Chapter #
This indicates the interest evinced by readers the world over in the subject of performability engineering. Encouraged by this interest, Scrivener Publishing LLC undertook the publication of a series of books in the area of performability engineering with Prof. Krishna B. Misra and Prof. John Andrews of Nottingham University, UK as Series Editors in 2014, and it is matter of satisfaction that as many as 10 books have been published under this series so far. The books published under this series have solicited good number of citations in spite of the fact that much time has not elapsed after their publication.
As an Editor of this series, I solicit manuscript of books from academicians, engineers and researchers in any area of Performability Engineering. For details, please contact the Series Editor.
Krishna B. Misra, Series Editor
June 2020
1 [1] Misra, Krishna B. (Editor), Handbook of Performability Engineering, Springer, London, 2008.
Preface
Network reliability evaluation for complex communication networks is a time-consuming task. Complex networks such as computer communication networks, telecommunication networks, transport networks, etc. are an integral part of our daily life. Complex networks interconnect multiple processing units in distributed systems. Performance of such networks can be measured by evaluating the reliability index, which represents the probability that the network operates satisfactorily for a given period of time when used under stated operating conditions. A network designer has to choose suitable fault-tolerant and highly reliable network architecture to avoid performance degradation and carry out multiple communication tasks concurrently.
Parallel computers with multiple processors have emerged to meet energy-efficient performance demands of current and future challenging computing applications. These multiprocessor systems need interconnection networks for connecting processors and memory modules. Advances in parallel and distributed computing have made interconnection networks a potential networking alternative to meet the growing demands of high-performance computing applications. These networks need to perform continuously without repair for long periods of time. Therefore, these need to be reliable and fault-tolerant as these networks are quite complex and big in size. Failure of its components should not lead to complete network failure as it may be catastrophic.
The main objective of this book is to device approaches for reliability modeling and evaluation of such complex networks. Such evaluation helps to understand which network can give us better reliability by their design. New designs of fault-tolerant interconnection network layouts are proposed, which are capable of providing high reliability through path redundancy and fault tolerance through reduction of common elements in paths. This book covers the reliability evaluation of various network topologies considering multiple reliability performance parameters (two terminal reliability, broadcast reliability, all terminal reliability, and multiple sources to multiple destinations reliability).
For decades, work on network reliability evaluation has been going on. Our S.C. School of Quality and Reliability, earlier known as Reliability Engineering Centre, has significantly contributed to this area of research during the 1980s and onwards. The school has been involved in developing models for reliability evaluation of complex networks but static and ad hoc. The authors of this paper have significant contributions in this area.
Interconnection networks pose another challenge to reliability evaluation due to their complex and large architectures. There are very few in this area. This subject matter has not been much addressed in easy-to-follow book materials. This book tries to bring the aspects of latest research conducted at S.C. School of Quality and Reliability to have better reach to audiences. This is the first edition of the book, so there are many improvements possible. The authors can be provided with feedback to improve the contents of the book in the possible next edition.
This book first introduces the basic concepts of network reliability and the popular interconnection network architectures in the first two chapters. Then it discusses approaches used for reliability evaluation of such a network in Chapters 3, 4, and 5. Some new novel architectures providing better reliability and fault tolerance while being cost effective are presented in Chapter 6.
The authors would like to thank everyone (faculty, students, and staff) from S.C. School of Quality and Reliability, IIT Kharagpur, India, for their encouragement, suggestions, and support while carrying this analysis. We would also like to thank Prof. K.B. Mishra and Martin Scrivener for their guidance and support in writing this book.
Dr. Neeraj Kumar Goyal and Dr. S. Rajkumar
August 2020
1
Introduction
1.1 Introduction
In recent years’ human beings have become largely dependent on communication networks, such as computer communication networks, telecommunication networks, mobile switching networks etc., for their day-to-day activities [1]. In today’s world, humans and critical machines depend on these communication networks to work properly. Failure of these communication networks can result in situations where people may find themselves isolated, helpless and exposed to hazards. It is fact that every component or system can fail and its failure probability increases with size and complexity.
Therefore, it is essential to compute and assure the reliability of these networks, which are growing and becoming complex. Reliability modeling and computation is necessary for reliability and safety assurance of these networks [2]. It also helps to identify weak links. These weak links can be improved cost effectively using reliability design techniques. Recent developments in communication hardware industry has
