| ECRIT | H. Schulzrinne |
| Internet-Draft | Columbia University |
| Intended status: Standards Track | H. Tschofenig |
| Expires: September 10, 2012 | Nokia Siemens Networks |
| C.H. Holmberg | |
| Ericsson | |
| M. Patel | |
| InterDigital Communications | |
| March 11, 2012 |
Public Safety Answering Point (PSAP) Callback
draft-ietf-ecrit-psap-callback-04.txt
After an emergency call is completed (either prematurely terminated by the emergency caller or normally by the call taker) it is possible that the call taker feels the need for further communication. For example, the call may have been dropped by accident without the call taker having sufficient information about the current situation of a wounded person. A call taker may trigger a callback towards the emergency caller using the contact information provided with the initial emergency call. This callback could, under certain circumstances, be treated like any other call and as a consequence it may get blocked by authorization policies or may get forwarded to an answering machine.
The IETF emergency services architecture specification already offers a solution approach for allowing PSAP callbacks to bypass authorization policies to reach the caller without unnecessary delays. Unfortunately, the specified mechanism only supports limited scenarios. This document discusses shortcomings of the current mechanisms and illustrates additional scenarios where better-than-normal call treatment behavior would be desirable.
Note that this version of the document does not yet specify a solution due to the lack of the working group participants agreeing on the requirements.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/.
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."
This Internet-Draft will expire on September 10, 2012.
Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Summoning police, the fire department or an ambulance in emergencies is one of the fundamental and most-valued functions of the telephone. As telephone functionality moves from circuit-switched telephony to Internet telephony, its users rightfully expect that this core functionality will continue to work at least as well as it has for the legacy technology. New devices and services are being made available that could be used to make a request for help, which are not traditional telephones, and users are increasingly expecting them to be used to place emergency calls.
An overview of the protocol interactions for emergency calling using the IETF emergency services architecture are described in [RFC6444] and [I-D.ietf-ecrit-phonebcp] specifies the technical details. As part of the emergency call setup procedure two important identifiers are conveyed to the PSAP call taker's user agent, namely the Address-Of-Record (AoR), and, if available, the Globally Routable User Agent (UA) URIs (GRUU). RFC 3261 [RFC3261] defines the AoR as:
In SIP systems a single user can have a number of user agents (handsets, softphones, voicemail accounts, etc.) which are all referenced by the same AOR. There are a number of cases in which it is desirable to have an identifier which addresses a single user agent rather than the group of user agents indicated by an AOR. The GRUU is such a unique user- agent identifier, which is still globally routable. RFC 5627 [RFC5627] specifies how to obtain and use GRUUs. [I-D.ietf-ecrit-phonebcp] also makes use of the GRUU for emergency calls.
Regulatory requirements demand that the emergency call setup procedure itself provides enough information to allow the call taker to initiate a call back to the emergency caller. This is desirable in those cases where the call got dropped prematurely or when further communication need arises. The AoR and the GRUU serve this purpose.
The communication attempt by the PSAP call taker back to the emergency caller is called 'PSAP callback'.
A PSAP callback may, however, be blocked by user configured authorization policies or may be forwarded to an answering machine since SIP entities (SIP proxies as well as the SIP user equipment itself) cannot differentiate the PSAP callback from any other SIP call. "Call barring", "do not disturb", or "call diversion"(aka call forwarding) are features that prevent delivery of a call. It is important to note that these features may be implemented by SIP intermediaries as well as by the user agent.
Among the emergency services community there is the desire to offer PSAP callbacks a treatment such that chances are increased that it reaches the emergency caller. At the same time a design must deal with the negative side-effects of allowing certain calls to bypass call forwarding or other authorization policies. Ideally, the PSAP callback has to relate to an earlier emergency call that was made "not too long ago". An exact time interval is difficult to define in a global IETF standard due to the variety of national regulatory requirements.
To nevertheless meet the needs from the emergency services community a basic mechanism for preferential treatment of PSAP callbacks was defined in Section 13 of [RFC6444]. The specification says:
This approach mimics a stateful packet filtering firewall and is indeed helpful in a number of cases. It is also relatively simple to implement even though it requires state to be maintained by the user agent as well as by SIP intermediaries. Unfortunately, the solution does not work in all deployment scenarios. In Section 3 we describe cases where the currently standardized approach is insufficient.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].
Emergency services related terminology is borrowed from [RFC5012]. This includes terminology like emergency caller, user equipment, and call taker.
This section illustrates a number of scenarios where the currently specified solution, as specified in [I-D.ietf-ecrit-phonebcp], for preferential treatment of callbacks fails. As explained in Section 1 a SIP entity examines an incoming PSAP call back by comparing the domain of the PSAP with the destination domain of the emergency call.
In some deployment environments it is common to have incoming and outgoing SIP messaging routed through different SIP entities. Figure 1 shows this graphically whereby a VoIP provider uses different SIP proxies for inbound and for outbound call handling. Unless they two devices are state synchronized the callback hitting the inbound proxy would get treated like any other call since the emergency call established state information at the outbound proxy only.
