This site uses cookies. More information presents the page The policy for cookies and similar technologies. CLOSE Close the warning message

Piotr Nazimek sustained his PhD thesis

On February 28, 2012: Piotr Nazimek sustained his PhD thesis: "Discovering, efficiency measurement and usage optimization of software assertions", supervised by Professor Janusz Sosnowski.

Software assertions are used to improve software reliability. One way of determining them is to use algorithms for dynamic detection of assertions that serve to identify high number of different conditions, based only on informations collected during execution of the program, without the static analysis of its source code.  The nature of these algorithms and the limited amount of information to analyze needs to investigate the usefulness of discovered assertions and method of selecting a subset of them to ensure efficient detection of errors.
Assertions (also called invariants, properties or conditions) dynamic detection algorithms are designed to find different types of dependences in programs based only on informations collected during their execution without static analysis. Specificity of those algorithms and limited quantity of data to analyze needs to investigate the usage principle of discovered dependences for software dependability increase.
This thesis defines different types of measures to express efficiency and inefficiency of discovered assertions during error tolerance inspection or fault detection process. Also other measures like static cost, dynamic cost, error detection latency were defined. The methods of measurement are described for proposed values. There were also presented general optimization models in the form of integer programming problems with goal functions like efficiency, inefficiency, cost, number of assertions or error coverage.
This dissertation introduces techniques for increasing efficiency of detected assertions through using program execution trace. Conceptions of trace, assertion with trace and trace operations were defined.  The work describes several algorithms that allow discovering assertions with trace, reducing number of traces in assertions with trace set, shortening traces in assertions with trace set and reducing number of observation point id's for assertions with trace sets. Different ways of assertions with trace verification are presented.
Presented algorithms were investigated using original experimental methods. For several applications, like manufacture line control driver or implementations of compressing algorithm from ZIP family, experiments description and results have been presented.

Last modified: Wednesday, June 13, 2012 - 3:43:50 PM, Bożenna Skalska

x x News (7) - by publication date

‹‹ December 2011 ››
Mon Tue Wed Thu Fri Sat Sun
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31