A probabilistic diagnostic model in malicious environments and its algorithms

Presents a new model, noted as (p,m)-diagnosable, which can probabilistically model malicious environments in a uniform manner. A heuristic-inference algorithm, an approximate diagnosability algorithm and two characterization theorems of (p,m)-diagnosable systems are given. The complexity of (p,m)-diagnosability is examined to be co-NP-complete. An example (p,m)-diagnosable system is given