Network densification becomes inevitable due to rapidly increasing mobile data traffic in the evolution of future cellular networks. Therefore, Heterogeneous Wireless Networks eliminate the challenges that are not covered in 4G-Long Term Evolution. The extensive contribution of the proposed work is to develop a cluster based algorithm that establishes the maximum node connections in the heterogeneous wireless network and also to implement effective Multi-User Generalized Frequency Division Multiplexing in the physical layer for future wireless communications. The presence of unconnected nodes (Base Stations) in the wireless network reduces the network efficiency and coverage capacity. Whenever a wireless device is connected to at least one isolated neighbor node (BS), it acts as a Virtual Clustered Head based on shortest path node and residual energy of the node. A Virtual Cluster (VC) based multipath routing is implemented for Heterogeneous Networks to maximize the Energy Efficiency (EE), Spectral Efficiency (SE), Channel Capacity (CC), and Network Capacity (NC) and minimize the Bit Error Rate and latency under different constraints in the network. The simulation results with GFDM modulation shows the improvement in the network performance and compared with existing OFDM results.