A Distributed Learning-based Interference Coordination and Resource Allocation Scheme

In cellular networks, improving throughput by resource allocation is a challenging problem due to the interdependence of resource allocation schemes among different cells. In this paper, we propose a distributed interference coordination and resource allocation scheme based on deep learning, aiming to maximize throughput by allocating appropriate orthogonal and non-orthogonal resources to users. The designed interference coordination can decouple the impact of resource allocation schemes of interfering cells on the data rate, enabling each cell to independently optimize its resource allocation scheme without performance loss. We propose an unsupervised learning-based resource allocation scheme, which not only achieves near-optimal resource allocation but also effectively reduces the communication overhead for interference coordination. Simulation results demonstrate that our scheme achieves performance comparable to exhaustive search (i.e., the centralized optimization) under different interference distribution conditions, with lower computational complexity and communication overhead.