Machine-to-Machine (M2M) technology enables to interconnect machines with the least possible human intervention. The nature of M2M traffic, small and periodic, is different from the traditional traffic originating from computers or mobile phones, big and bursty. However, the number of nodes involved in such technology impacts the network as it has mainly been designed for Human-to-Machine (H2M) communications. Due to the number of involved M2M nodes and their intrinsic constraints, the retrieval of M2M traffic with existing networks is a real challenge. In order to do so, the authors propose a lightweight service announcement architecture, re-using already deployed Wi-Fi networks. This solution extends information announced in the IEEE 802.11 beacon frame. It enables M2M devices to automatically detect the most appropriate network among the one discovered, reduces the amount of signaling messages exchanged to do so and consequently, the delay needed to collect this traffic. The feasibility study, based on real-world data, shows that the frequency of announcement, the amount of data to send and the available density of Access Points, have to be taken into account in order to use Wi-Fi networks to retrieve M2M traffic.