Journal of Software Engineering and Applications, 2013, 6, 48-52
http://dx.doi.org/10.4236/jsea.2013.62008 Published Online February 2013 (http://www.scirp.org/journal/jsea)
Deadlock Detection and Avoidance in Static Step Topology
under Distributed Environment
Taskeen Zaidi
1
, Vipin Saxena
2
Department of Computer Science, Babasaheb Bhimrao Ambedkar University, Lucknow, India.
Email: taskeenzaidi867@gmail.com, vsax1@rediffmail.com
Received December 6
th
, 2012; revised January 5
th
, 2013; accepted January 14
th
, 2013
ABSTRACT
During the past years, the distributed computing approach has become very popular due to various advantages over
centralized approach. In the distributed approach, the execution of a process has reduced and also it requires low cost
for installation. Many of the researchers are using the modeling approach for solution of the software and hardware ar-
chitecture research problems. The most popular approach of modeling is known as Unified Modeling Language based
on the object-oriented technology. In the present work, a method of deadlock detection is explained for the newly pro-
posed static step topology for the distributed network. In the step topology, the processes are taken as a task, sub task,
macro, subroutine, etc which are executed in reflexive and symmetric manners when the systems are interconnected to
each other under distributed environment and avoidance technique is also presented for the same. The deadlock detec-
tion technique is presented through a UML class model.
Keywords: Distributed System; UML Class Model; Step Topology; Deadlock Detection; Deadlock Avoidance
1. Introduction
In the current scenario, many of the software companies
are changing the layout of computing labs from the old
centralized computing approach towards the distributed
computing approach as it involves low cost and saves the
execution time of the tasks. Another advantage is that
distributed computing system does not share the global
clock. In the centralized approach, computing time of the
tasks is much more in comparison of the distributed ap-
proach as the deadlock occurs frequently in the central-
ized approach. In the distributed computing approach
deadlock situation can be handled by the use of mutual
exclusion i.e. synchronization of the tasks. The present
work is based upon a distributed kind of network con-
nected through the static step topology designed by the
authors; deadlock situations are detected and methods for
avoidance of these situations are elaborated. In the pre-
sent work, a static step topology is used for the high
speed National Knowledge Network (NKN) connectivity
and deadlock detection technique is defined for the exe-
cution of the processes/tasks or one transmits the infor-
mation in terms of files from one device to another de-
vice attached through the topology. A method of avoid-
ance is also explained. On the another aspects, modeling
is necessary before executing the research problems as it
saves the time and errors can be detected at the early
stages of the software development, therefore, in the cur-
rent scenario, many of the researchers are using the ob-
ject-oriented Unified Modeling Language to construct
the various models for the solution of the research prob-
lems, therefore, in the present work UML class model is
also designed for finding the conditions of deadlock and
a procedure to minimize the deadlock situations occur-
ring in the distributed environment.
2. Related Work
In the current scenario and due to evolution of high speed
network connectivity, many of the researchers share the
information at the remote by using the remote login or
transferring the information over the distributed network.
Sharing of information in form of files, data, audio, video
as well as mutual exclusion of tasks are well explained in
[1]. The static interconnection of the network plays an
important role for sharing the files and it is known as the
topology and an important reference Tanenbaum [2] has
described various types of topologies used to intercom-
nect the devices under distributed computing environ-
ment. Various kinds of topologies like centralized, de-
centralized and hybrid, which are used for modeling, and
internetworking of distributed systems are explained in
[3]. Another important reference is Hwang [4] which
describes all the aspects related to the software and hard-
ware architecture including the parameters affecting the
execution of the processes under the distributed comput-
ing environment. Due to evolution of the windows pro-
gramming, the tasks called as processes are synchronized
Copyright © 2013 SciRes. JSEA