5G networks are expected to drastically enhance the existing broadband wireless connectivity in many ways including higher bandwidth, lower delay, and wider coverage. To provide the required spectral efficiency, a key technology that could support the 5G vision is a transmission scheme based on Cooperative Multi Point Transmission (CoMP) combined with Orthogonal Frequency-Division Multiplexing (OFDM). This, however, requires an optimum resource allocation in a CoMP-OFDM system, which is shown to be a highly complex problem. Our proposal, referred to as scalable CoMP (SCoMP), tackles this complexity by dividing the optimal resource allocation problem into four subproblems such that CoMP becomes scalable and applicable in a practical 5G network without a huge overhead in terms of protocol signaling and backhaul requirements. Namely, these four subproblems are as follows: base station (BS) group formation (clustering), assignment of mobile stations (MS) to BS groups, MS group formation and OFDM frame construction, which are solved independently and therefore at lower complexity. Our approach improves its scalability even further by applying a static but overlapping BS grouping scheme which is MS-unaware. We decrease the potential performance loss due to MS-unawareness that is needed for improved scalability by two mechanisms. First, we propose to use overlapping BS clusters supporting variable cluster sizes which mitigates inter-cluster interference. Second, we apply dynamic grouping of MSs to make sure, only spatially-compatible MSs are being served simultaneously in the same group. Finally, our approach is able to deal with bursty network traffic. The realisation of SCoMP is enabled through the deployment of the software defined networking approaches in the backhaul network.