PI combines detection probability at each step of an adversary path with the remaining path delay and the response force time. Sandia's EASI method computes it for a single path:
Its main lesson: once detection is reasonably early, response time and delay dominate. Our worked model shows extra late-path detection changing PI by under 0.01.
Inputs you need
Credible adversary paths; detection probability and delay for each element on each path; and the distribution of response force time, including nights and weekends. The path with the lowest PI is the critical path, and it defines the system's effectiveness.
Levers
Detect earlier on the path; add delay after the first likely detection point; reduce mean and variance of response time. Adding detection late in the path rarely helps.