Brian, Thanks for your quick response. I may have some more questions later, but let's see how far we get now.
On 12/11/13 3:14 PM, Rosen, Brian wrote: > In most regulatory environments, the device that has the radio in the shared > spectrum must contact the database. You could imagine having a whitespace > radio in a next get cell phone, but generally the answer to the question is > no. Having said that, a database could, if it wanted to, grant anyone access > to the database. ok > >> Also, have we liaised this document to the IEEE 802.11 people and >> do/would they care (see below about licensed/unlicensed)? > We have connected with the 802.22 folks, not 802.11. ok > > >> >> >> 4.1. Database Discovery (and elsewhere): >> >> >>> 4.1. Database Discovery >>> >>> Different regulators may have different requirements for the approval >>> and operation of databases, such as: >>> >>> o A regulator may only allow a limited number of certified databases >>> to operate. It also may require the certification of each device- >>> to-database pairing. >>> o A regulator may maintain a trusted website that lists all approved >>> databases, known as the Listing Server. It also may mandate how >>> devices use the listing server. >>> >>> o A regulator may allow each database to define its own terms of >>> use, so that, for example, an approved device may not be able to >>> access all approved databases. >>> >> The basis of this document is that the regulators are in direct contact >> with the Master Device. > No. The database is in contact with the device, not the regulators. The > regulators authorize the database, have access to it, but do not operate or > manipulate it directly. ok > >> I have some concerns about that approach: >> >> 1. It would seem to me computationally intensive on the part of the >> regulator to compute a geometry/contour to determine which frequencies >> are available. As much as I would be worried about a single request, I >> don't see why regulators would want to take such aggregate load. Has a >> regulator agreed to do so? Or am I wrong about the computation? > The database does the computation. The contour calculations can be done in > advance, but would have to change as protected primary users change their > use. The database operation locates the device within the set of > precalculated protected contours to determine which ones reduce the spectrum > available to the whitespace device. Ok, so the issue here is the data capacity requirements, in terms of why the calculation is done by the database and not by the (master) Device? > >> 2. Much of this seems to be predicated on micro-auctions in secondary >> spectrum. Interesting idea and worth pursuing, however, the one week >> period below seems to lead us to question what "micro" means in this >> context. See below for that > You are incorrect. All secondary users (whitespace devices) in the same > location get the same spectrum availability. There is no mechanism to > allocate or share that available spectrum. There is some desire to provide > that kind of service, but it is beyond our charter. Why then provide DeviceOwner if all secondary users are treated the same? That's how I leapt to my conclusion. Also see below. > >> . >> >> 3. I may be demonstrating my parochial nature, but there is a lot of >> complexity in this document that is predicated on the assumption that >> licensed use will be permitted. That is- a specific station license is >> required to transmit. Wouldn't this protocol simplify considerably if >> it focused on unlicensed use? In fact, wouldn't it boil down to a file >> transfer protocol and a database format as opposed to bouncing all of >> these queries back and forth? My question really is how the authors >> came to this approach. I can imagine some arguments, such as wanting to >> track down bad actors, but I raise the issue because of the plethora of >> PII that could end up getting shipped to regulators on a regular basis >> over this. > You have an incorrect model. > There is a set of licensed, protected users. Right. > The database knows about them, and calculates contours to assure their use > is protected. There is a set of unlicensed, but type-approved whitespace > devices (which includes masters and slaves). Sure. > They are secondary users - they can use spectrum if none of the primary, > licensed users use it. Right. > The secondary, unlicensed users ask the database what spectrum is available > at their location at that time. Ok, and see above about my question as to where the calculation takes place. > The database calculates which primary users would be affected by a > secondary transmitter at that location and remove the primary user’s spectrum > from the available secondary spectrum. The database returns what is left > (what spectrum is not used by any primary user at that location at that time). ok > > All of this is provided in a regulated environment. The regulator decides > what the protections are, and what the unlicensed devices have to do to use > the available spectrum. The protocol is very well generalized. But my leap of logic was again due to DeviceOwner. > >> >> Section 4.1.1: >> >>> Within a regulatory domain that has a Database Listing Server, a >>> Device MUST use it to determine the URLs of databases for the domain. >>> The URI of the Listing Server for a regulatory domain MAY be >>> preconfigured in the device. Where allowed by the regulator, the >>> Device MAY save the database list and SHOULD contact the Database >>> Listing Server periodically to update its list. The time between >>> such updates MUST be no longer than one week, or any update interval >>> required by the applicable regulatory domain, whichever is shorter. >> I presume that in this paragraph, when you say "Device" you mean "Master >> Device". going further, it seems to me that the point should not be >> whether or not such a device has direct Internet connectivity, but >> whether or not it can receive updates. Those updates could conceivably >> come from many sources, including ones in which IP is not used. > This section is referring to the process by which the device (a master device > if there are master/slave devices) locates the database. Not every use has > slaves. In some countries, there is a listing service of multiple competing > database operators. Only a database on that list may be used. This is the > control the regulator has on de-authorizing databases. ok > > The PAWS protocol is based on IP transports. We don’t consider models where > there are alternatives to IP to access the listing service or the database. ok > >> Moreover, there seems to have been an arbitrary period of time chosen (1 >> week). Why not state that in terms of the EventTime in AVAIL_SPECTRUM_RESP? > Again, this is the frequency at which the listing service must be requeried, > not the database. The frequency is normally set by the regulatory domain, > and devices have to be built for one or more regulatory domains. It’s much > less often than the spectrum query. This basically determines how quickly a > regulator can de-authorize a database. Got it. Question: what happens if the database that authorized a device to use a frequency with some parameters is de-authorized before EventTime has expired? Must the device submit a new request? Is that something that would be contained within a RuleSet (e.g., external)? Thanks, Eliot _______________________________________________ paws mailing list [email protected] https://www.ietf.org/mailman/listinfo/paws
