Unsupervised Power Allocation Based on Combination of Edge Aggregated Graph Attention Network With Deep Unfolded WMMSE

To address the challenge of optimizing power distribution within ad hoc wireless networks, deep learning techniques have been employed to navigate the complexities of evolving network topologies and variable channel characteristics. Moreover, there's an anticipation for the deep learning model to unearth more generalized and distinct features. In this paper, we introduce the Unsupervised Power Allocation based on Attentive Graph Representation (UPAR) method. UPAR combines an Edge Aggregated Graph Attention Regression Network (EGRN) with Deep Unfolded Weighted Minimum Mean Square Error (DU-WMMSE) within an unsupervised learning framework. Initially, EGRN leverages an attention mechanism and Gate Recurrent Units (GRU) to amalgamate features from adjacent nodes and their connecting edges, enhancing the identifiability and generalization of the graph's features. Subsequently, DU-WMMSE applies unsupervised learning to refine these graph features through regression training, transforming WMMSE iterations into efficient, interpretable layers within the neural network. This process introduces adjustable parameters, employing both feed forward and back propagation phases during end-to-end training. The proposed method significantly narrows the solution space for the objective function, accelerating the training process and enhancing the efficiency of the deep learning model. Simulation results confirm the UPAR method's superior performance and adaptability across a variety of scenarios.