I keep getting a "convenc" error. Line 88. Could be "varargin" missing argument? Not sure.

clc
clear all
%% initialization
para=52;
noc=52;
fftlen=64;
gi=0;
nd=128; % number of OFDM frames to send
nb=para*nd/2; % total bits to send
np=1; % number of paths
% snr=50; %dB
tblen=7; % traceback length for viterbi
trellis = poly2trellis(3,[7 5]);
allsnr=[20:40];
ii=0;
msg = randi(1,nb); % generate message stream.
for snr=allsnr
    ii=ii+1; % keep track of the loop
    %% transmitter
    % convolutional encoding
    encoded = convenc(msg,trellis,);
    % serial to parallel data
    encodedFrames=reshape(encoded,para,nd);
    modulated=encodedFrames*2-1;
    % mapping
    crmapped(2:27,:) = modulated(1:26,:);
    crmapped(39:64,:)= modulated(27:52,:);
    %% ifft
    transmitted = ifft(crmapped);
    %% Fading channel model
    % start fading model. This is the part you will modify or complete
    fd = 50; % Doppler shift
    M = 2; % Number of delayed signal paths
    Mdelay = [1 3]; % delay of each delayed signal
    K=4; % rice factor
    dt = 1/250000; % minimum time resolution
    N = 6; % number of oscillators for Jake’s fading model
    for sc=1:fftlen, % sc is the subcarrier number
        phi = rand*1000; % fading counter
        faded(sc,:) = fadingModel(transmitted(sc,:),fd,M,Mdelay,K,dt,N,phi);
        % function to build
    end
    % end fading model
    %% ------------------------------------
    % AWGN
    noisyFaded = awgn(faded,snr);
    % noisyFaded=toChannel;
    %% receiver
    % fft
    received = fft(noisyFaded);
    % channel estimation
    tr0 = crmapped(:,1); % original "training" data;
    tr1 = received(:,1); % training symbol
    h = tr0./tr1; % reverse rotation
    H = h*ones(1,nd); % extend the reverse rotation matrix to cover all symbols
    compensated=received.*H; % signal compensation
    % compensated = received; % disable signal compensation
    % demapper
    crdemapped = zeros(para,nd);
    crdemapped(1:26,:) = compensated(2:27,:);
    crdemapped(27:52,:) = compensated(39:64,:);
    % demodulate
    demodulated = double(crdemapped>0);
    % parallel to serial
    demapped = reshape(demodulated,1,[]);
    % decoder
    decoded = vitdec(demapped,trellis,tblen,'trunc','hard'); % viterbi decoder
    sdecoded = decoded;
    % calculate ber
    [be(ii) ber(ii)] = biterr(msg,decoded)
end
% plot ber
plot(allsnr,ber);
title('BER vs. SNR')
xlabel('SNR')
ylabel('BER')

[Jan]

Please post the complete error message. Then the readers do not have to count, where "line 88" is.

Answer 1

[Jan]

The line is incomplete:

encoded = convenc(msg,trellis,);
%                             ^ ???

[Marcel]

% Check msg

outType = class(msg); % for output data type

if ~isempty(msg)

validateattributes(msg, {'numeric', 'logical'}, ... <--- Line 88

{'vector', 'binary'}, 'convenc', 'MSG');

coder.internal.errorIf(mod(length(msg), k) ~=0, ...

'comm:convenc:InvalidMsgLength');

end

Answer 2

[Steven Lord]

If you're calling the convenc function from Communications Toolbox, the section of that documentation page that describes the msg input argument states that its data type must be either double or logical and it must be a vector of binary values. The code you've posted does call it with a double array, but the line of code you've used to create that array:

nb = 4; % Added so this code runs
msg = randi(1,nb); % generate message stream.

creates an nb-by-nb matrix of values where each element is an integer in the range [1, 1]. Unless nb is 1, msg is not a vector of 0's and/or 1's.

msg

msg = 4×4

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

You could make this return a 1-by-nb vector by adding one input to randi listing the range of values from which you want to generate the random data.

msg = randi([0 1], 1, nb)

msg = 1×4

0 1 0 1