Abstract
The quest to achieve optimum bandwidth usage, lower transmission errors, and increase data
throughput is creating a lot of pressure on the improvement of orthogonal frequency division
multiplexing (OFDM) systems. OFDM transmission is the backbone of modern radio
technologies. Moreover, it is being used for the conceptualisation of more advanced
transmission setups (e.g. orthogonal time-frequency spacing networks). Conventionally,
OFDM signals are obtained by performing a Fourier transform on a quadrature phase shift
keying (QPSK) signal and subsequently adding a cyclic prefix. QPSK is a stable modulation
technique with good spectral efficiency. However, during transmission, the QPSK carrier
signal can undergo numerous phase changes resulting in phase ambiguities and carrier offset
drift. Moreover, the QPSK encoding process gives rise to bit and block errors on the transmitted
signal. Cumulatively, all these weaknesses decrease the efficiency of the channel. In this
research, three hybrid genetic algorithms (GA) are proposed for improving OFDM
transmission. Two of the GAs are used to enhance the QPSK technique whilst the third GA is
used to enhance a pulse amplitude modulation (PAM) carrier signal. PAM transmission is used
for comparison when evaluating the performances of the GA-assisted QPSKs in an OFDM
network. The GA is a very reliable algorithm that gives accurate optimisation results. However,
it struggles with proper gene concatenation and selecting the best individuals for future
generations. It often encounters partial optimism and premature stagnation during its search for
solutions. For the first hybrid algorithm, the wind-driven optimisation (WDO) is proposed as
the selection operation of the GA resulting in a hybrid GAWD algorithm. The WDO has fast
convergence and very few parameters that need adjustments which makes it suitable as an
internal selection operation. The proposed hybrid structure can also improve the quality of the
GA’s mutated populations and their search efficiency. For the second algorithm, a custom
three-point crossover is proposed as the genetic recombination operator for the GAWD
resulting in a hybrid GAWD3 algorithm. This proposed structure can guarantee optimum gene
synthesis within the hybrid algorithm through efficient gene integrations. The third proposed
algorithm constitutes the particle swarm optimisation (PSO) as the fitness scaling and selection
operations of the GA resulting in a hybrid GAPSO algorithm. The PSO is good at scattering its
particles across the search space and quickly converging at the most optimum solution. During
OFDM transmission, the GAWD and the GAWD3 were used to deduce and assign optimum
phase values for a QPSK modulating signal. The GAPSO was used for oversampling a PAM
modulating signal. When compared to the conventional QPSK, the developed QPSKs had
better OFDM channel BER performance. Similarly, the enhanced PAM also outperformed the
conventional QPSK. The simulation results showed that the incorporation of the hybrid genetic
algorithms helped to lower BER and maintain the signal waveform since the received signal
greatly corresponded with the transmitted signal.
Keywords: Bit error rate, Genetic Algorithm, Orthogonal frequency division multiplexing,
Oversampling, Particle swarm optimisation, Phase ambiguity, Pulse amplitude modulation,
Quadrature phase shift keying, Wind-driven optimisation.