
- Where Does Mobile IP Reside?
- Generally How Does Mobile IP Works?
- Summary
20Is Mobile IP an Official Standard?
- Mobile IP was approved by the Internet
Engineering Steering Group (IESG) in June 1996
and published as a Proposed Standard in November
1996. - Main reference document Request for Comments
(RFC) 2002 - There are other RFCs defining specific aspects of
Mobile IP, such as tunneling, applicability,
Management Information Base
21What Is the Scope of the Mobile IP Solution?
- Mobile IP is a network-layer solution to node
mobility in the Internet - It accomplishes its task by setting up the
routing tables in appropriate nodes, such that IP
packets can be sent to mobile nodes not connected
to their home link - Can be considered to be a routing protocol, which
has a very specialized purpose of allowing IP
packets to be routed to mobile nodes which could
potentially change their location very rapidly. - Mobile IP is unique in its ability to accommodate
heterogeneous mobility in addition to homogeneous
mobility. - Solves the primary problem of routing IP packets
to mobile nodes, which is a first step in
providing mobility on the Internet. A complete
mobility solution would involve enhancements to
other layers of the protocol stack.
22What Are the Requirements for Mobile IP?
- A mobile node must be able to communicate with
other nodes after changing its link-layer
point-of-attachment to the Internet - Must be able to communicate using its home
(permanent) IP address, regardless of its current
link-layer point-of-attachment to the Internet - Must be able to communicate with other computers
that do not implement the Mobile IP mobility
functions - The Mobile IP implementation should be limited
only to the mobile nodes themselves and the few
nodes which provide special routing functions on
their behalf - Must not be exposed to any new security threats
over and above those to which any fixed node on
the Internet is exposed
23What Assumption Does Mobile IP Make?
- Mobile IPs fundamental assumption is that
unicast packets those destined to a single
recipient are routed without regard to their IP
Source Address. - We will see how that assumption, though
theoretically valid, might not be operationally
valid under certain circumstances (Denial-of
Service)
24Where Does Mobile IP Reside?
- There are 3 functional entities where it is
implemented - Mobile Node a node which can change its
point-of-attachment to the Internet from one link
to another while maintaining any ongoing
communications and using its (permanent) IP home
address - Home Agent router with an interface on the
mobile nodes home link, which - Is informed by the mobile node about its current
location, represented by its care-of-address - In some cases, advertises reachability to the
network-prefix of the mobile nodes home address,
thereby attracting IP packets that are destined
to the mobile nodes home address - Intercepts packets destined to the mobile nodes
home address and tunnels them to the mobile
nodes current location, i.e. to the
care-of-address
25Where Does Mobile IP Reside?
- Foreign Agent a router on a mobile nodes
foreign link which - Assists the mobile node in informing its home
agent of its current care-of address - In some cases, provides a care-of address and
de-tunnels packets for the mobile node that have
been tunneled by its home agent - Serves as default router for packets generated by
the mobile node while connected to this foreign
link
26Mobile IP Entities and Relationships
27IP Tunneling
- A tunnel is a path followed by a fist packet
while it is encapsulated within the payload
portion of a second packet
28Properties of Care-of Address
- A care-of address is an IP address associated
with mobile node that is visiting a foreign link - A care-of address is specific to the foreign link
currently being visited by a mobile node - Generally changes every time the mobile node
moves from one foreign link to another - No Mobile IP-specific procedures are needed in
order to deliver packets to a care-of address - Is used as the exit-point of a tunnel from the
home agent toward the mobile node - Is never returned by DNS when another node looks
up the mobile nodes hostname
29Two Conceptual Types of Care-of Addresses
- A foreign agent care-of address is an IP address
of a foreign agent which has an interface on the
foreign link being visited by a mobile node. Can
be shared by many mobile nodes simultaneously - A collocated care-of address is an IP address
temporarily assigned to an interface of the
mobile node itself. The network-prefix of a
collocated care-of address must equal the
network-prefix that has been assigned to the
foreign link being visited by a mobile node. This
type of c/o address might be used by mobile node
in situations where no foreign agents are
available on a foreign link. A collocated c/o
address can be used by only one mobile node at a
time
30(No Transcript)
31Generally How Does Mobile IP Works?
- Home Agents and Foreign Agents advertise their
presence on any attached links by periodically
multicasting or broadcasting special Mobile IP
messages called Agent Advertisements - Mobile Nodes listen to these Agent Advertisements
and examine their contents to determine whether
they are connected to their home link or a
foreign link - A Mobile Node connected to a foreign link
acquires a care-of address. A foreign agent
care-of address can be read from one of the
fields within the foreign agents Agent
Advertisement. A collocated care-of address must
be acquired by some assignment procedure, such as
Dynamic Host Configuration Protocol (DHCP), the
Point-to-Point Protocols IP Control Protocol
(IPCP), or manual configuration
32How Does Mobile IP Works (cont.)?
