A Quantization-based ODE Approximation and HPP-LGCA Approach to ARGESIM Benchmark C17 ‘SIR-type Epidemic’ in a DEVS Environment based on MATLAB

Simulation Notes Europe SNE 21(1), 2011, 57-60
DOI: 10.11128/sne.21.bn17.10053


The benchmark shows on the one hand that both QSS approaches deliver the same results as MATLAB’s ODE45 solver. Moreover the QSS approach offers additional methods for solving stiff-systems. These methods use future derivatives similar to implicit time integration methods. Because the QSS approach can be seamless integrated in a discrete event simulation environment like the PDEVStbx., its application is particularly useful in hybrid system simulation. On the other hand the benchmark shows that even CAs can be modelled on the basis of the common PDEVS specification. Hence, it is possible to simulate models regarding to time and space. However, the generation of the domain is difficult and the simulation quite slow. The reason for this is the fact that PDEVS coordinator algorithms focus only on the coupling relations between individual atomic DEVS and not on their position on a domain or their neighbourhood. This leads to ineffectiveness and difficulties if the model is com-posed of many thousand square lattices which are only coupled with its four nearest neighbours. These drawbacks can be overcome by using the Cell-DEVS formalism [4] which allows the definition of cell spaces based on the DEVS formalism and CA models.