This thesis investigates various detection processes that operate
with known time invariant channels. The investigations are divided
into two main areas, the first of which involves uncoded digital signals.
Three different detection processes have been studied here and some
promising systems have been developed from these. The first of the
detection processes is an iterative detection process whereas the second
detection process involves the linear filtering of the received signal.
Binary signals are considered in the investigations here. The third
detection process achieves the near-maximum likelihood detection of
a 16-point QAM digital signal transmitted over a telephone circuit
at 9600 bits/second. The detector here operates on the received sample
values directly without using any complex prefiltering. The second
area of investigation covered in this thesis involves coded digital
signals. Binary and 16-point QCM signals have been considered here.
Rate2 and 3 non-systematic convolutional codes with optimum free distance
have been used in conjunction with the appropriate Gray codes for the
encoding of the signals. At the receiving end, a joint near-maximum
likelihood detection/decoding process is used. Computer simulation
tests have shown that the system improves the tolerance to Gaussian
noise over the corresponding uncoded system at low error rates.
History
School
Mechanical, Electrical and Manufacturing Engineering