In this case, the discrete time representation of the model is straightforward.
The discrete time output of the sampler synchronized to the kth user
following the chip-matched filter 
is given by
![]()
since the pulse shaping filter 
 has duration 
.
Given 
 of length 
, we wish to represent the chip-spaced input
to the tapped-delay line as the 
-length vector
 
in order to express the output of the linear filter as
![]()
where 
 is the delay in chips after the first chip in the
spreading sequence.  If 
 then 
.  This means that the
first element of 
 lines up with the first chip in the
spreading sequence.  If 
, then 
 may be negative and
if 
 then 
 may be positive.  This makes the model more
general by not requiring the first element of 
 to line up
with the first element of the spreading sequence of bit n.
In the discrete time model, we construct baud-spaced channel convolution
matrices of dimension 
 for each user, i.e., 

