[Bluez-devel] multi-frequency scanning & US $2,000 'bounty'

[Bluez-devel] multi-frequency scanning & US $2,000 'bounty'

[Paul Huber]

[-- Attachment #1.1: Type: text/plain, Size: 3895 bytes --]

Thanks for all the technical input.  I'm an experienced C coder, but don't
know much about bluetooth technical stuff.  So a lot of this went over my
head.  I'm kind of busy to learn everything I'd need to know, and don't want
to leach off the bluez devs -- I'm a coder myself, so I know the value of
everyone's time.  I have a small budget of $2,000 already approved for me to
solve this issue.  If that's not enough, let me know, and I can try to get
more approved.  Since I'm really busy with another project, I'll offer it as
a bounty if someone can provide me a solution so I don't need to invest the
time to learn all this stuff.  I was going to try to do it myself for that
amount that I quoted the client, but I realize there are too many things I
don't know, like the acceptable level of radiation, that it's better for me
just to pass it on.

First, regarding the concerns previously stated in the thread: Don't worry
about 'spamming'; the customer has to explicitly sign up online first.
There's also no concern about getting the Bluetooth logo on the box; this is
a closed system for a particular retail customer.  It's a black box
solution.  Also, there's no concern about needing to actually 'pair' with
the device that fast and transmit data; we just need to quickly detect the
phones as they pass through.  And any mods to bluez will be released under
GPL.

I'm under NDA so I can't give out all the details on the business model.
But I'll offer the bounty to anyone that can offer me the following
solution:

1)  Spec out the hardware: With a budget of no more than US $40 per unit in
quantities of 1,000, I need someone to say which USB Bluetooth dongle plus a
antenna can be used to detect mobile phones as they pass through an area
that is 6 meters x 6 meters.  I've ordered the following antenna already,
which may be all that's needed: http://www.hyperlinktech.com/web/hg2412p.php
I would send you the antenna, the dongle, etc., so you could confirm it all
works.  If it doesn't, and if you need to buy other dongles, antennas, or
other hardware to do the testing, I will pay for that in addition to the
bounty.  The hardware needs to be legal, such as not exceeding laws about
radiation, but doesn't need to be Bluetooth certified.

2)  Code the software: I need a piece of software that can detect a mobile
phone and report the bluetooth id and the mac address of the phone reliably
in 2 seconds or less when that phone is passing under the antenna at a speed
of up to 2.5 meters per second (ie a brisk walk) so that each phone will
always be detected every time as it passes through the coverage area.

The software app can just write the mac + bluetooth id to stdout as the
phones are detected.  That is sufficient for this, and I can do the rest.
I'd like the software to be in C or C++.  Ideally I'll probably use
the 2.4kernel for this since it seems lighter weight.  But I'd need
you to confirm
that the bluetooth stuff works sufficiently well on 2.4, and we can go to
2.6 if needed.  It has to be reliable and accurate and detect a standard
bluetooth phone that is not obscured by anything except perhaps fabric (ie
in a pocket) 99% of the time.

The deliverables are: 1) specify what usb dongle and antenna to use
(make/model/supplier), 2) code that compiles with 'make' and runs and
performs the tasks described.  I'd like to have the finished solution within
about 4 weeks.

If someone thinks that the requirements are doable in the time frame, I'll
get the client to provide the contract confirming payment when the
deliverables are met.  If it's not doable or will require more time/money,
let me know, and I'll try to get the client to approve more.  I told them I
thought $2k was reasonable, even though I don't really know all that's
involved.  If you e-mail me privately, I'll give you a phone number too so
you can call if you'd like to discuss.

Thanks.

[-- Attachment #1.2: Type: text/html, Size: 4326 bytes --]

[-- Attachment #2: Type: text/plain, Size: 314 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/

[-- Attachment #3: Type: text/plain, Size: 164 bytes --]

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] multi-frequency scanning & US $2,000 'bounty'

[Marcel Holtmann]

Hi Paul,

> First, regarding the concerns previously stated in the thread: Don't
> worry about 'spamming'; the customer has to explicitly sign up online
> first.  There's also no concern about getting the Bluetooth logo on
> the box; this is a closed system for a particular retail customer.
> It's a black box solution.  Also, there's no concern about needing to
> actually 'pair' with the device that fast and transmit data; we just
> need to quickly detect the phones as they pass through.  And any mods
> to bluez will be released under GPL. 

since the system is using Bluetooth, you have to get the Bluetooth logo
on your box. Otherwise you are not allowed to sell it. You really wanna
look into the Bluetooth IP license stuff that comes along with the
Bluetooth logo. You should not ignore this matter.

