### Table 4. Evaluation of the Algorithm with one Faulty Sensor

"... In PAGE 3: ... Sensor Failure The effect of the failures depends a lot of their location. The results presented in the Table4 are divided into two parts whether if the failure is inside or out side of the hot spot. The values presented in the table are not the absolute performance of the algorithm but are relative to the ideal case without failure.... ..."

### TABLE I GROSS SYNDROMES FOR BISTER-1 AND THEIR DIAGNOSIS UNDER ASSUMPTION OF AT MOST ONE FAULTY PLB.

### Table 2: Gross Y1a4 Y2 syndromes for BISTer-2 for one faulty PLB assuming that a faulty PLB also fails as a TPG and as an ORA.

2004

"... In PAGE 4: ... Proof: We rst show that BISTer-2 is 1-diagnosable for the case that the faulty PLB also fails as a TPG and ORA; the proof for the case it passes as a TPG and as an ORA is similar. Table2 which is self-explanatory shows that the gross syndromes Y1 for each faulty PLB is unique. Hence BISTer-2 is 1-diagnosable.... In PAGE 5: ... Similarly, any detailed syndrome equality between any of the other three faulty pairs is extremely unlikely. Finally, all the gross syndromes (Y1, Y2) of Table2 (1 faulty PLB) are distinct from those of Table 3 (2 faulty PLBs). We can thus distinguish between the syndromes for single and dou- ble faults.... ..."

Cited by 3

### Table IV enumerates every posible combination of error states in any component. In the three first columns R stands for Running (the running pseudo state of Fig. 6) and E for Error (the running error state in Fig. 6). Any of the errors shown in Table IV does not lead the system to a fatal situation, except if we consider the last case. The asynchronicity of execution, initiative and communications between the components which form the systems allows it to keep the systme working until the faulty one is recovered, or the whole system is explicitly stopped.

### Table 1: Computed resistance ranges of potentially faulty resistors.

"... In PAGE 8: ...Table1 summarizes the results of diagnosis. The example illustrates the basic idea of using CLP( lt;) to diagnose soft faults: each individual component is considered in turn and its actual value is computed from the measurements.... In PAGE 17: ...3 Example 3: R5 100k instead of 10k In the last experiment, we inserted a deviation fault of R5 (100 k instead of correct 10 k ) into the circuit. Like in previous examples, measurement results (Table 9), computed values of the suspected components for the four modes of operation ( Table1 0), and the computed mean values and ranking of the suspected components are given (Table 11). From the computed values of components for the normal mode, R3 can be eliminated due to its negative value.... In PAGE 18: ...423 x x x x 1005 0.002 Table1 0: Computed values of components for di erent test modes for R5 = 100k . values of gain and phase in the all-test mode indicate that the faulty element is one of the resistors.... In PAGE 19: ...0698 R3 997 3.371 Table1 1: Ranking of the suspected components and their predicted values for R5 = 100k . CLP( lt;) and model-based diagnosis techniques.... ..."

### Table 4. print_tokens (test pool size: 4130) V Faulty

2002

"... In PAGE 7: ....2.3. Data and Analysis. Table4 -10 presents the results of the experiment for each subject. The first column shows the version number and the second one shows the abbreviations of faulty function names.... ..."

Cited by 5

### Table 3: Mean percentage performances of patient KT and versions of faulty/non-faulty PbA for 161 regular words (RW) and 161 irregular words (IW). Patient KT Faulty PbA Non-Faulty PbA

"... In PAGE 5: ... The simulations were obtained for a model with damage in the third and fourth multi- strategies (see Marchand and Damper, 2000, for detailed speci cation) and only substrings of length between 2 and 4 can be segmented in pattern matching. Table3 shows KT apos;s mean reading accuracy over the various tests per- formed by McCarthy and Warrington together with our corresponding simulation results for impaired and non-faulty PbA. Clearly, it is pos- sible to reproduce quite well the patient apos;s car- dinal symptoms: his ability to pronounce regu- lar words much better than irregular ones.... ..."

