TBD                                                       H. Schulzrinne
Internet-Draft                                               Columbia U.
Expires: April 17, 2005                                 October 17, 2004


   Requirements for Session Initiation Protocol (SIP)-based Emergency
                                 Calls
               draft-schulzrinne-sipping-emergency-req-01

Status of this Memo

   This document is an Internet-Draft and is subject to all provisions
   of section 3 of RFC 3667.  By submitting this Internet-Draft, each
   author represents that any applicable patent or other IPR claims of
   which he or she is aware have been or will be disclosed, and any of
   which he or she become aware will be disclosed, in accordance with
   RFC 3668.

   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 April 17, 2005.

Copyright Notice

   Copyright (C) The Internet Society (2004).

Abstract

   This document enumerates requirements for emergency calls in VoIP and
   general Internet multimedia systems.  We divide the requirements into
   "trunk replacement" and "end-to-end".  Trunking solutions only
   exchange the emergency call center's circuit-switched access by an
   IP-based system.  The requirements for end-to-end IP-based emergency
   calling address functional and security issues for determining the
   correct emergency address, for identifying the appropriate emergency



Schulzrinne              Expires April 17, 2005                 [Page 1]


Internet-Draft           Emergency requirements             October 2004


   call center and for identifying the caller and its location.  While
   we focus on systems that employ the Session Initiation Protocol
   (SIP), many of the requirements may also apply to other environments,
   such as those using H.248/Megaco, MGCP or H.323.

Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.  Terminology  . . . . . . . . . . . . . . . . . . . . . . . . .  4
   3.  Definitions  . . . . . . . . . . . . . . . . . . . . . . . . .  5
   4.  Trunk Replacement  . . . . . . . . . . . . . . . . . . . . . .  6
   5.  End-to-End IP-Based Emergency Calls  . . . . . . . . . . . . . 10
     5.1   Emergency Address  . . . . . . . . . . . . . . . . . . . . 10
     5.2   Identifying the Caller Location  . . . . . . . . . . . . . 11
     5.3   Identifying the Appropriate Emergency Call Center  . . . . 12
     5.4   Identifying the Caller . . . . . . . . . . . . . . . . . . 15
     5.5   Call Setup and Call Features . . . . . . . . . . . . . . . 15
   6.  Security Considerations  . . . . . . . . . . . . . . . . . . . 17
   7.  Acknowledgments  . . . . . . . . . . . . . . . . . . . . . . . 18
   8.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 19
   8.1   Normative References . . . . . . . . . . . . . . . . . . . . 19
   8.2   Informative References . . . . . . . . . . . . . . . . . . . 19
       Author's Address . . . . . . . . . . . . . . . . . . . . . . . 19
       Intellectual Property and Copyright Statements . . . . . . . . 20



























Schulzrinne              Expires April 17, 2005                 [Page 2]


Internet-Draft           Emergency requirements             October 2004


1.  Introduction

   Users of telephone-like services expect to be able to call for
   emergency help, such as police, the fire department or an ambulance,
   regardless of where they are, what (if any) service provider they are
   using and what kind of device they are using.  Unfortunately, the
   mechanisms for emergency calls that have evolved in the public
   circuit-switched telephone network (PSTN) are not quite appropriate
   for evolving IP-based voice and real-time multimedia communications.
   This document outlines some of the requirements that end systems and
   network elements such as SIP proxies need to satisfy in order to
   provide emergency call services that offer at least the same
   functionality as existing PSTN services, while hopefully making
   emergency calling more robust, cheaper to implement and
   multimedia-capable.

   In the future, users of other real-time and near real-time services
   may also expect to be able to summon emergency help.  For example,
   instant messaging (IM) users may want to use such services.  IM is
   particularly helpful for hearing-disabled users (RFC 3351 [2]) and in
   cases where bandwidth is scarce.  For lack of a better term, we will
   use the term "caller" or "emergency caller" to refer to the person
   placing an emergency call or sending an emergency IM.

   Emergency callers and ECCs expect calls to be completed reliably.
   Where possible, a callback number and the current caller location
   shouls be delivered to the ECC to speed up emergency response and to
   limit prank calls.

