Bernoulli sensing matrix for compressed sensing using matlab

I am trying to implement compressed sensing in matlab, also trying different types of sensing matrix (Gaussian, Bernoulli, Fourier), and I have problems implementing -+1 Bernoulli random matrix as a sensing matrix, I am generating it as follows:

p=0.5;
A=(rand(M,256)<p);
A=A*2-1;

where M is

M => C*K*log(N/K)

N is the vector length, K is the non-zero coefficients, is that correct?

[Walter Roberson]

Your M does not appear to be restricted to non-negative integers, which would be required to use as the number of rows for rand()