Dynamic leader–follower consensus with adversaries: A multi-hop relay approach