   The emergency calls described in this document differ from the
   emergency telecommunications service (ETS) described in XXX.  In ETS,
   relatively small numbers of emergency workers need to maintain
   communication even when parts of the infrastructure are destroyed or
   disabled.  Emergency calls, on the other hand, are placed by
   civilians to call for emergency services such as fire, ambulance and
   police services.  Thus, these two services are complementary.

   We distinguish two sets of requirements, one for ECC trunk
   replacement use of SIP (Section XXX), where VoIP emergency callers
   still use the existing PSTN, and end-to-end SIP emergency calls
   (Section XXX) that terminate SIP-originated emergency calls without
   transitioning the PSTN.

   There is a third approach, where SIP-originated calls terminate on a
   PSTN gateway in each emergency calling area.  This architecture is
   left for future consideration and discussed in other standardization
   organizations, such as NENA, as it is strongly dependent on the
   currently-deployed emergency services network architecture.



Schulzrinne              Expires April 17, 2005                 [Page 3]


Internet-Draft           Emergency requirements             October 2004


2.  Terminology

   In this document, the key words "MUST", "MUSTNOT", "REQUIRED",
   "SHALL", "SHALLNOT", "SHOULD", "SHOULDNOT", "RECOMMENDED", "MAY", and
   "OPTIONAL" are to be interpreted as described in RFC 2119 [1] and
   indicate requirement levels for compliant implementations.

   Since a requirements document does not directly specify an
   implementable protocols, these compliance labels should be read as
   indicating requirements for the protocol or architecture, rather than
   an implementation.








































Schulzrinne              Expires April 17, 2005                 [Page 4]


Internet-Draft           Emergency requirements             October 2004


3.  Definitions

   Emergency call center (ECC): An emergency call center (ECC) receives
      emergency calls within a specific geographic area and dispatches
      emergency services, such as fire, police and rescue services.  An
      ECC may also serve as a backup for another ECC and, in backup
      mode, dispatch emergency services outside of its normal service
      region.  In the United States and Canada, ECCs are called Public
      Safety Answering Points (PSAPs).
   Internet Protocol ECC (IECC): An Internet protocol emergency call
      center (IECC) is an ECC that uses Internet protocols, such as SIP
      for call signaling, RTP for media delivery, to receive emergency
      calls.
   Call taker: A call taker is an agent, typically a government
      employee, at the ECC that accepts calls and may dispatch emergency
      help.  (Sometimes the functions of call taking and dispatching are
      handled by different groups of people, but these divisions of
      labor are not generally visible to the outside and thus do not
      concern us here.)
   Basic emergency service: Basic emergency service allows a user to
      reach an ECC serving its current location, but the ECC may not be
      able to determine the identity or geographic location of the
      caller (except by having the call taker ask the caller).
   Enhanced emergency service: Enhanced emergency services add the
      ability to identify the caller identity and/or caller location to
      basic emergency services.  (Sometimes, only the caller location
      may be known, e.g., from a public access point that is not owned
      by an individual.)
   Trunk replacement: In the trunk replacement architecture, the caller
      uses the existing PSTN infrastructure to place an emergency call.
      Only the path from the "selective router", or the equivalent
      functionality outside North America, to the ECC uses IP-based
      communications.  The call may well be placed from a VoIP device,
      but is assumed to enter the PSTN very close to the location of the
      caller.  The use of Internet protocols is invisible to the caller.
   End-to-end emergency service: In end-to-end emergency service, the
      caller and ECC both use Internet protocols end-to-end.
   Selective router: A selective router or enhanced emergency call
      control office.  The enhanced emergency call control office is
      "[t]he Central Office that provides the tandem switching of 9-1-1
      calls.  It controls delivery of the voice call with ANI to the
      PSAP and provides Selective Routing, Speed Calling, Selective
      Transfer, Fixed Transfer, and certain maintenance functions for
      each PSAP.  Also known as 9-1-1 Selective Routing Tandem or
      Selective Router." (NENA Glossary) The term may be specific to
      North America.  (TBD:  Find out if there are other terms.)





