RFC 5039






Network Working Group                                       J. Rosenberg
Request for Comments: 5039                                   C. Jennings
Category: Informational                                            Cisco
                                                            January 2008


             The Session Initiation Protocol (SIP) and Spam

Status of This Memo



   This memo provides information for the Internet community.  It does
   not specify an Internet standard of any kind.  Distribution of this
   memo is unlimited.

Abstract



   Spam, defined as the transmission of bulk unsolicited messages, has
   plagued Internet email.  Unfortunately, spam is not limited to email.
   It can affect any system that enables user-to-user communications.
   The Session Initiation Protocol (SIP) defines a system for user-to-
   user multimedia communications.  Therefore, it is susceptible to
   spam, just as email is.  In this document, we analyze the problem of
   spam in SIP.  We first identify the ways in which the problem is the
   same and the ways in which it is different from email.  We then
   examine the various possible solutions that have been discussed for
   email and consider their applicability to SIP.

























Rosenberg & Jennings         Informational                      [Page 1]

RFC 5039                        SIP Spam                    January 2008


Table of Contents



   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.  Problem Definition . . . . . . . . . . . . . . . . . . . . . .  3
     2.1.  Call Spam  . . . . . . . . . . . . . . . . . . . . . . . .  4
     2.2.  IM Spam  . . . . . . . . . . . . . . . . . . . . . . . . .  7
     2.3.  Presence Spam  . . . . . . . . . . . . . . . . . . . . . .  7
   3.  Solution Space . . . . . . . . . . . . . . . . . . . . . . . .  8
     3.1.  Content Filtering  . . . . . . . . . . . . . . . . . . . .  8
     3.2.  Black Lists  . . . . . . . . . . . . . . . . . . . . . . .  9
     3.3.  White Lists  . . . . . . . . . . . . . . . . . . . . . . .  9
     3.4.  Consent-Based Communications . . . . . . . . . . . . . . . 10
     3.5.  Reputation Systems . . . . . . . . . . . . . . . . . . . . 12
     3.6.  Address Obfuscation  . . . . . . . . . . . . . . . . . . . 14
     3.7.  Limited-Use Addresses  . . . . . . . . . . . . . . . . . . 14
     3.8.  Turing Tests . . . . . . . . . . . . . . . . . . . . . . . 15
     3.9.  Computational Puzzles  . . . . . . . . . . . . . . . . . . 17
     3.10. Payments at Risk . . . . . . . . . . . . . . . . . . . . . 17
     3.11. Legal Action . . . . . . . . . . . . . . . . . . . . . . . 18
     3.12. Circles of Trust . . . . . . . . . . . . . . . . . . . . . 19
     3.13. Centralized SIP Providers  . . . . . . . . . . . . . . . . 19
   4.  Authenticated Identity in Email  . . . . . . . . . . . . . . . 20
     4.1.  Sender Checks  . . . . . . . . . . . . . . . . . . . . . . 21
     4.2.  Signature-Based Techniques . . . . . . . . . . . . . . . . 21
   5.  Authenticated Identity in SIP  . . . . . . . . . . . . . . . . 22
   6.  Framework for Anti-Spam in SIP . . . . . . . . . . . . . . . . 23
   7.  Additional Work  . . . . . . . . . . . . . . . . . . . . . . . 24
   8.  Security Considerations  . . . . . . . . . . . . . . . . . . . 24
   9.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 24
   10. Informative References . . . . . . . . . . . . . . . . . . . . 25





















Rosenberg & Jennings         Informational                      [Page 2]

RFC 5039                        SIP Spam                    January 2008


1.  Introduction



   Spam, defined as the transmission of bulk unsolicited email, has been
   a plague on the Internet email system.  Many solutions have been
   documented and deployed to counter the problem.  None of these
   solutions is ideal.  However, one thing is clear: the spam problem
   would be much less significant had solutions been deployed
   ubiquitously before the problem became widespread.

   The Session Initiation Protocol (SIP) [2] is used for multimedia
   communications between users, including voice, video, instant
   messaging, and presence.  Consequently, it can be just as much of a
   target for spam as email.  To deal with this, solutions need to be
   defined and recommendations put into place for dealing with spam as
   soon as possible.

   This document serves to meet those goals by defining the problem
   space more concretely, analyzing the applicability of solutions used
   in the email space, identifying protocol mechanisms that have been
   defined for SIP that can help the problem, and making recommendations
   for implementors.

2.  Problem Definition



   The spam problem in email is well understood, and we make no attempt
   to further elaborate on it here.  The question, however, is what is
   the meaning of spam when applied to SIP?  Since SIP covers a broad
   range of functionality, there appear to be three related but
   different manifestations:

   Call Spam:  This type of spam is defined as a bulk unsolicited set of
      session initiation attempts (i.e., INVITE requests), attempting to
      establish a voice, video, instant messaging [1], or other type of
      communications session.  If the user should answer, the spammer
      proceeds to relay their message over the real-time media.  This is
      the classic telemarketer spam, applied to SIP.  This is often
      called SPam over Ip Telephony, or SPIT.

   IM Spam:  This type of spam is similar to email.  It is defined as a
      bulk unsolicited set of instant messages, whose content contains
      the message that the spammer is seeking to convey.  IM spam is
      most naturally sent using the SIP MESSAGE [3] request.  However,
      any other request that causes content to automatically appear on
      the user's display will also suffice.  That might include INVITE
      requests with large Subject headers (since the Subject is
      sometimes rendered to the user), or INVITE requests with text or
      HTML bodies.  This is often called SPam over Instant Messaging, or
      SPIM.



Rosenberg & Jennings         Informational                      [Page 3]

RFC 5039                        SIP Spam                    January 2008


   Presence Spam:  This type of spam is similar to IM spam.  It is
      defined as a bulk unsolicited set of presence requests (i.e.,
      SUBSCRIBE requests [4] for the presence event package [6]), in an
      attempt to get on the "buddy list" or "white list" of a user in
      order to send them IM or initiate other forms of communications.
      This is occasionally called SPam over Presence Protocol, or SPPP.

   There are many other SIP messages that a spammer might send.
   However, most of the other ones do not result in content being
   delivered to a user, nor do they seek input from a user.  Rather,
   they are answered by automata.  OPTIONS is a good example of this.
   There is little value for a spammer in sending an OPTIONS request,
   since it is answered automatically by the User Agent Server (UAS).
   No content is delivered to the user, and they are not consulted.

   In the sections below, we consider the likelihood of these various
   forms of SIP spam.  This is done in some cases by a rough cost
   analysis.  It should be noted that all of these analyses are
   approximate, and serve only to give a rough sense of the order of
   magnitude of the problem.

2.1.  Call Spam



   Will call spam occur?  That is an important question to answer.
   Clearly, it does occur in the existing telephone network, in the form
   of telemarketer calls.  Although these calls are annoying, they do
   not arrive in the same kind of volume as email spam.  The difference
   is cost; it costs more for the spammer to make a phone call than it
   does to send email.  This cost manifests itself in terms of the cost
   for systems that can perform telemarketer call, and in cost per call.

   Both of these costs are substantially reduced by SIP.  A SIP call
   spam application is easy to write.  It is just a SIP User Agent that
   initiates, in parallel, a large number of calls.  If a call connects,
   the spam application generates an ACK and proceeds to play out a
   recorded announcement, and then it terminates the call.  This kind of
   application can be built entirely in software, using readily
   available (and indeed, free) off-the-shelf software components.  It
   can run on a low-end PC and requires no special expertise to execute.

