[OpenTRV-dev] Mobile phone detector

Thu Nov 5 12:51:46 GMT 2015

Deniz:

The feedback from a (Bluetooth) engineer at the Cambridge meetup was that for both power and security reasons WiFi and BT chatter a lot less from idle phones than they used to, contrary to some of my other sources.

Which is why we’re on this track.  But I’d be happy to be wrong!

Rgds

Damon

> On 5 Nov 2015, at 12:47, Deniz Erbilgin <deniz.erbilgin at gmail.com> wrote:
> 
> Thanks Kevin,
> 
> I have a couple of extra questions:
> - Do you have any idea what the most popular bands are in London?
> - The higher frequencies you mention have fairly small wavelengths. How feasible would it be to use a directional antenna to discriminate between, say mobiles directly under the bus shelter and those nearby/the cell site?
> - As far as I'm aware, a lot people walk around with their wifi left on. Would it be more practical have a similar system checking for wifi connections instead/as well?
> 
> I guess it would be worth going around bus stops with a spectrum analyzer and a couple of different antennas...
> 
> Regards,
> 
> Deniz
> 
> On 05/11/15 11:50, Damon Hart-Davis wrote:
>> Hi,
>> 
>> Thanks for that analysis!  Wow!
>> 
>> Key points to note: our initial solution does not have to be perfect nor work at every shelter (nor all of the time).
>> 
>> I would be delighted if we were right on occupancy 50% of the time; combining the outputs of multiple sensors in a sensible way is part of our aim.
>> 
>> We also can in principle record typical sensor output levels over time to set a noise floor or even to disable a sensor entirely if it hears loud continuous chatter from a nearby base station!
>> 
>> We can also choose only to fire up the cellular detector if other sensors are currently drawing a blank, ie we can selectively throw more energy at some sensors dynamically.
>> 
>> ***
>> 
>> So, what is the simplest thing that could we do to have a (say) 50% chance of catching an average Londoner’s mobile chatting to the base station, initially ignoring energy consumption?
>> 
>> Could we say have a big rectenna that we put in the seat and is even powered by the energy from the phone itself and drives a GPIO or more?
>> 
>> Next option up we have a nice current sense op amp in our valve motor driver circuit with VERY low quiescent current which can give a decent boost on a voltage from the rectenna without killing us.
>> 
>> Rgds
>> 
>> Damon
>> 
>> 
>>> On 5 Nov 2015, at 10:21, Kevin Wood <kevin at the-wood-family.com> wrote:
>>> 
>>> Hi All,
>>> 
>>> As promised, a few thoughts:
>>> 
>>> Mobile phones are also battery critical devices that like to sleep as
>>> often as possible. Making 2 such devices detect each other is always going
>>> to be a challenge!
>>> 
>>> We have quite a few technologies operating at a few different frequencies
>>> in the UK:
>>> 
>>> GSM/GPRS:
>>> 880 - 915 MHz
>>> 1710 - 1785 MHz
>>> 
>>> WCDMA
>>> 1920 - 1980 MHz
>>> Some in the GSM bands too
>>> 
>>> LTE
>>> 832 – 862 MHz
>>> 1710 – 1785 MHz
>>> 2500 – 2570 MHz
>>> 3400 - 3800 MHz (TDD)
>>> 
>>> The above are uplink frequencies (mobile TX) but there will be adjacent
>>> downlink bands where cell sites will be constantly transmitting. Places
>>> like bus stops are a popular location for cell sites, so discriminating
>>> between them and the mobile stations might be a challenge.
>>> 
>>> With TDD network configurations, cell site transmissions are on the same
>>> channel as mobile transmissions!
>>> 
>>> What is transmitted is too complex to even try to demodulate, so a basic
>>> rf detector is about all you can do to detect presence.
>>> 
>>> Such a simple detector is prone to false triggering from any other radio
>>> signal it receives. About the only defence we have is to make it frequency
>>> selective around one or more of the above bands.
>>> 
