Dependent Failure Analysis (DFA)

Dependent failure analysis is the method defined in ISO 26262-9, Clause 7 for verifying that elements which are claimed to be independent, or merely free from interference, really are. Statistical independence is an assumption, not a physical guarantee: a shared supply, clock, memory, or environment can let a single root cause defeat several elements at once. DFA systematically identifies dependent failure initiators (DFIs) and the coupling factors connecting supposedly independent elements, then confirms that mitigations break the coupling. It underpins ASIL decomposition, redundancy claims, and mixed-ASIL coexistence, and it runs iteratively from the concept phase through hardware and software design.

A common-cause failure strikes multiple elements from one shared root - for example, a shared 5 V reference drifting and corrupting both channels of a redundant sensor pair at the same time. A cascading failure propagates from one element to the next: the first element fails, and its failure causes the second to fail. Both are dependent failures, and both defeat independence claims, but they need different mitigations - a common cause is broken by removing the shared resource or adding diversity, while a cascade is broken by blocking the propagation path.

Whenever your safety concept depends on independence or freedom from interference. The classic triggers are ASIL decomposition (the decomposed elements must be independent), redundant channels credited as fault tolerance, diverse monitoring such as lockstep or watchdog arrangements, and mixed-ASIL software coexisting on a shared processor. The rigor of the analysis scales with the ASIL of the safety goal being protected, and DFA is most valuable when started at the concept phase, where changing an architecture is still cheap.

Coupling factors are the shared attributes that connect supposedly independent elements and give a dependent failure initiator its path. The course organizes them into four ordered categories spanning physical to design coupling: shared resources such as power, clock, and memory; shared environment such as temperature, vibration, and a common compartment; communication such as a shared bus segment; and common design factors such as shared IP blocks, libraries, and development tooling. Each checklist item is resolved as mitigated, accepted, or open.

Ten chapters covering the six-step method, the four coupling-factor categories, hardware and software DFA, and the terminology split between DFA, FFI, common-cause, and cascading failure. Six diagrams visualize the concepts, and the worked example runs a full DFA over a dual-channel brake-by-wire architecture down to a closed DFI register. A free account starts you off, and the Pro and Expert plans unlock more of the library.