RAT selection based on association probability in 5G heterogeneous networks

The use of multiple radio access technologies (RATs) is inevitable in future heterogeneous cellular networks. Various RATs can offer different throughputs, and thus RAT selection plays an important role in quality of service provisioning. In this paper, considering a heterogeneous network with two throughput classes, we introduce a new practical probabilistic RAT selection approach. In contrary to the common deterministic approaches, in this scheme, each user performs the RAT selection periodically in a random manner using an association probability determined by a central unit in the network. The user remains connected to the selected RAT for a specific amount of time and then again performs the selection. To find the association probabilities, the network throughput maximization is formulated. Since the resulting problem is not convex, we propose two alternatives, one with sub-optimal solution but lower complexity and one with the use of change of variables with optimal solution. Finally, numerical simulations demonstrate the superior performance of our proposed schemes in comparison to several possible approaches. Also, these numerical investigations indicate that our sub-optimal method works very close to the optimal one.