[Ecrit] (114) Comments on ECRIT Reqs ID (part I)

All

[*** this is part 1 of a 2 part message to the list - since I received an 
error posting a message that exceeded 40k, the break between parts is at 
section 6 of the ID]

I reviewed the requirements ID draft-schulzrinne-ecrit-requirements-00.txt 
for the meeting this coming week, and have a few comments.... ok, 114 
comments.... listed.... below....

For those of you attending the interim, I will bring these up there. For 
those of you not attending the interim, chime in and be heard about this 
comments.

My overall comment to this ID is it is taking on waaaaaay more than the WG 
charter intended, but I am not surprised. I know a lot of folks want to get 
specific items addressed here. Some of those items I feel should be 
addressed elsewhere.

Anyway, to the comments - here they are in the order they appear in the doc 
with one exception, the last one. The last one should be addressed 
somewhere, and I do not know if it should only be addressed in

http://www.ietf.org/internet-drafts/draft-ietf-sipping-location-requirements-02.txt 
and its next version in the SIP WG.

The comments:

[note to ID authors - I know this is a group effort, so do not take my use 
of the word "you" personally]

The following is a review of draft-schulzrinne-ecrit-requirements-00 on May 
14th, 2005

Comment #1
Section 2 - Terminology - top paragraph
"MUSTNOT", "SHALLNOT", "SHOULDNOT" are all 2 words each

Comment #2
Section 2 - Terminology - 2nd paragraph
"protocols" should be singular

Comment #3
Section 2 - Terminology - term
When did IAP become AIP?

IAP and ISP flowed and should be retained.

Comment #4
Section 2 - Terminology - term
basic emergency service - should be capitalized

Comment #5
Section 2 - Terminology - term
"domain" should be renamed "administrative domain"

Comment #6
Section 2 - Terminology - term
You mention "IPSAP" before defining it, and don't state you will define it 
below like you do other terms. Some call this an IP-PSAP, not an IPSAP, BTW...

Comment #7
Section 2 - Terminology - term
    emergency caller: The user or user device entity needing sending his/
       her location to another entity in the network.

"...needing sending..." is poor wording

Comment #8
Section 2 - Terminology - term
    emergency caller: The user or user device entity needing sending his/
       her location to another entity in the network.

The "emergency caller" is "the individual in distress desiring help".

Comment #9
Section 2 - Terminology - term
geographic coordinate system - you fail to mention "a defined meridian", 
which should be added to the definition

Comment #10
    R2.  International:  The protocols and protocol extensions developed
       MUST support regional, political and organizational differences.

       Motivation: It must be possible for a device or software developed...

This "must" in the motivation should be a MUST as it talks normatively

Comment #11
    R5.  Minimum Connectivity:  NEED REQUIREMENT HERE

       Motivation:  If there is network connectivity between the
       emergency caller and the PSAP, and routing information is
       available, the call should be completed, even if other parts of
       the network are not reachable.

This makes little sense, because if there is connectivity, there is a 
route, yet the wording seems to believe there might not be (which means 
there is "no connectivity")

Comment #12
    R6.  Incremental Deployment Emergency calls from IP-based devices
       MUST be incrementally supported.

This wording doesn't make sense; should it be:

    "R6  Incremental Deployment - incremental deployment of IP-based 
emergency calling MUST be supported"

Comment #13
   "R6 Motivation:  Any mechanism must be deployable incrementally and
       work even if not all entities support IP-based emergency calling.
       For example, User agents conforming to the SIP specification [1],
       but unaware of this document, must be able to place emergency
       calls, possibly with restricted functionality."

This ID professes that it all about IP e2e connectivity, yet this states it 
isn't really necessary.  Which is it? This motivation, or the wording in 
the Intro section?

Comment #14
    R7.  Middlebox Reliance:  For a transient time the device and the UA
       MAY use the help of servers (e.g.  ESRP) to provide the
       connectivity to ECC, especially for ECC not yet connected to the
       Internet.

Same as comment to #13 - This ID professes that it all about IP e2e 
connectivity, yet this states it isn't really necessary.  Which is it? This 
motivation, or the wording in the Intro section?

Comment #15
   R7   Motivation:  Emergency calling mechanisms must support existing
       emergency call centers based on circuit-switched technology as
       well as future ECCs that are IP-enabled.

This motivation statement clearly contradicts the Intro section (stating 
this is for IP e2e scenarios).

Comment #16
A2.  Local: Since many countries have already deployed national
       emergency identifiers, such as 911 in North America and 112 in
       large parts of Europe, UAs, proxies and call routers MUST
       recognize these universal emergency identifiers, but MAY NOT
       recognize lower level local emergency identifiers, including those
       such as 999, 122, 133, etc.  In addition, these same call routing
       entities SHOULD recognize emergency identifiers that are used in
       other jurisdictions

I don't understand this one as written, as there are ~60 different numbers 
around the world, we should list which ones are recommended for support and 
discuss the list.

