The previous Scatternet formation algorithms can be broadly divided into two categories: proactive algorithms and reactive (on-demand) algorithms. The proactive Scatternet formation algorithms are to build a static Scatternet. Although they can consider the complete conditions in the network, they neglect the characteristics of each stand-alone Bluetooth node. On the other hand, reactive Scatternet formation algorithms are to build a more adaptive Scatternet according to the demands of each Bluetooth device dynamically. Therefore, the reactive algorithms receive much attention in research communities in recent years. In this thesis, we propose a two-phase on-demand Scatternet formation algorithm called TPSF+C based on a previous study of TPSF+. TPSF+C considers the connectivity and the living time of Bluetooth devices. Besides, we use fuzzy logic to decide a proper route. In the simulation results, TPSF+C is superior to TPSF+ in the number of reconstruction and the reconstruction time.