Edinburgh Napier University

  Coordinated multi-point (CoMP) transmission is one of the key features for long term evolution advanced (LTE-A) and a promising concept for interference mitigation in 5th generation (5G) and beyond future densely deployed wireless networks. Due to the cost of coordination among many transmission points (TP), radio access network (RAN) needs to be clustered into smaller groups of TPs for coordination. In this paper, we develop a novel, load-aware clustering model by employing a merge/split concept from coalitional game theory. A load-aware utility function is introduced to maximize both spectral efficiency (SE) and load balancing (LB) objectives. We show that proposed load-aware clustering model dynamically adapts into the network load conditions providing high SE in low-load conditions and results in better load distribution with significantly less unsatisfied users in over-load conditions while keeping SE at comparable levels when compared to a greedy clustering model. Simulation results show that the proposed solution can reduce the number of unsatisfied users due to over-load conditions by 68.5% when compared to the greedy clustering algorithm. Furthermore, we analyze the stability of the proposed solution and prove that it converges to a stable partition in both homogeneous network (HN) and random network (RN) with and without hotspot scenarios. In addition, we show the convergence of our algorithm into the unique clustering solution with the best payoff possible when such a solution exists.