Schulzrinne              Expires April 17, 2005                 [Page 5]


Internet-Draft           Emergency requirements             October 2004


4.  Trunk Replacement

   In the trunk replacement architecture, an ECC replaces an analog
   (CAMA) or digital (ISDN) trunk with packet-based access, typically
   over one or more high-speed access lines such as DSL or leased lines.
   The packet-based access terminates in the "selective router" that
   normally hands off calls to the ECC.  Thus, the ECC becomes an EICC,
   but no larger scale infrastructure changes are required.  To amplify,
   in the trunk-replacement model, a SIP user agent calling for
   emergency assistance can NOT dial reach the ECC directly via a SIP
   session; rather, the SIP session terminates on a PSTN gateway,
   traverses the PSTN as in today's circuit-switched environment and is
   only converted to VoIP at the selective router handling the ECC.

   Motivation: Trunk replacement is motivated by cost and call setup
   considerations.  It may be cheaper to use IP-based technology for the
   access link and ECC-internal communications.  Also, many existing
   (US) PSAPs use analog technology (CAMA trunks), to receive emergency
   calls.  These trunks, originally designed for operator positions, can
   pulse out the ten or 20-digit (for wireless) caller's number, but as
   dialed digits.  Thus, they add several seconds of call setup delay.
   This can be particularly disconcerting since it affects the time
   until the call taker can pick up the call.  IP-based communications,
   using, for example, SIP as a call signaling protocol, can effectively
   eliminate this extra caller identification delay.  (Additional delays
   are caused by the often very low speed access to the mapping database
   that maps caller identity to geographic location.) Finally, since
   pending calls do not consume access network resources, such systems
   may be more robust in the face of overload.

   M1:  Coexistence: Due to the investment required, not all ECCs will
      convert to IP-based access at the same time.  Thus, emergency
      calls MUST work in a network where some ECCs use existing (analog)
      technology, some ISDN, others IP.  In particular, existing back-up
      relationships between ECCs must continue to work.
   M2:  Call setup delay: The call setup delay MUST NOT be no larger
      than for existing analog trunks and SHOULD be significantly
      smaller.  Call setup times of two seconds or less are RECOMMENDED.
   M3:  Call identification: Signaling from the PSTN switch must be able
      to convey both ten and 20-digit caller identities (ANI --
      automatic number identification) used in North America and other
      digit strings used elsewhere.
   M4: Call transfer: Call takers MUST be able to transfer active
      sessions to other call takers within the same ECC and to other
      ECCs, even those not using Internet.






Schulzrinne              Expires April 17, 2005                 [Page 6]


Internet-Draft           Emergency requirements             October 2004


   M5: Simultaneous alerting: A given set of call takers must be alerted
      to any incoming emergency call.
   M6:  Call routing: The call may be awarded to the first call taker to
      answer or it may be routed to call agents based on policies, such
      as least-busy.  Agents must be able to be assembled into multiple
      groups according to policies specified by ECC authorities.  These
      groupings must be changeable by the ECC authority [4].
   M7: Call queueing: It must be possible to queue calls, either in
      answered or unanswered state.  Queued calls must be able to
      receive recorded announcements.  ECC personnel, as directed by
      policy, should be able to modify the announcements.  The call
      queue should allow automatic or manual transfer to another
      location of calls that exceed a particular expected waiting time
      [4].
   M8: Call identification: The call taker MUST be able to distinguish
      the following incoming call types [4]:
      *  emergency calls dialed via a univeral emergency number;
      *  direct-dialed emergency calls;
      *  transfers from other ECCs;
      *  anonymous calls;
      *  administrative calls;
      *  call origination (wireline, wireless, telecommunication devices
         for the deaf (TDD));
      *  default-routed calls (These are calls for which selective
         routing information was unavailable, resulting in the call
         being routed to a "default" ECC based on other criteria.)
   M9: Information delivery: The call setup request MUST be able to
      deliver the following information [4]:
      *  called party number (to identity ECC or type of call);
      *  calling party number, including any numbering plan digits;
      *  delivery of indication of caller ID blocking for non-emergency
         calls;
      *  location information or lookup keys;
      *  ANI on abandoned calls;
      *  indication that a terminating emergency call has been alternate
         routed from another PSAP.
   M10: Agent sign-on: Agents must be able to log on and log off;
      workstations conditions should at least include "ready", "not
      ready" and "busy" [4].
   M11  Conferencing: Occasionally, supervisors, translators or other
      specialists need to participate in an emergency call.  Thus, it
      MUST be possible to add one or more parties, not necessarily
      located in the IECC, to any emergency call at any time.
   M12: Announcements: Callers may receive automated announcements or
      other indications of call status [4].