   The cost per call is also substantially reduced.  A normal
   residential phone line allows only one call to be placed at a time.
   If additional lines are required, a user must purchase more expensive
   connectivity.  Typically, a T1 or T3 would be required for a large-
   volume telemarketing service.  That kind of access is very expensive
   and well beyond the reach of an average user.  A T1 line is
   approximately US $250 per month, and about 1.5 cents per minute for
   calls.  T1 lines used only for outbound calls (such as in this case)



Rosenberg & Jennings         Informational                      [Page 4]

RFC 5039                        SIP Spam                    January 2008


   are even more expensive than inbound trunks due to the reciprocal
   termination charges that a provider pays and receives.

   There are two aspects to the capacity: the call attempt rate, and the
   number of simultaneous successful calls that can be in progress.  A
   T1 would allow a spammer, at most, 24 simultaneous calls, and
   assuming about 10 seconds for each call attempt, about 2.4 call
   attempts per second.  At high-volume calling, the per-minute rates
   far exceed the flat monthly fee for the T1.  The result is a cost of
   250,000 microcents for each successful spam delivery, assuming 10
   seconds of content.

   With SIP, this cost is much reduced.  Consider a spammer using a
   typical broadband Internet connection that provides 500 Kbps of
   upstream bandwidth.  Initiating a call requires just a single INVITE
   message.  Assuming, for simplicity's sake, that this is 1 KB, a 500
   Kbps upstream DSL or cable modem connection will allow about 62 call
   attempts per second.  A successful call requires enough bandwidth to
   transmit a message to the receiver.  Assuming a low compression codec
   (say, G.723.1 at 5.3 Kbps), this requires approximately 16 Kbps after
   RTP, UDP, and IP overheads.  With 500 Kbps upstream bandwidth, this
   means as many as 31 simultaneous calls can be in progress.  With 10
   seconds of content per call, that allows for 3.1 successful call
   attempts per second.  If broadband access is around $50/month, the
   cost per successful voice spam is about 6.22 microcents each.  This
   assumes that calls can be made 24 hours a day, 30 days a month, which
   may or may not be the case.

   These figures indicate that SIP call spam is roughly four orders of
   magnitude cheaper to send than traditional circuit-based telemarketer
   calls.  This low cost is certainly going to be very attractive to
   spammers.  Indeed, many spammers utilize computational and bandwidth
   resources provided by others, by infecting their machines with
   viruses that turn them into "zombies" that can be used to generate
   spam.  This can reduce the cost of call spam to nearly zero.

   Even ignoring the zombie issue, this reduction in cost is even more
   amplified for international calls.  Currently, there are few
   telemarketing calls across international borders, largely due to the
   large cost of making international calls.  This is one of the reasons
   why the "do not call list", a United States national list of numbers
   that telemarketers cannot call -- has been effective.  The law only
   affects U.S. companies, but since most telemarketing calls are
   domestic, it has been effective.  Unfortunately (and fortunately),
   the IP network provides no boundaries of these sorts, and calls to
   any SIP URI are possible from anywhere in the world.  This will allow
   for international spam at a significantly reduced cost.




Rosenberg & Jennings         Informational                      [Page 5]

RFC 5039                        SIP Spam                    January 2008


   International spam is likely to be even more annoying than national
   spam, since it may arrive in languages that the recipient doesn't
   even speak.

   These figures assume that the primary limitation is the access
   bandwidth and not CPU, disk, or termination costs.  Termination costs
   merit further discussion.  Currently, most Voice over IP (VoIP) calls
   terminate on the Public Switched Telephone Network (PSTN), and this
   termination costs the originator of the call money.  These costs are
   similar to the per-minute rates of a T1.  It ranges anywhere from
   half a cent to three cents per minute, depending on volume and other
   factors.  However, equipment costs, training, and other factors are
   much lower for SIP-based termination than a T1, making the cost still
   lower than circuit connectivity.  Furthermore, the current trend in
   VoIP systems is to make termination free for calls that never touch
   the PSTN, that is, calls to actual SIP endpoints.  Thus, as more and
   more SIP endpoints come online, termination costs will probably drop.
   Until then, SIP spam can be used in concert with termination services
   for a lower-cost form of traditional telemarketer calls, made to
   normal PSTN endpoints.

   It is useful to compare these figures with email.  VoIP can deliver
   approximately 3.1 successful call attempts per second.  Email spam
   can, of course, deliver more.  Assuming 1 KB per email, and an
   upstream link of 500 Kbps, a spammer can generate 62.5 messages per
   second.  This number goes down with larger messages of course.
   Interestingly, spam filters delete large numbers of these mails, so
   the cost per viewed message is likely to be much higher.  In that
   sense, call spam is much more attractive, since its content is much
   more likely to be examined by a user if a call attempt is successful.

   Another part of the cost of spamming is collecting addresses.
   Spammers have, over time, built up immense lists of email addresses,
   each of the form user@domain, to which spam is directed.  SIP uses
   the same form of addressing, making it likely that email addresses
   can easily be turned into valid SIP addresses.  Telephone numbers
   also represent valid SIP addresses; in concert with a termination
   provider, a spammer can direct SIP calls at traditional PSTN devices.
   It is not clear whether email spammers have also been collecting
   phone numbers as they perform their Web sweeps, but it is probably
   not hard to do so.  Furthermore, unlike email addresses, phone
   numbers are a finite address space and one that is fairly densely
   packed.  As a result, going sequentially through phone numbers is
   likely to produce a fairly high hit rate.  Thus, it seems like the
   cost is relatively low for a spammer to obtain large numbers of SIP
   addresses to which spam can be directed.





Rosenberg & Jennings         Informational                      [Page 6]

RFC 5039                        SIP Spam                    January 2008


2.2.  IM Spam



   IM spam is very much like email, in terms of the costs for deploying
   and generating spam.  Assuming, for the sake of argument, a 1KB
   message to be sent and 500 Kbps of upstream bandwidth, that is 62.5
   messages per second.  At $50/month, the result is .31 microcents per
   message.  This is less than voice spam, but not substantially less.
   The cost is probably on par with email spam.  However, IM is much
   more intrusive than email.  In today's systems, IMs automatically pop
   up and present themselves to the user.  Email, of course, must be
   deliberately selected and displayed.  However, most popular IM
   systems employ white lists, which only allow IM to be delivered if
   the sender is on the white list.  Thus, whether or not IM spam will
   be useful seems to depend a lot on the nature of the systems as the
   network is opened up.  If they are ubiquitously deployed with white-
   list access, the value of IM spam is likely to be low.

   It is important to point out that there are two different types of IM
   systems: page mode and session mode.  Page mode IM systems work much
   like email, with each IM being sent as a separate message.  In
   session mode IM, there is signaling in advance of communication to
   establish a session, and then IMs are exchanged, perhaps point-to-
   point, as part of the session.  The modality impacts the types of
   spam techniques that can be applied.  Techniques for email can be
   applied identically to page mode IM, but session mode IM is more like
   telephony, and many techniques (such as content filtering) are harder
   to apply.

