Re: VARA HF Modem Speeds


Rick Muething
 

Graham,

During the development of the Simulator there were many mods to the code doing the audio processing.  Specifically my initial code (and some of the initial tests) did not correctly model the multipath (created by the delay and modulation (with a randomized Real + Imaginary veector) of the audio. This was found to be due in part to problems with the C++ implementation of the narrow band (~3 Hz) multipath filters which were implemented using IIR filters.   It took a couple of months to get the implementation to match/track other simulators (hardware and software).  Also the latest simulator firmware versions (2.03)  gave users the option of using 1 path (WGN), 2 path multipath or a 4 path multipath. The results published in Tom W's testing writeup:

https://winlink.org/sites/default/files/downloads/a_winlink_digital_mode_performance_comparison_based_on_the_ionis_sim_hf_vhf_channel_simulator_-_november_2_2020_0.pdf

use only 2 path multipath (two I + Q paths) along with the most recent revisions of VARA, Pactor 4 and the Winlink Express client program.  Some mods in the Winlink  Express modem drivers allowed slightly (10-15%) more throughput in both Pactor 4 and VARA.  VARA 2.8 KHz BW was not available for testing.  That should allow slightly faster throughput at High S:N levels (> 25 dB) .  In practice however it is rare (without Linears or high gain antennas) to often see digital connections > 25 dB  S:N.

The simulator of course does not know or care what protocol it is processing it simply takes the audio from the modem/sound card,  checks to make sure it is not overdriving the simulator ,and processes the audio at the requested simulator bandwidth (300-3300 Hz   or 300-6300 Hz)  and then mixes WGN at the requested S:N and the desired level of multipath (using the recommended path delay values by the CCIR/ITU ) .  The extensive testing showed repeatable results [net throughputs for long (5-15 minute) ARQ sessions] for all protocols.

The Ionos Simulator also computes and displays the Crest Factor which can help determine the optimum drive level for the transmitter.  With very low crest factors (< 2:1 or < 3dB) it is quite possible to drive most amateur SSB transmitters  to exceed its thermal limits even if it is not driven to maximum rated PEP.  With high crest factors (>5-6 dB) most amateur transmitters will be PEP limited before thermal limited and this will result in lower net Energy per bit and therefore lower throughput for a given PEP drive level.

The main value of the simulator is that it allows duplicating statistically similar paths (S:N and multipath) that are almost impossible to do using using over the air testing. This is why they have been used extensively in testing and characterizing protocols such as Mil STD 188 and STANAG.

73,

Rick KN6KB


On 1/22/2021 9:17 AM, Graham wrote:

Rick 

I , just noticed the  P4  and V4 speeds , had , in effect  reversed, where as the  other  modes , look to be similar ,

interesting  would be  comparison of  OLIVIA  and  ROS-HF-4  modes 

P4 is  used on HF out side the  part97  regulatory area , various on air comparisons  , have been  made , one  ive posted below 
Test via HB9AVK
   592 km  7.050  , shows a  11:7   compound advantage , test '2' shows  P4  advantage , 10 seconds in  120 
link survival, V4  appears also to offer advantage ?

In the  simulation, the  audio is presented  directly ? ,  Vara has been  developed  using live  links, 
notably , recent versions are  bandwidth  tailored to maximise  use of available bandwidth , peak ratios also have been  
dynamically matched .. all these contribute to  optimisation 


73 -Graham
g0nbd




COMPARE VARA and PACTOR4  upload via Gateway HB9AK

 

Test via HB9AVK   592 km  7,050 MHz DIAL

 

 

Term 1

 

*** Completed send of VARA message JE2O4B0QBRP2

*** Sent 1 message.  Bytes: 20230,  Time: 01:03,  bytes/minute: 19030

 

*** Completed send of Pactor message MATG09PBPW98

*** Sent 1 message.  Bytes: 20218,  Time: 03:11,  bytes/minute: 6332

 

Term 2

 

*** Completed send of VARA message 6Z4SZQFPRIJL

*** Sent 1 message.  Bytes: 20213,  Time: 02:16,  bytes/minute: 8870

 

*** Completed send of Pactor message 2XWY9G2MCS0N

*** Sent 1 message.  Bytes: 20219,  Time: 02:07,  bytes/minute: 9517

 

Term 3

 

*** Completed send of VARA message TC67GB6RLZ1S

*** Sent 1 message.  Bytes: 20213,  Time: 02:46,  bytes/minute: 7301

 

*** Completed send of Pactor message Z6PSHUFB8HZ3

*** Sent 1 message.  Bytes: 20217,  Time: 02:21,  bytes/minute: 8573

 

Term 4

 

*** Completed send of VARA message K9K11XBW9D3F Vara WL2K

*** Sent 1 message.  Bytes: 20215,  Time: 02:37,  bytes/minute: 7692

 

*** Completed send of Pactor message 58KVLSL4OUSX    fast QRM Disconnecting Timeout

*** Sent 1 message.  Bytes: 20219,  Time: 05:24,  bytes/minute: 3740

 

Result:

 

Term

MODE

Total bytes sent

Time

bytes/minute

Message ID

 

1

VARA

20230

01:03

19030

JE2O4B0QBRP2

 

 

P4

20218

03:11

6332

MATG09PBPW98

 

 

 

 

 

 

 

 

2

VARA

20213

02:16

8870

6Z4SZQFPRIJL

 

 

P4

20219

02:07

9517

2XWY9G2MCS0N

 

 

 

 

 

 

 

 

3

VARA

20213

02:46

7301

Z6PSHUFB8HZ3

 

 

P4

20217

02:21

8573

 

 

 

 

 

 

 

 

 

4

VARA

20215

02:37

7692

K9K11XBW9D3F

 

 

P4

20219

05:24

3740

58KVLSL4OUSX

 

 

 

 

 

 

 

 

 

 

 

 

VARA Mean Value:  10723

P4 Mean Value:       7040

Join main@digitalradio.groups.io to automatically receive all group messages.