The emergence of Bitcoin cryptocurrency sparked interest among businesses and researchers in exploring the technical aspects of blockchain and distributed ledger technology systems. However, present-day blockchain technologies still face notable limitations in terms of scalability and flexibility, particularly when dealing with large-scale and dynamically reconfigurable scenarios. To address the scalability challenges, the sharding technique has shown promise in horizontally scaling out blockchain systems by dividing the network into multiple shards or clusters.

However, there is a need for further research and investigation to enhance the flexibility and reconfigurability of these clusters, especially in application scenarios like microgrids and edge-based systems where clusters may require structural changes. Moreover, when applying blockchain in highly mobile environments like vehicular communication, special attention must be given to the stability of consensus nodes due to node mobility.

In this work, innovative mechanisms have been developed that substantially streamline the re-clustering process within blockchain frameworks. By optimising previously redundant procedures, the approach proposed in this work has achieved significant time reductions for merging and splitting operations, quantified by a reduction factor of 1/22000. This breakthrough not only improves operational efficiency but also reduces the computational demands during these critical phases.

Furthermore, this thesis introduces the Mobility-aware Reputation-based Blockchain Framework (MRBF), incorporating an Intelligent Consensus Node Selection Algorithm (ICNS) that utilises Artificial Neural Networks (ANNs). This is the first framework that systematically uses mobility and stability parameters to enhance the effectiveness of consensus procedures, achieving a 6.8% improvement in the average node reputation and a 17.5% increase in the average Stability of Node (SoN) values. Additionally, it reduces the average message counts by 33.9%, demonstrating the practical benefits of the tailored selection algorithm in highly mobile environments.

Building on these innovations, enhancements were made to ICNS, especially in contexts involving vehicular communications. Utilising a unique simulation setup that incorporates realistic mobility datasets, a 34% decrease in block confirmation latency is demonstrated. Incorporating network connectivity conditions, the refined algorithm improved node reputation by 45% and SoN values by 38%.

This research enhances the fields of blockchain and mobile communications by introducing a robust framework for dynamic cluster-based scenarios. It provides insights into the flexibility challenges of blockchain systems and has wide applicability in microgrids, edge-based systems, and vehicular communications, demonstrating its potential for real-world implementation and adoption.