[Namaste-dev] Physical Downlink Paper Feedback

Michelle w5nyv at yahoo.com
Fri Jun 13 18:13:49 PDT 2008

This feedback is for the paper at:
The document needs a name, revision, and date to better help identify who wrote it and when.
There seems, to me, to be a conflict between the document and an emailed explanation to the list.
In the document, 
"One of the key aspects which defines the ACP is that there is a single downlink stream which will be...
required to service users with widely varying capabilities, the most important of which for this
discussion is antenna gain. Initial discussions have suggested a spread of as much as 20dB between
well-equipped fixed stations with a moderately sized dish, and smaller mobile ones with something
more like a patch or helix. It would be very easy, but also very wasteful, to simply send all bits with
enough energy such that the weakest stations could always receive them. Instead, we need to find a
way to impart as much as 20dB per bit of extra energy to those bits which are destined for the weaker
stations, and this represents most of the challenge for the downlink PHY design.
This (varying EbN0) allows the link can be designed for the highest capability users, and throttled back for the lower
rate users. By allowing for a variable number of repetitions of each block of data, we can smoothly
vary the rate in accordance with the actual capabilities of the weaker stations. We could even take into
account the current utilization level of the satellite to smoothly trade off robustness vs. capacity. When
the system is lightly loaded, everything might be repeated. When utilization is heavy, we could reduce
the repetition factor to accommodate more users. The only real cost to this system is that synchronization 
and the information about what is a repetition of previous data need to be robust enough for the weaker 
stations to be able to track."
Later on, Matt wrote an email (May 29th)
"When I talked about a continuum of data rates depending on the size of 
the receiver, I didn't really mean that anybody could use any antenna 
and the satellite would send at the appropriate rate.  What I really 
meant was that we could set that repetition rate to whatever was 
necessary for the standard small antennas that you guys shipped, and 
this was a variable we didn't need to decide on early in the process."
Ok, I think I need some help understanding. When I read the document, and then read the email, the two descriptions don't seem to agree. "In accordance to the actual capabilities of the weaker stations", as well as the repeated term "weaker stations", to me, means (first) that there are weaker microwave stations (which I think were ruled out at the 2006 San Diego meeting?), and (second) that the link changes based on the capabilities of the stations in the system in real time, in real use. 
Are the classes of stations therefore only a development strategy? Is there no plan to dynamically allocate bandwidth and/or coding gain to support a variety of types of stations? There's an a priori problemo in here that I think Paul KB5MU already brought up, as well as the problem of when a station identified as weaker gets a new antenna plugged in and is suddenly in the black, so to speak, with respect to the link. Station capability isn't fixed, in other words, since many people upgrade whenever and whatever they can. 
Continuing with other feedback received from others:
"I'd expect usage to follow local time, so the peakedness of usage would
likely follow geographical user density. Practically, this likely means peak
usage will move across the United States.

In peak usage times, both low capability and high capability stations will
be in use, probably in a mix following the percentage of the type stations
in use.  If low capability stations are less expensive, they may form a
serious majority.

Ignoring admission controls, the possibility that low capability stations
will fully dominate the satellite's capacity might should be modeled and
How does the system assign resources to a mix of low and high capability stations in congestion?
Some comments about congestion control can be read here:
These are Namaste discussions from March and assumes that we intend to support stations of varying capability.

Feedback continues:
"Maybe this can be done with some sort of variable "frame
boundary," where some quanity of timeslots are reserved for different
capability users?

It's difficult to analyze the downlink without considering the uplink
structures.  I guess this comes later. 

If this is a true "router in the sky," then some interesting things may
happen with a mix of user capabilities and retransmissions.  Some smart
person could probably model this.  Most data transmission systems seem to
start falling apart at about 66% of maximum capacity.  The added latency due
to hop time on retransmissions may be an issue.

I feel some control capability may be necessary to handle variable traffic
levels, control admission, and so on.

I recall, and my references are at the office, that some of the "XPSK"
modulations are not true constant envelope as some of the state transitions
pass thru zero amplitude. E.g., TDMA cellular amps were linear amps."
more soon!
-Michelle W5NYV

More information about the Namaste-dev mailing list