This paper deals with distributed algorithms for monitoring the topology of a dynamic group of mobile wireless sensor networks. We propose two major extensions of a distributed static group consensus algorithm and an experimental implementation. Group consensus algorithms are exploited to let each node obtain the knowledge of its connected com-ponents. The proposed extensions provide a more accurate information about the proximity of nodes and allow to deal with dynamic networks using a periodical reevaluation of the group detection. We validate these algorithms by implementing them in an original and challenging application scenario, in the context of a real bicycle race. The real traces thus obtained and analyzed show the effectiveness of our live group detection implementation.