2.3.  Presence Spam



   As defined above, presence spam is the generation of bulk unsolicited
   SUBSCRIBE messages.  The cost of this is within a small constant
   factor of IM spam so the same cost estimates can be used here.  What
   would be the effect of such spam?  Most presence systems provide some
   kind of consent framework.  A watcher that has not been granted
   permission to see the user's presence will not gain access to their
   presence.  However, the presence request is usually noted and
   conveyed to the user, allowing them to approve or deny the request.
   In SIP, this is done using the watcherinfo event package [7].  This
   package allows a user to learn the identity of the watcher, in order
   to make an authorization decision.

   Interestingly, this provides a vehicle for conveying information to a
   user.  By generating SUBSCRIBE requests from identities such as
   sip:please-buy-my-product@spam.example.com, brief messages can be
   conveyed to the user, even though the sender does not have, and never
   will receive, permission to access presence.  As such, presence spam
   can be viewed as a form of IM spam, where the amount of content to be



Rosenberg & Jennings         Informational                      [Page 7]

RFC 5039                        SIP Spam                    January 2008


   conveyed is limited.  The limit is equal to the amount of information
   generated by the watcher that gets conveyed to the user through the
   permission system.

   This type of spam also shows up in consent frameworks used to prevent
   call spam, as discussed in Section 3.4.

3.  Solution Space



   In this section, we consider the various solutions that might be
   possible to deal with SIP spam.  We primarily consider techniques
   that have been employed to deal with email spam.  It is important to
   note that the solutions documented below are not meant to be an
   exhaustive study of the spam solutions used for email but rather just
   a representative set.  We also consider some solutions that appear to
   be SIP-specific.

3.1.  Content Filtering



   The most common form of spam protection used in email is based on
   content filtering.  Spam filters analyze the content of the email,
   and look for clues that the email is spam.  Bayesian spam filters are
   in this category.

   Unfortunately, this type of spam filtering, while successful for
   email spam, is completely useless for call spam.  There are two
   reasons.  First, in the case where the user answers the call, the
   call is already established and the user is paying attention before
   the content is delivered.  The spam cannot be analyzed before the
   user sees it.  Second, if the content is stored before the user
   accesses it (e.g., with voicemail), the content will be in the form
   of recorded audio or video.  Speech and video recognition technology
   is not likely to be good enough to analyze the content and determine
   whether or not it is spam.  Indeed, if a system tried to perform
   speech recognition on a recording in order to perform such an
   analysis, it would be easy for the spammers to make calls with
   background noises, poor grammar, and varied accents, all of which
   will throw off recognition systems.  Video recognition is even harder
   to do and remains primarily an area of research.

   IM spam, due to its similarity to email, can be countered with
   content analysis tools.  Indeed, the same tools and techniques used
   for email will directly work for IM spam.








Rosenberg & Jennings         Informational                      [Page 8]

RFC 5039                        SIP Spam                    January 2008


3.2.  Black Lists



   Black listing is an approach whereby the spam filter maintains a list
   of addresses that identify spammers.  These addresses include both
   usernames (spammer@example.com) and entire domains (example.com).
   Pure blacklists are not very effective in email for two reasons.
   First, email addresses are easy to spoof, making it easy for the
   sender to pretend to be someone else.  If the sender varies the
   addresses they send from, the black list becomes almost completely
   useless.  The second problem is that, even if the sender doesn't
   forge the From address, email addresses are in almost limitless
   supply.  Each domain contains an infinite supply of email addresses,
   and new domains can be obtained for very low cost.  Furthermore,
   there will always be public providers that will allow users to obtain
   identities for almost no cost (for example, Yahoo or AOL mail
   accounts).  The entire domain cannot be blacklisted because it
   contains so many valid users.  Blacklisting needs to be for
   individual users.  Those identities are easily changed.

   As a result, as long as identities are easy to manufacture, or
   zombies are used, black lists will have limited effectiveness for
   email.

   Blacklists are also likely to be ineffective for SIP spam.
   Mechanisms for inter-domain authenticated identity for email and SIP
   are discussed in Section 4 and Section 5.  Assuming these mechanisms
   are used and enabled in inter-domain communications, it becomes
   difficult to forge sender addresses.  However, it still remains cheap
   to obtain a nearly infinite supply of addresses.

3.3.  White Lists



   White lists are the opposite of black lists.  It is a list of valid
   senders that a user is willing to accept email from.  Unlike black
   lists, a spammer cannot change identities to get around the white
   list.  White lists are susceptible to address spoofing, but a strong
   identity authentication mechanism can prevent that problem.  As a
   result, the combination of white lists and strong identity, as
   described in Section 4.2 and Section 5, are a good form of defense
   against spam.

   However, they are not a complete solution, since they would prohibit
   a user from ever being able to receive email from someone who was not
   explicitly put on the white list.  As a result, white lists require a
   solution to the "introduction problem" - how to meet someone for the
   first time, and decide whether they should be placed in the white
   list.  In addition to the introduction problem, white lists demand
   time from the user to manage.



Rosenberg & Jennings         Informational                      [Page 9]

RFC 5039                        SIP Spam                    January 2008


   In IM systems, white lists have proven exceptionally useful at
   preventing spam.  This is due, in no small part, to the fact that the
   white list exists naturally in the form of the buddy list.  Users
   don't have to manage this list just for the purposes of spam
   prevention; it provides general utility, and assists in spam
   prevention for free.  Many popular IM systems also have strong
   identity mechanisms since they do not allow communications with IM
   systems in other administrative domains.  The introduction problem in
   these systems is solved with a consent framework, described below.

   The success of white lists in IM systems has applicability to SIP as
   well.  This is because SIP also provides a buddy list concept and has
   an advanced presence system as part of its specifications.  The
   introduction problem remains.  In email, techniques like Turing tests
   have been employed to address the introduction problem.  Turing tests
   are considered further in the sections below.  As with email, a
   technique for solving the introduction problem would need to be
   applied in conjunction with a white list.

   If a user's computer is compromised and used a zombie, that computer
   can usually be used to send spam to anyone that has put the user on
   their white list.

3.4.  Consent-Based Communications



   A consent-based solution is used in conjunction with white or black
   lists.  That is, if user A is not on user B's white or black list,
   and user A attempts to communicate with user B, user A's attempt is
   initially rejected, and they are told that consent is being
   requested.  Next time user B connects, user B is informed that user A
   had attempted communications.  User B can then authorize or reject
   user A.

   These kinds of consent-based systems are used widely in presence and
   IM.  Since most of today's popular IM systems only allow
   communications within a single administrative domain, sender
   identities can be authenticated.  Email often uses similar consent-
   based systems for mailing lists.  They use a form of authentication
   based on sending cookies to an email address to verify that a user
   can receive mail at that address.

   This kind of consent-based communications has been standardized in
   SIP for presence, using the watcher information event package [7] and
   data format [8], which allow a user to find out that someone has
   subscribed.  Then, the XML Configuration Access Protocol (XCAP) [10]
   is used, along with the XML format for presence authorization [11] to
   provide permission for the user to communicate.




Rosenberg & Jennings         Informational                     [Page 10]

RFC 5039                        SIP Spam                    January 2008


   A consent framework has also been developed that is applicable to
   other forms of SIP communications [12].  However, this framework
   focuses on authorizing the addition of users to "mailing lists",
   known as exploders in SIP terminology.  Though spammers typically use
   such exploder functions, presumably one run by a spammer would not
   use this technique.  Consequently, this consent framework is not
   directly applicable to the spam problem.  It is, however, useful as a
   tool for managing a white list.  Through the PUBLISH mechanism, it
   allows a user to upload a permission document [13] that indicates
   that they will only accept incoming calls from a particular sender.

