This long, heavily theoretical paper introduces and analyzes a general hierarchical framework for the decentralized diagnosis of discrete event systems, consisting of detecting unobservable significant events (such as faults).

After an introduction and literature review, the paper is organized into six sections: diagnosing unobservable events, decentralized architecture for diagnosis, decentralized diagnosis with one local decision, decentralized diagnosis with two local decisions, disjunctive and conjunctive architectures with m decisions, and other global functions with two local decisions. Conclusions are provided in another section before a comprehensive list of 27 references. Each section contains theoretical definitions, theorems, and proofs. Altogether, the paper presents nine theoretical definitions and 21 theorems, each with a proof.

This is a very difficult paper to read and understand. Only those theoretical researchers working in the topic of the paper will be able to comprehend its mathematical content and benefit from it.