Schulzrinne              Expires April 17, 2005                 [Page 7]


Internet-Draft           Emergency requirements             October 2004


   M13: Call queues: Supervisors MUST be able to manage call queues.
   M14: Call metrics: Supervisors and/or agents can measure call delays
      and other performance metrics [4].
   M15:  Monitoring and recording: In many jurisdictions, both sides of
      all emergency calls are automatically recorded as potential legal
      evidence.  Thus, it MUST be possible to record and timestamp all
      signaling and media from all successful, queued, failed and
      aborted calls.
   M16: Abandoned calls: ECCs need to be notified of abandoned calls,
      i.e., emergency calls that are dropped by the caller before being
      answered by a call taker.
   M17:  Transition to end-to-end: Protocols and architecture SHOULD be
      chosen so that a trunk-replacement IECC can receive emergency
      calls placed by IP endpoints without major system changes or
      hardware upgrades.
   M18: Authentication of incoming calls: The IECC MUST be able to
      ascertain that the calls it receives are indeed originating from
      the selective router.
   M19: Authentication of the IECC: The selective router MUST be able to
      be assured that the calls it places reach the desired IECC rather
      than an impostor.
   M20:  Confidentiality: Implementations MUST support confidentiality
      for call signaling and media streams, to protect them against
      unauthorized disclosure to third parties.
   M21:  Robustness: An IECC SHOULD be able to automatically route all
      incoming calls to another backup IECC, even if the access link(s)
      to the primary IECC are inoperative.  Any such redirection MUST be
      authenticated.
   M22:  Overflow handling: An IECC SHOULD be able to automatically
      route calls to another IECC if the (expected) waiting time exceeds
      a configured threshold.
   M23:  Hold: The call taker MUST be able to place the a call in a
      status that allows him/her to handle other calls without
      disconnecting from the caller.  A visual/audible notification
      should be available for the call taker to alert them that a call
      is on hold.  The call should continue to be recorded and an
      optional voice message should be made available for the caller so
      they are aware of the status of their call [4].
   M24:  Forced disconnect of caller: The "forced disconnect of caller"
      feature allows the ECC call taker to disconnect a call when the
      call is in an off hook status at the calling parties end.  This
      eliminates the possibility that emergency resources are needlessly
      tied up by emergency calls made and then left off hook [4].
   M25:  Called party hold: This feature allows a call taker to continue
      to stay connected to the calling party even if the calling party
      attempts to place their phone in an on-hook status [4].





Schulzrinne              Expires April 17, 2005                 [Page 8]


Internet-Draft           Emergency requirements             October 2004


   M26:  Caller ring back: Caller ring-back allows the call taker to be
      able to ring a phone back even if the destination phone is in an
      off-hook status [4].
















































Schulzrinne              Expires April 17, 2005                 [Page 9]


Internet-Draft           Emergency requirements             October 2004


5.  End-to-End IP-Based Emergency Calls

   End-to-end emergency calls originate on an Internet device, traverse
   IP networks and terminate on an IP-capable ECC (IECC).

   As noted, emergency calls need to be identified as such Section 5.1
   and be routed to the appropriate emergency call center (see Section
   5.3).  The ECC needs to determine who (Section 5.4) placed the call
   from where (Section 5.2).  Emergency calls may not be subject to
   access restrictions placed on non-emergency calls.  Also, some call
   features may interfere with emergency calls, particularly if triggerd
   accidentally (Section 5.5).

5.1  Emergency Address

   The emergency address is used by the emergency caller to declare a
   call to be an emergency call and to guide the call to an ECC.  The
   emergency address could a be "sip", "sips" or "tel" URI, or some
   other, yet-to-be-defined URI scheme.

