One document matched: draft-montenegro-mipv6sec-bit-method-00.txt



Network Working Group                                      G. Montenegro
Internet-Draft                                          Sun Labs, Europe
Expires: October 4, 2002                                     P. Nikander
                                            Ericsson Research NomadicLab
                                                           April 5, 2002


                Protecting Against Bidding Down Attacks
                draft-montenegro-mipv6sec-bit-method-00

Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups.  Note that
   other groups may also distribute working documents as Internet-
   Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at http://
   www.ietf.org/ietf/1id-abstracts.txt.

   The list of Internet-Draft Shadow Directories can be accessed at
   http://www.ietf.org/shadow.html.

   This Internet-Draft will expire on October 4, 2002.

Copyright Notice

   Copyright (C) The Internet Society (2002).  All Rights Reserved.

Abstract

   Mobile IPv6 uses "return routability" to secure route optimization.
   Even after using this procedure, there are residual threats for which
   other stronger methods provide protection.  Since these are optional,
   and return routability is the default, an attacker may engage in
   "bidding down" attacks.  These attacks aim at coercing participants
   in Mobile IPv6 route optimization to forgo the stronger methods for
   the default return routability.  This document discusses what the
   participants in route optimization can do to deter or alleviate
   bidding down attacks: the "step down" procedure for the mobile node
   and the "bit method" at the correspondent node.



Montenegro & Nikander    Expires October 4, 2002                [Page 1]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


Table of Contents

   1.    Introduction . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.    Bidding Down Attacks . . . . . . . . . . . . . . . . . . . .  6
   2.1   Man-in-the-Middle Attack (MiTM)  . . . . . . . . . . . . . .  7
   2.1.1 General Case . . . . . . . . . . . . . . . . . . . . . . . .  7
   2.1.2 Attack in the Mobile IPv6 Case . . . . . . . . . . . . . . .  8
   2.1.3 Mallory situated between the Home Agent and Bob  . . . . . . 10
   2.2   Impostor Attack  . . . . . . . . . . . . . . . . . . . . . . 11
   3.    Defending Against Bidding Down Attacks . . . . . . . . . . . 12
   3.1   Step Down Procedure  . . . . . . . . . . . . . . . . . . . . 12
   3.2   Bit Method . . . . . . . . . . . . . . . . . . . . . . . . . 13
   4.    Security Considerations  . . . . . . . . . . . . . . . . . . 15
   5.    Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 16
         References . . . . . . . . . . . . . . . . . . . . . . . . . 17
         Authors' Addresses . . . . . . . . . . . . . . . . . . . . . 17
         Full Copyright Statement . . . . . . . . . . . . . . . . . . 18


































Montenegro & Nikander    Expires October 4, 2002                [Page 2]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


1. Introduction

   Mobile IPv6 [2] must support route optimization between mobile nodes
   and correspondent nodes that have no previous security relationship
   with each other.  It must do so without recourse to infrastructure
   based solutions.  The default mechanism used is called "return
   routability" (RR).  Since RR is not based on strong cryptographic
   assurances there are residual threats (see Residual_Threats.txt [1]),
   in particular, the so called "future attacks" [3][4], as compared to
   the baseline case of IPv6 without mobility,

   Bidding down attacks raise serious concerns in two special areas:

   1.  Effects on Stationary Nodes

      A stationary node may not wish to be involved in route
      optimization as a mobile node (i.e.  to redirect its address by
      asking its correspondent nodes to instal binding cache entries on
      its behalf).  Such a node would benefit from the ability to
      unequivocally (and securely) indicate that they never redirect
      their address.  Otherwise, it can fall prey to attacks which
      exploit the residual threats allowed by RR.

