Re: [Discuss-gnuradio] gr-ieee 802.11 Wifi Transceiver Issue

[Abhinav Jadon]

Hi Bastian ,

There is another issue regarding the same topic I would like to bring to the fore.

I use a pair of X310. Due to some constraints, I shifted to one X310.

Upon running the transceiver code on a single X310, the transceiver shuts down after few seconds of action (the LED ceases to blink) while the RX continues to be up and decodes the packets.

If I run only the RX ( replacing the UHD sink with a null sink), it continues to decode the packet.

If I run only the TX ( replacing the UHD source with a null source), it continues to transmit, the LED stays on until I stop the flowgraph.

I also ran simple tests (single tone transmission and reception) on the same device. The TX and RX LEDs remain up until the flowgraph is stopped. This was done at the behest of James Humpheris of Ettus Research.

I raised the issue on the Ettus mailing list and they asked me to put this up on the GNURadio Mailing List.

Regards

Abhinav PS  Jadon

2012122
Electronics and Communication Engineering Undergraduate
IIIT - Delhi
IASc Summer Research Fellow 2015
E: address@hidden

M: +919650936845

On Mon, Feb 15, 2016 at 9:51 AM, Bastian Bloessl <address@hidden> wrote:

Hi,

On 14 Feb 2016, at 14:46, Abhinav Jadon <address@hidden> wrote:

There are no overruns and the connections to the antenna ports are correct.

The frame detection is working

Note, that this also means that frame detection is only triggered once per frame. Sometimes, it can be triggered all the time (if there is a DC offset or LO leakage for example).

Since you connected the devices via cable I would try to change LO offset of sender and receiver.

(Btw, I guess you use attenuators)

I tried all the stuff you told me to try ie I tried LMS ansd LO offset, they worked as in I saw some packets being decoded by the wireshark connector. But the packet content was incorrect . Each packet decoded looked like this 

new mac frame  (length 10)
=========================================
frame too short to parse (<20)
WIRESHARK: received new message
message length 10
WIRESHARK: d_msg_offset: 0   to_copy: 43   d_msg_len 43
WIRESHARK: output size: 32768   produced items: 43

You should check in Wireshark if the content makes sense. I just implemented a very minimal parser for demo purposes.

While the packet that is being transmitted has the following characteristics

WIRESHARK: received new message
message length 624
WIRESHARK: d_msg_offset: 0   to_copy: 657   d_msg_len 657
WIRESHARK: output size: 32768   produced items: 657

Is the signal field being wrongly interpreted ?

That would be one thing to find out. The easiest way is to enable the log option of the Decode Signal block (not the Wireshark block) to print what it decoded.

In general, I would recommend to try to find out where things break. (is the frame detected, is the signal field decoded, does the constellation plot look OK, etc.)

Best,

Bastian

On Thu, Feb 11, 2016 at 6:13 AM, Bastian Bloessl <address@hidden> wrote:

Hi

> On 10 Feb 2016, at 15:13, Abhinav Jadon <address@hidden> wrote:
> Next I replaced the loopback with a uhd source and sink and connected the RF frontends using a SMA cable. It fails to give me any packet (the wireshark connector). I tried playing with the value of rx_gain and tx_gain. I also tried playing with the value of the the correlation threshold.
> I then chose to debug using the data flow. The data in the flowgraph flows until the Decode MAC block where it gets dropped with checksum wrong message.
>

Your debugging sounds already pretty good. Some more stuff you could try

- assert that there are no overruns (‘O’s or ‘D’ on the console)
- check that frame detection is working (there are no frames detected when the transmitter is turned off)
- test with antennas
- assert that you connected the correct antenna ports (RX and TX use a different ports by default)
- set a different LO offset
- use the LMS equaliser
- try a different sample rate / bandwidth
- check if the signal field is decoded correctly (log or debug option)

Best,
Bastian