   A1:  Universal: Each device and all network elements MUST recognize
      one or more global emergency call identifiers, regardless of the
      location of the device, the service provider used (if any) or
      other factors.
      Motivation:  SIP and other call signaling protocols are not
      specific to one country or service provider and devices are likely
      to be used across national or service provider boundaries.  Since
      services such as disabling mandatory authentication for emergency
      calls (S1) requires the cooperation of outbound proxies, the
      outbound proxy has to be able to recognize the emergency address
      and be assured that it will be routed as an emergency call.  Thus,
      a simple declaration on a random URI that it is an emergency call
      will likely lead to fraud and possibly attacks on the network
      infrastructure.  A universal address also makes it possible to
      create user interface elements that are correctly configured
      without user intervention.  UA features could be made to work
      without such an identifier, but the user interface would then have
      to provide an unambiguous way to declare a particular call an
      emergency call.
   A2:  Local: Since many countries have already deployed national
      emergency numbers, such as 911 in North America and 112 in large
      parts of Europe, UAs, proxies and call routers MUST recognize
      local emergency numbers.  In addition, they SHOULD recognize
      emergency numbers that are found elsewhere.
      Motivation:  The latter requirement is meant to help travelers
      that may not know the local emergency number and instinctively
      dial the number they are used to from home.  However, it is
      unlikely that all systems could be programmed to recognize any



Schulzrinne              Expires April 17, 2005                [Page 10]


Internet-Draft           Emergency requirements             October 2004


      emergency number used anywhere as some of these numbers are used
      for non-emergency purposes, in particular extensions and service
      numbers.
   A3:  Recognizable: Emergency calls MUST be recognizable by user
      agents, proxies and other network elements.  To prevent fraud, an
      address identified as an emergency number for call features or
      authentication override MUST also cause routing to an ECC.
   A5:  Minimal configuration: Any local emergency numbers SHOULD be
      configured automatically, without user intervention.
      Motivation:  A new UA "unofficially imported" into an organization
      from elsewhere should have the same emergency capabilities as one
      officially installed.
   A6:  Secure configuration: Devices SHOULD be assured of the
      correctness of the local emergency numbers that are automatically
      configured.
      Motivation:  If we assume a fixed, global emergency service
      identifier that requires no configuration and only configure local
      "traditional" emergency numbers, users are not likely to suddenly
      dial some random number if a rogue configuration server introduces
      this as an additional emergency number.  The ability to override
      all locally configured emergency number is of more concern.

5.2  Identifying the Caller Location

   This section supplements the requirements outlined in RFC 3693 [3].
   Thus, the requirements enumerated there are not repeated here.  In
   general, we can distinguish two modes of operation:  direct and
   indirect location provision.  In direct location provision, the
   calling end system knows its own location and can convey this
   location to the ECC.  In an indirect system, the caller is identified
   by a permanent or temporary identifier, which the ECC then uses to
   map the caller to a current location.  (In the current North American
   enhanced emergency calling system, the landline terminal phone number
   is mapped to a location using the so-called ALI (Automatic Location
   Identification) database.  For wireless phones, a temporary
   identifier is created and then mapped to the location information.)

   (This is somewhat similar to terminal-based and network-based
   location services in wireless emergency calling services.  However,
   even in direct location provision, the terminal may well acquire the
   location information from a third party, e.g., a wireless location
   beacon or a DHCP server.)

   L1:  Multiple location providers: For indirect locations, ECCs MUST
      be able to access different location providers.  The location
      provider may be tied to the service provider or may be independent
      of the service provider.




Schulzrinne              Expires April 17, 2005                [Page 11]


Internet-Draft           Emergency requirements             October 2004


      Motivation:  This requirement avoids that all users have to rely
      on a single location provider.  This requirement is hard to avoid
      if there are no traditional national application-layer service
      providers.
   L2:  Civic and geographic: Where available, both civic (street
      address) and geographic (longitude/latitude) information SHOULD be
      provided to the ECC.
      Motivation:  While geographic information can usually be
      translated into civic coordinates, some coordinates, such as
      building numbers and floors, are more easily provided as civic
      coordinates since they do not require a detailed surveying
      operation.  For direct location determination, it may also be
      easier for the user to check civic coordinates for correctness.
   L3:  Location source identification: Sources and translations of
      location data MUST be indicated to the ECC.  (Motivation:  This
      allows the ECC to better judge the reliability and accuracy of the
      data and track down problems.)