- The mobile IP Registers the care-of address
acquired previously with its home agent, using a
message-exchange defined by Mobile IP. It asks
for service from a Foreign Agent, if one is
present on the link. In order to prevent
Denial-of-Service attacks, the registration
messages are required to be authenticated - The Home Agent or some other router on the home
link advertises reachability to the
network-prefix of the Mobile Nodes home address,
thus attracting packets that are destined to the
Mobile Nodes home address. The Home Agent
intercepts these packets, and tunnels them to the
care-of address that the mobile node registered
previously - At the care-of address at either the Foreign
Agent or one of the interfaces of the mobile node
itself the original packet is extracted from
the tunnel and then delivered to the Mobile Node - In the reverse direction, packets sent by the
Mobile Node are routed directly to their
destination, without any need for tunneling. The
Foreign Agent serves as a default router for all
packets generated by visiting node
33Mobile IP Summary
- Allows node mobility across media of similar or
dissimilar types - Uses the Mobile Nodes permanent home address
when it changes its point of attachment to the
Internet - Not requires any hardware and software upgrades
to the existing, installed base of IPv4 hosts and
routers other than those nodes specifically
involved in the provision of mobility services - Mobile Node must provide strong authentication
when it informs its Home Agent of its current
location - Uses tunneling to deliver packets that are
destined to the Mobile Nodes home address - 3 main entities Mobile Nodes, Foreign Agents and
Home Agents - 3 basic functions Agent Discovery, Registration,
Packet Routing
34Part 3a. Security Issues Simple Mobile IP
Application (Intranet without connection to the
Internet)
- How is Mobile IP deployed?
- Insider Attack
- Mobile Node Denial-of-Service
- Replay Attacks
- Theft of Information Passive Eavesdropping
- Theft of Information Session-Stealing (Takeover)
Attack - Other Active Attacks
35How is Mobile IP Deployed?
- All hosts are wholly owned by the enterprise
- Each router performs both home agent and foreign
agent functionality
36Insider Attacks
- Usually involve a disgruntled employee gaining
access to sensitive data and then forwarding it
to a competitor - Enforce strict control who can access what data
- Use strong authentication of users and computers
- Encrypt all data transfer on an end-to-end basis
between the ultimate source and ultimate
destination machines to prevent eavesdropping
37Mobile Node Denial-of-Service
- A Bad guy sends a tremendous number of packets to
a host (e.g., a Web server) that brings the host
CPU to its knees. In the meantime, no useful
information can be exchanged with the host while
it is processing all of nuisance packets - A Bad Guy somehow interferes with the packets
that are flowing between two nodes on the
network. Generally speaking, the Bad Guy must be
on the path between the two nodes on order to
wreak any such havoc
38Denial-of-Service Attack
- A Bad Guy generates a bogus Registration Request
specifying his own IP address as the care-of
address for a mobile node. All packets sent by
correspondent nodes would be tunneled by the
nodes home agent to the Bad Guy
39How Does Mobile IP Prevents this
Denial-of-Service Attack?
- Note In case of mobility a Bad Guy could attack
from anywhere in the network, it does not have to
be on the way. - Solution to require cryptographically strong
authentication in all registration messages
exchanged by a mobile node and its home agent. - Mobile IP by default supports MD5 Message-Digest
Algorithm (RFC 1321) that provides secret-key
authentication and integrity checking
40Authentication of Registration Messages via Keyed
MD5
- A mobile node generates a Registration Request,
consisting of the fixed length portion and the
Mobile-Home Authentication Extension, it fills in
all the fields of the request and extension
except for the Authenticator field. Then it
computes 16-byte MD5 message digest over the
shared secret key, the fixed length portion, all
extensions without Authenticator field, and the
shared secret key again. The Mobile IP
authentication extensions provide both
authentication and integrity checking
41Replay Attacks
- A Bad Guy could obtain a copy of a valid
Registration Request, store it, and then replay
it at a later time, thereby registering a bogus
care-of address for the mobile node - To prevent that the Identification field is
generated is a such a way as to allow the home
agent to determine what the next value should be - In this way, the Bad Guy is thwarted because the
Identification field in his stored Registration
Request will be recognized as being out of date
by the home agent (timestamps or nonces are used
for Identification field)
42Summary
- Mobile IP registration has has built-in
prevention of denial-of-service attacks.
