To improve the theoretical understanding of the byzantine model and enable a modular design of algorithms, we propose to decompose the byzantine behaviour into a data failure behaviour and a communication failure behaviour. We argue that the two failure types are orthogonal and we point out how they generate a range of several new interesting failure models, which are less difficult than byzantine, but different than the already well understood crash model. Such intermediate models are relevant and subject to recent studies, e.g. .
|Title of host publication||Distributed Computing (22nd International Symposium, DISC 2008, Arcachon, France, September 22-24, 2008. Proceedings)|
|Place of Publication||Berlin|
|Publication status||Published - 2008|
|Name||Lecture Notes in Computer Science|