System and software cost correlation to reliability

Abstract

While technology evolves, challenges in system requirements, design and reliability modeling continue to keep pace with future underlying technology innovations and domain changes. Complex-software intense web-based information systems, with personal banking, retail marketing and retail sales evolved to online transactions and so did the need for secure information assurance to protect from illegal hacking. New electronics, computer technologies and software and system methodologies evolve to keep pace with military, aerospace, medical, human, manufacturing and cyber security requirements, system reliability evolve. In addition, shortened product development life-cycle in industry and the Department of Defense (DoD) went from three years to 18-months incorporating system reliability analysis and modeling for hardware, electronics, and software into the requirements, design, development, test and evaluation, and operations and support (O&S) sustainment phases. The challenge of technology evolution has increased the demand for systems to be built to meet reliability requirements and designed to meet the human need at least possible cost. Cost Estimation Model II (COCOMOII) estimates software engineering effort and cost of a software system based on source lines of code, engineering team capabilities, reliability and process maturity [1]. Cost Estimation System Engineering Model (COSYSMO) estimates system engineering effort and cost of a system based on system requirement counts, engineering team capabilities, reliability and process maturity.[1] Reliability and cost modeling for a continuous complex system using common reliability models, COCOMOII and COSYSMO, demonstrated a correlation between reliability with cost and process maturity with cost during developmental phases of the product lifecycle. The empirical reliability and developmental cost data analysis using COCOMOII and COSYSMO is recognized with limitations since the cost correlation, process maturity, reliability impact was built on only one system during developmental phases. Despite limited samples, these preliminary results indicate that it is possible to estimate developmental cost increase as a function of reliability improvements and developmental cost decrease as a function of process maturity. Additional research and analysis should be done to replicate the analysis. Since these COCOMOII and COSYSMO do not estimate effort and cost during the O&S phase, an O&S cost model was provided to demonstrate the correlation of system and software cost to reliability for an evaluation of total ownership costs (TOC) and return on investment. To model total TOC which includes O&S costs, the Army Logistics Support Activity Cost Analysis Strategy Assessment (CASA) model was run on Predator Unmanned Aerial Vehicle (UAV), Global Hawk UAV, MH-60S Fleet Combat Support Helicopter, CH-47F Improved Cargo Helicopter, and Force XXI Battle Command, and Brigade-and-Below (FBCB2) systems [2]. The CASA model estimates TOC for the entire life-cycle of a system, implemented by Navy, Army, Air Force, NASA, U.S. Coast Guard, and Marines. The CASA model with the five systems established a relationship between achieved reliability improvement and reduction in O&S cost. The empirical reliability, investment, and O&S data analysis is recognized with limitations since the reliability investment and reliability impact was built on only five systems. Additional research and analysis should be done to replicate the analysis. Despite limited samples, these preliminary results indicate that it is possible to estimate O&S cost reductions as a function of reliability investment and reliability improvements. This is significant since O&S costs account for over two-thirds of the TOC [3]. Further research can be done to evaluate process maturity impact on the entire acquisition life-cycle TOC. It is also recognized that the applicability of the results may be limited to the specific systems under evaluation, although this data are typical of those for mission critical systems comprised of hardware and software. Further research is necessary to determine whether the results are applicable to systems in general or better specified to particular types of systems. Whether the results would be applicable to other DoD or commercial systems could only be determined by additional research.