Quantitative Test Techniques for Advanced Automatic Checkout

High-speed, digital, internally-programmed, automatic checkout equipment is a powerful test tool whose impact has not yet been felt in testing large complex systems. It has, in certain ways, much higher test capabilities than either a sequentially programmed machine or a human operator. For these capabilities to be harnessed, new test techniques are required to ensure that failures or weaknesses are detected and separated from normal drifts. These techniques involve the design-of-the-prime system to secure through instrumentation, wiring, test points, and other built-in provisions, precisely the kind and quantity of test data needed. The test techniques involve improvement of several present-day test practices. The selection of test points, for example, is based upon both broad and detailed quantitative factors related to probability-of-mission-success, and reliability, rather than upon intuition, but with ample opportunity for the exercise of the designer's judgment. In general, more data are required, but the pay-off is in greatly increased reliability and availability, together with lower net costs. The immediate objectives of failure detection, prediction, prevention, and isolation are defined and a test program is described whose purpose is to maximize the failure information content of the test data rather than to reconfirm abstract design parameters. One technique described is called ``inferential testing'' or ``introspective testing'' which, complemented by end-to-end testing, is especially powerful in testing closed-loop systems. Another approach and technique described is that of wringing all possible test data from a fixed set of test points by ``network testing.