And just to be clear. Not only the modifications to BlueZ, all programs
liking in libbluetooth must be made GPL. The Bluetooth library itself is
GPL and thus your programs must be licensed under GPL, too.

> 2)  Code the software: I need a piece of software that can detect a
> mobile phone and report the bluetooth id and the mac address of the
> phone reliably in 2 seconds or less when that phone is passing under
> the antenna at a speed of up to 2.5 meters per second (ie a brisk
> walk) so that each phone will always be detected every time as it
> passes through the coverage area.

If you mean class of device with the bluetooth id, then the D-Bus signal
RemoteDeviceUpdated gives you exactly that information.

> The software app can just write the mac + bluetooth id to stdout as
> the phones are detected.  That is sufficient for this, and I can do
> the rest.  I'd like the software to be in C or C++.  Ideally I'll
> probably use the 2.4 kernel for this since it seems lighter weight.
> But I'd need you to confirm that the bluetooth stuff works
> sufficiently well on 2.4, and we can go to 2.6 if needed.  It has to
> be reliable and accurate and detect a standard bluetooth phone that is
> not obscured by anything except perhaps fabric (ie in a pocket) 99% of
> the time. 

I don't know who comes up with this crap that a 2.4 kernel is more
lightweight. Because the source code is smaller? Yeah, that is a pretty
bad indicator. The kernel to use is 2.6.23 and there is no real reason
to use any older kernel.

> The deliverables are: 1) specify what usb dongle and antenna to use
> (make/model/supplier), 2) code that compiles with 'make' and runs and
> performs the tasks described.  I'd like to have the finished solution
> within about 4 weeks. 
> 
> If someone thinks that the requirements are doable in the time frame,
> I'll get the client to provide the contract confirming payment when
> the deliverables are met.  If it's not doable or will require more
> time/money, let me know, and I'll try to get the client to approve
> more.  I told them I thought $2k was reasonable, even though I don't
> really know all that's involved.  If you e-mail me privately, I'll
> give you a phone number too so you can call if you'd like to discuss. 

Just a small piece of advise for everybody who is thinking about taking
this bounty. Don't post to the mailing list a request on how to do this
and expect us to write the code for you. Get it done by yourself since
it is actually pretty simple.

Regards

Marcel



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] multi-frequency scanning & US $2,000 'bounty'

[Paul Huber]

[-- Attachment #1.1: Type: text/plain, Size: 3552 bytes --]

>> all programs liking in libbluetooth must be made GPL

That's no problem.  We planned that already anyway.

>> Just a small piece of advise for everybody who is thinking about taking
this bounty...

Don't take this the wrong way.  I'm not trying to bribe anyone or anything.
It's just that I'm getting paid to provide a solution, and it doesn't seem
right for me to keep pestering the devs on the list with 'how do I?'
questions, and I could waste a lot of time.  I just figured it was better to
be generous and pass it along to anybody who has the know-how to do it right
and do it fast.

>> you have to get the Bluetooth logo on the box

The customer is buying a few hundred off-the-shelf cheap PC's, loading my
software, and then separately they're buying some USB Bluetooth dongles to
stick in the USB port.  They're not then re-selling it.  I wasn't expecting
any issues with Bluetooth licensing because if I go to the store and buy a
PC and separately buy a bluetooth dongle I don't need to put the logo on the
box.  Since my customer is just doing the same thing, I didn't think there
was an issue.  They're not selling it as a Bluetooth-enabled solution, and
they're not reselling the PC + dongle together.  To your knowledge this is
ok, right?

