Betweenness centrality is a popular metric in social science, and recently it was adopted also in computer science. Betweenness identifies the node, or the nodes, that are most suitable to perform critical network functions, such as firewalling and intrusion detection. However, computing centrality is resource-demanding, we can not give for granted that it can be computed in real time at every change in the network topology. This is especially true in mesh networks that generally use devices with few computation resources. This paper shows that using the fastest state-of-the-art heuristic algorithm it is indeed possible to compute network centrality even in real, low-power networking hardware in a network made of up to 1000 nodes. Not only, observing the behavior of a real mesh network it shows that centrality does not need to be updated at every topology change, but it can be safely re-computed with an interval in the order of the tens of minutes. Our findings confirms that centrality can be effectively and successfully used as a building block for security functions in mesh networks.