HOME PUBLICATIONS PATENTS RESEARCH WORKING PAPERS ON-LINE PUBLICATIONS LINKS

FOREWORD This page contains links to some publications that have not yet been submitted for publication in any other outlet. The author copyrights these materials and reserves all rights with respect to the materials, including the right to remove, revise, submit for publication later, or otherwise, without notification. You are welcome to quote briefly from these materials as long as they are cited. If you wish to quote extensively, kindly email the author mtortore@rci.rutgers.edu for permission. I hope you find these papers interesting and useful. SERVICE RELIABILITY In about 1996-1997, it became apparent to me that when people talk about “network reliability,” what they often mean is reliability of some function or task that the network is supposed to perform. These tasks are usually understood¾and sold¾as services, so it was natural to inquire whether there was any theory or engineering framework for service reliability. While there were a few studies in the literature that had this flavor (see the references in the paper), results were scattered and it could not be said that there existed a coherent theory for service reliability. I therefore undertook to create this theory. In a sense, this begins as a relatively straightforward application of the mathematical theory of reliability to services. The unusual wrinkle, though, is that services are intangible and perhaps this accounts for the lack of attention that reliability of services had received. But following the design for reliability roadmap (see the 1998 Chan and Tortorella tutorial reference on the PUBLICATIONS page) that starts with failure modes and failure mechanisms and proceeds through quantitative modeling and risk analysis does yield useful results. You can read about the details in the paper, which focuses mainly on applications in telecommunications (because that’s the industry I was working in when this all began). In fact, thanks in part to the efforts of my colleagues Dave Hoeflin and Hossein Eslambolchi, AT&T has begun to adapt and use these ideas in designing and managing the telecom and datacom services they sell. However, you should remember that the service reliability theory and engineering ideas are generic and can also be applied in other diverse areas such as logistics, health care, financial services, and anywhere there is a service delivery infrastructure that is used to deliver a service to customers who are external to the service vendor. A future paper for journal publication will expose the material from the more generic point of view. To view this paper, which is in postscript format, you need a viewer like Ghostview. NUMERICAL SOLUTION OF RENEWAL-TYPE INTEGRAL EQUATIONS To compute some of the reliability figures of merit in SUPER, the system reliability modeling software developed at Bell Labs around 1985, it was necessary to solve some renewal-type integral equations. We had the additional wrinkle that the underlying CDFs were only known at discrete points, so we wanted to avoid methods that required continuity or differentiability. So I developed some new quadrature methods for Stieltjes integrals and used them to get a new method for solving renewal-type integral equations numerically when no densities were available. The paper you see here is an enlarged version of one that is submitted for publication in the INFORMS Journal on Computing. This version contains a full list of references to many different numerical techniques for dealing with these equations as well as a few more examples that were omitted from the published paper. To view this paper, which is in postscript format, you need a viewer like Ghostview.