5.3  Identifying the Appropriate Emergency Call Center

   From the previous section, we take the requirement of a single (or
   small number of) emergency addresses which are independent of the
   caller's location.  However, since for reasons of robustness,
   jurisdiction and local knowledge, ECCs only serve a limited
   geographic region, having the call reach the correct ECC is crucial.
   While an ECC may be able to transfer an errant call, any such
   transfer is likely to add tens of seconds to call setup latency and
   is prone to errors.  (In the United States, there are about 6,000
   PSAPs.)

   There appear to be two basic architectures for translating an
   emergency address into the correct IECC.  We refer to these as
   caller-based and mediated.  In caller-based resolution, the caller's
   UA consults a directory and determines the correct IECC based on its
   location.  We assume that the UA can determine its own location,
   either by knowing it locally or asking some third party for it.  A UA
   could conceivably store a complete list of all ECCs across the world,
   but that would require frequent synchronization with a master
   database as ECCs merge or jurisdictional boundaries change.

   For mediated resolution, a SIP (outbound) proxy or redirect server
   performs this function.  Note that the latter case includes the
   architecture where the call is effectively routed to a copy of the
   database, rather than having some non-SIP protocol query the
   database.  Since servers may be used as outbound proxy servers by
   clients that are not in the same geographic area as the proxy server,
   any proxy server has to be able to translate any caller location to
   the appropriate ECC.  (A traveller may, for example, accidentally or



Schulzrinne              Expires April 17, 2005                [Page 12]


Internet-Draft           Emergency requirements             October 2004


   intentionally configure its home proxy server as its outbound proxy
   server, even while far away from home.)

   Note that the first proxy doing the translation may not be in the
   same geographic area as the UA placing the emergency call.

   The problem is harder than for traditional web or email services.
   There, the originator knows which entity it wants to reach,
   identified by the email address or HTTP URL.  However, the emergency
   caller only dialed an emergency address.  Depending on the location,
   any of several ten thousand ECCs around the world could be valid.  In
   addition, the caller probably does not care which specific ECC
   answers the call, but rather that it be an accredited ECC, e.g., one
   run by the local government authorities.  (Many ECCs are run by
   private entities.  For example, universities and corporations with
   large campuses often have their own emergency response centers.)

   I1:  Correct IECC: The system MUST reach the correct IECC, that is,
      an IECC that serves the location of the caller.  In particular,
      the location determination should not be fooled by the location of
      IP telephony gateways or dial-in lines into a corporate LAN (and
      dispatch emergency help to the gateway or campus, rather than the
      caller), multi-site LANs and similar arrangements.
   I2:  Early routing: In mediated mode, the first proxy server along a
      request path MUST attempt to route the call to the appropriate
      IECC.
      Motivation:  Proxy servers close to the caller can be expected to
      have better call routing knowledge, particularly if international
      boundaries are being crossed.
   I3:  Choice of IECCs: The system SHOULD offer the emergency caller a
      choice as to whether he wants to reach a local private emergency
      response center, e.g., on a corporate campus, or the
      government-run emergency call center responsible for his current
      location.
      Motivation:  This choice is often, but not always, provided today.
      For example, in some cases, the local campus emergency center is
      reachable by a different number or 9-911 reaches the external ECC,
      while 911 reaches campus security.
   I4:  Assuring IECC identity: The emergency caller SHOULD be able to
      determine conclusively that he has reached an accredited emergency
      call center.
      Motivation:  This requirement is meant to address the threat that
      a rogue, possibly criminal, entity pretends to accept emergency
      calls.
      Implementations SHOULD allow callers to proceed, with appropriate
      warnings or user confirmations, if the identity of the destination
      IECC cannot be verified.




Schulzrinne              Expires April 17, 2005                [Page 13]