      This is potentially a very serious threat, because it can make any
      "legacy" node in the Internet vulnerable to the future attacks.
      This includes nodes which are always stationary and which would
      never do that which mobile nodes do when they move: use RR to ask
      their peers to insert binding cache entries on their behalf (to
      redirect their addresses).  Popular servers would be prime targets
      for attack given their well-known addresses: rogue nodes could
      "redirect" them by installing bogus binding cache entries.  Of
      course, all IPv6 nodes are expected to implement route
      optimization as correspondent nodes, but not as mobile nodes.

      Therefore, it is highly desirable to set up a mechanism that
      allows correspondent nodes to easily check whether a given node's
      address is redirectable or not.

   2.  Implications for Improved IPv6 Signaling Security

      There are implications with respect to a future Internet with
      improved IPv6 control signaling security.  For example, it is
      conceivable that Neighbour Discovery security could be improved.
      Due to existing issues with Neighbor Discovery [5] RR does not
      make current on-link security much worse.  However, this could
      change as a result of improvements in ND security, and as a result
      RR might be left as the weakest link in Internet security.  There
      is a need to future-proof IPv6 by (1) putting mechanisms in place



Montenegro & Nikander    Expires October 4, 2002                [Page 3]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


      from the very start, and (2) requiring that all IPv6 nodes
      implement them.  Only such an arrangement would allow a secure
      migration towards more secure Binding Update authorization
      mechanisms, for example, at the same time as ND is improved.
      However, this is very hard if attackers can claim that the nodes
      in question only support RR.

      Because of this, it is essential to allow other (non-default)
      stronger mechanisms to be used instead of RR.

   Thus, the selection mechanism should be impervious to the "bidding
   down" attack, in which an attacker forces parties that are initially
   interested in engaging in a more secure procedure, to fall back on a
   less secure one.

   All Mobile IPv6 implementations should protect against the "bidding
   down" attack.  Several such mechanisms have been proposed.  This
   document reviews the step down procedure used in conjunction with the
   "bit method" and examines their effectivity.

   The "bit method" essentially divides the IPv6 interface identifier
   space into two.  Addresses from one space allow a node to express the
   following to its peers:

      "(1) I do not engage in redirection operations on my address, or,
      if I do, (2) I only do so with cryptographically strong mechanisms
      in which I will prove to your satisfaction that I do have
      authorization to use that address."

   Addresses from the other space allow a node to express the following
   to its peers:

      "(3) I engage in redirection operations on my address and do not
      intend to prove conclusively that I do have authorization to use
      this address (e.g.  RR is ok)"

   (3) is a viable tradeoff for a mobile node to make: yes, there are
   some vulnerabilities, but it obtains route optimization by being able
   to redirect its address with a relatively simple mechanism free of
   intellectual property.

   For an unsuspecting stationary node that is not interested in
   redirecting its address, (3) may not be a viable tradeoff.  It
   essentially means that the node only obtains the negative side of the
   tradeoff (it becomes a potential victim of the vulnerabilities in
   RR), as it is not at all interested in the benefit (route
   optimization of its addresses).




Montenegro & Nikander    Expires October 4, 2002                [Page 4]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


   An attacker could just claim that the address is his and insert a
   binding cache entry (a host route) into any correspondent node.  This
   will break communications between the correspondent node and the
   stationary node.  If, on the other hand, the address expresses either
   (1) or (2) above to the correspondent node, this attack is no longer
   possible: the correspondent node will expect a cryptographic proof of
   authorization.  If the attacker can produce it, there are much more
   serious problems at hand.

   Nevertheless, it is not absolutely necessary to separate the address
   space as described above by using the "bit method." It is enough to
   require all nodes interested in redirecting their addresses to prove
   cryptographically that their address is redirectable.  This implies
   that ALL redirectable addresses must be "Hash Based Addresses," that
   is, they must be derived from a secure hash of some bit sequence (not
   necessarily a Public Key), and require proof of that derivation as
   part of the redirection procedure (e.g.  route optimisation).  Since
   stationary hosts do not produce their addresses in this manner, they
   are protected by the difficulty in inverting the secure hash: it is
   prohibitively expensive for attackers to produce derivation proofs
   for "stationary" addresses.