Specifically, it is impossible for a Bad Guy to
lie to a mobile nodes home agent about that
mobile nodes current care-of address, because
all registration messages provide authentication
of the messages source, integrity checking and
replay protection
43Theft of Information Passive Eavesdropping
- Assumption unauthorized persons will inevitably
gain wired or wireless access to the network
infrastructure - Use of Link-Layer Encryption
- We also assume that key management for the
encryption is performed without disclosing the
keys to any unauthorized parties - Use of End-to-End Encryption (SSH, SSL)
44End-to-End Encryption vs. Link Encryption
- The Encapsulating Security Payload (RFC 1827) can
provide end-to-end encryption to other
application programs not supporting it themselves
45Theft of Information Session-Stealing (Takeover)
Attack
- A Bad Guy waits for a legitimate node to
authenticate itself and start an application
session - Then it takes over the session by impersonating
the identity of the legitimate node - Usually he must send a tremendous number of
nuisance packets to the legitimate node in order
to prevent it from realizing that its session was
hijacked
46Session-Stealing on the Foreign Link
- The Bad Guy waits for a mobile node to register
with its home agent - The Bad Guy eavesdrops to see if the mobile node
has any interesting conversation taking place
(remote login session to another host, connection
to the electronic mailbox) - The Bad Guy floods the mobile node with nuisance
packets - The Bad Guy steals the session by sending the
packets that appear to have come from the mobile
node and by intercepting packets destined to the
mobile node
47Session-Stealing Prevention
- Same method as in the case of Passive
Eavesdropping - minimally link-layer encryption between the
mobile node and the foreign agent
(session-stealing on the foreign link) - With the preference of end-to-end encryption
between the mobile node and its corresponding
node (elsewhere) - Note a good encryption scheme provides a method
by which a decrypting node can determine whether
the recovered plaintext is gibberish or whether
it is legitimate (integrity checking)
48Other Active Attacks
- The Bad Guy connects to the network jack, figures
out he IP address to use, and tries to break to
the other hosts on the network - He figures out the network-prefix that has been
assigned to the link on which the network jacks
connected - The Bad Guy guesses a host number to use, which
combined with the network-prefix gives him an IP
address to use on the current link - The Bad Guy proceeds to try to break into the
hosts on the network guessing user-name/password
pairs
49Protection against such attacks
- All publicly accessible network jacks must
connect to foreign agent that demands any nodes
on the link to be registered (authenticated). - Remove all non-mobile nodes from the link and
require all legitimate mobile nodes to use
(minimally) link-layer encryption
50Summary Intranet Model Security
- We described a simple deployment of Mobile IP on
individual corporate campus (intranet) - All of the routers were upgraded to be both home
agents and foreign agents, all reasonably
portable host were upgraded to mobile hosts - Home addresses were assigned according to the
users department - Mobile IP authentication Keys were configured
between the mobile nodes and their respective
home agents - Assumed the existence of physical security flaws
- Used link encryption over the foreign link to
minimally protect the internal data, but
generally preferred end-to-end encryption - Considered Denial-of Service attack in which a
Bad Guy lie to a mobile nodes home agent about
mobile nodes current care-of address - Showed how a combination of the Mobile-Home
Authentication Extension and Identification field
are designed to provide Authentication, Integrity
Checking, and Replay Protection for all
Registration Requests and Replies
51Part 3b. Internet-Wide Mobility A more
Complicated Application
- This Mobil IP application allows a user to move
anywhere throughout the entire Internet without
exposing his private network to additional
security threats - We will consider the problem of mobile nodes
getting packets past the firewall when they are
outside of the private network boundary (the
subject of active research in Mobile IP Working
Group of Internet Engineering Task Force)
52Model for This Application
53The Requirements
- There must be a firewall between the corporate
network and the global Internet - Authorized mobile nodes belonging to employees of
the corporation must not suffer any loss of
connectivity to resources inside the firewall,
even when connected to a foreign link outside the
firewall - The corporate network must not be exposed to any
new security threats over and above those that
face any network connected to the Internet
(through a firewall) - A visitor must be able to communicate with the
global Internet (and presumably his own private
network) from public areas such as conference
rooms, training facilities, etc.