>> using an internal antenna

Previously I've done another unrelated application with mobile phones and a
normal usb bluetooth dongle and tried to demo it at a few trade shows, using
the 'piecemeal' discovery.  It worked fine in the office, but every time the
show got busy with hundreds of phones all around, the scanning slowed to a
crawl.  It took several minutes to detect a phone.  Running hcitool scan
would also lock up for several minutes.  It seems the generic bluetooth usb
dongle was picking up too many phones in a broad area or something.  I never
dug into the real reason, and just wrote this off as a problem with crowded
areas.  But for this application it will be used in crowded areas, so that's
why I wanted to focus the scan area tightly so there's only a few phones in
the scan area at once.  Unfortunately this is really tough to test unless I
get a few hundred phones.  So I was hoping that if I got an antenna where
the signal dropped off sharply outside the detection zone then it would
work.

>> D-Link DBT-120  worked well

Yes, that's what I used in the previous trade show application

>> The phones are gonna be in visible mode?

Yes

>> Scanning times can't be less than 4 secs, but that doesn't mean that the
phone has to be there for 4 secs, with one or two secs is more than enough,
but scanning will not return results in less than that time.

Can the phones be reported as soon as they're detected, so you don't have to
wait the full 4 seconds?  The ideal goal is that people can briskly walk
through this 6 meter area (yes, understand it's a circle not a square) and
be detected before the leave the area.

>> 2.4 is not lighter weight than 2.6

Okay.  We can use 2.6.  I know that around 2.6.15 usb and some other things
were badly broken, and after talking around to some devs on kernel org the
consensus was that 2.4 was more stable and simpler because the multi-tasking
architecture was simpler, and that mission-critical, robust appliances
usually used 2.4.  But, that was a while ago, and if bluetooth doesn't work
well in 2.4, then we'll just use 2.6.

>> Are you planning to use any embedded device running linux? Or just a pc?

It's just a PC in that's a basic X86 box.  But it will be a really cheap
one, like a 500Mhz AMD Geode, or maybe a Via C3.

[-- Attachment #1.2: Type: text/html, Size: 3979 bytes --]

[-- Attachment #2: Type: text/plain, Size: 314 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/

[-- Attachment #3: Type: text/plain, Size: 164 bytes --]

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] multi-frequency scanning & US $2,000 'bounty'

[Christoph Brill]

Am Mittwoch, den 31.10.2007, 19:56 -0800 schrieb Paul Huber:
> 1)  Spec out the hardware: With a budget of no more than US $40 per
> unit in quantities of 1,000, I need someone to say which USB Bluetooth
> dongle plus a antenna can be used to detect mobile phones as they pass
> through an area that is 6 meters x 6 meters.  I've ordered the
> following antenna already, which may be all that's needed:
> http://www.hyperlinktech.com/web/hg2412p.php   I would send you the
> antenna, the dongle, etc., so you could confirm it all works.  If it
> doesn't, and if you need to buy other dongles, antennas, or other
> hardware to do the testing, I will pay for that in addition to the
> bounty.  The hardware needs to be legal, such as not exceeding laws
> about radiation, but doesn't need to be Bluetooth certified. 

I think most if the dongles are able to to that. I used a cheap Anycom
dongle (~10 euros) that is able to detect mobile devices within an area
of 10 meters and even through walls. Reply times were very small when
the devices could see each other.

I did a similar software recently and it is able to find passing by
devices. The trick behind was to discover devices and cancel the
discovery after 3 to 5 seconds. This way you only get the "near ones".
After that immediately restart the discovery. The Anycom dongle was able
to handle this even with an internal antenna.

Maybe (ok, I'm sure!) there are better solutions out there (like
periodic discovery) but this one worked really fine.

> The software app can just write the mac + bluetooth id to stdout as
> the phones are detected.  That is sufficient for this, and I can do
> the rest.  I'd like the software to be in C or C++.  Ideally I'll
> probably use the 2.4 kernel for this since it seems lighter weight.
> But I'd need you to confirm that the bluetooth stuff works
> sufficiently well on 2.4, and we can go to 2.6 if needed.  It has to
> be reliable and accurate and detect a standard bluetooth phone that is
> not obscured by anything except perhaps fabric (ie in a pocket) 99% of
> the time.

I think using 2.4 because it's "lightweight" is nonesense. I use a 200
MHz epatec eTC-2300[1] with a 2.6.23 that is really stripped down. The
box is up in ~10 seconds and responses are great. It's even able to
handle the bluetooth 2 dongle.

[1] http://store.epatec.net/de/product_info.php?cPath=1&products_id=42


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel