Sean de's answer unfortunately is not sufficient for your purposes.
There is no command or setting that you can use in MATLAB to force the use of all integers. For example
K = K + 1
would interpret the 1 as a double precision number (but would convert it to integer if K is an integer data type.)
Some operations such as mean() internally convert values to double precision, as adding a number of integers is likely to produce integer overflow.
Your options are to use the Fixed Point Toolbox, or to use Embedded MATLAB.