   The problem is that these "Hash Based Addresses" have IPR concerns.
   At the time of this writing it is not at all clear if those issues
   will be resolved.  Accordingly, this document deals with what the
   Mobile IPv6 Security Design Team believes to be the second best
   method: the "bit method."  Nonetheless, other methods are mentioned
   in Section 3.2.

   Section 2 discusses the Bidding Down attack, examining its
   implications with respect to the mobile node and the correspondent
   node.  Thus, it motivates a subsequent Section 3 on the defense
   mechanisms.


















Montenegro & Nikander    Expires October 4, 2002                [Page 5]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


2. Bidding Down Attacks

   The problem at hand consists of a node that wishes to affect how
   another node routes packets  to it.  Given that the problem is not
   limited to Mobile IP, it is worthwhile to adopt a more general model.
   Alice is the node that wishes to alter how Bob routes packets to it.
   We also assume that there is an attacker, Mallory, that launches the
   bidding down attack.

   Bidding down attacks exist in general outside the domain of Mobile
   IP.  The analysis starts with a general model in which it may be more
   difficult to implement defenses, and then proceeds to the more
   specific Mobile IP model.

   Assume Alice and Bob are capable of two flavors of secure exchanges:

   "strong" flavor: As part of the exchange, Alice can provide
      cryptographical assurances to Bob about some of its properties,
      including its address (every single bit in it).  This can be
      obtained in a variety of manners, infrastructure-less (CGA [6][7])
      or infrastructure-based (PKI, key distribution center, trusted
      third party, AAA, etc).  The exact method used does not matter.
      What matters is the requirement that in a strong exchange Alice
      can cryptographically prove to Bob certain properties about itself
      (including its address).  The exchange itself is also integrity
      protected.  "Strong" schemes under consideration for Mobile IP all
      do, of course, satisfy these requirements.

   "weak" flavor: As part of the exchange, Alice *cannot* provide
      cryptographical assurances to Bob about some of its properties,
      including its address.

   Alice wants to engage Bob in an exchange, and Mallory wishes to
   affect the result of the exchange such that they end up using the
   weak flavor.

   Let's assume a fixed bit (or bit pattern) in Alice's address (the
   "bit method") makes an explicit distinction between addresses used
   with the "strong" and "weak" exchanges.  The objective is to avoid
   bidding down from the "strong" to the "weak" flavor.

   Of course, if Alice implements the "strong" exchange, a very valid
   policy would be for it not to engage any more in "weak" exchanges.
   This simplifies Alice's protocol processing and is more secure
   because Alice avoids any risk of falling victim to a bidding down
   attack.  For a mobile node, this translates to requiring a "strong"
   mechanism for route optimization.  The mobile node simply forgoes the
   benefits of route optimization and limits itself to bidirectional



Montenegro & Nikander    Expires October 4, 2002                [Page 6]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


   tunneling [2] when it communicates with correspondent nodes that only
   implement the "weak" RR mechanism.

   If, however, Alice continues to employ the "strong" and "weak"
   exchanges, she cannot use the same address for both.  Because of the
   fixed bit pattern in the address, in essence, she appears to her
   peers as two different hosts ("strong" Alice versus "weak" Alice),
   depending on which of her two different addresses she uses:

   As:  address used with the "strong" exchange.

   Aw:  address used with the "weak" exchange.

   At least the interface ID portions of these addresses (lower 64 bits)
   will most probably be completely different and uncorrelated.  This is
   true of CGA [6][7] mechanisms under consideration as "strong"
   infrastructureless mechanisms for route optimization.  Therefore, it
   is not always possible for an observer to derive one address from the
   other, although some methods may allow it.  If, in fact, the routing
   prefix portions of both addresses (higher 64 bits) are also
   uncorrelated, the possibility that Mallory can remain "in the middle"
   is negligible.  An attacker that can achieve such a feat has already
   subverted the routing infrastructure to such a degree that it will be
   in a position to launch much more serious attacks.  Accordingly, if
   Alice uses both the "strong" and "weak" mechanisms, it is in her best
   interest to make sure that the corresponding addresses are completely
   uncorrelated (both the routing prefixes as well as the interface
   ID's).

