Re: [Discuss-gnuradio] Please Help in trellis gnuradio

[Achilleas Anastasopoulos]

The output of the FSM belongs to the set {0,1,2,3}.
The Viterbi decoder requires a vector of size 4 for each FSM
symbol. So if the output of the FSM is [0,3,3,2,1,0] (6 symbols)
then you need to provide the Viterbi decoder with a stream
of  6 x 4 floats, each indicating the "distance" between
the "received" signal and each of the 4 symbols in the alphabet.
If you want to simulate a noiseless transmission, the input to the
viterbi should look like:
[0 a a a   a a a 0   a a a 0   a a 0 a   a 0 a a   0 a a a ]
where a is any positive number.

The simplest way to implement this is to setup a lookup table
(in gnuradio is called chunks_to_symbols of length 4 and dimensionality 4)
so that
0 maps to [0 a a a]
1 maps to [a 0 a a]
2 maps to [a a 0 a]
3 maps to [a a a 0]
after the encoder and before the Viterbi decoder.

Achilleas

On Sat, Jun 4, 2011 at 5:58 AM, smith mark <address@hidden> wrote:
> Hi Achilleas,
> I hope that you are fine. I want to implement convolutional coding using the
> trellis block. I don't want to use any modulation scheme or anything else
> after the encoder. The flow graph I want is shown below
> vector source---->trellis encoder----> viterbi or any decoder------->sink
> Part of the code is shown below
> ----------------------------------------------------------------------------
> src_d = (1,0,1,0,0,0,1,1,0,0,1)
> sorc = gr.vector_source_b (src_d)  //source
> encoder = trellis.encoder_bs(f,0)    //encoder
> s2f= gr.short_to_float()
> dec=trellis.viterbi_s(f,len(src_d),0,-1)  //decoder
> dst = gr.vector_sink_s ()                   //sink
> tb.connect (sorc,encoder,s2f,dec,dst)
> return(dst.data())
> ----------------------------------------------------------------------------
>  When I run the above flow graph I get nothing just empty, ( ), list.
> I am using "awgn1o2_4.fsm" file. I got the correct result after encoder. I
> can't use the trellis.metric or trellis.viterbi_combined_X to implement
> decoder as I am not using any modulation :( .
> Please help me in this ??
> Regards
> Smith
>