Hi Ben,

This is an important point. The protocol has a AVAIL_SPECTRUM_BATCH_REQ query, 
with which the master can request spectrum by specifying multiple locations, 
and the response to the request needs to specify the available channels for 
each location separately.
Now, assuming a master wants to operate in an area which covers 1000m. That 
means that it has to specify ~ (1000/50)^2=400 different locations in the 
AVAIL_SPECTRUM_BATCH_REQ. Is that feasible?
We discussed this issue last year, I was asking the protocol to include a query 
 type which would allow the master to specify a location and a radius of 
operation, and ask for available spectrum for the specified location, but folks 
in the meeting argued the master should receive the available spectrum for each 
location, then figure out itself the channels available for the given area.

-Gabor

From: [email protected] [mailto:[email protected]] On Behalf Of ext 
Benjamin A. Rolfe
Sent: Thursday, October 17, 2013 12:47 PM
Cc: [email protected]
Subject: Re: [paws] Support for including Slave Device location

There needs to be a way for a "master" (I truly dislike that term on many 
levels) to be able to request information for multiple locations on  behalf of 
other devices.  This may *not* be a problem with the protocol and may only be a 
problem of my understanding, but it is *not* safe to assume that all "slave" 
devices (I dislike 'slave' even more) are within a 50m sphere of the master 
device geolocation where the channel availability information is valid (per 
FCC).
A very likely use case is there is are internet connected devices at the "edge" 
of a network (fixed, mains powered) communicating with a number of devices 
without direct internet connection, or possibly a complete IP stack. Think M2M 
over low data rate links, with resource  constrained nodes operating from 
batteries and, possibly, moving about. They may be fixed with location 
per-configured at install or, more likely, using some means of radio location 
to determine a position relative to the internet-connected peer that can source 
channel information.  The internet-connected edge device may serve as a proxy 
source of channel availability information.
Such a use case is supported by FCC rules, and last time I looked OfCom and 
elsewhere as well.

It may be that the protocol already supports this via multiple requests made by 
the internet-connected node. In which case I apologize for the diversion.

Ben
On 10/17/2013 12:28 PM, Vincent Chen wrote:
Ben,

No. The Database only returns spectrum for the primary device descriptor.
The "master device descriptor" is for reference only when the primary device is 
a slave.

I.e., When the Master requests spectrum for itself, it will use the primary 
device descriptor.

-vince

On Thu, Oct 17, 2013 at 12:23 PM, Benjamin A. Rolfe 
<[email protected]<mailto:[email protected]>> wrote:
Then the database returns two channel lists, one for each location?
It is feasible that they would be different at least by FCC rules.
B

On 10/17/2013 11:57 AM, Don Joslyn wrote:
Yes, that makes sense to me.

From: Vincent Chen [mailto:[email protected]]
Sent: Thursday, October 17, 2013 2:53 PM
To: Don Joslyn
Cc: [email protected]<mailto:[email protected]>
Subject: Re: [paws] Support for including Slave Device location

I think, for consistency, the SPECTRUM_USE_NOTIFY should also have a "slave 
location".

In other words:
 - Whether or not the slave has geo-location capability, "location" continues 
to be the Master device location
 - For Slaves that have geo-location capability, the "slave location" would be 
included

Does that make sense?

-vince

On Thu, Oct 17, 2013 at 10:50 AM, Don Joslyn 
<[email protected]<mailto:[email protected]>> wrote:
Thanks Vince,

In addition, It's my current understanding that Ofcom requires slave devices to 
report "Channel Usage". In PAWS it would be accomplished via master device 
sending a SPECTRUM_USE_NOTIFY on behalf of the slave device. We might need to 
add slave device location to that message, or indicate that the location 
parameter contains the slave device's location whenever etsiEnDeviceCategory is 
equal to "slave". Does that make sense?
Don

From: [email protected]<mailto:[email protected]> 
[mailto:[email protected]<mailto:[email protected]>] On Behalf Of 
Vincent Chen
Sent: Thursday, October 17, 2013 1:26 PM
To: Benjamin A. Rolfe
Cc: [email protected]<mailto:[email protected]>
Subject: Re: [paws] Support for including Slave Device location

Thanks Don,

The in-progress draft has already changed the definition of "Slave" to match 
that in the use-case RFC, which does not reference geo-location capability.

Adding the optional slave location to the AVAIL_SPECTRUM_REQ seems to make 
sense.

Ben. There is already support in PAWS to include both the Slave and Master's 
device descriptors.

-vince



On Thu, Oct 17, 2013 at 9:29 AM, Benjamin A. Rolfe 
<[email protected]<mailto:[email protected]>> wrote:
There is a similar requirement, though not as explicitly stated, in the FCC use 
case.  A device not directly connected to the database works through a 
connected device. For the connected ("master" in OfCom terms) to provide the 
data  to another, it must verify that the other device is authorized.  This can 
be done by having the connected device make a request using the device 
identification information of the "slave".   I realize that I had *assumed* the 
protocol as drafted  supported  this, i.e. the device making the request  could 
fill  in the ID information of another device in the request.  IF this is not 
true, then the protocol does not support a very likely use case in the US.

FWIW.
Ben

On 10/17/2013 8:34 AM, Don Joslyn wrote:
After reviewing several Ofcom TVWS operational requirements documents, it is my 
current understanding that Ofcom operation in TVWS includes a use case where 
the slave device's location may be included in the available spectrum request 
sent via the master device to the database. It appears that the current PAWS 
protocol specification (version 6) does not support inclusion of the slave 
device's location as a parameter in requests, and furthermore the PAWS 
specification assumes by slave definition that slave devices are without 
geo-location capability.

To support Ofcom's use case that includes slave device location, I would like 
to suggest that we consider adding an optional parameter for "Slave Device 
Location", and update the slave definition to support slave devices that 
include geo-location capability. The new "Slave Device Location" parameter 
could be added directly to the AVAIL_SPECTRUM_REQ message format, or added via 
another ETSI-specific parameter.

Thank you,
Don




_______________________________________________

paws mailing list

[email protected]<mailto:[email protected]>

https://www.ietf.org/mailman/listinfo/paws


_______________________________________________
paws mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/paws



--
-vince



--
-vince



_______________________________________________

paws mailing list

[email protected]<mailto:[email protected]>

https://www.ietf.org/mailman/listinfo/paws


_______________________________________________
paws mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/paws



--
-vince

_______________________________________________
paws mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/paws

Reply via email to