Simulation Systems Department.
Elaboration of instrumental systems for simulation was the result of our participating in development and exploitation of simulation models of definite complex systems.
The instrumental simulation systems as a rule are almost as flexible as general programming languages are, and the bigger part of the work with such a system is writing in a general programming language. But the instrumental system proposes (and often even imposes) a strict and consistent concepts as of the model design and simulation, either of the work distribution among the team of simulationists, then gives the facilities to carry this concepts through the project, and very many useful tools for all the stages of the work, including the debugging. So modeling with an instrumental system is easier and clearer then writing the entire model in a general programming language, but still it requires a high qualification from the simulation team. To continue the above musical comparison, the instrumental systems may be compared with an electronic synthesizer or even with a small studio. To manage it, is of cause easier and cheaper than an orchestra, though sometimes it allows reaching almost an orchestral sound quality, even when playing a melody with one finger.
In the Simulation System Department the Multilingual Instrumental System of Simulation (MISS) was elaborated. The characteristic features of MISS are multimodeling and object-event approach.
The heart of the Object Approach is the idea of decomposition. Nobody can clear comprehend how such a complex system as the Strategic Defense Initiative works entirely. But one can describe how works an observe station, or a command center, or a missile. The idea is to decompose the whole system into set of objects and simulate each object separately. Then to link the simulated objects into the whole model, run it, and see what happens. The problem is to find rules of synthesis, which would permit to link separately simulated objects into the entire model. But the reality itself encourages solving this problem: most of complex systems consist of their components, which will be the objects in our approach, and interrelations among them are what they are (i.e. are known as a part of arrangement of the system). So such decomposition is to conserve the system arrangement, for the future synthesis.
Multimodeling Approach. Modeling complex systems requires a “model of models”, or a hierarchy of models where each model represents the systems at a given level of abstraction. The need for this kind of model was first brought out when blending discrete event methods with continuous methods. As the essence of the multimodeling is the decomposition again, it may be combined with an object-oriented approach: a model is supposed to be a group, and each group may consist of groups and/or objects. So there is a hierarchy of object-oriented models. The multimodel is computationally more attractive than the single level model because the analyst may weave through the abstraction network while focusing the computation of dynamics only in those areas that require additional computation. To design more effective models, one needs to have models containing more than one level of abstraction. Such a model may be more complex, but it can answer a larger class of questions than a single-layer model. It gives an ability to choose the lowest level of detail on those parts of a model which are of special interest.
A further step in the simulation concepts development was the Event Approach. The idea to serve an event whenever it occurs is known in programming from the time of the first Interrupt Service Routines (ISR). A computer processes its routine job, and when an inquiry from some external device occurs, the computer interrupts its job and an ISR serves the inquiry, then the computer resumes its interrupted job, though it may be now influenced by the results of the ISR call. But the most popular this idea became in programming of 90th, when event-oriented languages appeared. In simulation of complex systems the Event Approach was used from the end of 80th. From the simulation point of view, the main idea of the Event Approach is the decomposition again, but here it is the decomposition of time. This decomposition conserves the set of system events. If there are no events, all elements of all the processes do their routine job, but it is known about every element what kind of system events it may cause, and before the element begins its routine job, it schedules the nearest system event it will cause in future. A system event may change the order of elements execution, and then the routine job of all the elements repeats, till the next system event occurs.
This scheme of alternation of the routine job and the system events is the base of the synthesis, discussed in the section devoted to the Object Approach. Taking it along with the Object Approach and Multimodeling, one can get the Multimodel Object/Event Approach ? the most powerful of contemporary methods for the Simulation of Complex Systems. The main advantage of this approach is that all the interrelations of the elements become carried over the system events – the points of synchronization of all the processes – while all the elements, while running its routine jobs, are independent from each other. The power of this decomposition allows the members of simulationists’ team to elaborate the routines of elements concurrently and independently, knowing near nothing about the neighbor’s work and the entire project, but being sure that the entire model would start to work properly, just after the programming and debugging of all the elements was completed. So this approach is profitable from the point of view of the team programming technology, and the last fact is a sufficient advantage of the method, as the Simulation of Complex Systems is always a task for a team. Further, as between the two adjacent system events the elements run independently, so they may run concurrently, on parallel processors or on several workstations in a computer network. So the Multimodel Object/Event Approach leads to the idea of Parallel, Distributed and Web-based Simulation. (Hyperreference to the text in Russian)
In more detail MISS is described in the following publications:
Yu. Brodsky, V. Lebedev, V. Ogaryshev, Yu. Pavlovsky, G. Savin General problems of simulation of complex systems. In the book "Cybernetics questions"; Moscow: Acad. of Sci. of USSR, 1990, pp.42-48.(in Russian)
Yu. Brodsky, V. Lebedev Instrumental System for Simulation MISS Computer Center of Russian Science Academy (in Russian). 1991, 179 p.
Back to CCAS page
Copyright 2000, Simulation Systems.