This paper analyzes the scanning process in IEEE 802.11 networks in an urban setting characterized by a high Access Point (AP) density. Most of these APs belong to a community network, known as a collection of APs announcing the same network name (Service Set Identifier or SSID). The owner of an AP can optionally configure the community network of his/her AP, resulting in an irregular topology for each community network as there is no central planning authority. We investigate the relationship between the time spent in each channel while scanning for available APs and the number of AP actually detected. In particular we show that, in order to discover all available APs at a given location, we need to combine the results of multiple scans. Based on this result we argue that the efficiency of the scanning process could be greatly improved by using a database shared by all the users of a community network, containing the available APs at different locations.