54Threats That Are the Same As Before
- Threats from insiders (restrict access to info)
- Denial-of-service attacks (use of strong
authentication) - Passive eavesdropping and active takeover attacks
(encryption) - Physical Intrusion to the restricted portion of
the campus (control of physical access)
55Firewalls
- 3 basic types of firewalls packet-filtering
routers, application-layer relays, and secure
tunnelers
56Packet-Filtering Router as Firewall
57Example of Access Control List (ACL)
- Forward all packets belonging to connections
initiated by internal machines - Forward all packets belonging to email
connections initiated by outside machines - Forward all DNS messages
- Discard all other packets
58Advantages of Packet-Filtering Routers
- Fast (simple processing involving examining of IP
Source and Destination Address fields, and TCP
and UDP header fields) - Independent of applications
- Inexpensive to upgrade
59Problems
- Difficult to configure correctly
- Obscure syntax of ACLs (usually there is no GUI)
- Any mistake leaves the private network
vulnerable to security attacks - No reliable way to check ACLs correctness
- IP addresses of the machines in the private
network are visible to the public network - Little or no disk space to log a suspicious
activity - Do not support users authentication before being
allowed to communicate outside the firewall
60Application-Layer Relays
- The two routers are configured with ACLs which
allow packets only to and from the relay host
61Advantages
- Ability to enforce more sophisticated security
policies since they understand not only packet
headers, but also the applications themselves - Auditing and logging capabilities
- Authentication support
62Disadvantages
- Slow and visible to end users (might have many
TCP connections open at the same time) - Some applications might not be supported by
firewall possibly because it does not work
symmetrically in both directions
63Secure Tunnelers (picture)
64Secure Tunnelers
- If the packet is tunneled to the firewall and has
valid authentication (and usually encryption), it
is de-tunneled and routed transparently to the
destination node within the private network - Otherwise, the packet is submitted to
application-layer relay and is processed
accordingly
65Virtual Private Network (VPN)
66Host1 sends a packet to Host2 (see the previous
picture)
- Host1 builds an IP packet its own IP address as
the Source Address and Host2s IP address as the
Destination Address - The packet is ultimately forwarded to the
firewall on the left - The firewall prepends an IP Encapsulating
Security Payload header to the original IP packet
and encrypts the original IP header and payload
(the encryption algorithm provides authentication
and integrity checking as well) - The firewall places the resultant Encapsulating
Security Payload header plus encrypted original
packet within the payload portion of a new IP
packet. The new IP packet has a Source Address of
the leftmost firewall and a Destination Address
of the rightmost firewall - The new packet is transmitted over the Internet,
where it is ultimately received by the firewall
on the right
67Host1 sends a packet to Host2 (cont.)
- The firewall consumes the outermost IP packet
header and examines the IP Encapsulating Security
Payload header. The Security Parameters Index
field within that header informs the firewall how
to process the received cipher-text. The firewall
proceeds to decrypt and verify the authentication
and integrity of the packet - If the packet is authentic, the firewall removes
the IP Encapsulating Security Payload header to
recover the original IP packet - The firewall forwards the packet, which is
ultimately delivered to Host2 via conventional
routing
68How do we protect a Mobile Node That Is Outside
the Firewall?
- Mobile Node as a Special Case of Virtual Private
Networks (VPN) - The firewall is a software module running on
the mobile node
69Requirements for Secure Firewall Traversal in
Mobile IP
- Must protect the mobile node and the private
network from passive eavesdropping and active
takeover attack - Must work for organizations that have private
addresses (that are not advertised to the rest of
the Internet) on their networks - Must not require the firewall to implement or
understand Mobile IP - Must resolve the problem of the mobile node
Registration through the firewall - Must work in presence of internal private network
firewalls
70Firewall Traversal Using VPN Questions to Answer
- How does a mobile node establish the
authenticated and/or encrypted tunnel to the
firewall? - Does the mobile node establish this tunnel before
or after it registers with its home agent? - Is the mobile nodes home agent inside or outside
the firewall? - How do we establish keys between the mobile node
and its firewall? - How do the mobile node and the firewall agree on
a set of encryption and/or authentication
algorithms to use? - How does the mobile node know whether it is
inside or outside the firewall?
71Conclusions
- Firewall Traversal is a work in progress in the
Mobile IP community - It usually implements the IP Authentication
Header, IP Encapsulation Security Payload and
ISAKMP/Oakley for key management - The general solution can be formulated as
establishing an encrypted and authenticated
tunnel between the mobile node and the firewall
72Summary
- We described a more complicated deployment of
Mobile IP on individual corporate campus that was
characterized by placing all publicly accessible
network jacks outside of the corporations
firewall - We also sketched in general a solution for
firewall traversal using Virtual Private Networks
73Model for Commercial, Mobile IP service
74References
- James D. Solomon, Mobile IP The Internet
Unplugged, Prentice Hall, 1998. - David B. Johnson. Mobile IP in the Current and
Future Internet, Tutorial for MobiCom 2000. - Charles Perkins, Mobile Networking with Mobile
IP, IEEE Internet Computing, 2(1)58-69,
January/February 1998.