Life-cycle analyses software

Compiled by Dr. Candace Wheeler (Staff Research Scientist, Vehicle Emissions and Life Cycle Analysis, General Motors Corporation) using GaBi Software System for Life Cycle Engineering (University of Stuttgart and PE Europe GmbH). 

Robust analysis phase The construction of Catalysis business and requirements models covers more than in the more conventional style. In Catalysis, more of the important decisions are pinned down. As a result, there is less work later in the design stage and less work over the maintenance part of the life cycle, the part that accounts for most of a software system s cost. (To cope with any uncomfortable feeling of risk that this approach may generate, see the remarks in Sections 1.11.2 and 1.11.3.)... 

Evidence of sufficient control of these issues should be demonstrated in the validation documentation. Compliance must be integrated using a formal methodology and an appropriate system life-cycle approach that is clearly identified in the user requirements phase for any new computerized systems. The priority for validation activities can be established by analyzing the control scheme system and subsystem inventory for the criticality, validation stams, software category, and system type. This analysis aids validation planning and prioritization. 

Computer Aided Software Environments (CASE) Tools designed to support the analysis and design phases of the software development life cycle. The tools are usually oriented toward the support of graphical notations. [Defined for this book.]... 

The software life cycle starts with a requirements analysis and product definition. These define the require-... 

NIST. 2006b. Life-Cycle Cost Analysis Tool for Chem/Bio Protection of Buildings. Accessed on March 15, 2007, at http //www2.bfrl.nist.gov/software/LCCchembio/index.htm. 

Software hazard analysis is extremely important and will be of growing importance in the future. The software hazard analysis effort should parallel the system safety program for system hardware, be a life cycle effort, and use a combination of methods and approaches. 

Waterfall ModeL The waterfall model of software development is characterized hy a sequential software development life cycle. The key characteristic of the waterfall model is that each step of the software development life cycle is completed before going to the next step. The first step is to perform an initial study of the problem to be sure it is feasible. In the second step, a detailed analysis of what the project entails is completed. This is one of the most important components of the waterfall model and often uses a specialist called a systems analyst. In the third step, the information contained in the systems analyst s report... 

The last question is to determine if we could use such data for reliability prediction. Early in the process, for example, at the requirements definition and analysis phase, our knowledge of the attributes and characteristics of the software development process is limited coding personnel and coding languages are undetermined, the design approach may not be defined, etc. Hence, our reliabihty assessment will be uncertain. The more we move into the life cycle process, the more information becomes available, and hence our uncertainty bound reduces. Let us call M the model of the process, that is, the set of characteristics that define the process. The characteristics of such model may include... 

Our uncertainty is in M. Let p(Al) be the probabihty that the model followed is M. This probability is a function of the time into the life cycle, p M,t). A reasonable approximation of p M,t) is p(M,k) where k is the fcth life cycle phase. Let , be the rehability of the software at the end of the software development process if it was developed under model M,. Initially, at the end of the requirements definition and analysis phase (fc = 1), the expected reliability ( ) is... 

Therefore, the estimation of a medical device s safety lies the determination of the probability that a medical device would become hazardous for the user or patient, resulting in safety problems and harm. This activity is known as risk management and is regulated by International Organization for Standardization (ISO) standard 14,971 [63,64]. This process is divided in two activities, risk analysis and risk evaluation, that assist in a detailed assessment of the risks associated to the medical device s entire life cycle, from manufacturing to disposal. The estimation of the risk is a complex procedure that considers the probability of device failure or nonconformity, software, user errors, and type of injury possibly resulting to people or users. 

Authorization is necessary for software modification under the procedures specified during planning. Major issues involved in authorization shall include but are not limited to Hazard to be affected, and proposed change with necessary reason (duly documented). It is necessary to ensure that the required SIL is maintained. In this connection for detailing, clause number 7.8 of the standard may be referenced. The modification process involves an analysis on the impact of the proposed on functional safety and how much of the safety life cycle must be repeated. 

The assessor then quantifies the inputs and outputs for each step (or unit process) in the product or process life cycle. Inputs may derive from direct process knowledge, literature values, databases maintained by industry organizations or governmental agencies, or measurements. Outputs are estimated using mass balance calculations, noting uncertainties and assumptions. These calculations quickly become quite complicated due to the number of steps in the life cycle and the inputs and outputs that occur at different times and places. As a result, LCA commonly requires software tools [99,100]. Further, it may be worthwhile to perform a sensitivity analysis at the conclusion of this phase and reiterate the calculations with a refined set of assumptions. 

In this paper, a method of software safety verification at the system level based on STPA is proposed. We investigated the application of the STPA structure to software, and we found that STPA can be directly used for software. We mapped the results of the STPA safety analysis to a formal specification to be able to verify safety requirements at the software code level. The limitation of the method is that the formal specification is done manually which may lead to much effort to construct and check the potential combinations of relevant states. Therefore, we are exploring the automation of this step and integrate it with our A-STPA tool as future work. Furthermore, we plan in-depth case studies to improve the method by applying it to real safety-critical software in industry. We plan also to investigate the effectiveness of using the proposed method during an ISO 26262 life cycle in the automotive industry. 

---