calculate checksum for the input vector

i need to calculate checksum value for given vector. when i try to compile the code below i receive 14 error all related to the following Function 'Embedded MATLAB Function' (#18.1210.1225), line 49, column 18: "bitand(value,1)"

similar error type is for bitshift

appreciate the solution for this problem

thanks in advance

function CRC = fcn(input_value, CRC_int)

CRC = CRC_int;

l = numel(input_value);

for j = 1:l

    if j == 1
    value = bitshift(input_value(j),-1);
    else
    value = input_value(j)
    m = (bitand(value,1) && bitand(CRC,1));
    n = (bitand(value,1) || bitand(CRC,1));

        if m ==  1
        CRC(j) = bitshift(CRC, -1);
        value = bitshift(value, -1);
        elseif m == 0 && n ~=1

            CRC = bitshift(CRC, -1);
            value = bitshift(value, -1);

        elseif n == 1

            CRC_temp = bitshift(CRC, -1);
            CRC = bitxor(CRC_temp,CRC_int);
            value = bitshift(value, -1);
        end
    end
CRC = CRC;

for i = 1:7

    if (bitand(value,1) && bitand(crc_var,1))

        CRC = bitshift(CRC, -1);
        value = bitshift(value, -1);

        elseif ((bitand(value,1) == 0) && (bitand(CRC,1) == 0))

            CRC = bitshift(CRC, -1);
            value = bitshift(value, -1);

         elseif (bitand(value,1) || bitand(CRC, 1))

            CRC_temp = bitshift(CRC, -1);
            CRC = bitxor(CRC_temp,CRC_int);
            value = bitshift(value, -1);

    end

end

    %crc_var = crc_var

   % CRC = dec2hex(CRC);

end

Answer 1

[Fangjun Jiang]

In R2007b, when used in Embedded MATLAB Function, bitand() doesn't support floating-point input. The arguments must belong to an integer class.

[M]

thank you Fangiun...how can i avoid it?

[Fangjun Jiang]

You need to define your variable "value" as integer class, as well as the constant,e.g. bitand(int32(value),int32(1))

[Jan]

Or you use FLOOR(x/2) instead of BITSHIFT(x, -1). And REM(x, 2) is faster than BITAND(x, 1).

Answer 2

[M]

Fangjun / Jan thank you for your hints....

the final answer that works for me is the following

function crc_var = fcn(input_value)

ply =uint16(hex2dec('8408'));
crc_var =uint16(0);
value = uint16(0);
l = numel(input_value);

for j = 1:l
    value = uint16(input_value(j))
for i = 1:8
%     x = and(value,one)
%     y = and (crc_var,one)
    m = (bitand(value,1) && bitand(crc_var,1))
    n = (bitand(value,1) || bitand(crc_var,1))

      if (m ==  1 || (m == 0 && n ~=1))

          crc_var = bitshift(crc_var, -1)
          value = bitshift(value, -1)

     elseif (n == 1)            
              crc_var_temp = bitshift(crc_var, -1)
              crc_var = bitxor(crc_var_temp,ply)
              value = bitshift(value, -1)

      end

end

      %crc_var = crc_var

      crc_var_h = dec2hex(crc_var)

end

[Fangjun Jiang]

So I assume the cause is due to the fact that you need to use integer class inputs for bitand() in EMC. Once you defined 'value' and 'crc_var' as uint16, the constant 1 is automatically casted.

[M]

seams to be the case....