Software Testing Verification

Verification refers to a set of activities that ensure that software correctly implements a specific function. "Are we building the product right?"
"Confirmation by examination and provisions of objective evidence that specified requirements have been fulfilled."
Using the above definitions in software development
Validation, in its simplest terms, is the demonstration that the software implements each of the software requirements correctly and completely. In other words, the "right product was built." Verification is the activity, which ensures the work products of a given phase fully implement the inputs to that phase, or "the product was built right."

Levels of Verification
There are four levels of verification:
i) Component Testing
Testing conducted to verify the implementation of the design for one software element (unit, module) or a collection of software elements.
ii) Integration Testing
An orderly progression of testing in which various software elements and/or hardware elements are integrated together and tested. This testing proceeds until the entire system has been integrated.
iii) System Testing
The process of testing an integrated hardware and software system to verify that the system meets its specified requirements.
iv) Acceptance Testing
Formal testing conducted to determine whether or not a system satisfies its acceptance criteria and to enable the customer to determine whether or not to accept the system.

Types of Verification
There are four types of verification that can be applied to the various levels outlined above:
i) Inspection
Typical techniques include desk checking, walkthroughs, software reviews, technical reviews, and formal inspections (e.g., Fagan approach).
ii) Analysis
Mathematical verification of the test item, which can include estimation of execution times and estimation of system resources.
iii) Testing
Also known as "white box" or "logic driven" testing. Given input values are traced through the test item to assure that they generate the expected output values, with the expected intermediate values along the way. Typical techniques include statement coverage, condition coverage, and decision coverage.
iv) Demonstration
Also known as "black box" or "input/output driven" testing. Given input values are entered, and the resulting output values are compared against the expected output values. Typical techniques include error guessing, boundary-value analysis, and equivalence partitioning.
Explanation
The four methods for verification can be used at any of the levels although some work better than others for a given level of verification.
As an example, the most effective way to find anomalies at the component level is inspection. On the other hand, inspection is not applicable at the system level (you don't look at the details of code when performing system level testing).
A logical approach to testing is to utilize techniques and methods that are most effective at a given level.
Component level verification can easily get very expensive.
Companies need to avoid making statements like "all paths and branches will be executed during component testing."
These statements make for a very expensive test program, as all code developed is required to have one of the most labor-intensive type of testing performed on it.
To minimize the costs of component verification, the V&V group develops rules for determining the type of verification method(s) needed for each of the software functions.
As an example, very low complexity software function, which is not on the safety critical list, may only need informal inspection (walkthrough) performed.
Other complicated functions typically require white box testing since the functions become difficult to determine how they work.
We recommend performing inspections before doing the white box testing for a given module as it is less expensive to find the errors earlier in the development.
The resulting V&V effort has become a significant part of the software development effort for a medical device.
One of the key pieces to demonstrate that the system is implemented completely is a Requirements Traceability Matrix (RTM), which documents each of the requirements traced to design items, code, unit, integration and system test cases.
The RTM is an easy and effective way for documenting - what are the requirements, where are they implemented, and how have you tested them.