A complex system is comprised of a (usually large) number of low level agents (simple machine with limited capability) that interact concurrently within an environment and exhibit high level behaviour. The understanding of which requires “development”, “nonlinear models”, “out-of equilibrium descriptions” and “simulations”. Basically, the environment provides the medium for the interaction with the agents. The collective effort of such low level agents can be expressed as some sort of pattern of high level behaviour. It is not possible to trace the origin of high level behaviour from the low level behaviour. So simulation is way to investigate the behaviour of such systems. Before simulating a complex system, we have to model the complex system by observing the various agents and their behaviour and the way in which they communicate and interact with the environment.

In numerous scientific discipline complex systems modelling and simulation is becoming important because the real time experiment could be disaster such as human patient simulation as well as its time and cost involve. Powerful tools can me made by using the accurate simulation and that result would help for the future experimental work. Though number of modelling and simulation are already exist such as Breve and Repast for environments which allow agent based complex system simulation. As this research was influence by CoSMoS Project, which is funded by EPSRC and Microsoft research. University of Kent and York aims to build generic modelling and simulation for complex systems that can be use for analysing, designing and exploring complex systems under a single unified framework. This modelling process will be for the generic complex systems and the environment would be heavily concurrent and distributed by using a process oriented programming model. Final infrastructure will be supported for the number of processing platform such as FPGAs, general PCs and cluster. Figure-1 shows the modelling and simulation process for CoSMoS project.

In order to implement distributed complex system simulation, I can be done through designing distributed simulation framework. We can use a number of multi agent based model to develop distributed simulation framework such as genetic-algorithms, ant food foraging, cellular-automaton, boids, annular sorting and L-systems. In order to design the distributed simulation we select bird flocking model, boids as a case study.

Implementation of Boid Simulation

Network design

It is not enough from a single simulated framework to decide which model is efficient for complex systems simulation. Complex system is based on large number of agent interacting concurrently in a distributed environment. These large numbers of agent interaction exhibit high level behaviour. To get close to the complex systems, simulation is the way. Different simulation might provide different behaviour of the complex systems. So, in future we can create number of simulation using multi agent base model such as genetic-algorithms, ant food foraging, cellular-automaton, annular sorting and L-systems in process oriented language JCSP.net. From the number of resulting simulation we can model an unified generic model which can used to simulate any complex systems.

Text