   Can a consent framework, like the ones used for presence, help solve
   call spam?  At first glance, it would seem to help a lot.  However,
   it might just change the nature of the spam.  Instead of being
   bothered with content, in the form of call spam or IM spam, users are
   bothered with consent requests.  A user's "communications inbox"
   might instead be filled with requests for communications from a
   multiplicity of users.  Those requests for communications don't
   convey much useful content to the user, but they can convey some.  At
   the very least, they will convey the identity of the requester.  The
   user part of the SIP URI allows for limited free form text, and thus
   could be used to convey brief messages.  One can imagine receiving
   consent requests with identities like
   "sip:please-buy-my-product-at-this-website@spam.example.com", for
   example.  Fortunately, it is possible to apply traditional content
   filtering systems to the header fields in the SIP messages, thus
   reducing these kinds of consent request attacks.

   In order for the spammer to convey more extensive content to the
   user, the user must explicitly accept the request, and only then can
   the spammer convey the full content.  This is unlike email spam,
   where, even though much spam is automatically deleted, some
   percentage of the content does get through, and is seen by users,
   without their explicit consent that they want to see it.  Thus, if
   consent is required first, the value in sending spam is reduced, and
   perhaps it will cease for those spam cases where consent is not given
   to spammers.

   As such, the real question is whether or not the consent system would
   make it possible for a user to give consent to non-spammers and
   reject spammers.  Authenticated identity can help.  A user in an
   enterprise would know to give consent to senders in other enterprises
   in the same industry, for example.  However, in the consumer space,
   if sip:bob@example.com tries to communicate with a user, how does
   that user determine whether Bob is a spammer or a long-lost friend
   from high school?  There is no way based on the identity alone.  In
   such a case, a useful technique is to grant permission for Bob to
   communicate but to ensure that the permission is extremely limited.



Rosenberg & Jennings         Informational                     [Page 11]

RFC 5039                        SIP Spam                    January 2008


   In particular, Bob may be granted permission to send no more than 200
   words of text in a single IM, which he can use to identify himself,
   so that the user can determine whether or not more permissions are
   appropriate.  It may even be possible that an automated system could
   do some form of content analysis on this initial short message.
   However, this 200 words of text may be enough for a spammer to convey
   their message, in much the same way they might convey it in the user
   part of the SIP URI.

   Thus, it seems that a consent-based framework, along with white lists
   and black lists, cannot fully solve the problem for SIP, although it
   does appear to help.

3.5.  Reputation Systems



   A reputation system is also used in conjunction with white or black
   lists.  Assume that user A is not on user B's white list, and A
   attempts to contact user B.  If a consent-based system is used, B is
   prompted to consent to communications from A, and along with the
   consent, a reputation score might be displayed in order to help B
   decide whether or not they should accept communications from A.

   Traditionally, reputation systems are implemented in highly
   centralized messaging architectures; the most widespread reputation
   systems in messaging today have been deployed by monolithic instant
   messaging providers (though many Web sites with a high degree of
   interactivity employ very similar concepts of reputation).
   Reputation is calculated based on user feedback.  For example, a
   button on the user interface of the messaging client might empower
   users to inform the system that a particular user is abusive.  Of
   course, the input of any single user has to be insufficient to ruin
   one's reputation, but consistent negative feedback would give the
   abusive user a negative reputation score.

   Reputation systems have been successful in systems where
   centralization of resources (user identities, authentication, etc.)
   and monolithic control dominate.  Examples of these include the large
   instant messaging providers that run IM systems that do not exchange
   messages with other administrative domains.  That control, first of
   all, provides a relatively strong identity assertion for users (since
   all users trust a common provider, and the common provider is the
   arbiter of authentication and identity).  Secondly, it provides a
   single place where reputation can be managed.

   Reputation systems based on negative reputation scores suffer from
   many of the same problems as black lists, since effectively the
   consequence of having a negative reputation is that you are
   blacklisted.  If identities are very easy to acquire, a user with a



Rosenberg & Jennings         Informational                     [Page 12]

RFC 5039                        SIP Spam                    January 2008


   negative reputation will simply acquire a new identity.  Moreover,
   negative reputation is generated by tattling, which requires users to
   be annoyed enough to click the warning button -- a process that can
   be abused.  In some reputation systems, "reputation mafias"
   consisting of large numbers of users routinely bully or extort
   victims by threatening collectively to give victims a negative
   reputation.

   Reputation systems based on positive reputation, where users praise
   each other for being good, rather than tattling on each other for
   being bad, have some similar drawbacks.  Collectives of spammers, or
   just one spammer who acquires a large number identities, could praise
   one another in order to create an artificial positive reputation.
   Users similarly have to overcome the inertia required to press the
   "praise" button.  Unlike negative reputation systems, however,
   positive reputation is not circumvented when users acquire a new
   identity, since basing authorization decisions on positive reputation
   is essentially a form of white listing.

   So, while positive reputation systems are superior to negative
   reputation systems, they are far from perfect.  Intriguingly, though,
   combining presence-based systems with reputation systems leads to an
   interesting fusion.  The "buddy-list" concept of presence is, in
   effect, a white list - and one can infer that the users on one's
   buddy list are people whom you are "praising".  This eliminates the
   problem of user inertia in the use of the "praise" button, and
   automates the initial establishment of reputation.

   And of course, your buddies in turn have buddies.  Collectively, you
   and your buddies (and their buddies, and so on) constitute a social
   network of reputation.  If there were a way to leverage this social
   network, it would eliminate the need for centralization of the
   reputation system.  Your perception of a particular user's reputation
   might be dependent on your relationship to them in the social
   network: are they one buddy removed (strong reputation), four buddies
   removed (weaker reputation), three buddies removed but connected to
   you through several of your buddies, etc.  This web of trust
   furthermore would have the very desirable property that circles of
   spammers adding one another to their own buddy lists would not affect
   your perception of their reputation unless their circle linked to
   your own social network.

   If a users machine is compromised and turned into a zombie, this
   allows SPAM to be sent and may impact their reputation in a negative
   way.  Once their reputation decreases, it becomes extremely difficult
   to reestablish a positive reputation.





Rosenberg & Jennings         Informational                     [Page 13]

RFC 5039                        SIP Spam                    January 2008


3.6.  Address Obfuscation



   Spammers build up their spam lists by gathering email addresses from
   Web sites and other public sources of information.  One way to
   minimize spam is to make your address difficult or impossible to
   gather.  Spam bots typically look for text in pages of the form
   "user@domain", and assume that anything of that form is an email
   address.  To hide from such spam bots, many Web sites have recently
   begun placing email addresses in an obfuscated form, usable to humans
   but difficult for an automata to read as an email address.  Examples
   include forms such as, "user at example dot com" or "j d r o s e n a
   t e x a m p l e d o t c o m".

   These techniques are equally applicable to prevention of SIP spam,
   and are likely to be as equally effective or ineffective in its
   prevention.