### Table 4. Simulation results w.r.t. the identification of faulty processors - second run

"... In PAGE 20: ...presence of permanent/intermittent faults significantly diminishes the probability of executing the second phase for this strategy. Table4 shows the behaviour of SCP and 2+1 with respect to the identification of faulty processors in the present case. Numbers in Table 4 are smaller than the corresponding ones in Table 3 because of the smaller number of faulty processors now present in the system.... In PAGE 20: ... Table 4 shows the behaviour of SCP and 2+1 with respect to the identification of faulty processors in the present case. Numbers in Table4 are smaller than the corresponding ones in Table 3 because of the smaller number of faulty processors now present in the system. However, the same comment as in Table... ..."

### Table 1. Eigenvalues: Normal State vs. Faulty States.

"... In PAGE 5: ... The normal state of operation is modi ed by a linear trend added over the entire pro le from which ten snap- shots are collected, introducing a nonstationary pattern (Faulty State 4). Simulation Results Table1 presents the eigenvalues (Equation 2) for the normal state of operation, the set of simulations with sta- tionary changes, and the set of simulations with nonstation- ary changes over the snapshots. Note that the eigenvalues correspond to the mean-squared value of the coe cient vec- tors.... In PAGE 5: ... A change in the severity of a particular pattern can be indicated by a corresponding change in the signi cance of the eigenvalues. The signif- icance of the resulting eigenvalues in Table1 will be dis- cussed in the following subsections. Notice that the eigen- values are always listed in a descending order.... In PAGE 6: ...the multi-component signal is decomposed into 4 signi cant eigenvalues, shown in Table1 , summing to 98% of the total energy in the signal. The rst two eigenvalues (#1,#2) cor- respond to low-frequency sinusoidal eigenpro les, while the next two eigenvalues (#3,#4) correspond to high-frequency sinusoidal eigenpro les.... In PAGE 6: ... They can be used to determine whether changes in the am- plitude of one of the sinusoidal components of the multi- component signal can be detected with the KL transform. The eigenvalues from faulty states 1a and 1b, compared to the normal state, are shown in Table1 . As in the case of the normal state of operation, for faulty states 1a and 1b, we obtain 4 principal eigenvalues and eigenpro les.... In PAGE 6: ... As in the case of the normal state of operation, for faulty states 1a and 1b, we obtain 4 principal eigenvalues and eigenpro les. When we increase the magnitude of the high-frequency component of the input signal, we notice that the eigenvalues correspond- ing to the high-frequency component increase in magnitude (Eigenvalues #1 and #2, Faulty 1a and 1b, Table1 ), com- pared to the corresponding eigenvalues in the normal state of operation (Eigenvalues #3 and #4, Normal, Table 1). Notice that the magnitude of the eigenvalues for Faulty states 1a and 1b re ect the increase in the magnitude of the high-frequency sinusoidal component.... In PAGE 6: ... Notice that the magnitude of the eigenvalues for Faulty states 1a and 1b re ect the increase in the magnitude of the high-frequency sinusoidal component. The eigenvalues corresponding to the low-frequency component (Eigenval- ues #3 and #4, Faulty 1a and 1b, Table1 ) remain ap- 6 Copyright c... In PAGE 7: ... In the simulations, the rst ten snapshots in- dicate normal status, while the next ten snapshots indicate an o set change. As shown in Table1 , this fault manifests itself as an added principal eigenpro le, with a relatively large eigen- value (Eigenvalue #1); the remaining eigenvalues are ap- proximately the same as the normal state of operation (Eigenvalues #2 and #3 in the faulty state 2 compare to eigenvalues #1 and #2 in the normal state, etc.).... ..."

### Table 1, where Injected Faults is the number of injected SEUs, and Wrong Answer is the number of SEUs for which the faulty circuit produces outputs that differ from the fault-free one. During our experiments we adopted a workload composed of 100,000 randomly generated input stimuli.

"... In PAGE 5: ...Add8 15,000 14,587 1,352 30 Add16 15,000 14,598 1,692 41 Mul8 15,000 14,603 1,977 23 Table1 . Fault Injection results From these results we can observe that most of the in- jected faults provoke erroneous behaviors in the plain, un- hardened circuits.... ..."

Cited by 2