Indoor Wi-Fi positioning?

[Andre Navarro]

What do you guys know about who's doing the wifi positioning thing?
like investing, or have actual implementations out there?

I asked one of the corelocation guys at apple if they were doing
something with this, and he hinted that they were working on it, but
obviously couldn't say anything. I know Nokia is doing something, and
I think Navteq as well.

I think it would be pretty valuable to match accurate indoor
positioning with accurate layouts/floorplans or whatever that's needed
about the building. Sorry if this obvious. I was just wondering if
this data can be wiki-fied and open?

Can the data be crowd-sourced? Could we simply abstract a major subset
of buildings and create templates for users to pick? They can create a
location, add rooms/areas/floors/etc and anybody can survey the wifi
spectrum? Then buildings can get an overall rating, or health from
everybody's readings. I'm not talking about very detailed floorplans,
just cookie-cutter templates of different types of buildings
subdivided in X floors / Y rooms, etc.

Ok this is a lot of work for the user, but somebody will do it. We
could at least get in each city big buildings like conference halls,
malls, hotels, office complex/campus, or even their own home (for the
cool factor). Health can then be corrected in real-time, so it would
be self-healing in events such as wifi AP rentals at a conference.

What the user gets in return is accurate hyper-local location data.
Even if it's as basic as "Section Ab" of Mall X, I think that would be
valuable?

I know someone mentioned in the API proposals a way to capture more
information about the wifi reading? let's just attach a bit more info
on there (the above crowd-sourced data). Are there any standards out
there on how to represent geo places in more detail?

If we could put all of this under a stupidly simple UI, or game
(something mainstreamable), then I think that would be really cool,
but please critique the shit out of this idea if you have time please.
Am I completely out of touch with what's going on?

--
Andre Navarro

[Jorge Silva]

On Jul 1, 6:19 pm, Andre Navarro <andre.nava...@gmail.com> wrote:
> What do you guys know about who's doing the wifi positioning thing?
> like investing, or have actual implementations out there?

We have been piloting some algorithms for the past year and a bit.
Here are the project urls:

http://scyp.atrc.utoronto.ca/projects/wips
http://scyp.atrc.utoronto.ca/projects/tagin

> I asked one of the corelocation guys at apple if they were doing
> something with this, and he hinted that they were working on it, but
> obviously couldn't say anything. I know Nokia is doing something, and
> I think Navteq as well.

The most promising commercial solution seems to be glopos:
http://www.glopos.com/site/ but they don't use WiFi, rather they send
custom packets to nearby cell towers and do some kind of time of
arrival-based triangulation.

> I think it would be pretty valuable to match accurate indoor
> positioning with accurate layouts/floorplans or whatever that's needed
> about the building. Sorry if this obvious. I was just wondering if
> this data can be wiki-fied and open?
>
> Can the data be crowd-sourced? Could we simply abstract a major subset
> of buildings and create templates for users to pick? They can create a
> location, add rooms/areas/floors/etc and anybody can survey the wifi
> spectrum? Then buildings can get an overall rating, or health from
> everybody's readings. I'm not talking about very detailed floorplans,
> just cookie-cutter templates of different types of buildings
> subdivided in X floors / Y rooms, etc.

We've had a few attempts at crowd-sourcing text-only data:
http://scyp.atrc.utoronto.ca/projects/tagin/demos but the data barely
trickles in. Indoor data is really difficult to gather/use because
there are no absolute references like geo-coordinates that we can
refer to, we've mostly gone with human-readable tags (and floorplan
pixel coordinates with WIPS). One of the issues we face is that there
is no incentive for people to fill in all the info for nothing in
return. But maybe you guys have better ideas, we really haven't found
the right model yet.

> Ok this is a lot of work for the user, but somebody will do it. We
> could at least get in each city big buildings like conference halls,
> malls, hotels, office complex/campus, or even their own home (for the
> cool factor). Health can then be corrected in real-time, so it would
> be self-healing in events such as wifi AP rentals at a conference.

Funny you mention conferences... organizers are looking for indoor
navigation solutions badly. We've been contacted by airports too. From
their feedback we think that instead of crowd-sourcing, we'd be better
off creating an engine that admins can use to implement their own
local positioning solution. This would include an application that
allows them to add their own floorplans and do their own "warwalking"
to attach meaningful data to each location of interest. That is our
focus now and we are starting with an Android app:
https://code.launchpad.net/~jorge-silva/wips/beta (still early code).

> What the user gets in return is accurate hyper-local location data.
> Even if it's as basic as "Section Ab" of Mall X, I think that would be
> valuable?

That is what we thought, but most people already know where they are,
the value comes with knowing where to go, and without a global
reference such as GPS, indoor positioning really only becomes
meaningful within a single building or venue, there doesn't seem to be
a need for a global indoor positioning DB, but for a tool that can be
used to implement local positioning.

> I know someone mentioned in the API proposals a way to capture more
> information about the wifi reading? let's just attach a bit more info
> on there (the above crowd-sourced data). Are there any standards out
> there on how to represent geo places in more detail?

Yes, I think there was a suggestions to add RSSI (received signal
strength) which is what we use to do indoor location. In terms of meta-
data, the XMMP presence protocol seems the most complete of the open
standards: http://xmpp.org/extensions/xep-0080.html

> If we could put all of this under a stupidly simple UI, or game
> (something mainstreamable), then I think that would be really cool,
> but please critique the shit out of this idea if you have time please.
> Am I completely out of touch with what's going on?

You are definitely in the right track... I am sharing our experiences
on the same line of thought with our couple of attempts. Our
conclusion is that we think making a tool to let ppl do their own
local indoor positioning is the better way to go, but we would
definitely be open to contribute a more crowd-sourced approach to
populating an open DB. That was our original goal, it just doesn't
seem to be the best way forward. Hope this helps!

Jorge

> --
> Andre Navarro