"... In PAGE 5: ...9 98.8 Table5 summarizes the results of fault detection. Columns two to four give the percentage of these faults detected by parity, Berger and weight-based codes respectively.... In PAGE 5: ...The results presented in Table5 show that the fault detection capability of weight-based codes is much higher than that of the other codes. It can be made significantly higher than that of a Berger code (special member of weight-based codes) by a slight increase in the number of check bits.... ..."

"... In PAGE 5: ...9 98.8 Table5 summarizes the results of fault detection. Columns two to four give the percentage of these faults detected by parity, Berger and weight-based codes respectively.... In PAGE 5: ...The results presented in Table5 show that the fault detection capability of weight-based codes is much higher than that of the other codes. It can be made significantly higher than that of a Berger code (special member of weight-based codes) by a slight increase in the number of check bits.... ..."

"... In PAGE 12: ...Based on the statechart diagrams given in Figure 10 and Figure 11 all legal/faulty TP can be identified for each state of the system. Table2 lists the pairs of (legal) in- coming and legal/faulty outgoing transition for the states of the substates. A transition in a statechart diagram can be triggered by more than one event, e.... In PAGE 13: ...odeling as introduced in [10, 19] vs. statechart modeling as introduced in this paper. For the case study #1 the same tester created the ESGs and statecharts assuring that the models describe the same functionality of the SUT. This case study was carried out in two different ways: In the first version (Case Study A in Table2 ), the tester started with the construction of statecharts and then the ESGs were constructed. Ac- cordingly, Case Study B denotes the way around: First were created ESGs and then statecharts.... In PAGE 13: ...onstructing the ESGs and statecharts, i.e., each tester created the model independ- ently from each other. Table2 summarizes the results of the both strategies. Table 2.... In PAGE 13: ...#2 12 11 5 As visible in Table2 , case study #1 detected 30+ faults, no matter which model was constructed first. Unexpectedly, constructing the statecharts and ESG separately by different testers (case study #2) lead to a smaller total number of faults detected by both models.... ..."

"... In PAGE 6: ... 4 Experimental Results and Analysis As a baseline for our experiments, we ran the full test- suites as well as a randomly generated sets-set. The results are summarized in Table1 . The table shows the number of test-cases in each test-suite and their fault finding capability (total fault finding capability as well as broken down per fault-class).... In PAGE 6: ... Thus, the randomly generated tests serve as a simple baseline for the other test suites; one would ex- pect the tests carefully crafted to provide a certain cover- age to perform better than the randomly generated test-set. As can be seen in Table1 , the randomly generated test per- form surprisingly well compared to the test-suites providing structural coverage. We have discussed the reasons behind the poor performance of Variable Domain and Transition Coverage in a previous study [17] and a discussion of this topic is outside the scope of this paper.... In PAGE 6: ... We have discussed the reasons behind the poor performance of Variable Domain and Transition Coverage in a previous study [17] and a discussion of this topic is outside the scope of this paper. From the results in Table1 one can also observe that the more rigorous the test criteria, the better the fault finding capability. For instance, MCDC with usage detects more faults (72%) than any other coverage criteria considered and also outperforms random testing (66%).... ..."

"... In PAGE 1: ... A Linux-based data acquisition and signal processing system was used to pre- process the data stream in real time and to analyze start transients and harmonic content of fan and compressor motor power signals. Fault-specific patterns were identified as indicated in Table1 . Selected fault signatures are shown in Figures 2 through 7.... ..."

"... In PAGE 7: ... The test suites developed utilizing fleld data also gener- ated gains over the in-house test suite in terms of fault de- tection capabilities. Table2 reports the fault seeded in each version, the faults detected through the in-house test suite, and the additional faults found through the test cases gener- ated based on fleld data. The test suite generated through simple translation of fleld data discovered a new fault in v4.... ..."

"... In PAGE 5: ...9 99.1 Table4 presents the percentage of faults detected with four weights. As there can be considerable aliasing within four weights, some of the circuits show significant increase in fault detection capability from {1,2,3,4} to {2,3,4,5} to {3,4,5,6}.... ..."

"... In PAGE 5: ...9 99.1 Table4 presents the percentage of faults detected with four weights. As there can be considerable aliasing within four weights, some of the circuits show significant increase in fault detection capability from {1,2,3,4} to {2,3,4,5} to {3,4,5,6}.... ..."

"... In PAGE 1: ... This paper develops an alternative approach to self-checking fault detection based on random pairings and temporal voting to reduce such exposures. Table1 lists characteristics of selected fault-handling strategies. Specialized encoding schemes... ..."

"... In PAGE 20: ...ore failures occurred. The number of faults detected was recorded and used in the analysis. 5.3 Results and Analysis The faults are summarized in Table 2, a summary of the results is given in Table 3, and detailed testing results from the coupling-based and category-partition techniques are listed in Table4 . Several of the units were developed as reusable components, and had functionalitythatwas not used in Mistix.... ..."