   It is worth mentioning that the source of addresses need not be a Web
   site - any publicly accessible service containing addresses will
   suffice.  As a result, ENUM [9] has been cited as a potential gold
   mine for spammers.  It would allow a spammer to collect SIP and other
   URIs by traversing the tree in e164.arpa and mining it for data.
   This problem is mitigated in part if only number prefixes, as opposed
   to actual numbers, appear in the DNS.  Even in that case, however, it
   provides a technique for a spammer to learn which phone numbers are
   reachable through cheaper direct SIP connectivity.

3.7.  Limited-Use Addresses



   A related technique to address obfuscation is limited-use addresses.
   In this technique, a user has a large number of email addresses at
   their disposal, each of which has constraints on its applicability.
   A limited-use address can be time-bound, so that it expires after a
   fixed period.  Or, a different email address can be given to each
   correspondent.  When spam arrives from that correspondent, the
   limited-use address they were given is terminated.  In another
   variation, the same limited-use address is given to multiple users
   that share some property; for example, all work colleagues, all
   coworkers from different companies, all retailers, and so on.  Should
   spam begin arriving on one of the addresses, it is invalidated,
   preventing communications from anyone else that received the limited
   use address.

   This technique is equally applicable to SIP.  One of the drawbacks of
   the approach is that it can make it hard for people to reach you; if
   an email address you hand out to a friend becomes spammed, changing
   it requires you to inform your friend of the new address.  SIP can
   help solve this problem in part, by making use of presence [6].



Rosenberg & Jennings         Informational                     [Page 14]

RFC 5039                        SIP Spam                    January 2008


   Instead of handing out your email address to your friends, you would
   hand out your presence URI.  When a friend wants to send you an
   email, they subscribe to your presence (indeed, they are likely to be
   continuously subscribed from a buddy list application).  The presence
   data can include an email address where you can be reached.  This
   email address can be obfuscated and be of single use, different for
   each buddy who requests your presence.  They can also be constantly
   changed, as these changes are pushed directly to your buddies.  In a
   sense, the buddy list represents an automatically updated address
   book, and would therefore eliminate the problem.

   Another approach is to give a different address to each and every
   correspondent, so that it is never necessary to tell a "good" user
   that an address needs to be changed.  This is an extreme form of
   limited-use addresses, which can be called a single-use address.
   Mechanisms are available in SIP for the generation of [16] an
   infinite supply of single use addresses.  However, the hard part
   remains a useful mechanism for distribution and management of those
   addresses.

3.8.  Turing Tests



   In email, Turing tests are mechanisms whereby the sender of the
   message is given some kind of puzzle or challenge, which only a human
   can answer (since Turing tests rely on video or audio puzzles, they
   sometimes cannot be solved by individuals with handicaps).  These
   tests are also known as captchas (Completely Automated Public Turing
   test to tell Computers and Humans Apart).  If the puzzle is answered
   correctly, the sender is placed on the user's white list.  These
   puzzles frequently take the form of recognizing a word or sequence of
   numbers in an image with a lot of background noise.  The tests need
   to be designed such that automata cannot easily perform the image
   recognition needed to extract the word or number sequence, but a
   human user usually can.  Designing such tests is not easy, since
   ongoing advances in image processing and artificial intelligence
   continually raise the bar.  Consequently, the effectiveness of
   captchas are tied to whether spammers can come up with or obtain
   algorithms for automatically solving them.

   Like many of the other email techniques, Turing tests are dependent
   on sender identity, which cannot easily be authenticated in email.

   Turing tests can be used to prevent IM spam in much the same way they
   can be used to prevent email spam.

   Turing tests can be applied to call spam as well, although not
   directly, because call spam does not usually involve the transfer of
   images and other content that can be used to verify that a human is



Rosenberg & Jennings         Informational                     [Page 15]

RFC 5039                        SIP Spam                    January 2008


   on the other end.  If most of the calls are voice, the technique
   needs to be adapted to voice.  This is not that difficult to do.
   Here is how it could be done.  User A calls user B and is not on user
   B's white or black list.  User A is transferred to an Interactive
   Voice Response (IVR) system.  The IVR system tells the user that they
   are going to hear a series of numbers (say 5 of them), and that they
   have to enter those numbers on the keypad.  The IVR system reads out
   the numbers while background music is playing, making it difficult
   for an automated speech recognition system to be applied to the
   media.  The user then enters the numbers on their keypad.  If they
   are entered correctly, the user is added to the white list.

   This kind of voice-based Turing test is easily extended to a variety
   of media, such as video and text, and user interfaces by making use
   of the SIP application interaction framework [14].  This framework
   allows client devices to interact with applications in the network,
   where such interaction is done with stimulus signaling, including
   keypads (supported with the Keypad Markup Language [15]), but also
   including Web browsers, voice recognition, and so on.  The framework
   allows the application to determine the media capabilities of the
   device (or user, in cases where they are handicapped) and interact
   with them appropriately.

   In the case of voice, the Turing test would need to be made to run in
   the language of the caller.  This is possible in SIP, using the
   Accept-Language header field, though this is not widely used at the
   moment, and meant for languages of SIP message components, not the
   media streams.

   The primary problem with the voice Turing test is the same one that
   email tests have: instead of having an automata process the test, a
   spammer can pay cheap workers to take the tests.  Assuming cheap
   labor in a poor country can be obtained for about 60 cents per hour,
   and assuming a Turing test of a 30-second duration, this is about
   0.50 cents per test and thus 0.50 cents per message to send an IM
   spam.  Lower labor rates would reduce this further; the number quoted
   here is based on real online bids in September of 2006 made for
   actual work of this type.

   As an alternative to paying cheap workers to take the tests, the
   tests can be taken by human users that are tricked into completing
   the tests in order to gain access to what they believe is a
   legitimate resource.  This was done by a spambot that posted the
   tests on a pornography site, and required users to complete the tests
   in order to gain access to content.

   Due to these limitations, Turing tests may never completely solve the
   problem.



Rosenberg & Jennings         Informational                     [Page 16]

RFC 5039                        SIP Spam                    January 2008


3.9.  Computational Puzzles



   This technique is similar to Turing tests.  When user A tries to
   communicate with user B, user B asks user A to perform a computation
   and pass the result back.  This computation has to be something a
   human user cannot perform and something expensive enough to increase
   user A's cost to communicate.  This cost increase has to be high
   enough to make it prohibitively expensive for spammers but
   inconsequential for legitimate users.

   One of the problems with the technique is that there is wide
   variation in the computational power of the various clients that
   might legitimately communicate.  The CPU speed on a low-end cell
   phone is around 50 MHz, while a high-end PC approaches 5 GHz.  This
   represents almost two orders of magnitude difference.  Thus, if the
   test is designed to be reasonable for a cell phone to perform, it is
   two orders of magnitude cheaper to perform for a spammer on a high-
   end machine.  Recent research has focused on defining computational
   puzzles that challenge the CPU/memory bandwidth, as opposed to just
   the CPU [26].  It seems that there is less variety in the CPU/memory
   bandwidth across devices, roughly a single order of magnitude.

   Recent work [28] suggests that, due to the ability of spammers to use
   virus-infected machines (also known as zombies) to generate the spam,
   the amount of computational power available to the spammers is
   substantial, and it may be impossible to have them compute a puzzle
   that is sufficiently hard that will not also block normal emails.  If
   combined with white listing, computational puzzles would only be
   utilized for new communications partners.  Of course, if the partner
   on the white list is a zombie, spam will come from that source.  The
   frequency of communications with new partners is arguably higher for
   email than for multimedia, and thus the computational puzzle
   techniques may be more effective for SIP than for email in dealing
   with the introduction problem.