   Additionally, if Alice is a mobile node it will also use a care-of
   address at its visiting link.

2.1 Man-in-the-Middle Attack (MiTM)

   Suppose the following situation:

      Alice  ====  Mallory  ==== Bob

   Mallory's convenient location may have been acquired by exploiting
   vulnerabilities in neighbor discovery, or by gaining control of a
   conveniently located network element (a router, for example).  It is
   important to point out that routers that carry both incoming and
   outgoing traffic for Alice are in an ideal position to launch DoS
   attacks.

2.1.1 General Case

   The simplest attack is as follows:



Montenegro & Nikander    Expires October 4, 2002                [Page 7]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


      Alice(As,strong) ===> Mallory         X  Bob

      Alice            <==  Mallory (weak)  X  Bob

   Alice sends, among other things, its "strong" address, itself an
   indication that it wishes to engage in the strong exchange.  Mallory
   throws this packet away and replies spoofing Bob, indicating a
   preference for the weak variant.  Notice that if Mallory is on the
   path ingress filtering does not apply to this spoofed packet.

   Also, this is much simpler than rewriting Alice's address with a
   "weak" address and then sending the packet to Bob.  Mallory simply
   repeats this for any signaling packets to Bob in which Alice uses its
   strong address.

   The hope is that eventually, Alice will give up and use its weak
   address, at which point, Mallory will let the traffic through,
   presumably, because it can break the protocol:

      Alice(Aw,weak)   ===> Mallory        ==> Bob

      Alice(Aw,weak)   <==  Mallory        <== Bob

   The above step is not as obvious as it seems.  As explained above,
   generally there is no correlation whatsoever between the previous As
   address observed by Mallory, and the subsequent Aw address once Alice
   gives up and settles for the weak exchange.  Mallory would need to
   use heuristics, or look at link layer information in order to
   establish that this newly observed address Aw belongs to Alice.
   Alternatively, Mallory could just be on the lookout for any signaling
   exchange carried out with weak addresses in order to exploit them.

   The attack depends on Mallory's being able to selectively drop all
   the exchange signaling packets from Alice and substitute them with
   its own.  This also depends on As and Aw using the same route (both
   going through Mallory).

2.1.2 Attack in the Mobile IPv6 Case

   The above attack is not nearly as easy in MobileIPv6, because it is
   much harder for Mallory to selectively drop the exchange signaling
   packets from Alice and substitute with its own.

   This difficulty stems from the fact that in the RR mechanism, the
   mobile node sends simultaneous RR "initialization" messages through
   two different routes.  The figure below shows a summarized rendition
   of the RR exchange.  In it, messages 1 (HoTI) and 2 (CoTI) initialize
   the RR process.  The mobile node reverse tunnels the HoTI via its



Montenegro & Nikander    Expires October 4, 2002                [Page 8]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


   home agent (first route), and it sends the CoTI directly (second
   route).  Likewise, the correspondent node responds with parallel
   messages by sending the HOT to the mobile node's home address HoA
   (which is then intercepted and tunneled to the mobile node's current
   location by the home agent), and the COT directly to the mobile node
   at its care-of address CoA.  The mobile node is able to send a valid
   binding update (message 5) only after it receives valid HOT and COT
   messages [2].

       1. MN(HoA) -> CN: HoTI(HoA, ...)
       2. MN(CoA) -> CN: CoTI(CoA, ...)
       3. CN -> MN(HoA): HoT(...)
       4. CN -> MN(CoA): CoT(...)
       5. MN(CoA) -> CN: BU(HoA, CoA, ...)

