Re: [Geopriv] Some comments on draft-thomson-geopriv-uncertainty-08

[James Winterbottom <[e-mail]>]

Okay.

Sent from my iPad

On 09/11/2013, at 6:39 AM, Brian Rosen <[e-mail]> wrote:

> Martin and I had a chat and we agreed upon a solution.  I’ll write a draft, which will be very short.
> What we’re going to propose is an extension to LoST that add a flag to a geo that explicitly states that the area represented by the PIDF represents the area of uncertainty of location in one state, or that the area represents the target with unknown uncertainty in the other state.  If the flag is not present, the assumption is that the sender is clueless the location is that of the target with unknown certainty.
> 
> This makes it clear that an area representation of location either does, or does not specify uncertainty.
> 
> Brian
> 
> On Nov 8, 2013, at 11:15 AM, James Winterbottom <[e-mail]> wrote:
> 
>> Sure you can.
>> We have an explicit radius parameter for circle, you may set this to zero.
>> We also allow,the definition of just a point.
>> I think that you will find that the constructs between MLP and GeoShape are almost directly interchangeable.
>> 
>> Cheers
>> James
>> 
>> Sent from my iPad
>> 
>> On 09/11/2013, at 4:33 AM, Brian Rosen <[e-mail]> wrote:
>> 
>>> Not claiming I know more than you, but I think E2 and MLP report position AND they report uncertainty by describing the position as a point, and the uncertainty as the radius of a circle or ellipse.  They don’t just sent a circle.  It’s point plus circular uncertainty, not just a circle.  In PIDF, you can’t tell the difference between a location reported as a circle with no uncertainty specified from a point with circular uncertainty.
>>> 
>>> The definition of the reported position is the lat/lon of the position and the radius of uncertainty.  We have no similar way to report uncertainty in PIDF.
>>> 
>>> Brian
>>> 
>>> On Nov 7, 2013, at 2:00 PM, James Winterbottom <[e-mail]> wrote:
>>> 
>>>> Hi Brian,
>>>> 
>>>> MLP and E2 are essentially the same as what we use in Geopriv
>>>> 
>>>> Sent from my iPad
>>>> 
>>>> On 08/11/2013, at 8:49 AM, Brian Rosen <[e-mail]> wrote:
>>>> 
>>>>> Sure:
>>>>> 
>>>>> When you get a location from a GPS receiver, it typically uses the NMEA interface, which reports a lat/lon and a Dilution of Precision method of specifying uncertainty.
>>>>> MLP usually reports location as point with circle, ellipse or arc uncertainty
>>>>> E2 usually reports location as point with uncertainty
>>>>> 
>>>>> Brian
>>>>> 
>>>>> On Nov 7, 2013, at 1:02 PM, James Winterbottom <[e-mail]> wrote:
>>>>> 
>>>>>> Hi Brian,
>>>>>> 
>>>>>> Can you give an example of point 2 please?
>>>>>> I guess all the systems that I am familiar with use the area to express the uncertainty so I am keen to understand how it might be done differently.
>>>>>> 
>>>>>> Cheers
>>>>>> James
>>>>>> 
>>>>>> Sent from my iPad
>>>>>> 
>>>>>> On 08/11/2013, at 7:48 AM, Brian Rosen <[e-mail]> wrote:
>>>>>> 
>>>>>>> I’ve read this draft, and think it is an excellent explanation of the issues we face in geopriv with measurements.  I highly recommend that we adopt the draft and process it as a work group item.  I am particularly happy with the definitions it offers for the various quantities.
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> I find it lacking in two areas:
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> 1. I think the draft needs to discuss the issues of comparing measurements with different confidence.  I think the bottom line on that is: don’t do it, use one agreed upon confidence.
>>>>>>> 
>>>>>>> 2. I think we need to stop expressing measurements as just the uncertainty area.  While I understand why doing so is attractive, I think it is not what real systems do.  Real systems express a measurement and explicitly state the uncertainty of the measurement.  I think we need a way to express that.
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> Brian
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> Geopriv mailing list
>>>>>>> [e-mail]
>>>>>>> https://www.ietf.org/mailman/listinfo/geopriv
>>>>> 
>>> 
>