   These techniques are an active area of research right now, and any
   results for email are likely to be usable for SIP.

3.10.  Payments at Risk



   This approach has been proposed for email [27].  When user A sends
   email to user B, user A deposits a small amount of money (say, one
   dollar) into user B's account.  If user B decides that the message is
   not spam, user B refunds this money back to user A.  If the message
   is spam, user B keeps the money.  This technique requires two
   transactions to complete: a transfer from A to B, and a transfer from
   B back to A. The first transfer has to occur before the message can
   be received in order to avoid reuse of "pending payments" across



Rosenberg & Jennings         Informational                     [Page 17]

RFC 5039                        SIP Spam                    January 2008


   several messages, which would eliminate the utility of the solution.
   The second one then needs to occur when the message is found not to
   be spam.

   This technique appears just as applicable to call spam and IM spam as
   it is to email spam.  Like many of the other techniques, this
   exchange would only happen the first time you talk to people.  Its
   proper operation therefore requires a good authenticated identity
   infrastructure.

   This technique has the potential to make it arbitrarily expensive to
   send spam of any sort.  However, it relies on cheap micro-payment
   techniques on the Internet.  Traditional costs for Internet payments
   are around 25 cents per transaction, which would probably be
   prohibitive.  However, recent providers have been willing to charge
   15% of the transaction for small transactions, as small as one cent.
   This cost would have to be shouldered by users of the system.  The
   cost that would need to be shouldered per user is equal to the number
   of messages from unknown senders (that is, senders not on the white
   list) that are received.  For a busy user, assume about 10 new
   senders per day.  If the deposit is 5 cents, the transaction provider
   would take 0.75 cents and deliver 4.25 cents.  If the sender is
   allowed, the recipient returns 4.25 cents, the provider takes 0.64
   cents, and returns 3.6 cents.  This costs the sender 0.65 cents on
   each transaction, if it was legitimate.  If there are ten new
   recipients per day, that is US $1.95 per month, which is relatively
   inexpensive.

   Assuming a micro-payment infrastructure exists, another problem with
   payment-at-risk is that it loses effectiveness when there are strong
   inequities in the value of currency between sender and recipient.
   For example, a poor person in a Third World country might keep the
   money in each mail message, regardless of whether it is spam.
   Similarly, a poor person might not be willing to include money in an
   email, even if legitimate, for fear that the recipient might keep it.
   If the amount of money is lowered to help handle these problems, it
   might become sufficiently small that spammers can just afford to
   spend it.

3.11.  Legal Action



   In this solution, countries pass laws that prohibit spam.  These laws
   could apply to IM or call spam just as easily as they could apply to
   email spam.  There is a lot of debate about whether these laws would
   really be effective in preventing spam.

   As a recent example in the US, "do not call" lists seem to be
   effective.  However, due to the current cost of long-distance phone



Rosenberg & Jennings         Informational                     [Page 18]

RFC 5039                        SIP Spam                    January 2008


   calls, the telemarketing is coming from companies within the US.  As
   such, calls from such telemarketers can be traced.  If a telemarketer
   violates the "do not call" list, the trace allows legal action to be
   taken against them.  A similar "do not irritate" list for VoIP or for
   email would be less likely to work because the spam is likely to come
   from international sources.  This problem could be obviated if there
   was a strong way to identify the sender's legal entity, and then
   determine whether it was in a jurisdiction where it was practical to
   take legal action against them.  If the spammer is not in such a
   jurisdiction, the SIP spam could be rejected.

   There are also schemes that cause laws other than anti-spam laws to
   be broken if spam is sent.  This does not inherently reduce SPAM, but
   it allows more legal options to be brought to bear against the
   spammer.  For example, Habeas <http://www.habeas.com> inserts
   material in the header that, if it was inserted by a spammer without
   an appropriate license, would allegedly causes the spammer to violate
   US copyright and trademark laws, possibly reciprocal laws, and
   similar laws in many countries.

3.12.  Circles of Trust



   In this model, a group of domains (e.g., a set of enterprises) all
   get together.  They agree to exchange SIP calls amongst each other,
   and they also agree to introduce a fine should any one of them be
   caught spamming.  Each company would then enact measures to terminate
   employees who spam from their accounts.

   This technique relies on secure inter-domain authentication - that
   is, domain B can know that messages are received from domain A.  In
   SIP, this is readily provided by usage of the mutually authenticated
   Transport Level Security (TLS)[22] between providers or SIP Identity
   [17].

   This kind of technique works well for small domains or small sets of
   providers, where these policies can be easily enforced.  However, it
   is unclear how well it scales up.  Could a very large domain truly
   prevent its users from spamming?  At what point would the network be
   large enough that it would be worthwhile to send spam and just pay
   the fine?  How would the pricing be structured to allow both small
   and large domains alike to participate?

3.13.  Centralized SIP Providers



   This technique is a variation on the circles of trust described in
   Section 3.12.  A small number of providers get established as "inter-
   domain SIP providers".  These providers act as a SIP-equivalent to
   the interexchange carriers in the PSTN.  Every enterprise, consumer



Rosenberg & Jennings         Informational                     [Page 19]

RFC 5039                        SIP Spam                    January 2008


   SIP provider, or other SIP network (call these the local SIP
   providers) connects to one of these inter-domain providers.  The
   local SIP providers only accept SIP messages from their chosen inter-
   domain provider.  The inter-domain provider charges the local
   provider, per SIP message, for the delivery of SIP messages to other
   local providers.  The local provider can choose to pass on this cost
   to its own customers if it so chooses.

   The inter-domain SIP providers then form bi-lateral agreements with
   each other, exchanging SIP messages according to strict contracts.
   These contracts require that each of the inter-domain providers be
   responsible for charging a minimum per-message fee to their own
   customers.  Extensive auditing procedures can be put into place to
   verify this.  Besides such contracts, there may or may not be a flow
   of funds between the inter-domain providers.

   The result of such a system is that a fixed cost can be associated
   with sending a SIP message, and that this cost does not require
   micro-payments to be exchanged between local providers, as it does in
   Section 3.10.  Since all of the relationships are pre-established and
   negotiated, cheaper techniques for monetary transactions (such as
   monthly post-paid transactions) can be used.

   This technique can be made to work in SIP, whereas it cannot in
   email, because inter-domain SIP connectivity has not yet been broadly
   established.  In email, there already exists a no-cost form of inter-
   domain connectivity that cannot be eliminated without destroying the
   utility of email.  If, however, SIP inter-domain communications get
   established from the start using this structure, there is a path to
   deployment.

   This structure is more or less the same as the one in place for the
   PSTN today, and since there is relatively little spam on the PSTN
   (compared to email!), there is some proof that this kind of
   arrangement can work.  However, centralized architectures as these
   are deliberately eschewed because they put back into SIP much of the
   complexity and monopolistic structures that the protocol aims to
   eliminate.

4.  Authenticated Identity in Email



   Though not a form of anti-spam in and of itself, authenticated or
   verifiable identities are a key part of making other anti-spam
   mechanisms work.  Many of the techniques described above are most
   effective when combined with a white or black list, which itself
   requires a strong form of identity.





