Friend recommendation has also become popular on mobile devices, where location often also plays a role and makes the recommendations more transient or ad-hoc. Quercia et al. \cite{Quercia_2009} discussed “friendsensing”, sensing friends based on Bluetooth information on mobile devices. Friends were recommended based on co-location, while two basic approaches were attempted, taking into account the duration of co-location and its frequency, respectively. A weighted graph was built accordingly and recommendations were generated using that graph based on link analysis (shortest path, page rank, k-markov chain, and HITs). Simulation-based evaluation indicated both basic approaches perform similarly well and way beyond a random baseline.