In contrast to synthetic user mobility models, user movements in real-world scenarios are restricted and typically conform to location-specific street layouts. Further in urban areas, user movements depend on traffic laws and behavior of other users. For example, vehicular users are said to stop at red traffic lights or should brake, if a vehicle ahead suddenly stops. Moreover, cellular users moving in these urban environments may face severe and abrupt changes in receive signal levels, which in the worst case result in connection drops. In order to pro-actively prevent these situations, where link rate decreases, handover execution is triggered too late, or even the connection dropped, a context-enhanced user movement prediction scheme is presented in this paper. Further, achievable performance gains using user movement prediction and modeling network deployment, user mobility, and radio propagation in a more realistic manner as envisioned for the next generation of wireless networks 5G are presented.