Internet-Draft           Emergency requirements             October 2004


      Motivation:  Verification can fail for any number of reasons, such
      as lack of a common certificate chain, especially when traveling,
      call forwarding, or the expiration of certificates.
      Accreditation, e.g., in the case of corporate or university
      campuses, may not exist.
   I5:  Traceable resolution: Particularly for mediated resolution, the
      caller SHOULD be able to definitively and securely determine who
      provided the resolution answer.
   I6:  Assuring directory identity: The querier (UA or server) MUST be
      able to assure that it is querying the intended directory.
   I7:  Query response integrity: The querier MUST be able to be
      confident that the query or response has not been tampered with.
   I8:  Assuring update integrity: Any update mechanism for the
      directory MUST ensure that only authorized users can change
      directory information.  An audit trail MUST be provided.
   I9:  Call setup latency: The directory lookup SHOULD add minimal
      delay to the call setup.  Since outbound proxies will likely be
      asked to resolve the same geographic coordinates repeatedly, a
      suitable time-limited caching mechanism SHOULD be supported (see
      also "Ix").
   I10:  Multiple directories: A UA or proxy SHOULD be able to use
      multiple different directories to resolve the emergency address.
      We do not assume that a single directory has worldwide or even
      nationwide coverage.
      (Motivation: This allows competing or regional data sources.)
   I11: Referral: All directories SHOULD refer out-of-area queries to an
      appropriate default or region-specific directory.
      Motivation:  This requirement alleviates the potential for
      misconfigurations to cause calls to fail, particularly for
      caller-based queries.
   I12:  Multiple protocols: It MAY be useful if directories support
      multiple query protocols, such as SIP (for proxying), IRIS, LDAP,
      a SOAP-based query and others.  A mandatory-to-implement protocol
      MUST be specified and an over-abundance of similarly-capable
      choices appears undesirable.
      (Motivation:  It appears likely that the resolution mechanism will
      be needed by a variety of session protocols and user
      applications.)
   I13:  Robustness: The resolution mechanism MUST allow to deploy
      systems that are robust in the face of partial network and
      directory server failures.  Caching MAY be used to mitigate
      temporary unavailability of directories or network connectivity.
   I14:  Incrementally deployable: An Internet-based emergency call
      system MUST be able to deployed incrementally.  In the initial
      stages of deployment, an emergency call may not reach the optimal
      ECC.





Schulzrinne              Expires April 17, 2005                [Page 14]


Internet-Draft           Emergency requirements             October 2004


   I15:  Testable: A user SHOULD be able to test whether a particular
      address reaches emergency help, without actually causing emergency
      help to be dispatched or consuming ECC call taker resources.  Such
      tests MUST indicate the source of any problems, including the
      validity and plausibility of civic and geospatial location
      addresses.

5.4  Identifying the Caller

   Enhanced emergency call systems provide the ECC with the identity and
   location of the caller.  In PSTN-based systems, the identity is
   represented by the number of the terminal the call is placed from.
   In a SIP-based system, we have two distinct identities, namely the
   address of the terminal (SIP Contact header field) and the identity
   (name and/or AOR) of the person using the terminal.  Depending on the
   circumstances, only one of them may be available.  For example, from
   a public terminal (Internet payphone), only the Contact address may
   be useful.

   In most jurisdictions, callers do not have a choice as to whether
   they want to reveal their location or identity; such disclosure is
   typically mandated by law.

   C1:  Identity: The system SHOULD allow (but not force) to identify
      both the caller's identity and his or her terminal network
      address.
   C2:  Privacy override: The end system MUST be able to automatically
      detect that a call is an emergency call so that it can override
      any privacy settings that conflict with emergency calling.
      (Whether this override can be configured by the user or is
      considered a condition of service is considered a legal matter,
      not a protocol issue.)
      Motivation:  Since emergency calls are often placed by children,
      by people using somebody else's end system or by people in panic,
      any configuration should be automated rather than relying on user
      interaction at the time of the call.  Delaying a call until the
      user discovers that they have to answer some screen prompt or deal
      with a voice prompt in an unfamiliar language is likely to lead to
      large call setup delays or call failures.  This does not preclude
      that end systems can allow, on a call-by-call basis, to configure
      special call parameters.

5.5  Call Setup and Call Features

   S1:  Authentication override: In many jurisdictions, emergency calls
      can be placed by any device, regardless of whether it has
      subscribed for service.  Similarly, outbound proxies and other
      call filtering elements MUST be able to be configured so that they



Schulzrinne              Expires April 17, 2005                [Page 15]


