Logo

Satellite Internet Forum.
Please note that this forum is now READ ONLY and closed for login and changes
The new forum is here Space and Satellite Communications forum

Welcome, Guest.
Welcome to this satellite broadband discussion forum.  Wherever you are and whatever your problem we are here to help each other. Connecting to the internet via satellite is not always easy but is critically important to those in remote places or with poor terrestrial infrastructure.  Both service providers and customers are encouraged to contribute.  Read the Forum rules.
      Satellite internet forum          

Optimizing data rate on VSAT-link - Homework problems

(Read 4826 times)
jorel
Ex Member


Mar 23rd, 2014 at 1:15pm  
Hello all,

New to the forum, I'm a student in Sweden and currently styding satellites and having a problem. Trying 10 hours per day for week to get this right but I'm failing.

I have a homework to optimize a Point to Point VSAT-link (a european to an other european city) the sat is at 36000 km height so roughly 240ms delay up/down. All parameters except these are exchangable qoute: "VSAT link between city 1 and city 2 Transponder bandwidth is limited to the same as a QPSK modem at 4096 kb/s needs." And measured BER is 5x10E-6.

Carrier frequencies and bandwith is not defined. I have struggled with dealing with Eb/N0 and BER, flow control and interleave, and pondered time to time about the locked parameters ie if I need to estimate frequencies a QPSK modem at 4096 kb/s uses. At the end I want to plot this in a graph to show differences. Also I guess I need to change modulation schemes to optimize link further.

I want to use Selective repeate ARQ, and change modulation to HQPSK in the end but perhaps starting with QPSK to see if it's possible to get better data rate.

I hope there are somebody feeling like pinting me in the right direction, this is a distance course and I seems to have gotten lost in all the books (William Stalling wireless book).

Kind regards.
Back to top
 
 
IP Logged
 
Admin1
YaBB Administrator
*****
Offline


Personal text from: Profile,
Options, Top line

Posts: 844
Reply #1 - Mar 23rd, 2014 at 4:21pm  
I think they want you to work out if you can get a higher information bit rate at the same BER using some permutation of modulation method and FEC rate.

Find graphs or formulas that relate BER and Eb/No for various modulation methods.

Uncoded QPSK at BER 5x10E-6 needs 9.8 dB Eb/No.   "Uncoded" means no FEC is in use, which is very rare.

NB: Your starting transmission rate is 4096kbit/s.  Your starting information rate is 4096kbit/s.

Whatever you come up with must achieve the same BER 5x10E-6 with the same total energy.

It would be very useful to write out a table (use spreadsheet) of modulation methods (BPSK, QPSK, 8QAM, 16QAM, 32QAM, 64QAM etc), symbol rate, the bits per symbol, and the transmission bit rate for each, based on 4096 for QPSK.  In the column 'symbol rate' write 2048 in every box, all the way down.

The row for QPSK line will say symbol rate 2048, 2 bits per symbol, and transmission rate 4096 ksps.

Then add many extra columns (in sets of 3) on the right, headed
"Info rate at Ratio 0.95", "Ratio 0.95 avail Eb/No", "Ratio 0.95 required Eb/No",  "Info rate at Ratio 7/8 ", "Ratio 7/8 avail Eb/No", "Ratio 7/8 required Eb/No",  "Info rate at Ratio 3/4 ", "Ratio 3/4 avail Eb/No", "Ratio 3/4 required Eb/No",  "Info rate at Ratio 1/2 ", "Ratio 1/2 avail Eb/No", "Ratio 1/2 required Eb/No",

Fill these boxes with information rates, using the transmission rate and FEC ratios.
The available Eb/No is the 9.8 - 10log(info rate/4096).
The required Eb/No will need to come from graphs or equations or manufacturers data sheets according to modem performance.

Remember that allowed energy is constant, so if you increase the information rate to say 8192 your allowed Eb/No is down by 3 dB to 6.8 dB. Note that the "b" bit in Eb/No refers to information rate bits, not transmission rate bits.

Compare the Available and Required Eb/No.  Which just works and gives the best results?. That is your answer.

