Kostanai State University A.Baitursynov, Kazakstan
Analytical, simulative, natural modeling and its methods.
One of the most important methods of scientific
knowledge is the simulation. During the term of modeling the object of study is
replaced by a simple object called a model. The most important species of the
modeling process are considered two of its kind - the physical and mathematical
modeling.
With the
physical (so-called. Full-scale) modeling system under study is replaced by a
similar material system, which reproduces the main features of the system while
preserving their structure. For example, the trial network, which is determined
by the possible organization of a network based on a variety of computer
systems, a variety of routers, common operating systems and
applications.
However, there are some limitations of physical
modeling capabilities. Although this method allows us to solve a variety of
problems with a small number of combinations of parameters of the system under
study. In fact, with full-scale, ie, the precise modeling of the computer
network is almost impossible to control its operation for a variety of options
in terms of diversity and differences in the types of network devices -
switches, routers, etc. Checking
even several different routers, objectively differ in functionality and usage
environment, is associated not only with the temporary loss and great efforts,
but also with serious material costs, which are undesirable.
But at a time when network optimization requires
no changes of types of devices or operating systems, but change only parameters and operating components of any conduct experiments
in real time for an incalculable number of different settings of these
parameters is not practical for how-ever a reasonable time. Even the usual
increase/decrease the size of the package in any of the existing protocols
require a separate operating system configuration in a larger number of
stations included in the network, which will require Herculean effort from the
administrator.
In this case, network optimization is preferable to use a method of
mathematical modeling. The main feature of the mathematical model is a
relatively complex set of commands and set conditions (equations and
inequalities, the formulas, the various logical conditions) determining changes
in the system according to the current settings, the initial conditions,
operating configurations, the input data and time.
In a separate category of mathematical models
should be allocated simulation models. These models are mainly its diversity -
is a computer program that incrementally and systematically generates events
theoretically occur in real time. The objectives of simulation models of
computer networks are in the process of connecting your computer to the
network; separate commands into packets and frames necessary protocols;
reproduction of the generation of important messages application, time and
queue delays associated with the processing of packets or commands within the
operating system to accept or to refuse to processing and dispatch of all
incoming packets, etc.
While using a simulation network environment does not
make sense to purchase expensive equipment - all work is simulated in software,
satisfactorily reproduce all the options and features of the equipment
available.
The main advantage of simulation models is the
ability to replace the processing of the commands in the system studied in real
time on a much accelerated process, limited only taming rate of the program.
Thus, within a few minutes, you can simulate the operation of the network for a
period of up to several days, allowing the experimenter an opportunity to
expand the boundaries of variation of parameters to evaluate and predict the
network accessible way.
The result of the simulation model is reflected in
the statistical data collected in the course of monitoring the work of the
network of the most important characteristics: reaction time of receiving,
processing and shipping, the usage and utilization of channels or nodes that
share the probability of packet loss, etc.
There are some special languages
used in the simulation are compared to common purpose programming
language significantly accelerate and streamline the process of creating
software model. These languages can be confidently called, for
example Simula, GPSS, SIMDIS.
Today also developed a variety of simulation system,
which focus on a narrow class of systems under study and allow to build models
without programming. More details about these systems can be found at the
following links:
http://citforum.oldbank.com/nets/optimize/locnop_09.shtm