Primary examples are 115, 116, 117 etc for different emergency services 
within the same jurisdiction.

Comment #17
A2.  Local - [Ed.  The requirement A2 is really a set of 3 requirements, and
       needs to have the question answered which is: "what does the term
       "local" mean?".]

Good question...

Comment #18
    A2 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
       emergency number used anywhere as some of these numbers are used
       for non-emergency purposes, in particular extensions and service
       numbers.

What is meant by this last phrase: "... in particular extensions and service
       numbers."

Comment #19
  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 a PSAP.

Recommend replacing some of the last sentence with: "...any call recognized 
as an emergency call MUST be delivered expeditiously to an ECC".

Comment #20
    A5.  Secure configuration: Devices SHOULD be assured of the
       correctness of the local emergency numbers that are automatically
       configured.

How do you "assure a device" without asserting something that requires a 
key the remote and local domains understand, as well as the UA understands?

Comment #21
    A6.  Backwards-compatible: Existing devices that predate the
       specification of emergency call-related protocols and conventions
       MUST be able reach a PSAP.

I've read this above somewhere else in the ID. If a PIDF-LO is required to 
complete a 911 call, and it isn't present, and the device doesn't support 
the error messages from  draft-ietf-sipping-location-requirements-02.txt, 
how can this work?

I think this can work in residential and SMB scenarios, but I'm not sure 
about any others. VPNs in those other network types is but one reason for 
these problems.

Comment #22
    A7.  Common Identifier:  User initiated requests using local
       initiation methods (e.g. 9-1-1) MUST be supported across non-local
       domains (e.g. foreign countries).

This is stated too US-centric. The IETF isn't about a country. This should 
state that all/most recognizable local emergency numbers will be converted 
to (perhaps) one SIP user-part for all SIP intermediaries to know to 
process accordingly.

Comment #23
   A7  Motivation: While traveling, users must be able to use their
       familiar "home" emergency identifier.  Users should also be able
       to dial the local emergency number in the country they are
       visiting.

Looking at the range of potential numbers, this will be interesting to meet 
- including just dialing '0' and meaning an emergency call

Comment #24
Section 5
    Proxy-inserted: A proxy along the call path inserts the location or
       location reference.

This violates 3261 and draft-ietf-sipping-location-requirements-02.txt

Comment #25
    L1.  Multiple location services: For UA-referenced locations, PSAPs
       MUST be able to access different location providers.  The location
       provider may be tied to the ASP, AIP or ISP or may be independent
       of these entities.

This may be inside an enterprise, and that should not be underscored or 
overlooked due to the trust-boundary issues known there

Comment #26
   L1      Motivation:  This requirement avoids that all users have to rely
       on a single location service provider.  This requirement is hard
       to avoid if there are no traditional national application-layer
       service providers.

I don't understand the first sentence as written, especially since this ID 
just got done talking about internationalization scope.

Comment #27
    L2.  Civic and Geographic: Where available, both civic (street
       address) and geographic (longitude/latitude) information SHOULD be
       provided to the PSAP.

This necessitates a PIDF-LO extension to indicate if one was derived from 
the other. That XML element doesn't exist to my knowledge.

Comment #28
   L2        Motivation:  While geographic coordinate information can usually
       be translated into civic address location information, "some
       specific information, such as building number and floor, is more
       easily provided as civic location information since it does not
       require a detailed surveying operation".  For direct location
       determination, it may also be easier for the user to check civic
       location information to assure verity.

I don't understand the quoted comment (I put in the quotes BTW) in the 
first sentence above. Is this suggesting a detailed survey *will* yield the 
cube number from a GPS coordinate on a non-ground floor? If it won't, it 
doesn't need to be brought up as motivation.

Comment #29
    L3.  Location source identification: The source of a location data,
       whether measured, derived (e.g geocoding or reverse geocoding
       transformation), or manually input, MUST be indicated to the PSAP.

How is this not covered by the <method> element in the PIDF-LO, and if so, 
why is the requirement not just "...the <method> element MUST be present 
when known"...

Comment #30
   L3        Motivation:  This allows the PSAP to better judge the reliability
       and accuracy of the data and track down problems.

I don't believe the word "judge" is appropriate here, as the PSAPs first 
need to "learn" (a better word for above) from real world experience what 
data indicates what results.  Down the road, they will then analyze trends 
and such to determine how best to interpret data.

