none
Question on HWView RRS feed

  • Question

  • Hello!! I use SoraSDK1.5 and tested HW with HWView. In snr test of HWView window, I found that SNR is shown at the bottom of window.

    I have a question on this value. How can you calculate SNR value? As I know, received signal is mixed with tx signal + noise. So, in order to calculate SNR,  we have to know noise power. So, I want to know how Sora get noise power for SNR calculation.

    Second question is performance discrepancy between SORA. I use two Sora devices. (dev1, dev2) When I send 16-QAM signal (SNRtest) from dev1 to dev2, dev2 shows its SNR is 25~30db. However, from dev2 to dev1, dev1 shows that SNR is just 8~9db. Then, why does this pheonomena happen?

    Third question is decodability. I use HWView as a guideline for gain setting in SORA. Then, when I use 802.11a, for correctly decoding in receiver SORA device, what is minimum value for SNR? (16QAM case)  Is there any data for this minimum SNR for SORA?

    Thanks.




    • Edited by Jaeims Thursday, November 10, 2011 7:06 PM
    Thursday, November 10, 2011 5:38 PM

All replies

  • (1) There're many ways to calculate SNR, we use the way which reflects all the factors in the system, including channel, hardware, DSP algorithm, etc. That is, in the constellation mapping step, we calculate noise as *real constellation point - ideal constellation point*, signal strength is the average from all symbols in a packet.

    (2) It's natually asymmetric between (dev1,dev2) and (dev2, dev1), because transmission on these two directions are acutally using different/separated set of hardware modules, including TX radio chain, wireless channel and RX radio chain. In commercial WiFi radio, it looks the two directions are symmetric because the hardware has been calibrated in factory. In software radio (not limited to Sora), you have the flexibility to access PHY and even radio hardware, the cost you should pay is that the system becomes more complicated, more factors could affect end-to-end performance. Usually the end-to-end performace is not well calibrated. Particularly for Sora, the suggestion is to using a cable to connect two radios, then tune tx and rx gain. If the radio hardware works well, by tuning the parameters, usually you will be able to acheive similar (30db) SNR on both direction.

    (3) I don't have the data in hand, but I think you can easily find the data in literature.

    -Jiansong

    Friday, November 11, 2011 7:42 AM
  • Thank you for your reply. Regarding to question (1), I found some usual situation.

    At first, I use HWView where dev1 is sender and dev2 is receiver in snr test. After checking fine constellation map, I increase tx power in dev1. Contrary to my expectation, SNR is decreased. (by 1~2 db). I don't explain this situation.

    Thanks.

    Friday, November 11, 2011 9:48 PM