Résumé
This paper concerns probabilistic distributed graph algorithms to solve classical graph problems such as colouring, maximal matching or maximal independent set. We consider anonymous networks (no unique identifiers are available) where vertices communicate by single bit messages. We present a general framework, based on coverings, for proving lower bounds for the bit complexity and thus the execution time to solve these problems. In this way we obtain new proofs of some well known results and some new ones. The last part gives impossibility results on the existence of Las Vegas distributed algorithms to break symmetries at distance $k$ for $k\geq 3.$