User-centric wireless access virtualization (WAV) allows each user to be served by a set of carefully selected transmission points (TP)s forming a user-specific virtual base station (uVBS) adapted to its environment and quality-of-service (QoS) requirement. In this way, this new concept breaks away from the conventional cell-centric architecture to provide boundary less communications in future fifth-generation (5G) networks. This fundamental structural 5G evolution and the ultra-dense multi-tier heterogenous context foreseen in such networks require an inevitable rethinking of efficient scalable TP clustering. As such, this paper proposes three innovative low-cost clustering approaches that enable user-centric WAV and provide dynamic, adaptive, and overlapping TP clusters while requiring not only negligible overhead cost, but also minimum signaling changes at both network and user sides. In contrast to existing clustering techniques, the new ones we propose better leverage 5G features such as extreme densification and massive connectivity as well as new concepts such as millimeter wave (mmWave) spectrum and massive multiple-input multiple-output (MIMO. Simulations show that they may achieve until 154% and 282% of throughput and coverage gains, respectively. Furthermore, these approaches are flexible enough to be adapted to different network dimensions (i.e., space, time, etc.), thereby paving the way for achieving the dramatic performance improvements required by 5G networks to cope with the upcoming mobile data deluge.