Comment #31
    L4.  Certifiable: In some cases, the source and generation time of
       the location object used for call routing and caller location
       display MUST be verifiable, e.g., by a digital signature.  The
       security requirements describe this in more detail.

If my wireline phone has a timestamp of 2002, is that bad/untrusted? I 
don't believe so. This requirement needs to be adjusted to state this is a 
function of the <method> the location was derived from. Further, a future 
document (perhaps years from now) needs to be generated to give 
recommendations of those functions per <method> value.

Comment #32
    L5.  Multiple locations: Multiple locations MAY be associated with
       the caller

How is this requirement different that L2?

If it can be different, then this text needs to state when/how it can be 
different, yet still be considered good.

Comment #33
   L5      Motivation:  Multiple locations may occur either because the
       caller has provided more than one civic or geographic
       (coordinates) location, supplies both civic and geospatial
       location information, or because different location determination
       entities make different assessments of the caller's location."

There needs to be a requirement stating if there is location information in 
both formats, and both are depicting the same location (as far as the UA is 
concerned), they MUST be in the same PIDF-LO; if different - or not sure 
they are the same location, they need to be in separate PIDF-LOs (typically 
with different <method> values).

Comment #34
   L5      Motivation:  Multiple locations may occur either because the
       caller has provided more than one civic or geographic
       (coordinates) location, supplies both civic and geospatial
       location information, or because different location determination
       entities make different assessments of the caller's location."

If there are more than one, and they are different, which is trusted to be 
true?

Comment #35
    L6.  Validation of civic location: It MUST be possible to validate an
       address prior to its use in an actual emergency call.

How is "validate" here in L6 different from "certifiable" in L4?

Comment #36
    L6.  Validation of civic location: It MUST be possible to validate an
       address prior to its use in an actual emergency call.

"... prior to its use in an actual emergency call."

should be replaced with:

"at any time" only.

And...it's an implementation issue when this occurs, so I'm not sure this 
belongs in the ID. Someone like NENA should be defining this type of 
requirement.

Comment #37
    L7.  Provide location: Calls using VoIP or subsequent methods MUST
       supply location with the call.

"... with the call"

should be replaced with:

"...in each Request message (if SIP is used)."

Comment #38
    L8.  Accept two location types: PSAPs shall accept location as civic
       and/or geo specified.

       [Ed.  Suggest deleting above requirement since it doesn't deal
       with routing]

I disagree with the Ed. Comment because this (as written) limits how many 
different formats a routing Proxy forwards, and how many formats an ECC has 
to deal with.

The same piece of location information (if in a PIDF-LO) will be used for 
routing and dispatching. Limiting the number to two here is good.

Comment #39
    L9.  Altitude included with location: All representations of location
       SHALL include the ability to carry altitude.  This requirement
       does not imply altitude is always used or supplied.

Either it does or it doesn't, but "include the ability..." is wishy-washy

Comment #40
    L10.  Preferred datum: The preferred geographic coordinate system for
       emergency calls SHALL be WGS-84.

2D or 3D?

Comment #41
    L11.  Multiple locations: If multiple locations are provided with a
       call, it SHOULD be possible to identify the most accurate,
       current, appropriate location information to be used for routing
       emergency calls and dispatching emergency responders.

Exactly how will this be accomplished, especially by a Proxy? This is 
fruitless, therefore pointless. L2 should be all that is stated for this 
type of requirement.

Comment #42
    L14.  Imprecise location: Imprecise location information MUST be
       available for emergency call routing and location delivery in
       cases where precise measurement based location determination
       mechanisms fail.

How is "imprecise location" indicated to the PSAP as "imprecise"? If it 
cannot be, it shouldn't be a requirement.

Comment #43
    L15.  Default identification: PSAPs MUST be made aware when imprecise
       location information was used to route a call.

Same comment here that was to L14 - How is "imprecise location" indicated 
by the proxy to the PSAP as "imprecise"? If it cannot be, it shouldn't be a 
requirement.

Comment #44
    L16.  Location Responsibility: Location determination MUST assume a
       responsible party.

This is begging for liability statements to be included every time a 
location is given to a device, that it should not be used unless the user 
agrees "not to hold the deliverer responsible" before usage.

Comment #45
   L17      Motivation:  The location information MUST be attributed to a
       specific point in time.  That is, the location used for routing
       and which is reported to the PSAP call taker, must be the actual
       location of the caller at the time of making the call.

This makes learning/getting/retrieving location a real-time event as the 
emergency call is being initiated - and this is not practical. If my home 
wireline phone has a timestamp of 2002, is that bad?

Comment #46
    L18.  Location, Emergency Caller: Location provided with call MUST be
       associated with an Emergency Caller.