   Of course, if Mallory is on Alice's local link it can discard all its
   packets.  But this constitutes a DoS attack.  To carry out a bidding
   down attack, Mallory must be more subtle.  It must not to alert Alice
   that something out of the ordinary is happening.  Thus it needs to be
   very selective as to which packets it will act upon.

   Because of this it is much safer for Alice to encrypt messages
   tunneled through its home agent, as MIPv6 says it SHOULD.
   Furthermore, it is much safer to encrypt the tunneled packets, and
   not merely the Mobility Header messages.  This way Mallory cannot
   distinguish between data and signalling packets being tunneled by
   Alice through her home agent (HA).

   Here, both the strong and weak exchanges share the RR (return
   routability) procedure in which Alice and Bob exchange packets in
   parallel via topologically independent routes.  Since the MN to HA
   path should be encrypted, this makes it very difficult for Mallory to
   actually remain in the middle (i.e.  be able to see and selectively
   modify traffic) for all the packet exchanges.  Suppose Alice using a
   strong address MN is the mobile node, and Bob is the correspondent
   node.

   Alice initiates the RR exchange by sending two packets simultaneously
   (HoTI via the home agent, and CoTI directly to the correspondent
   node), both of which express preference for a strong exchange.  This
   preference is indicated in two ways by the HoTI:

   o  by the use of a "strong" address, and

   o  by explicit fields within the signaling packet

   The CoTI only uses the latter indication.  The above attack would be:




Montenegro & Nikander    Expires October 4, 2002                [Page 9]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


   strong HoTI: CoA(encrypted tunnel)  ===> HA         ==> Bob
   strong CoTI: CoA                    ===> Mallory     X  Bob

   Notice that Mallory is no longer "in the middle" because the HoT
   packet is encrypted.  Mallory may discard only the packet it sees.
   In the second phase of the attack, the RR responses to the above
   arrive at Alice:

   strong HoT: CoA(encrypted tunnel)  <=== HA         <== Bob
   weak   CoT: CoA                    <=== Mallory     X  Bob

   Alice must wait to receive both the HoT and the CoT before proceeding
   with the RR procedure.  It will see two conflicting replies from Bob.
   In particular, it is quite suspicious that the safe reply (the HoT
   via the encrypted tunnel from its HA) indicates a preference for a
   strong exchange, while the CoT indicates the opposite.  This is
   already a very strong indication that an attack is under way.

   However, if Alice is not encrypting, or if it is selectively
   encrypting only the signaling portion of the packets, it is possible
   for Mallory to launch the attack.  Namely, it can discard signaling
   packets from Alice (HoTI and CoTI) and reply with its own indicating
   a preference for a weak exchange.  In this case, both HoT and CoT
   would be consistent, and Alice might restart the RR procedure using
   its weak address in a subsequent HoTI.

2.1.3 Mallory situated between the Home Agent and Bob

   The above procedure is particularly important in this case.  In this
   situation, RR is known to be breakable, and if Mallory is able to
   place itself in the middle (with the ability to discard packets
   selectively) then there is no further protection.  However, if
   neighbor discovery is secure in the future, this may be very
   difficult for Mallory to achieve.

   In this case, the required retransmissions will give the mobile node
   a chance to hear back from the real correspondent node in case it
   does in fact support the strong exchange.  Simply put, Alice must
   handle bidding down attacks by being very stubborn and insisting on
   (and re-requesting) a strong exchange.  Protection on Bob's side is
   much simpler.  It simply hinges on the fact that Bob will never
   engage in weak exchanges with nodes whose address have the bits set
   to indicate preference for a strong exchange.








Montenegro & Nikander    Expires October 4, 2002               [Page 10]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


2.2 Impostor Attack

       Mallory === Bob
         Alice  .../

   Here, Alice may not be actively communicating with Bob.  It need not
   even be on line.  Mallory poses as Alice and tries to convince Bob.

