We deal with the threats to user privacy in the context of wireless local networks. We focus on location privacy where an adversary tries to learn a user's past and current locations. The current Wi-Fi standard is vulnerable to location privacy and mobility profiling attacks due to the transmission of personally identifying information such as the MAC address in plain text. We provide a generic mathematical model to quantify and express the privacy and elaborate a decentralised algorithm that allows users to attain their desired levels of privacy while lowering its effect on the QoS perceived by them. We evaluate our proposal with numerical simulation and mobility traces collected from Wi-Fi users in an office environment. We show that higher privacy can be obtained with a variable effect on the throughput available to users.