,-------.
,' `.
,-------. / Emergency \
,' `. | Services |
/ VoIP \ I | Network |
| Provider | n | |
| | t | |
| | e | |
| +-------+ | r | |
+--+---|Inbound|<--+-----m | |
| | |Proxy | | e | +------+ |
| | +-------+ | d | |PSAP | |
| | | i | +--+---+ |
+----+ | | | a-+ | | |
| UA |<---+ | | t | | | |
| |----+ | | e | | | |
+----+ | | | | | | |
| | | P | | | |
| | | r | | | |
| | +--------+ | o | | | |
+--+-->|Outbound|--+---->v | | +--+---+ |
| |Proxy | | i | | +-+ESRP | |
| +--------+ | d | | | +------+ |
| | e || | |
| | r |+-+ |
\ / | |
`. ,' \ /
'-------' `. ,'
'-------'
Consider the following emergency call routing scenario shown in Figure 2 where routing towards the PSAP occurs in several stages. In this scenario we consider a SIP UA that uses LoST to learn the next hop destination closer to the PSAP. This call is then sent to the user's VoIP provider. The user's VoIP provider receives the emergency call and creates state based on the destination domain, namely state.com. It then routes it to the indicated ESRP. When the ESRP receives it it needs to decide what the next hop is to get it closer to the PSAP. In our example the next hop is the PSAP with the URI psap@town.com.
When a callback is sent from psap@town.com towards the emergency caller the call will get normal treatment by the VoIP providers inbound proxy since the domain of the PSAP does not match the stored state information.
,-------.
+----+ ,' `.
| UA |--- esrp1@foobar.com / Emergency \
+----+ \ | Services |
\ ,-------. | Network |
,' `. | |
/ VoIP \ | +------+ |
( Provider ) | |PSAP | |
\ / | +--+---+ |
`. ,' | |
'---+---' | | |
| |psap@town.com |
esrp@state.com | | |
| | | |
| | | |
| | +--+---+ |
+------------+---+ESRP | |
| +------+ |
| |
\ /
`. ,'
'-------'
Imagine the following case where an emergency call enters an emergency network (state.org) via an ERSP but then gets forwarded to a different emergency services network (in our example to police-town.org, fire-town.org or medic-town.org). The same considerations apply when the police, fire and ambulance networks are part of the state.org sub-domains (e.g., police.state.org).
Similarly to the previous scenario the problem here is with the wrong state information being established during the emergency call setup procedure. A callback would originate in the police-town.org, fire-town.org or medic-town.org domain whereas the emergency caller's SIP UA or the VoIP outbound proxy has stored state.org.
,-------.
,' `.
/ Emergency \
| Services |
| Network |
| (state.org) |
| |
| |
| +------+ |
| |PSAP +--+ |
| +--+---+ | |
| | | |
| | | |
| | | |
| | | |
| | | |
| +--+---+ | |
------------------+---+ESRP | | |
esrp-a@state.org | +------+ | |
| | |
| Call Fwd | |
| +-+-+---+ |
\ | | | /
`. | | | ,'
'-|-|-|-' ,-------.
Police | | | Fire ,' `.
+------------+ | +----+ / Emergency \
,-------. | | | | Services |
,' `. | | | | Network |
/ Emergency \ | Ambulance | | fire-town.org |
| Services | | | | | |
| Network | | +----+ | | +------+ |
|police-town.org| | ,-------. | +----+---+PSAP | |
| | | ,' `. | | +------+ |
| +------+ | | / Emergency \ | | |
| |PSAP +----+--+ | Services | | | ,
| +------+ | | Network | | `~~~~~~~~~~~~~~~
| | |medic-town.org | |
| , | | |
`~~~~~~~~~~~~~~~ | +------+ | |
| |PSAP +----+ +
| +------+ |
| |
| ,
`~~~~~~~~~~~~~~~
The IETF emergency services architecture also considers cases where the resolution from the Service URN to the PSAP URI does not only happen at the SIP UA itself but at intermediate SIP entities, such as the user's VoIP provider.
Figure 4 shows this message exchange of the outgoing emergency call and the incoming PSAP graphically. While the state information stored at the VoIP provider is correct the state allocated at the SIP UA is not.
,-------.
,' `.
/ Emergency \
| Services |
| Network |
|police-town.org|
| |
| +------+ | Invite to police.example.com
| |PSAP +<---+------------------------+
| | +----+------------------+ ^
| +------+ |Invite from | |
| ,police.example.com| |
`~~~~~~~~~~~~~~~ v |
+--------+ ++-----+-+
| | query |VoIP |
| LoST |<-----------------------|Service |
| Server | police.example.com |Provider|
| |----------------------->| |
+--------+ +--------+
| ^
Invite| | Invite
from| | to
police.example.com| | urn:service:sos
V |
+-------+
| SIP |
| UA |
| Alice |
+-------+
In case an emergency call enters the PSTN, as shown in Figure 5, there is no guarantee that the callback some time later does leave the same PSTN/VoIP gateway or that the same end point identifier is used in the forward as well as in the backward direction making it difficult to reliably detect PSAP callbacks.
