The development of 5G communication technology has introduced new families of applications and use cases during just its standardization and some of them have brought peculiar system requirements, where existing channel coding in its current form might suffer from tackling them. As an example of this, it can be shown that deploying one of the powerful channel coding techniques might suffer in tackling stringent delay requirements while sustaining higher reliability in some mission-critical applications due to iterative decoding structure. Besides, an error-floor region might prevent turbo-coded systems from deploying in certain use cases. From this perspective, a channel coding technique, while currently less popular than the others, might have considerable potential over 5G and beyond after optimizing its modules.
Motivated by this fact, this thesis aims to present an SNR-adaptive convolutionally coded transmission model where the simplicity of a convolutional encoder has combined with current advanced optimization ability. Basically, the proposed transmission model combines one-shot decoding superiority of convolutional coder with the utilization of choosing an optimized group of symbol points, which are obtained specifically for a given convolutional encoder, channel characteristic and transmission model. The enabler of an SNR-adaptive optimization framework is the derivation of upper bound error performance expression by exploiting the product-state matrix technique, which is used in the calculation of generating function for a given convolutional encoder and it brings the superiority to work with any type of constellations. The importance of including fully arbitrary, irregular type, constellations in constellation search lies on that the uniform constellations can lead to suboptimal performance in many coded scenarios including bit-interleaved coded modulated and turbo-trellis coded modulations as already shown.
As an initial step towards the proposed SNR-adaptive transmission model, a generalized error performance calculation was first presented. Once optimized symbol point locations via particle swarm constellation optimizer are found for each operating point, the performance comparison of the proposed SNR-adaptive convolutionally coded transmission model with SNR-independent counterparts is given in terms of BER and spectral efficiency. The superiority of the proposed transmission model, in terms of algorithmic complexity and decoding latency, is also presented.