   We assume that Alice is either a stationary node that does
   participate in redirection procedures at all, or, if it does, that it
   wishes to use something other than the "weak" flavor: it does not use
   the baseline RR as its binding update authorization method, instead
   preferring a much stronger alternative.

   The attack starts by the attacker (Mallory) sending a fake request to
   the correspondent node (Bob), indicating the desire to use RR in the
   request.  The correspondent node (Bob), not having any knowledge of
   the true desires of the mobile node (Alice) accepts this.

   1.  Attacker sends first message(s) of RR to the correspondent node
       (the HoTI and CoTI messages).  In this network scenario both
       messages can be sent from the attacker's location towards the
       correspondent node.  This is because the attacker can place
       itself between the correspondent node and the CoA by carefully
       choosing the care-of address it uses in RR.  This can be done
       easily, for example, by selecting an address from the network
       where the attacker is located.

   2.  The correspondent node inspects the messages and continues the RR
       protocol by sending its COT and HOT messages.

   3.  Once the attacker receives the responses it is in a position to
       send a valid binding update.

   As a result, the attacker has established a binding cache entry for
   the mobile node's (Alice's) address at the correspondent node.  It
   only had to use RR regardless of the mobile node's preference for a
   stronger mechanism.













Montenegro & Nikander    Expires October 4, 2002               [Page 11]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


3. Defending Against Bidding Down Attacks

   As can be seen from the analysis above, bidding down attacks have
   different characteristics depending on whether they are looked at
   from the point of view of the mobile node or the correspondent node.
   Accordingly, they must use different mechanisms.  The two following
   sections detail the defenses used by the mobile node (Step Down
   Procedure) and the correspondent node (Bit Method).

3.1 Step Down Procedure

   By following certain rules a mobile node reduces the possibility of a
   bidding down attack and remains in control.  This is called the step
   down procedure.

   When Mallory is on the same link as mobile node Alice it can launch
   attacks to make her desist of the strong mechanism and settle for the
   weak one.  In order to prevent falling victim to these attacks, Alice
   must be very careful as to the conditions under which it will give up
   on using the strong mechanism.

   To begin with, above it has been noted that Alice MUST have valid HoT
   and CoT packets that indicate preference for the weak exchange.  In
   the event that Mallory also produces false HoT packets, Alice MUST be
   able to give preference to those received from its HA under the
   protection of the ESP tunnel.  It is not clear how this information
   is kept with the packet after it has been decapsulated and extracted
   from its ESP protection.  Because Mallory could be dropping packets
   from the HA, Alice MUST handle bidding down packets in similar manner
   to how it handles lost packets and the corresponding retransmissions.

   The rules for a mobile node to carry out a step down procedure are:

   o  Both a valid HoT and CoT are required to initiate a step down
      procedure.

   o  The procedure requires two more pairs of related HoT and CoT
      packets (should we only require one more?).

   o  Only after this complete step down procedure does a mobile node
      heed a correspondent's preference for a weak exchange and begin
      the corresponding weak RR process (in which the mobile node uses
      its weak address) or simply by desisting on using route
      optimization.

   If at any time during the step down procedure the mobile node
   receives any messages which specify preference for something other
   than RR (a stronger exchange), the step down procedure is aborted.



Montenegro & Nikander    Expires October 4, 2002               [Page 12]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


   This means that the mobile node will not engage in a weak exchange.
   Of course, it may also desist on using route optimization.

3.2 Bit Method

   The first time Bob receives a packet sent by Alice, he has no other
   knowledge about her but the packet itself.  In particular, the only
   information he has about Alice is the source IP address in the
   packet.  If Alice is a stationary node, she does not want her address
   to be redirectable, so Bob must not install a binding cache entry for
   it.  To prevent would-be attackers from "redirecting" Alice's
   address, Bob needs to learn if her address is indeed redirectable.
   He has several options:

   1.  He can query DNS, do a reverse lookup, and hope to learn
       something useful.  However, this may be expensive in terms of
       time and number of packets, causing packets sent to the DNS
       server that serves the reverse map for Alice's address, and
       thereby somewhat dangerous from the Denial-of-Service point of
       view.  Additionally, secure DNS is not widely deployed, and
       therefore he usually cannot fully trust the results.

