## A.2 Requirement classification for statistical testing

37.1713GPPRelease 16TSUniversal Terrestrial Radio Access (UTRA) and Evolved UTRA (E-UTRA)User Equipment (UE) performance requirements for RAT-Independent positioning enhancements

Requirements in the present document are either expressed as absolute requirements with a single value stating the requirement, or expressed as a success rate. There are no provisions for the statistical variations that will occur when the parameter is tested.

Annex A outlines the tests in more detail and lists the test parameters needed. The test will result in an outcome of a test variable value for the device under test (DUT) inside or outside the test limit. Overall, the probability of a "good" DUT being inside the test limit(s) and the probability of a "bad" DUT being outside the test limit(s) should be as high as possible. For this reason, when selecting the test variable and the test limit(s), the statistical nature of the test is accounted for.

The statistical nature depends on the type of requirement. Some have large statistical variations, while others are not statistical in nature at all. When testing a parameter with a statistical nature, a confidence level is set. This establishes the probability that a DUT passing the test actually meets the requirements and determines how many times a test has to be repeated and what the pass and fail criteria are. Those aspects are not covered by TSĀ 37.171. The details of the tests on how many times to run it and how to establish confidence in the tests are described in TS 37.571-1 [9]. This Annex establishes the variable to be used in the test and whether it can be viewed as statistical in nature or not.