Hi Tony, you will have to repeat the test next days when i improve lenght interleaver as I had expected :-)
The first think i said about ROS is that minimize the power "at the same character/minute rate", and that is just what you are tester :-). You cannot match Washl Function FEC with Viterbi Algorithm. If you want OLIVIA be equal of robust that ROS you will have to transmit much slower than ROS necessarily. You can see as the different between OLIVIA and ROS is the not inconsiderable number of 5 dBs (3.2 times less power at the same character/minute). Equally, the next mode ROS 8 baudios/2250 Hz will be 3dBS better than ROS 16 baudios, but, obviusly is half as slow than ROS 16. You can not go against mathematics. About frequency spreading of 25 Hz, thats is a value exaggerated. Normally, the spread doppler is usually of 1 or 2 Hz in the higher bands of the HF. About ROS threshold, ROS is designed in a way that the Initial Acquisition Sequence poses no funnel. That is, the Initial sequence always has better sensitivity than the data demodulator. Sometimes you will see the initial sequence is activated but the data cannot be demodulated. This is how it should be. Good job, and i expect you repeat the test with the new improvements I am making. ________________________________ De: Tony <[email protected]> Para: [email protected] Enviado: lun,1 marzo, 2010 23:29 Asunto: [digitalradio] ROS Path Simulations All, I ran several path tests with ROS-16 and Olivia 2K this evening. The simulator showed that Olivia is about as sensitive as ROS when configured to run at the same baud rate, but it is not as sensitive when configured to run at the same word-per-minute rate. Olivia 32/2K will runs about as fast as ROS, but it is roughly 5db less sensitive. Mode Sensitivity baud rate WPM Olivia 128/2K -14db 16 3 times slower than ROS-16 Olivia 32/2K -10db 64 same as ROS-16 ROS-16 -15db 16 That increase in sensitivity seems to help ROS cope with certain poor channel conditions (as per the path simulator) compared to Olivia running at the same speed. In CCIR poor channel tests, for example, where selective fading sweeps across the channel, ROS printed better than Olivia 32/2K with low signal-to-noise ratios. On the other hand, Olivia 128/2K (16 baud) had an edge over ROS under the same conditions, albeit, with much slower throughput. In high-latitude tests, severe Doppler spread caused throughput to fall off dramatically with ROS indicating that it will likely fail over severely distorted polar paths. This occurred when the frequency spreading was above 25Hz (ITU-R high-latitude severe distortion). Olivia was not affected. I found that ROS will not recover after the signal drops below it's minimum decode threshold and will not trigger ROS to start receiving if the signal is not strong enough at the beginning of the transmission. I'm not sure if this is something inherent in the mode or if it's a bug in the software. I'm sure Jose can answer that. I should noted that Olivia 32/1K compares well with both ROS-16 and Olivia 2K modes in terms of poor channel throughput. Olivia 16/500 does a fine job as well. I suspect that ROS would perform well in an 8 baud / 1000Hz mode version. Many thanks to Jose for the new mode. Tony -K2MO
