First, the software overview sensitivity
Sensitivity is the influence of different specific parameters and set values under different conditions Results of the software. Susceptibility testing is to find a valid input class may cause instability or some combination of some of the data is not processed normally carried out tests, the general need for sensitivity testing:
(1) may be found to be effective to cause the input test for a certain class of unstable combined data; (2) the input class found to be effective in the test data may cause some abnormal combined treatment. Testing process can be found in small, even insignificant mistakes.
The sensitivity of the software primarily embodies two aspects: environmental conditions and specific parameter settings.
Environmental conditions: the result of the same software in different software environments may not get the same result if the software functions to be achieved by outside environmental conditions and impact of relatively large, the sensitivity is higher.
Specific parameter settings: if a slight change these parameters on a particular impact on the operation of the software is relatively large, it is considered the great sensitivity of the software, on the contrary, if a little change these parameters to run a specific impact on the software is relatively small, is considered the sensitivity of the software is very small.
Second, the software sensitivity of the test method
Sensitivity analysis test method is a method of analyzing uncertain. Sensitivity analysis test analysis refers to one or a set of key indicators of the impact of an uncertain extent a certain change research relevant factors from the perspective of quantitative analysis occurs. The essence of the method is to change the values of the relevant variables to explain one by one through the key indicator of changes in the laws of these factors affect the size of the subject. Sensitivity analysis is an important basis for analyzing test the effects of different conditions and different setting values on the result of changes in the sensitivity analysis is to test the software operating results reflect the uncertainty of the impact of software functionality and performance is an integral part of the evaluation process. Due to the randomness of a large software or software model in the process of testing, factors and thus affect the test results of many, such as operating environment and some of the data, so there must be environmental conditions and parameters sensitivity test analysis under these conditions to arrive at the information on the important role of software design, as well as reference value, so as to provide a theoretical basis for the uncertainty of the results of the software.
Third, the software example shows sensitivity
Example 1:
Demand Requirements: Radio call quality should be clear intelligibility in normal circumstances.
Factors: frequency of the signal to noise ratio, sensitivity, etc.
Means of authentication: one carried out by changes in uncertainty, found little change in certain factors, there is a serious impact on call quality, this factor is an important factor affecting the call quality is explained.
Key factors in this verification test, to ensure that this factor does not affect the index in the range may vary.
In the design, the focus should also be to ensure that this factor can not be much change in scope.
Example Two:
Represents formula: Y = A * 0.001 + B * 0.999
The indicator: Y is a range of values
Uncertainty factors: A and B (PS: is a black-box testing under normal circumstances, do not know the above formula)
Gradually trimming uncertainties A and B, look for sensitivity factors;
The analysis sensitivity factor B.
You need to ensure range of Y under the same circumstances, changes in the scope of B can not be too large, if the B range is too large, resulting in the index, but off.