INTERNET-DRAFT                                                John Mangione
                                                      Competitive Computing
Title:           GPS^IP
Expiration Date: 12-31-96
filename:        draft-mangione-ipv6-gps-alt-00.txt


Status of This Memo:


This document is an Internet-Draft.  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.''

To learn the current status of any Internet-Draft, please check the
`1id-abstracts.txt'' listing contained in the Internet-Drafts Shadow
Directories on ftp.is.co.za (Africa), nic.nordu.net (Europe),
munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or
ftp.isi.edu (US West Coast).



ABSTRACT

GPS^IP is a suggested adjunct or alternate addressing scheme to IP,
version 6.  All of the remaining suggested, proposed, and ratified
standards would remain as they are.  As its name implies, GPS^IP would
use Global Positioning System (GPS) information, perhaps in one of the
currently unassigned regions of the IPv6 address space.  This global
position information would be placed somewhere in the 128 bit IP
address, to provide a reasonable level of uniqueness.  The remainder of
the IP address could remain as currently described, both in IPv4 and
IPv6, as described in other documents.  Alternatively, a portion of the
low order end of the IP address field could be populated with the MAC
address as supplied by the interface.


PERCEIVED BENEFITS

The benefits of using geographical position information in the IP
address are as follows:

Address Uniqueness.  With the accuracy of GPS locating equipment
constantly improving, and the inclusion of altitude information in the
address, each node that occupies a unique physical space will, by
definition, have a unique "address".  Although geographical
latitude/longitude position alone would suffice to provide relative
device uniqueness and routing knowledge, it is important to include
altitude information in the case of multifloored buildings, or racks of
equipment with unique identities.  To guarantee node address uniqueness,
MAC address information could be inserted into the IP address, in the
same way low order portions of IPX addresses are generated.

Mobility.  A node's address would change as it moved, constantly
providing a more accurate routable address, since the most local "GP
server" would also provide the most appropriate "default gateway"
address for traffic to and from the mobile node.

Transparency.  Addresses would not have to be established for individual
nodes, since they would gather their addresses from equipment either in
the node itself, or from nearby GP servers.   Furthermore, if a portion
of the low order end of the IP address field were to be populated with
the MAC address as supplied by the machine's interface circuitry, this
would guarantee uniqueness, even for machines that were right next to
each other.

Universality.  This scheme could be used to provide meaningful
information, by way of an IP address, for all types of equipment,
including computer equipment, mobile/cellular phones, pagers,
satellites, vending machines, etc.

Backward Compatibility in routing methods. Current routing methodologies
(i.e., lookup tables, more authoritative default gateways) could still
be used.

The Potential For Greater Routing Efficiency.  Currently, there is no
obvious method to correlate geographical position with a best route
metric, without artificially devising methods that would create such a
correlation (such as, "All nodes should be connected to their nearest
geographical router", or "All nodes that do connect to their nearest
geographical router can use the geography metric for routing".  As the
idea ruminates in the minds of the brilliant masses on the Net, there
may come a time where routers may use lat-long-alt information to move
datagrams toward a destination by algorithmic methods, rather than
current lookup methods, for determining best route.  Currently, however,
this is NOT being stated as a perceived benefit of use geographical
information in IP addressing.

Device Location Awareness.  Perhaps the greatest benefit would be that
administrators and users of networks alike would be able to know where
devices were on their networks.  .

DISCUSSION

GPS^IP is a suggested adjunct or alternate addressing scheme to IP,
version 6, referred to in RFC-1884.  As its name implies, GPS^IP would
use Global Positioning System (GPS) information.  The GPS is a system of
several space-based satellites that provide constantly transmitted
location information to Earth.  GPS receivers on the Earth (or
elsewhere), with this transmitted information, and through the use of
triangulation, can determine their own location, as latitude, longitude
and altitude.  This information, to a greater or lesser degree, provides
accurate location information with sub-meter accuracy.  Since other
documents describe how this is achieved and how the degrees of accuracy
are managed, this document will merely assume that this information can
be made available to a node through some relatively inexpensive
circuitry which would include RF receiver capabilities. The circuitry
would be able to electronically produce its own location coordinates.
This information will be referred to (in this document) as a node's
global position (GP).


This global position information would be placed in some currently
unassigned portion of the 128 bit IP address, to provide a reasonable
level of uniqueness, which is a basic requirement of any transport
addressing scheme.  The remainder of the IP address could remain as
currently described, both in IPv4 and IPv6, as the four octet system.
This would be done to provide administratively managed uniqueness, and,
to a lesser degree, routing information to the IP address.
Alternatively, a portion of the low order end of the IP address field
could be populated with the MAC address as supplied by the interface for
the guarantee of uniqueness.

