Neural Network Based Node Prioritization for Efficient Localization

Optimizing the resource utilization is essential for efficiently providing reliable location awareness in complex wireless environments. This paper presents a data-driven approach to node prioritization for efficient localization based on neural networks. We develop a node prioritization strategy for power allocation consisting of offline training and online operation. In the offline phase, we train a neural network to approximate a mapping of node prioritization decisions obtained via model-based optimization. In the online phase, the trained neural network is employed to determine the resource allocation. A case study validates the proposed approach and compares it against conventional methods based on uniform power allocation.