SIP in higher education

[Ben Teitelbaum <>]

No call today, but here are the minutes from last week (thanks to Jeff
Kuure).  The next call will be on Thursday, July 7th.

SIP.edu Conference Call Notes June 23, 2005

*Attendees*

Dennis Baron, MIT
Bob Bounds, Tello
Candace Holman, Harvard
Jerry Keith, UC Riverside
Megan Pengelly, Columbia
Yul Pyun, University of Hawaii
Ben Teitelbaum, Internet 2
Mike Van Norman, UCLA
Carlos Vicente, University of Oregon
Doug Walston, Cisco
Garrett Yoshimi, University of Hawaii
Dave Zimmerman, Berkeley

*Discussion*

Today's call began with Mike Van Norman mentioning that UCLA has been
added to Ed's ENUM trial.

The remainder of the call is devoted to Ben's presentation, which
focuses on various methods for contacting SIP addresses with 12-key
telephones.  Slides from the presentation are available:
http://people.internet2.edu/~ben/talks/200506-sipedu-by-numbers.pdf

There are currently a large number of existing 12-key phones which
make alphanumeric dialing difficult, requiring workarounds for SIP.edu
dialing based on email addresses.  Using E.164 dialing is one
solution, as it offers seamless integration with existing PSTN phones.
However, the E.164 numbers are administered by an outside entity, are
a finite resource, and using them raises regulatory concerns.  In
addition, PSTN interoperability could also be seen as a drawback,
hindering acceptance of modern SIP phones and better-than-POTS
communications.

Using non-E.164 numbers is another solution.  Numbers would be
administered by campuses, not an external entity.  Regulatory issues
may be less onerous than with E.164 and users would notice a change in
dialing.  Integration with E.164 could also happen in the future.
However, calling from legacy devices would require two-stage dialing
or escape codes.

Ben's opinion is that non-E.164 numbers are the preferable option.  He
provides three implementation options: IVR (interactive voice
response) gateway, domain redirection, and a central registrar of
hashed SIP.edu address registrations.  Campus dialing plans would have
to interpret escape codes for SIP.edu calling and addresses would have
to be structured in a specific manner.

The first option, an IVR gateway, sets up two stage dialing with the
gateway generating a SIP.edu INVITE.  This would be followed by some
sort of voice recognition or DTMF code to complete the call.  This
method would only work with voice clients.

The second option, the hash registry, would require that campuses
forward user agent registration to a central repository which would
generate an MD5 checksum of the SIP.edu address.  Collisions would be
a concern, but could be minimized.  The hash registrar could resolve
conflicts with some sort of intermediary software, or could simply
decline to resolve collisions at all.

This approach allows a caller to infer the SIP.edu number from the
email address (e.g. alphanumeric SIP.edu address), as the number would
be generated from the hash.  However, only registered user agents
could be reached and campus registrars would have to pass REGISTER
messages, which may not be supported by all campus SIP servers.
Additionally, numbers would be based on the hash and not on any
(potentially mnemonic) campus-assigned numbering scheme.

The third option is domain redirection. Each campus is assigned a
mnemonic prefix or postfix, and a central redirect server interprets
the prefix and redirects calls to the appropriate campus.  A drawback
would be the issuance of the prefixes - numbers based on school names
are inconsistent in length and some collisions would occur.  Conflicts
could be resolved by number length, by local campus numbering schemes,
or non-default prefixes.  Every campus could come up with its own
number scheme, based on existing extensions, username hashes, or
entirely new numbers.  All campus users/phones could be made
reachable.

Following Ben's presentation, discussion about the benefits and
drawbacks of the various numbering and implementation schemes occured
with no clear consensus reached by the group.  Dennis asks
participants to resume the discussion during the next call in order to
come up with a common approach, or more specifically to avoid multiple
incompatibly approaches to solving the same issue.  Participants who
are interested in framing the next discussion are asked to contact Ben
and Dennis via email off-list.

The next call will be Thursday, July 7th.

-- 
Ben Teitelbaum   http://people.internet2.edu/~ben/