Rosenberg & Jennings         Informational                     [Page 20]

RFC 5039                        SIP Spam                    January 2008


   In email, two types of authenticated identity have been developed -
   sender checks and signature-based solutions.

4.1.  Sender Checks



   In email, DNS resource records have been defined that will allow a
   domain that receives a message to verify that the sender is a valid
   Message Transfer Agent (MTA) for the sending domain [18] [19] [20]
   [21].  They don't prevent spam by themselves, but may help in
   preventing spoofed emails.  As has been mentioned several times, a
   form of strong authenticated identity is key in making many other
   anti-spam techniques work.

   Are these techniques useful for SIP?  They can be used for SIP but
   are not necessary.  In SIP, TLS with mutual authentication can be
   used inter-domain.  A provider receiving a message can then reject
   any message coming from a domain that does not match the asserted
   identity of the sender of the message.  Such a policy only works in
   the "trapezoid" model of SIP, whereby there are only two domains in
   any call - the sending domain, which is where the originator resides,
   and the receiving domain.  These techniques are discussed in Section
   26.3.2.2 of RFC 3261 [2].  In forwarding situations, the assumption
   no longer holds and these techniques no longer work.  However, the
   authenticated identity mechanism for SIP, discussed in Section 5,
   does work in more complex network configurations and provides fairly
   strong assertion of identity.

4.2.  Signature-Based Techniques



   Domain Keys Identified Mail (DKIM) Signatures [23] (and several non-
   standard techniques that preceded it) provide strong identity
   assertions by allowing the sending domain to sign an email, and then
   providing mechanisms by which the receiving MTA or Mail User Agent
   (MUA) can validate the signature.

   Unfortunately, when used with blacklists, this kind of authenticated
   identity is only as useful as the fraction of the emails that utilize
   it.  This is partly true for white lists as well; if any
   unauthenticated email is accepted for an address on a white list, a
   spammer can spoof that address.  However, a white list can be
   effective with limited deployment of DKIM if all the people on the
   white list are those whose domains are utilizing the mechanism, and
   the users on that white list aren't zombies.

   This kind of identity mechanism is also applicable to SIP, and is in
   fact, exactly what is defined by SIP's authenticated identity
   mechanism [17].




Rosenberg & Jennings         Informational                     [Page 21]

RFC 5039                        SIP Spam                    January 2008


   Other signature-based approaches for email include S/MIME[24] and
   OpenPGP[25].

5.  Authenticated Identity in SIP



   One of the key parts of many of the solutions described above is the
   ability to securely identify the sender of a SIP message.  SIP
   provides a secure solution for this problem, called SIP Identity
   [17], and it is important to discuss it here.

   The solution starts by having each domain authenticate its own users.
   SIP provides HTTP digest authentication as part of the core SIP
   specification, and all clients and servers are required to support
   it.  Indeed, digest is widely deployed for SIP.  However, digest
   alone has many known vulnerabilities, most notably offline dictionary
   attacks.  These vulnerabilities are all resolved by having each
   client maintain a persistent TLS connection to the server.  The
   client verifies the server identity using TLS, and then authenticates
   itself to the server using a digest exchange over TLS.  This
   technique, which is also documented in RFC 3261, is very secure but
   not widely deployed yet.  In the long term, this approach will be
   necessary for the security properties needed to prevent SIP spam.

   Once a domain has authenticated the identity of a user, when it
   relays a message from that user to another domain, the sending domain
   can assert the identity of the sender, and include a signature to
   validate that assertion.  This is done using the SIP identity
   mechanism [17].

   A weaker form of identity assertion is possible using the P-Asserted-
   Identity header field [5], but this technique requires mutual trust
   among all domains.  Unfortunately, this becomes exponentially harder
   to provide as the number of interconnected domains grows.  As that
   happens, the value of the identity assertion becomes equal to the
   trustworthiness of the least trustworthy domain.  Since spam is a
   consequence of the receiving domain not being able to trust the
   sending domains to disallow the hosts in the sending to send spam,
   the P-Asserted-Identity technique becomes ineffective at exactly the
   same levels of interconnectedness that introduce spam.

   Consider the following example to help illustrate this fact.  A
   malicious domain -- let us call them spam.example.com, would like to
   send SIP INVITE requests with false P-Asserted-Identity, indicating
   users outside of its own domain. spam.example.com finds a regional
   SIP provider in a small country who, due to its small size and
   disinterest in spam, accepts any P-Asserted-Identity from its
   customers without verification.  This provider, in turn, connects to
   a larger, interconnect provider.  They do ask each of their customers



Rosenberg & Jennings         Informational                     [Page 22]

RFC 5039                        SIP Spam                    January 2008


   to verify P-Asserted-Identity but have no easy way of enforcing it.
   This provider, in turn, connects to everyone else.  As a consequence,
   the spam.example.com domain is able to inject calls with a spoofed
   caller ID.  This request can be directed to any recipient reachable
   through the network (presumably everyone due to the large size of the
   root provider).  There is no way for a recipient to know that this
   particular P-Asserted-Identity came from this bad spam.example.com
   domain.  As the example shows, even though the central provider's
   policy is good, the overall effectiveness of P-Asserted-Identity is
   still only as good as the policies of the weakest link in the chain.

   SIP also defines the usage of TLS between domains, using mutual
   authentication, as part of the base specification.  This technique
   provides a way for one domain to securely determine that it is
   talking to a server that is a valid representative of another domain.

6.  Framework for Anti-Spam in SIP



   Unfortunately, there is no magic bullet for preventing SIP spam, just
   as there is none for email spam.  However, the combination of several
   techniques can provide a framework for dealing with spam in SIP.
   This section provides recommendations for network designers in order
   to help mitigate the risk of spam.

   There are four core recommendations that can be made:

   Strong Identity:  Firstly, in almost all of the solutions discussed
      above, there is a dependency on the ability to authenticate the
      sender of a SIP message inter-domain.  Consent, reputation
      systems, computational puzzles, and payments at risk, amongst
      others, all work best when applied only to new requests, and
      successful completion of an introduction results in the placement
      of a user on a white list.  However, usage of white lists depends
      on strong identity assertions.  Consequently, any network that
      interconnects with others should make use of strong SIP identity
      as described in RFC 4474.  P-Asserted-Identity is not strong
      enough.

   White Lists:  Secondly, with a strong identity system in place,
      networks are recommended to make use of white lists.  These are
      ideally built off existing buddy lists, if present.  If not,
      separate white lists can be managed for spam.  Placement on these
      lists can be manual or based on the successful completion of one
      or more introduction mechanisms.

   Solve the Introduction Problem:  This in turn leads to the final
      recommendation to be made.  Network designers should make use of
      one or more mechanisms meant to solve the introduction problem.



Rosenberg & Jennings         Informational                     [Page 23]

RFC 5039                        SIP Spam                    January 2008


      Indeed, it is possible to use more than one and combine the
      results through some kind of weight.  A user that successfully
      completes the introduction mechanism can be automatically added to
      the white list.  Of course, that can only be done usefully if
      their identity is verified by SIP Identity.  The set of mechanisms
      for solving the introduction problem, as described in this
      document, are based on some (but not all) of the techniques known
      and used at the time of writing.  Providers of SIP services should
      keep tabs on solutions in email as they evolve, and utilize the
      best of what those techniques have to offer.