   2.  He can try to use some other infrastructure, such as AAA, to
       obtain some information about Alice.  However, this method
       requires that Alice participates in the same infrastructure,
       which, in general, is unlikely for a number of years to come.
       Secondly, this method has the same drawbacks as DNS lookups,
       generating delay and extra traffic.

   3.  He can send a probe to Alice, checking that there indeed is
       somebody who answers at her address.  TCP SYN ACK is basically
       such a probe, and so are the new Mobile IPv6 RR packets (HoTI,
       CoTI, HoT, CoT).  The drawback of this method is that it does not
       say anything about Alice's address being redirectable or not (the
       whole point of this exercise).  Notice that once Bob knows that
       Alice's address is redirectable, this step is an integral part of
       that process, which is why it is part of RR.

   4.  He can require that, as part of the address redirection
       procedure, Alice prove in a cryptographically secure manner that
       her address is indeed redirectable.  The problem is that the only
       known way to do this in an infrastructureless manner raises
       intellectual property concerns whose implications are not clear.

   5.  He can have a look at the address, and try to deduce something
       from the address itself.  Currently, he can learn whether Alice
       is using a link local, site local or global address, whether
       Alice's address is supposed to contain a globally unique



Montenegro & Nikander    Expires October 4, 2002               [Page 13]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


       interface ID, etc.

   The whole idea of the "bit method" is to extend the last option above
   so that Bob can tell by looking at the address if it is redirectable
   or not.

   The "bit method" allocates a bit (or a bit pattern) in the IPv6 64
   bit Interface Identifier field, with the intention of later defining
   a method that allows peer hosts to securely find out what security
   protocols, and perhaps other functional features, the host using the
   address supports.  That is, if a host's address has the bit set, the
   host indicates to its peers either of the following:

      It does not use "redirection procedures" to influence how packets
      destined to itself are routed by a peer.  Currently, the only such
      "redirection procedure" proposed for widespread deployment is the
      Mobile IPv6 route optimization procedure.  Other similar
      applications of such procedures may appear in the future (for
      example for certain anycast implementations).  In the absence of
      any further signaling, the peer (correspondent node) assumes this
      is the case.

      If the node does use "redirection procedures", then it only does
      so by using a "strong" mechamism as defined above, i.e.  using
      further signaling and strong cryptographic mechanisms.

   If the bit is cleared in the host's address, it relies on the default
   ("weak") mechanisms.  In the case of Mobile IPv6, this is the RR
   procedure.

   Note that an active attacker on the path between Alice and Bob is
   able to clear a set bit.  However, that changes the address, and
   Alice is not going to answer to any possible replies sent by Bob.
   Thus, the bit prevents the attacker from impersonating as Alice and
   fooling Bob to use the less secure protocol.

   If the bit is set, the correspondent host MUST require a
   cryptographic assurance before heeding any redirects for the address
   in question.  Failure to do so means the correspondent host MUST NOT
   heed any redirects.  This eliminates any potential misuse of Return
   Routability to attack a given address, and prevents the correspondent
   node from being bid down.









Montenegro & Nikander    Expires October 4, 2002               [Page 14]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


4. Security Considerations

   This note deals almost exclusively with security issues.  The issue
   at hand is that the baseline Mobile IPv6 route optimization
   mechanism, RR, may "pollute" the Internet at large with some
   additional threats.  These affect all hosts (not just mobile hosts).
   Any protection must be mandated as part of the baseline RR mechanism.
   We discuss the "bit method" in which a bit or bit pattern in the
   address prevents the use of RR to attack stationary hosts.  Coupled
   with the "step down" procedure, it alleviates the attack against
   mobile nodes that implement route optimization mechanisms stronger
   than RR.  Unlike the latter which relies on trust in the routing
   infrastructure in order to authorize binding updates, the "strong"
   mechanisms all employ cryptographic assurances of some sort or other.

