Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Convolution Coding and Viterbi Decoding with Matlab: High BER with code rate of 5 / 8

Status
Not open for further replies.

Ernte1893

Newbie level 3
Newbie level 3
Joined
Jul 16, 2014
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Visit site
Activity points
43
Hey,

I am building a model for a transceiver and I am struggling atm with the convolution coding and especially its decoding ! I want to use the following code rates:
1/3; 1/2; 5/8; 3/4;

For the first two I use different trellis operators (still standard ones) and since I don't use any noise the BER is obviously 0.
For the 3rd and 4th rate I deleted some samples and added a zero pattern at the end (instead of the deleted samples). Am I doing something wrong ? I get really high BER values for the 3rd and 4th rate (0.2 - 0.5 !).


Code Matlab M - [expand]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
% FEC Code rate selection: k/n : k = information bits, n = total bits
coderate = 2; % 0 means no puncturing but coding rate of 1/3 = size: 3
              % 1 means puncturing (coding rate of 1/2 )     = size: 2
              % 2 means puncturing (coding rate of 5/8 )     = size: 8
              % 3 means puncturing (coding rate of 3/4 )     = size: 4
rate = [1/3 1/2 5/8 3/4];
 
 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% QPSK transmitter
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
%% Create SU payload with FEC
% Create Payload
rand_data = randn(1,data); 
Input = zeros(1,data);
for i = 1:data
        if rand_data(i) >= 0.5
            Input(i) = 1;
        else
            Input(i) = 0;
        end
end
 
% FEC
% Convolution Coding: for each input bit, there are 2 redundant bits concatenated: 
% a -> [ a a a ]; 
trellis = poly2trellis(7,[171 165 133]);
coded_data = convenc(Input,trellis);
% Puncturing: 
% depending on the code rate, the signal has to be changed as followed
switch coderate
    case 0 % Stays the same: 1/3        
    case 1 % a -> [ a a ] : 1/2
        coded_data(2:3:end) = [];
    case 2 % [ a b c d e ] -> [ a a b c c d e e ] : 5/8
        coded_data(3:15:end) = [];
        coded_data(3:14:end) = [];
        coded_data(3:13:end) = [];
        coded_data(6:12:end) = [];        
        coded_data(6:11:end) = [];        
        coded_data(6:10:end) = [];
        coded_data(9:9:end)  = [];
    case 3 % [ a b c d ] -> [ a a b c ] : 3/4
        coded_data(3:9:end) = [];
        coded_data(3:8:end) = [];
        coded_data(3:7:end) = [];
        coded_data(4:6:end) = [];
        coded_data(4:5:end) = [];        
    otherwise
end
 
Signal = coded_data;
 
% According to selected code rate, zeros are added to the signal
decoded = zeros(1,size(Signal,2)*rate(coderate+1)*3);
switch coderate
    case 0 % Stays the same: 1/3        
        decoded = Signal;        
        Output = vitdec(decoded,trellis,12,'trunc','hard');
    case 1 % [ a a ] New trellis operator: 1/2
        trellis2 = poly2trellis(7,[171 133]);
        decoded = Signal;  
        Output = vitdec(decoded,trellis2,12,'trunc','hard');
    case 2 % [ a a b c c d e e ] -> [ a a 0 0 0 b c c 0 0 0 d e e 0 ]  : 5/8
        decoded(1:15:end) = Signal(1:8:end);
        decoded(2:15:end) = Signal(2:8:end);
        decoded(6:15:end) = Signal(3:8:end);
        decoded(7:15:end) = Signal(4:8:end);
        decoded(8:15:end) = Signal(5:8:end);
        decoded(12:15:end) = Signal(6:8:end);
        decoded(13:15:end) = Signal(7:8:end);
        decoded(14:15:end) = Signal(8:8:end);        
        Output = vitdec(decoded,trellis,96,'trunc','hard');
    case 3 %  [ a a b c ] -> [ a a 0 0 0 b 0 0 c ]: 3/4
        decoded(1:9:end) = Signal(1:4:end);
        decoded(2:9:end) = Signal(2:4:end);
        decoded(6:9:end) = Signal(3:4:end);
        decoded(9:9:end) = Signal(4:4:end);          
        Output = vitdec(decoded,trellis,96,'trunc','hard');
    otherwise
end
 
% Viterbi Decoder for convolutianally encoded data
[number,ratio]=biterr(Output,Input); % Bit error rate

 

Raised cosine filtering is important to prevent ISI with symbol RLL span.
I did not look at your code.

Hey,
thanks for the reply, but since both convolution filters are working perfectly fine for both coderates 1/2 and 1/3, the ISI shouldn't be the reason, or what do you think ?
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top