>>> During use (speech call or internet activity) the mobile will regularly
>>> transmit. When idle, the timer that determines when it "phones home" is
>>> set in 6 minute intervals, and a typical value would probably result in a
>>> timer running for an hour or two between updates, so an idle phone is not
>>> easy to detect! Especially if you don't intend to sample 100% of the time.
>>> 
>>> If you're looking for a reasonably strong field strength, the detector
>>> needn't consume any battery power. A simple diode detector fed from a
>>> resonant antenna and tuned circuit could just wake the CPU using an
>>> analogue comparator channel when it "sees" RF.
>>> 
>>> But! Another issue is that power control of a mobile's transmission is
>>> very tight, again, because battery life of mobile is critical, and also
>>> because, in WCDMA and LTE technologies, received signals for each device
>>> must be as close to equal strength as possible at the cell site in order
>>> that one mobile doesn't swamp the signal from others. For this reason, if
>>> a cell site is close, we will have a double whammy of a strong signal from
>>> the cell site to reject, and low signals from any mobiles transmitting to
>>> it.
>>> 
>>> That's it for this "brain dump". Hopefully there's some useful stuff there!
>>> 
>>> 
>>> Best Regards
>>> 
>>> 
>>> Kevin
>>> 
>>> 
>>>> Fab, thanks!
>>>> 
>>>> All as Deniz says, and with the circuitry to be capable of running on at
>>>> most tens of microwatts average (we might be able to sample for a few
>>>> seconds every few minutes) at ~2.4V to work nicely with some variant of
>>>> our V0p2 board running from 2xAA NiMH or other similar low-power
>>>> microcontroller.
>>>> 
>>>> (We could even harvest a small amount of power to inject back into
>>>> batteries or a supercap as a secondary consideration!)
>>>> 
>>>> Rgds
>>>> 
>>>> Damon
>>>> 
>>>> 
>>>>> On 8 Oct 2015, at 22:16, Kevin Wood <kevin at the-wood-family.com> wrote:
>>>>> 
>>>>> Hi guys,
>>>>> 
>>>>> Currently on holiday but my day job is developing systems to test mobile
>>>>> phones.
>>>>> 
>>>>> I'll give this some thought.
>>>>> 
>>>>> Kevin
>>>>> 
>>>>> Hi All,
>>>>>> Damon has asked me to do a bit of research into using a mobile phone
>>>>>> detector as a presence sensor for the bus shelters.
>>>>>> 
>>>>>> The basic idea is an antenna, and an amplifier/filter tuned to the
>>>>>> appropriate frequency. Additionally we'd need some kind of conditioning
>>>>>> to get the output into a microcontroller readable format.
>>>>>> 
>>>>>> Does anyone have any thoughts on how practical this would be for
>>>>>> sensing
>>>>>> whether there are people waiting?
>>>>>> 
>>>>>> Regards,
>>>>>> 
>>>>>> Deniz
>>>>>> 
>>>>>> P.S. A quick google for how the detectors work gave me the following
>>>>>> results:
>>>>>> http://www.eeweb.com/project/circuit_projects/cell-phone-detector
>>>>>> http://www.electroschematics.com/1035/mobile-bug-detector-sniffer/
>>>>>> http://www.instructables.com/id/Free-Energy-Cellphone-detector-From-Cellphone-An/
>>>>>> http://www.instructables.com/id/VHF-UHF-RF-Sniffer/?ALLSTEPS
>>>>>> _______________________________________________
>>>>>> OpenTRV-dev mailing list
>>>>>> OpenTRV-dev at lists.opentrv.org.uk
>>>>>> http://lists.opentrv.org.uk/listinfo/opentrv-dev
>>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>> 
>>> _______________________________________________
>>> OpenTRV-dev mailing list
>>> OpenTRV-dev at lists.opentrv.org.uk
>>> http://lists.opentrv.org.uk/listinfo/opentrv-dev
>> _______________________________________________
>> OpenTRV-dev mailing list
>> OpenTRV-dev at lists.opentrv.org.uk
>> http://lists.opentrv.org.uk/listinfo/opentrv-dev
> 
> _______________________________________________
> OpenTRV-dev mailing list
> OpenTRV-dev at lists.opentrv.org.uk
> http://lists.opentrv.org.uk/listinfo/opentrv-dev