T. Deiß, "Test Generation for Codec Validation," ETSI TTCN-3 User Conference 2005: Sophia Antipolis (France), June 6-6, 2005.
Added by: Deleted user (25 Jul 2008 14:45:30 Europe/Berlin) Last edited by: Deleted user (13 Aug 2008 14:10:42 Europe/Berlin)
|Resource type: Conference Paper
BibTeX citation key: Deissb
View all bibliographic details
Publisher: ETSI (Sophia Antipolis (France))
Collection: ETSI TTCN-3 User Conference 2005
Views index: %
Popularity index: 9%
Usually the messages exchanged between a test system and the System Under Test need to be encoded when sending and decoded when receiving them. In some cases standardized encodings like ASN.1 BER or PER can be used and in these cases often mature tools to generate the encoding and decoding functions are available. Then, no special attention need to be paid to the question whether the encoding and decoding functions in the test system are correct themselves. But equally often, the messages have to be encoded with a specific encoding and either the encoding and decoding functions have to be written manually or codec generation tools have to be developed. In this situation the encoding and decoding functions in the test system might contain errors themselves, which is very disturbing when performing the actual tests and often raises questions about the validity of the tests at all.
But as the encoding and decoding functions provide a very well defined functionality, it is possible to generate test data to test the codecs in isolation and therefore raise the quality of the test system. We will present an approach that allows to generate test data and even complete test cases from given TTCN-3 or ASN.1 message definitions.
In the first part of the presentation we will present the general approach. We define what are the inputs to the test data generation and how does this affect the test data generation. The approach taken has effects on the class of errors that can be detected, we will discuss this in the presentation.
In the second part of the presentation we will present two strategies for test data generation. Testing the encoding and decoding of all possible values is usually not possible. Therefore strategies have to be applied to reduce the amount of generated test data, and still be able to detect errors in the code. We will present two strateges and compare them w.r.t. number of generated test data and error detection capability.
Nevertheless, the amount of test data can be large. This challenges the available TTCN-3 tools. In a third part of the presentation we will shortly investigate the effects of using specific TTCN-3 definitions on the resulting code size.
Added by: Deleted user Last edited by: Deleted user