Example:

Try 16QAM and 7/8 FEC.  This requires Eb/No = 7.5 dB (from graphs or equations or manufacturers data sheets according to modem performance)
The transmission rate is 4 x 2048.  The information rate is 7/8 * 8192 =  7168 kBit/s.
How much Eb/No is available ?  9.8 - 10 log (7168 / 4096) = 9.8 -2.4 = 7.4 dB
Compare 7.4 and 7.5.   Wow !  that just works!  It was a good starting guess by me.

You need to try other combinations, BPSK and QPSK will produce lower bit rates.
16 QAM, 32QAM and 64QAM which will give your much higher transmission rates (10240 and 12288) are worth investigating.

There are many FEC methods to choose from.
 
This modem http://www.newtec.eu/backend/files/leaflet/MDM6000%20High%20Speed%20Satellite%20...
offers up to Modulation method 256APSK with FEC ratios ranging from 20/45 to 11/15.

http://www.newtec.eu/frontend/files/userfiles/files/DIALOG/Whitepaper%20DVB_S2X....

http://www.dvb.org/resources/public/standards/a83-2_dvb-s2x_den302307-2.pdf

Best regards, Eric.
Back to top
« Last Edit: Mar 23rd, 2014 at 5:36pm by Admin1 »  
WWW  
IP Logged
 
jorel
Ex Member


Reply #2 - Mar 24th, 2014 at 10:54am  
Hello Eric,

I can not thank you enough! Very kind of you to go through so much detail.

I started looking in to this after your reply and I now have a better grasping of how to solve this. I will do as you suggest and also plot this in a graph for clear view of the differences.

I just have one follow up question about the value "5*10E-6". I got in to a discussion yesterday about how this should be interpred, the annonced course litterature isn't expaining BER-calculations in great detail and online sources is easily overwhelming or leave much to ask for.

I assume that E-notation is the same as 'scientific E-notation' meaning times ten raised to the power of minus six, ie (10^-6) which makes sense looking at BER Eb/No graphs. But Is this the correct way of looking at this notation or is it something else to it?

Thanks again!

Kind regards
Joel
Back to top
 
 
IP Logged
 
USN - Retired
Ex Member


Reply #3 - Mar 24th, 2014 at 12:14pm  
I used to calculate link budgets for a living. There is not enough information in your opening post to actually proceed. At a minimum, I'd have to know whether the 4096 k/bs was the data rate or the symbol rate. Conspicuous in are the frequency and gain figures relative to both uplink and downlink. It's also important to know how much transponder bandwidth is available so you're able to configure your carrier on a not-to-interfere (NIB) basis. 

The BER figure by the way, likely represents threshold error rate. In other words, that's the minimum acceptable operating parameter. Generally you want an Eb/No that is at least 6dB above threshold for an acceptable fade margin. Perhaps more, but that is determined by the max transponder bandwidth within which you're permitted to radiate.

There's probably more, it's been years since I actually built an operating link. But I can state with some confidence that it's not possible to do so with the information thus provided.

//greg//
Back to top
 
 
IP Logged
 
Admin1
YaBB Administrator
*****
Offline


Personal text from: Profile,
Options, Top line

Posts: 844
Reply #4 - Mar 25th, 2014 at 9:28pm  
You have the right idea about the E notation.

The vertical axis is always in log format, like a slide rule. Above the line for 1x10E-6 is a big gap, the next line is 2x10E-6, the next gap is smaller and so on, so the lines become crammed very close together, just as you reach the next main line at 1x10E-5.

Some graphs will not show any of the intermediate lines. Just the lines say from 1x10E-10 to 1x10E-1, in other words just 10 bold horizontal grid lines.
Back to top
 
WWW  
IP Logged
 
jorel
Ex Member


Reply #5 - Mar 26th, 2014 at 9:19pm  
Thanks both of you for your help and input, I have finished the task and learnt a lot mostly thanks to both of you and not really so much because of my teacher at a respected hightech university.

Now the hunt for knowledge moves on, trying to emulate satlinks in opensand, will be very interesting!

Thank so so much again.

Kind regards,
Joel
Back to top
 
 
IP Logged