   This document does not discuss the "bidding aside" attacks, in which
   an attacker forces the use of a "strong" mechanism instead of another
   also "strong" mechamism.  Presumably, it would do so because not all
   such mechanisms are equally strong, so by choosing judiciously, an
   attacker can lower the bar.  Nevertheless, this bar is sufficiently
   much higher than the RR level, since it requires carrying out
   successful cryptographic attacks.  If an attacker is able to do this,
   it can basically attack the Internet at will.  Furthermore, careful
   review before approving any future optional "strong" mechamisms can
   assure that the bar remains high enough.


























Montenegro & Nikander    Expires October 4, 2002               [Page 15]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


5. Acknowledgements

   This document "borrows" extensively from Mobile IPv6 security Design
   Team write-ups [1] and discussions.  Accordingly, it owes much to the
   other members (besides the authors) of the design team, Jari Arkko
   and Erik Nordmark.  It also benefits from discussion on the Mobile IP
   and IPv6 mailing lists.  Brian Carpenter's observations in the latter
   led to the the step down procedure.











































Montenegro & Nikander    Expires October 4, 2002               [Page 16]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


References

   [1]  Arkko, J., Nordmark, E., Nikander, P. and G. Montenegro, "Mobile
        IPv6 Security Design Team Writeups", see http://www.piuha.net/
        ~jarkko/publications/mipv6/, April 2002.

   [2]  Perkins, C. and D. Johnson, "Mobility Support in IPv6",
        Internet-Draft http://www.ietf.org/internet-drafts/draft-ietf-
        mobileip-ipv6-16, March 2002.

   [3]  Arkko, j. and T. Aura, "MIPv6 BU Attacks and Defenses", draft-
        aura-mipv6-bu-attacks-01 (work in progress), March 2002.

   [4]  Devarapalli, V., "Future Attacks and Threats", Thread on the
        Mobile IP alias http://playground.sun.com/mobile-ip/WG-archive/
        frm04733.html, January 2002.

   [5]  Kempf, J. and E. Nordmark, "Threat Analysis for IPv6 Public
        Multi-Access Links", draft-kempf-ipng-netaccess-threats-00 (work
        in progress), November 2001.

   [6]  Roe, M., Aura, T., O'Shea, G. and J. Arkko, "Authentication of
        Mobile IPv6 Binding Updates and Acknowledgments", draft-roe-
        mobileip-updateauth-02.txt (work in progress), February 2002.

   [7]  Montenegro, G. and C. Castelluccia, "Statistically Unique and
        Cryptographically Verifiable  (SUCV) Identifiers and
        Addresses.", NDSS 2002, February 2002.


Authors' Addresses

   Gabriel Montenegro
   Sun Labs, Europe
   29, chemin du Vieux Chene
   38240 Meylan
   France

   EMail: gab@sun.com


   Pekka Nikander
   Ericsson Research NomadicLab

   Espoo
   Finland

   EMail: Pekka.Nikander@nomadiclab.com



Montenegro & Nikander    Expires October 4, 2002               [Page 17]

Internet-Draft    Protecting Against Bidding Down Attacks     April 2002


Full Copyright Statement

   Copyright (C) The Internet Society (2002).  All Rights Reserved.

   This document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph are
   included on all such copies and derivative works.  However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for
   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than
   English.

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an
   "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
   TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
   BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
   HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement

   Funding for the RFC Editor function is currently provided by the
   Internet Society.



















Montenegro & Nikander    Expires October 4, 2002               [Page 18]


PAFTECH AB 2003-20262026-04-23 20:08:08