Abstract:
The main motivation of this thesis is to optimize the non-coherent detector
in code division multiple access (CDMA) communication system. The choice
to use non-coherent detection is made because of the fact that non-coherent
detectors does not utilize the phase information in receiver side, resulting in
lowering detector complexity. In the ultra wideband (UWB) technique both
direct-sequence spread spectrum (DS-SS) and time hopping SS (TH-SS) can
be combine, since UWB system has ability of combining the advantages of
direct-sequence (DS)-CDMA and time-hopping (TH)-CDMA systems and
avoiding their disadvantages. This hybrid DS-TH CDMA system can mitigate the channel effect in communication system, naming inter-symbol interference (ISI), multiuser interference (MUI) and multipath fading effects.
The correlation detector does not have ability to mitigate these channel effect while linear detector minimum mean-square error (MMSE) has ability
to mitigate these channel effect. In single user environment where there is
no obstacle between transmitter and receiver and are in line of sight with
each other, the correlation detector performance is near optimum and low
complexity. However, correlation detector can also be used for multiuser environment if channel effect is ignored.Furthermore, MMSE detector which
has ability to tackle channel effect can use hybrid DS-TH UWB system to
mitigate channel effect. This novel has three phases, in first phase both correlation and MMSE detector have been simulate in multiuser environment
ignoring channel effect, we find out that correlation detector perform well
with lower complexity and near ideal BER performance. In second phase
channel effect have been introduced and simulate the MMSE for hybrid DSTH UWB system, result shows MMSE produced good result by varying DS
iii
and TH spreading factor of DS-TH CDMA system. However, the complexity
for DS-TH CDMA increases for MMSE detector. In third phase, the adaptive
detector least-mean square (LMS) is investigated for DS-TH CDMA system
in term of their convergence speed, robustness, computational complexity
and BER performance. LMS has lower complexity then correlation detector and has BER performance better than MMSE, since it does not require
channel estimation. However, LMS result low BER performance in low SNR
region.