Within the network, certain nodes, such as routers, would have an
awareness of their geographic location.  This would be achieved by
equipping these nodes with inexpensive GPS devices that would
auto-locate.  Also, certain of these nodes would provide such location
information to other nodes that request it.  In effect, these nodes
would behave as Global Position Servers, "GP servers", or "GP daemons".
All "GP clients" would request and gather this information from GP
servers in their electronic horizon, or subnet, and use an algorithm,
such as quickest response time, to select the optimal default GP server.
The server that is chosen may or may not become the primary router, or
default gateway, for that client station. The router that would act as
the default gateway would become the destination router for all incoming
information to any nodes that are "registered with" that router.

Initially, only GP servers and routers would need to be configured,
either statically or dynamically, with accurate GP information. This
would be achieved statically, by using some hand-held lat-long-alt
device at the machine and manually entering it into the GP server, or
dynamically, by using a GPS location acquiring circuitry that would be
installed directly into the GP server, and would feed this information
directly and electronically to the system, perhaps through the BIOS,
like on-board clocks.

As time goes on, more client workstations would be outfitted with
position gathering circuitry, so that these nodes could also serve as
more local GP servers, and/or provide the actual lat-long-alt
information directly, for the internal use of the system to create its
own full IP address.

Not all routers would have to be GP aware at first. The core, or heart
of the network would be the devices that would have this information at
the outset.  Routers that were further away from the "backbone" would
not necessarily have Global positioning awareness. Those that weren't
would pad that portion of the address with zeros.  When the datagram
arrived at a GP aware router through standard routing methodologies, it
would be that router's responsibility to replace the appropriate zeros
with its own GP location coordinates. In this way, the GP portion of the
return address of the datagram would first become the nearest GP router
to the message originator.  Thus, the "heart" of the routing system
would be GP aware at first, with the more peripheral portions of the
Internet becoming aware through the attrition of older devices in favor
of GP aware newer technologies.


CURRENT ARGUMENTS AGAINST THIS IDEA:

"Geographical information cannot be used for routing purposes, since
wires and geography are two different mathematical systems."

Response:  Currently, there is no identifiable method of mapping
geographical information into an improved routing scheme than is
currently employed, but current routing schemes can still be used as
before, without alteration, so we are no worse off for having the
information there.  And, with geographical information is in the
address, at least the potential for the evolution of improved routing
algorithms based on geography is there.

"What about privacy?  I don't necessarily want anyone to know where I am
when I send a message."

Response:  A "defeat location information" can be included as a
configuration check box that users can manipulate for those holding this
concern.  Also, just as remailers are used, there would be similar ways
to ensure "locale privacy".

"There would be the cost overhead to include this equipment in
machines."

Response:  It would not have to go into all machines as it would be a
discretionary portion of the address. Over time, the machines that were
most in need of locating quickly would be the ones that would get the
additional hardware. Furthermore, with Global Position Servers able to
provide proximity information to other "clients", devices would be able
to provide general location information without adding equipment on a
device-by-device basis.

"There would be the addressing overhead to include global position
information in the actual addresses."

Response:  This overhead already exists, since the current proposal on
the table already accommodates 128 bits.  There is no less overhead
transmitting all zeros than in transmitting global positioning
information instead.


Why go through all the effort?

Currently, with the proliferation of devices on the Internet, as well as
the intranets, and with the addition of new types of devices, such as
vending machines and video cameras, as is currently being experimented
with on the Internet, keeping track of where these machines are will
become increasingly more challenging. With the addition of software that
would graphically display portions of the network in relation to a room,
a floor, a building, or a continent, administrators could find the
beaconing node, or the network enabled printer that is out of paper.
This could all be done by looking at a map on a computer monitor and
seeing the various devices in relation to the space that it occupies.
In a small network that is not reconfigured frequently, this information
may be trivial.  However, in networks of more than 100 nodes, a single
administrator or user would be hard-pressed to track all changes made to
the network, especially if that administrator or user were relatively
new to the network.  To know where the vending machine got moved to on
campus while someone was on vacation would be a very useful piece of
information, especially if that device were sending distress pages that
it needed servicing.




What do you think?


Author/Contact Information:

John Mangione
Competitive Computing
 Mountain View Drive
Suite 106
Colchester, Vermont 05446
voice: (802) 655-0757
fax:   (802) 655-6681
johnm@competitive.com