Internet-Draft           Emergency requirements             October 2004


      allow unauthenticated emergency calls.
   S2:  Mid-call features: The end system MUST be able to recognize an
      emergency call and allow configuration so that certain call
      features are not triggered accidentally.  For example, it may be
      inappropriate to transfer the ECC or put it on hold.  An end
      system MAY make it more difficult to disconnect an on-going
      emergency call or accept other incoming calls while in an
      emergency call.
      Motivation:  Call transfer initiated by the emergency caller is
      likely only to be a problem if a PSTN gateway or B2BUa is in the
      call path.  It is not clear how much effort should be expended on
      preventing intentional, as opposed to accidental, disconnection,
      since callers can typically find physical-layer means to terminate
      the call.  This feature is not generally available in the PSTN.
      For example, ANSI T1.628-2001 states that "E9-1-1 Call hold is an
      optional network feature provided to a PSAP which prevents a
      caller from disconnecting an ESC.  ....  However, there is no DSS1
      or SS7 support for this capability at this time."
   S3:  Testable: Users SHOULD be able to test the ability to place an
      emergency call without actually invoking an emergency response or
      tying up emergency call take resources.
      Motivation:  This capability is unfortunately missing from the
      current PSTN.
   S4:  Integrity: Implementations MUST provide mechanisms that ensure
      the integrity of SIP protocol component that are crucial to
      providing reliable emergency call service.  (This requirement
      implies authentication of the caller to allow integrity protection
      of the request and authentication of the ECC to allow integrity
      protection of responses.)






















Schulzrinne              Expires April 17, 2005                [Page 16]


Internet-Draft           Emergency requirements             October 2004


6.  Security Considerations

   Confidentiality, integrity and authentication are core requirements
   for multiple aspects of emergency calling.  Threats exist at the
   infrastructure and individual call level.  Security threats are
   identified throughout this document.

   An adversary could corrupt call information or ECC resolution to
   cause emergency calls to fail subtly, without the caller necessarily
   noticing.  This can be done on a call-by-call basis or by corrupting
   elements that perform the resolution, including the directory
   described in Section 5.3, Internet routing tables or DNS.
   (Obviously, there are typically other ways to make emergency calls
   fail completely, an approach phone-wire cutting burglars have
   practiced for years.  However, the ability to spoof an ECC requires
   physical access to the PSTN cable plant, while this may not be
   required in the IP case.)

   Here, we do not consider attacks on the emergency call infrastructure
   itself.  The techniques for dealing with such attacks are likely to
   be similar as those for protecting other network infrastructure,
   although the stakes may well be higher.  In particular, layered
   defenses against denial-of-service attacks, including return
   routability checks, are likely to be part of the defensive arsenal.



























Schulzrinne              Expires April 17, 2005                [Page 17]


Internet-Draft           Emergency requirements             October 2004


7.  Acknowledgments

   James Polk provided helpful comments on an earlier version of this
   document.















































Schulzrinne              Expires April 17, 2005                [Page 18]


Internet-Draft           Emergency requirements             October 2004


8.  References

8.1  Normative References

   [1]  Bradner, S., "Key words for use in RFCs to Indicate Requirement
        Levels", BCP 14, RFC 2119, March 1997.

8.2  Informative References

   [2]  Charlton, N., Gasson, M., Gybels, G., Spanner, M. and A. van
        Wijk, "User Requirements for the Session Initiation Protocol
        (SIP) in Support of Deaf, Hard of Hearing and Speech-impaired
        Individuals", RFC 3351, August 2002.

   [3]  Cuellar, J., Morris, J., Mulligan, D., Peterson, J. and J. Polk,
        "Geopriv Requirements", RFC 3693, February 2004.

   [4]  National Emergency Number Assocation, "NENA technical
        information document on the interface between the E9-1-1 service
        providers network and the Internet protocol (IP) PSAP", NENA
        NENA-08-501, February 2003.


Author's Address

   Henning Schulzrinne
   Columbia University
   Department of Computer Science
   450 Computer Science Building
   New York, NY  10027
   US

   Phone: +1 212 939 7004
   EMail: hgs+sip@cs.columbia.edu
   URI:   http://www.cs.columbia.edu
















Schulzrinne              Expires April 17, 2005                [Page 19]


Internet-Draft           Emergency requirements             October 2004


Intellectual Property Statement

   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.


Disclaimer of Validity

   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 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.


Copyright Statement

   Copyright (C) The Internet Society (2004).  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.


Acknowledgment

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




Schulzrinne              Expires April 17, 2005                [Page 20]