Graph-Based Algorithm Unfolding for Energy-Aware Power Allocation in Wireless Networks

We develop a novel graph-based trainable framework to maximize the weighted sum energy efficiency (WSEE) for power allocation in wireless communication networks. To address non-convex nature of problem, proposed method consists modular structures inspired by classical iterative suboptimal approach and enhanced with learnable components. More precisely, we propose deep unfolding successive concave approximation (SCA) method. In our unfolded SCA (USCA) framework, originally preset parameters are now via graph convolutional neural networks (GCNs) that directly exploit multi-user channel state information as underlying adjacency matrix. show permutation equivariance architecture, which is desirable property models applied network data. The USCA trained through stochastic gradient descent using progressive training strategy. unsupervised loss carefully devised feature monotonic objective under maximum constraints. Comprehensive numerical results demonstrate its generalizability across different topologies varying size, density, distribution. Thorough comparisons illustrate improved performance robustness over state-of-the-art benchmarks.