Reliability Analysis of Real-Time Multitasking Software Based on Neural Network - Advanced Research on Computer Education, Simulation and Modeling - page 347

Reliability Analysis of Real-Time Multitasking Software
Based on Neural Network
Xiang Chen 1 , Wei Hou 1 , and Yong Zhang 2
1 School of Civil Engineering, Xi'an Technological University,
710032 Xi'an, China
2 Wuhan University of Science and Technology,
430071 Wuhan, China
{xichen0801,houwei75}@163.com, [e-mail]
Abstract. Real-time multitasking software is core of large scale real-time sys-
tem in industrial control. Characteristics of real-time system determine that it
laid high requirements on software reliability. Based on analyzing on character-
istics of real-time multitasking software, the paper modeled and researched on
reliability of real-time multitasking software according to module mode based
on neural network and established neural network model of real-time multitask-
ing software. Simulation results based on application example shows that the
proposed model introduced running time of each task mode into reliability
model, which adapt to characteristic of real-time multitasking software and is
more accurate than the method that does not based on task modules.
Keywords: real-time multitasking, reliability, neural network.
1 Introduction
Real-time system is widely used in industrial control, the core of which is real-time
multitasking software. For characteristics of real-time system itself, it laid high re-
quirements on software reliability, especially real-time systems as large-scale oil
processing system. As to these complex real-time systems, the core real-time multi-
tasking software is a kind of complex software with complex control relationships,
high computation speed and rigid processing order. Error of system software may lead
to failure and even disastrous consequences [1]. Software reliability can be measured
by various parameters, such as reliability, failure rate, failure intensity, Mean Time To
Failure (MTTF) and etc, which can be interchangeable each other. Real-time multi-
tasking software can also be measured by these reliability indexes. In order to quanti-
tatively analyze on its reliability, it must clear that real-time multitasking software
also obey reliability increase rule for it has inherent characteristics of general software
firstly. Secondly, there are restrictions in different aspects between real-time multi-
tasking software and traditional software in many cases, which will affect its reliabil-
ity. Finally, reliability of hardware will also impact on reliability of whole system.
According to above analysis, we can know that analyzing on reliability of real-time
multitasking software can not follow traditional methods and models of ordinary