Современные информационные технологии/Компьютерная  инженерия

Master students Abeuov D.E., Dara A.S., Sakanova Zh.E., Umirbekova B.K.

Scientific supervisor Candidate of Technical Sciences Kurmanov B.K.

Kazakh-British Technical University, Kazakhstan

 

Analysis of the information processes in corporate systems and networks using GPSS World

 

Introduction. Currently, imitation models are becoming the most common class of mathematical models. These models represent a computer program that reproduces the events occurring in the real system step by step. Applied to computer networks their imitation models reproduce message generation processes by applications, partitioning messages into packets and frames of certain protocols, delays in processing of packets, and frames within the operating system, the process of obtaining access to the shared computer network environment, and etc. Imitation modeling does not require an expensive equipment, because its work is imitated by programs, quite accurately reproducing all the main features and parameters of such equipment.

The advantage of imitation models is the possibility to substitute the process of changing events in the system in real time to an accelerated process of changing events in the pace of program performance. As a result, network performance can be reproduced in several minutes instead of several days, which allows to evaluate network performance in a wide range of adjusting parameters. The result of the imitation model performance is statistical data about the most important characteristics of a network such as the reaction timely, coefficients of channels and nodes utilization, the probability of packet loss, and etc., which is collected during the observation of the proceeding events.

In this paper the problem of replenishment of imitation models library (IML) is considered. It is one of the important components of the imitation modeling subsystem of automation of local area networks design (LAN) system [1], and is designed for the evaluation and optimization of the designed LANs main parameters with the structure, synthesized using deterministic mathematical methods in the early stages, and is used in the final stage of the interactive LANs design process.

Relevance of the work is the fact that the developed imitation models contribute to the global goal of the corporate information system: providing the effective use of information resources, timely submission of necessary information for decision-making and effective management of the enterprise.

The novelty of this work is the development of imitation models for the analysis of information processes in corporate systems and networks based on the principles of structural and holistic approach to the design of complex systems [1], which consists of a comprehensive consideration of the structural and functional design of computer systems and networks.

Results. As a result of formed IML, imitation models for time index evaluation and LAN boot indicators were developed, using different access methods to a common transmission medium (random access with carrier at different levels of persistence, random access with Collision Detection, marker method , interval-marker method, combined methods).Furthermore, imitation models for communication protocols analysis via LAN are proposed. The elements and components of LAN are presented in the form of inter-related single and multichannel queuing systems with the following assumptions:

-                  Incoming stream of requests is Poisson;

-                  station number of a sender (receiver) is defined as a discrete random variable distributed according to the uniform law;

-                   delay of a frame before its retransmission is a random variable depending on the number of collisions;

-                  conflict detection and interruption of transmission occurs within a time, including the double-path signal in multidrop bus and logical circuits response time.

All models are developed using the modern object-oriented simulation system GPSS World [2]. The developed imitation models assume the assignment of the following initial data:

-                  the maximum signal propagation time within multidrop;

-                  station response time;

-                  data transmission rate in the mono channel;

-                  the length of messages, information and service frames;

-                  the number of subscriber stations;

-                   average interval of requests reception and others.

The following GPSS World elements are used to set the initial data: functions, stored values, and variables. As a result of the launch of imitation programs, the relations between stay (service) time of requests (posts, frames) in the network, stay time of requests in the queues, service interval of devices (mono channels, stations), holding time of devices, queue lengths, the number of incoming requests and the average length of pause from incoming requests, data transmission rate in the mono channel, the length of messages and the number of stations in the network can be obtained using different access methods and parameters of the communication protocols between subscriber stations.

Conclusion.  Developed models allow evaluation of such LAN important parameters  as data transmission rate and payload of the LAN mono channel. Through numerous imitation experiments with different parameters of communication protocols and access methods to the mono channel, developed models allow to choose the best option of network organization.

Proposed models are suggested to be included in the imitation modelsbase of automation system of LAN design. The utility of the developed models in a LAN design is significantly increased when it is combined with well-known analytical methods of LAN analysis such as analytical methods for the approximate evaluation of  local network effectiveness, suggested imitation models for accurate assessment.

Block structure of developed imitation models allows to increase the granularity of the analyzed processes to the required level during the modeling.

References:

1.    Sarypbekov Zh. S., Truman A.Ch., Kurmanov B.K. Models and methods of designing local area networks. Almaty, KazNTU, 1989.

2.    GPSS World Tutorial Manual. Copyright Minuteman Software. Holly Springs, NC, U.S.A. 2001.