China
Africa
Explore chapters and articles related to this topic
Testing Automotive Control Software
Published in Nicolas Navet, Françoise Simonot-Lion, Automotive Embedded Systems Handbook, 2017
If a “test oracle” by means of a model for the behavior of the test object exists, one can also evaluate the actual system reaction with the help of the oracle model. In this case, test evaluation is possible without explicit a priori determination of the expected outputs during test design.
Challenges in Testing Context-Aware Applications
Published in Syed Ijlal Ali Shah, Mohammad Ilyas, Hussein T. Mouftah, Pervasive Communications Handbook, 2017
Testing context-aware applications, especially testing for adaptation faults in the applications, are quite challenging due to factors mentioned in the previous section about context modeling and the followings [5,9]: There are several representations of contexts along with different formats used.The context variables are updated asynchronously at different rates by the middle-ware, causing transient inconsistencies between external physical context value and its internal representation within the application. A context variable can contain static information about local configuration of the mobile device (such as language preference) or dynamic information (such as GPS latitude and longitude). The dynamic context variables need to be refreshed (periodically or asynchronously) at different rates, which leads to synchronization problems when the adaptation manager tries to relate several context variables for rule triggering.The user may configure its own behavior and, hence, some of the context variables. This may lead to context inconsistencies and/or failure due to buggy user-defined configuration.The space of rules for adaptive actions become very complex to analyze due to several shared context variables, concurrent triggering of rules, and so on. Multiple rule predicates may be satisfied simultaneously and some predicates may be satisfied transiently.It is very difficult to define precise test oracle as the execution differs under various vectors of context input.The real environment to run application may not be available. There may not be, for example, enough sensors, or different types of networks available at the time of testing.As context may include sensitive information about people and their activities, some applications give opportunity to the user to protect their security and privacy. For example, the Context Toolkit [6] introduces the concept of context ownership. Under such case, it becomes even more difficult to test such context-aware application.
Neural network-based multi-view enhanced multi-learner active learning: theory and experiments
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2022
A Software Test Oracle is used during software testing to identify if the application under test (AUT) behaves as expected (Whittaker, 2000). It is a reliable source that provides the expected outcomes for any input combinations or user interaction specified by the software blueprints (Ran et al., 2009). A fully automated software testing framework is not possible unless the test oracle is automated to generate the AUT’s expected outcomes.