Latency-Aware Range Queries over Network Coordinates using Hilbert Curve Indexing
Latency-aware node discovery is a fundamental task in edge–cloud systems, where a node must identify nearby resources without comparing itself with every other node in the system. This paper addresses the latency range query problem, that is, finding all nodes whose latency to a given query node is below a specified threshold. The proposed method uses Vivaldi network coordinates, maps the coordinates to Hilbert keys, and stores the resulting keys in a B-tree index. Query processing then combines Hilbert-based range search with a pruning step that removes regions that cannot satisfy the latency bound. Experiments on an emulated 162-node FTTH topology show that, relative to Hilbert indexing alone, the pruning-enhanced method preserves the same true-positive and false-negative counts while greatly reducing false positives. As a result, precision and Jaccard similarity improve substantially, whereas recall remains unchanged. These findings show that Hilbert-based indexing, when paired with pruning, provides an effective way to support latency range queries in edge–cloud environments.