Learn best practices for creating verifiable, traceable requirements. The presentation also includes information about how Seapine's TestTrack supports streamlining better processes, data capture, reusability, and traceability in the requirements phase and a Q&A session.
Note the 2 nd two bullets are the V & V questions we ask ourselves.
Quickly A: requirement with no corresponding design, i.e. unfulfilled B: un-required design element how does that happen? un-necessary design (gold plating, etc.) more often, inadequate resolution of requirements
Quickly A: requirement with no corresponding system level software test (untested) B: test with no corresponding requirement how does that happen? inadequate resolution of requirements – requirements end up being documented in the tests - tests usually constructed from observed behavior (useless)
Top level arrowheads on wrong end. Two way arrows wrong .. A mess.
Note only software dealt with here. Similar traces for hardware, mechanical, etc. Reuse: If company makes many similar products, many of top level Design Inputs will be very similar. Applicable Standards may be same or slightly different. Imagine Project B copies top 2 levels (inputs, system requirements, and software requirements) from Project A. As differences in needs are identified, they can be traced to system requirement changes necessitated by the DI change because trace is in place. Similarly, changes in system requirement traces to needed changes in software requirements, etc.
Note – simplified situation for PPT purposes Quality System in top un-shaded area. Applies to all projects/devices, etc. High level, generic. Policy (top red) calls out the regulations, standards (general and device specific), and guidances with which the company complies. Each reg/std/guidance traces to specific element that belong at the Process level (e.g. the existence of plans, reviews, verification, validation, etc.) Note trace from SW Dev Process to SDP and from SW Test Process to SVVP The contents of project artifacts or specifics of activities that are spelled out in a guidance or standard are traced from that source doc to the project doc. In this case, note that 62304 traces to SDP and SVVP. It’s important to make these trace links obvious, visible, reportable, usable. Document users won’t comply with trace implications if they have to work too hard for them. Tools are needed so users don’t have to work so hard for creating and using the traces. So far, pretty generic and reusable device to device. When we get to green boxes the information gets more specific. Note that each of these could refer to sub-hierarchies. For instance, System level requirements/testing could be broken down as follows.
Depending on similarity of manufacturer devices, much can be reused project to project. Similar diagram could (should) apply to software requirements. Trace the requirements from 62304 for System Level Software Testing down as far as appropriate in the hierarchy. This takes a fair amount of design and planning, but worth the effort. There are no “set” ways to do it. The style and design depends on the manufacturer’s needs, tastes, and need to control. This kind of traceability not only applies to V&V artifacts, but also to SDP’s, SVVP’s, CMP’s, Change Management, Defect Management, etc.
Documents, in general should fan-out, i.e. one to many. Seldom many to one. If too much one to one, either not enough detail in child document, or too much detail in parent. Many to many may seem like it’s helpful, but in general is unusable.
Transition to Seapine Software with a high-level demonstration to some of the key points we spoke of earlier.