Huh? This is to a device, one in which a user might not be logged into.

Comment #47
   L19   Motivation:  Requirement 1 states that a level of authentication
       and validation for the source of the location is required.  This
       implies the need to for the Emergency Caller to determine the
       authenticating and validating entity for the emergency services
       domain in which they reside.  That is, it must be possible for an
       Emergency Caller to discover and utilize an answerable source of
       location in the access network they are using.

The term in the 1st sentence "level" needs to be defined in order to 
understand if it is met. Level is not currently defined within this ID.

Comment #48
    L19.  Location Domain Availability: Location domain MUST be
       obtainable by Emergency Caller.

       Motivation:  Requirement 1 states that a level of authentication
       and validation for the source of the location is required.  This
       implies the need to for the Emergency Caller to determine the
       authenticating and validating entity for the emergency services
       domain in which they reside.  That is, it must be possible for an
       Emergency Caller to discover and utilize an answerable source of
       location in the access network they are using.

"Emergency Caller" as used in the 1st sentence - I personally don't care 
who the authority is, and I doubt most (if any) do outside of this list and 
some regulators.

Comment #49
    L19.  Location Domain Availability: Location domain MUST be
       obtainable by Emergency Caller.

       Motivation:  Requirement 1 states that a level of authentication
       and validation for the source of the location is required.  This
       implies the need to for the Emergency Caller to determine the
       authenticating and validating entity for the emergency services
       domain in which they reside.  That is, it must be possible for an
       Emergency Caller to discover and utilize an answerable source of
       location in the access network they are using.

This last sentence isn't clear. Provide an example of me discovering an 
"answerable source" so I can visualize it.

Comment #50
    L20.  Location Certification: Location provided MUST be certified.

This doesn't account for a GPS chip in a PDA or cell phone, therefore this 
cannot be a MUST.

Comment #51
  L20      Motivation:  The Emergency Caller must be able to establish a
       session with the access domain authenticating and validating
       entity to obtain a certified location.  The authentication of the
       location is granted with an expiry time, after which the location
       within the domain is deemed invalid.

"INVALID"?  This is starting to look an awful lot like a DHCP lease 
operation, and we're not building a protocol to do this here.

Comment #52
    L21.  Location and Emergency Caller Identity: It MUST NOT be assumed
       that Emergency Caller identity provided with location is true
       identity of Emergency Caller.

This contradicts L18

Comment #53
    L22.  Location Acceptability: Location provided by Emergency Caller
       MUST be considered acceptable as input to authentication and
       validation entity.

This isn't really clear. Location should be in a certain format, and there 
is a 'the' missing in the requirement.

Comment #54
    L22.  Location Acceptability: Location provided by Emergency Caller
       MUST be considered acceptable as input to authentication and
       validation entity.

I see no requirement creating an "authentication and validation entity" 
which this requirement assumes already exists for every user.

Comment #55
    L24.  Location Query Authorization: The ability to query emergency
       caller location using a location key MUST be limited to authorized
       end points.

"end points" (which is one word BTW)

Should be replaced with:

"requests"

Comment #56
  L24      Motivation:  Where the Emergency Caller does not desire the
       transmission of their location in-band with their call setup, they
       shall have the option of requesting a unique query key such that
       only authorized end points may query the location directly from
       the domain.

As stated above: "end points" (which is one word)

Should be replaced with:

"requests"

As "endpoints" implies the actual device has an SA relationship with the 
database server, which may not be the case, and only the user created an SA 
from a log-in screen. This is a two level operation.

Comment #57
    L25.  Location Domain Authorization: Location Source entity MUST be
       authorized within the access domain.

This is starting to create an architecture now... and I think this is 
starting to overstep some bounds of the charter and belongs more in NENA 
than IETF

Comment #58
    L26.  Endpoint Location: Location MUST be tied to an endpoint within
       the access domain at the time of an emergency call.

GPSs violate this, so this cannot be a MUST - see L27 below

    L27.  Location Sources: Single source of location MUST NOT be
       assumed.

       Motivation:  To achieve this, the end user device MUST be able to
       retrieve its current location from the access provider, from the
       infrastructure, via GPS, ... or as last resort, from the user
       itself.

Comment #59
    L28.  Location Provided: Endpoint location SHOULD be provided to ECC.

This needs to be restated as "if location is provided to the routing proxy, 
it MUST be provided to the ECC.

cheers,
James

                                 *******************
                 Truth is not to be argued... it is to be presented.

                      Alas, few *truths* exist without the math.

                             ...all else is a matter of perspective

_______________________________________________
Ecrit mailing list
Ecrit@ietf.org
https://www1.ietf.org/mailman/listinfo/ecrit