   Don't Wait Until It's Too Late:  But perhaps most importantly,
      providers should not ignore the spam problem until it happens!  As
      soon as a provider inter-connects with other providers, or allows
      SIP messages from the open Internet, that provider must consider
      how they will deal with spam.

7.  Additional Work



   Though the above framework serves as a good foundation on which to
   deal with spam in SIP, there are gaps, some of which can be addressed
   by additional work that has yet to be undertaken.

   One of the difficulties with the strong identity techniques is that a
   receiver of a SIP request without an authenticated identity cannot
   know whether the request lacked such an identity because the
   originating domain didn't support it, or because a man-in-the-middle
   removed it.  As a result, transition mechanisms should be put in
   place to allow these to be differentiated.  Without it, the value of
   the identity mechanism is much reduced.

8.  Security Considerations



   This document is entirely devoted to issues relating to spam in SIP
   and references a variety of security mechanisms in support of that
   goal.

9.  Acknowledgements



   The authors would like to thank Rohan Mahy for providing information
   on Habeas, Baruch Sterman for providing costs on VoIP termination
   services, and Gonzalo Camarillo and Vijay Gurbani for their reviews.
   Useful comments and feedback were provided by Nils Ohlmeir, Tony
   Finch, Randy Gellens, Lisa Dusseault, Sam Hartman, Chris Newman, Tim
   Polk, Donald Eastlake, and Yakov Shafranovich.  Jon Peterson wrote
   some of the text in this document and has contributed to the work as
   it has moved along.




Rosenberg & Jennings         Informational                     [Page 24]

RFC 5039                        SIP Spam                    January 2008


10.  Informative References



   [1]   Campbell, B., Mahy, R., and C. Jennings, "The Message Session
         Relay Protocol (MSRP)", RFC 4975, September 2007.

   [2]   Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A.,
         Peterson, J., Sparks, R., Handley, M., and E. Schooler, "SIP:
         Session Initiation Protocol", RFC 3261, June 2002.

   [3]   Campbell, B., Rosenberg, J., Schulzrinne, H., Huitema, C., and
         D. Gurle, "Session Initiation Protocol (SIP) Extension for
         Instant Messaging", RFC 3428, December 2002.

   [4]   Roach, A., "Session Initiation Protocol (SIP)-Specific Event
         Notification", RFC 3265, June 2002.

   [5]   Jennings, C., Peterson, J., and M. Watson, "Private Extensions
         to the Session Initiation Protocol (SIP) for Asserted Identity
         within Trusted Networks", RFC 3325, November 2002.

   [6]   Rosenberg, J., "A Presence Event Package for the Session
         Initiation Protocol (SIP)", RFC 3856, August 2004.

   [7]   Rosenberg, J., "A Watcher Information Event Template-Package
         for the Session Initiation Protocol (SIP)", RFC 3857,
         August 2004.

   [8]   Rosenberg, J., "An Extensible Markup Language (XML) Based
         Format for Watcher Information", RFC 3858, August 2004.

   [9]   Faltstrom, P. and M. Mealling, "The E.164 to Uniform Resource
         Identifiers (URI) Dynamic Delegation Discovery System (DDDS)
         Application (ENUM)", RFC 3761, April 2004.

   [10]  Rosenberg, J., "The Extensible Markup Language (XML)
         Configuration Access Protocol (XCAP)", RFC 4825, May 2007.

   [11]  Rosenberg, J., "Presence Authorization Rules", RFC 5025,
         October 2007.

   [12]  Rosenberg, J., "A Framework for Consent-Based Communications in
         the Session Initiation  Protocol (SIP)", Work in Progress,
         October 2007.

   [13]  Camarillo, G., "A Document Format for Requesting Consent", Work
         in Progress, October 2007.





Rosenberg & Jennings         Informational                     [Page 25]

RFC 5039                        SIP Spam                    January 2008


   [14]  Rosenberg, J., "A Framework for Application Interaction in the
         Session Initiation Protocol  (SIP)", Work in Progress,
         October 2005.

   [15]  Burger, E. and M. Dolly, "A Session Initiation Protocol (SIP)
         Event Package for Key Press Stimulus (KPML)", RFC 4730,
         November 2006.

   [16]  Rosenberg, J., "Applying Loose Routing to Session Initiation
         Protocol (SIP) User Agents  (UA)", Work in Progress, June 2007.

   [17]  Peterson, J. and C. Jennings, "Enhancements for Authenticated
         Identity Management in the Session Initiation Protocol (SIP)",
         RFC 4474, August 2006.

   [18]  Allman, E. and H. Katz, "SMTP Service Extension for Indicating
         the Responsible Submitter of an E-Mail Message", RFC 4405,
         April 2006.

   [19]  Lyon, J. and M. Wong, "Sender ID: Authenticating E-Mail",
         RFC 4406, April 2006.

   [20]  Lyon, J., "Purported Responsible Address in E-Mail Messages",
         RFC 4407, April 2006.

   [21]  Wong, M. and W. Schlitt, "Sender Policy Framework (SPF) for
         Authorizing Use of Domains in E-Mail, Version 1", RFC 4408,
         April 2006.

   [22]  Dierks, T. and E. Rescorla, "The Transport Layer Security (TLS)
         Protocol Version 1.1", RFC 4346, April 2006.

   [23]  Allman, E., Callas, J., Delany, M., Libbey, M., Fenton, J., and
         M. Thomas, "DomainKeys Identified Mail (DKIM) Signatures",
         RFC 4871, May 2007.

   [24]  Ramsdell, B., "Secure/Multipurpose Internet Mail Extensions
         (S/MIME) Version 3.1 Message Specification", RFC 3851,
         July 2004.

   [25]  Elkins, M., Del Torto, D., Levien, R., and T. Roessler, "MIME
         Security with OpenPGP", RFC 3156, August 2001.

   [26]  Abadi, M., Burrows, M., Manasse, M., and T. Wobber, "Moderately
         Hard, Memory Bound Functions, NDSS 2003", February 2003.






Rosenberg & Jennings         Informational                     [Page 26]

RFC 5039                        SIP Spam                    January 2008


   [27]  Abadi, M., Burrows, M., Birrell, A., Dabek, F., and T. Wobber,
         "Bankable Postage for Network Services, Proceedings of the 8th
         Asian Computing Science Conference, Mumbai, India",
         December 2003.

   [28]  Clayton, R. and B. Laurie, "Proof of Work Proves not to Work,
         Third Annual Workshop on Economics and Information Security",
         May 2004.

Authors' Addresses



   Jonathan Rosenberg
   Cisco
   Edison, NJ
   US

   EMail: jdrosen@cisco.com
   URI:   http://www.jdrosen.net


   Cullen Jennings
   Cisco
   170 West Tasman Dr.
   San Jose, CA  95134
   US

   Phone: +1 408 421-9990
   EMail: fluffy@cisco.com























Rosenberg & Jennings         Informational                     [Page 27]

RFC 5039                        SIP Spam                    January 2008


Full Copyright Statement



   Copyright (C) The IETF Trust (2008).

   This document is subject to the rights, licenses and restrictions
   contained in BCP 78, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
   THE INTERNET ENGINEERING TASK FORCE DISCLAIM 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.

Intellectual Property



   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   http://www.ietf.org/ipr.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights that may cover technology that may be required to implement
   this standard.  Please address the information to the IETF at
   ietf-ipr@ietf.org.












Rosenberg & Jennings         Informational                     [Page 28]