| dc.description.abstract | This research proposes novel embedded Markovian queueing model-based quantitative models in order to establish a theoretical foundation to design a dynamic blockchain-based computing system with a specific interest in Ethereum. The proposed models commonly assume variable bulk arrivals of transactions in Poisson distribution, i.e., M^(1,n), where n the number of slots across all the mined transactions to be posted in a block or the current block. Firstly, a baseline model is proposed to have a static bulk service of transactions in exponential time, i.e., M^n, for posting the transactions in the current block, referred to as Variable Bulk Arrival and Static Bulk Service (VBASBS) queueing model of the M^(1,n)/M^n/1 type, in which note that n is fixed in order to demonstrate a static chain in terms of the size of the block. Secondly, an adaptive chain model, as a solution of dynamic blockchain in a reactive manner, is proposed based on a Variable Bulk Arrival and Variable Bulk Service (VBAVBS) queueing model of the M^(1,n)/M^(1,i,n)/1 type to provide a quantitative approach to design an adaptive chain that dynamically adapts the size of the block to varying performance trends, in which a state transitions from i back to 0, where 0<i</=n, are tracked in order to demonstrate the dynamically adaptive size of the block. Lastly, an asynchronous chain model, as a solution of dynamic blockchain in a proactive manner, is proposed based on a Variable Bulk Arrival and Asynchronous Bulk Service (VBAABS) queueing model is developed and presented to study and demonstrate the fully asynchronous and staged asynchronous chains. The analytical models are simulated extensively to compare the basic performances of the proposed models such as the average transaction waiting time, the average number of slots per block, and throughput. Further, extensive experiments are conducted in order to validate the analytical results by redesigning the source code of Ethereum to implement and demonstrate each of the proposed chains such as the baseline, the adaptive, the fully-asynchronous and the staged-asynchronous chains. The analytical results and the experimental results will be compared and discussed extensively. | |
