Hello, I did 3 more experiments to test the performance of Reed Solomon in gr.atsc, but I got some contradictory results. I don't know what else I can do on this. Hope Dr. Corgan could help me this time. 1,I did a simulation with Reed-Solomon coding scheme on my laptop and add std::cout << nerrors_corrrected << std::endl;, it functioned very well. When the number of errors is less than 10, the number will be output, otherwise, it will output -1, which means there are more than 10 symbols corrupted. 2,I added that coding scheme into packet_utils.py (I coded the payload+crc). While ten packets were sent out from the TX and all of the received packets were true (ok = True), the nerrors_corrected were all -1, which were supposed to be 0. 3, I also checked whether the output of the RS encoder matched
the input of the RS decoder, and the answer is YES, which means the packet before decoding was a correct packet, and the nerrors_corrected should not be -1! I think I did all of what I can......, but the result is really confusing me. Dr. Corgan, could u give me a little help? Bill --- On Sat, 5/9/09, Bill Stevenson <address@hidden> wrote:
From: Bill Stevenson
<address@hidden>
Subject: Re: [Discuss-gnuradio] a question about Reed-Solomon in GNURadio!
To: "Bill Stevenson" <address@hidden>
Cc: address@hidden
Date: Saturday, May 9, 2009, 12:44 AM
Hi!
I have successfully implemented the Reed-Solomon coding scheme in gr.atsc on USRP, the receiver can receive packets now. But after running tons of experimentation and comparing the performance between DQPSK without coding and DQPSK with Reed-Solomon, I cannot observe any performance improvement of the DQPSK with Reed-Solomon scheme. For the same PDR and BER, the RSSIs which are read using
probe.level() are similar, and Reed-Solomon did show any correcting effect.
Does the reed-solomon in gr.atsc take effect? I coded the payload+crc part. Thanks!
Bill
--- On Fri, 5/1/09, Bill Stevenson <address@hidden> wrote:
From: Bill Stevenson <address@hidden>
Subject: [Discuss-gnuradio]
a question about Reed-Solomon in GNURadio!
To: address@hidden
Date: Friday, May 1, 2009, 2:41 AM
I implemented the reed-solomon coding scheme in gr-atsc onto benchmark_tx.py and benchmark_rx.py. Following describes the results I got:
1, The transmitter can transmit reed-solomon coded stream with no problem. The receiver just keeps on waiting for the packets. There is no response at the receiver.
2, To fit the needs of USRP and the format of atsc data segment, the entire packet before the reed-solomon coding is packed in this way: MPEG_SYNC_BYTE+transport error byte+preamble+access code+header+ some padding (pad it into 256 bytes)+payload+ some padding+...+crc+..., the length of packet size is 1098, so after
packing process,
there are 6 data
segments sent out in each packet.
3, I uncommented the assert(in[i].pli.regular_seg_p()) in atsc_rs_decoder.cc and
atsc_derandomizer.cc because without uncommenting them, the corresponding error message will pop out the moment I run the benchmark_rx.py -f 2400M..... command at the receiver side.
4, I simulated this system in my laptop without connecting to any USRPs, and the result showed that the reed-solomon performance is apparently much better than that without reed-solomon.
Based on my limited knowledge, there might be two reasons: one is because of erasing the assert(in[i].pli.regular_seg_p()) phrase, but I don't think it is the culprit since the similer simulation did not indicate anying wrong with that method; the second one is MPEG_SYNC_BYTE+transport error byte ruined the correlating process of access code because it seemed that there were 3 preamble out there, but I am not sure about my conclusion.
I don't know what makes my receiving process down. Is that because of the MPEG_SYNC_BYTE in front of the preamble code? I really hope
someone could give me some help on this problem! Thanks a lot!!! :)
Bill
-----Inline Attachment Follows-----
|
|