+-----------+
| PSTN |-------------+
| Calltaker | |
| Bob |<--------+ |
+-----------+ | v
-------------------
//// \\\\ +------------+
| | |PSTN / VoIP |
| PSTN |---->|Gateway |
\\\\ //// | |
------------------- +----+-------+
^ |
| |
+-------------+ | +--------+
| | | |VoIP |
| PSTN / VoIP | +->|Service |
| Gateway | |Provider|
| |<------Invite----| Y |
+-------------+ +--------+
| ^
| |
Invite Invite
| |
V |
+-------+
| SIP |
| UA |
| Alice |
+-------+
Note: This scenario is considered outside the scope of this document. The specified solution does not support this use case.
[Editor's Note: The specification for a solution that meets the requirements will be placed in here.]
[Editor's Note: Instead of an abstract security description text will be provided with the solution description.]
[Editor's Note: IANA consideration text will be added once an agreement on the solution has been reached.
We would like to thank members from the ECRIT working group, in particular Brian Rosen, for their discussions around PSAP callbacks. The working group discussed the topic of callbacks at their virtual interim meeting in February 2010 and the following persons provided valuable input: John Elwell, Bernard Aboba, Cullen Jennings, Keith Drage, Marc Linsner, Roger Marshall, Dan Romascanu, Geoff Thompson, Janet Gunn.
At IETF#81 a small group of people got to together to continue the discussions started at the working group meeting to explore a GRUU-based solution approach. Martin Thomson, Marc Linsner, Andrew Allen, Brian Rosen, Martin Dolly, and Atle Monrad participated at this side-meeting.
Finally, we would like to thank Cullen Jennings for his discussion input. He was the first to propose a "token-based" solution.
In an attempt to describe the problem and to explore solution approaches the working group had also investigated alternative approaches. We document them here for completeness. The solutions fall into three categories: (1) Identity-based authorization, (2) Trait-based authorization, and (3) Call Marking. Even though these solutions are not mutually exclusive we describe them in separate sub-sections.
Beyond the disadvantages listed in each solution category none of them provides the emergency caller with the ability to restrict preferential PSAP callback handling to those cases where an earlier emergency call was initiated.
In Figure 6 an interaction is presented that allows a SIP entity to make a policy decision whether to bypass installed authorization policies and thereby providing preferential treatment. To make this decision the sender's identity is compared with a whitelist of valid PSAPs. The identity assurances in SIP can come in different forms, such as SIP Identity [RFC4474] or with P-Asserted-Identity [RFC3325]. The former technique relies on a cryptographic assurance and the latter on a chain of trust.
+----------+
| List of |+
| valid ||
| PSAP ids ||
+----------+|
+----------+
*
* whitelist
*
V
Incoming +----------+ Normal
SIP Msg | SIP |+ Treatment
-------------->| Entity ||=============>
+ Identity | ||(if not in whitelist)
+----------+|
+----------+
||
||
|| Preferential
|| Treatment
++=============>
(in whitelist)
This approach was not chosen because the establishment of a whitelist containing PSAP identities is operationally complex and does not easily scale world wide. Only when there is a local relationship between the VSP/ASP and the PSAP then populating the whitelist is far simpler. This would, however, constrain the applicability of the mechanism considerably.
An alternative approach to an identity based authorization model is outlined in Figure 7. In fact, RFC 4484 [RFC4484] illustrates a related emergency service use case.
+----------+
| List of |+
| trust ||
| anchor ||
+----------+|
+----------+
*
*
*
V
Incoming +----------+ Normal
SIP Msg | SIP |+ Treatment
-------------->| Entity ||=============>
+ trait | ||(no indication
+----------+| of PSAP)
+----------+
||
||
|| Preferential
|| Treatment
++=============>
(indicated as
PSAP)
In a trait-based authorization scenario an incoming SIP message contains a form of trait, i.e. some form of assertion. The assertion contains an indication that the sending party has the role of a PSAP (or similar emergency services entity). The assertion is either cryptographically protected to enable end-to-end verification or an chain of trust security model has to be assumed. In Figure 7 we assume an end-to-end security model where trust anchors are provisioned to ensure the ability for a SIP entity to verify the received assertion.
This solution was not chosen because trait-based authorization never got deployed in SIP. Furthermore, in order to ensure that the assertions are properly protected it is necessary to digitally sign, which requires some form of public key infrastructure for usage with emergency services. Finally, there need to be some policies in place that define which entities are allowed to obtain various roles. These policies and procedures do not exist today.
Call marking allows the PSAP to place a non-cryptographic label on outgoing calls that gives, when received by a SIP entity, preferential treatment for these callbacks.
When used in isolation this mechanism introduces considerable denial of service attacks due to the ability to bypass any authorization policies and could be utilized to distribute unwanted traffic.