Rethinking Traffic Matrix Completion: Estimate the Process, Not the Entries

2026-05-04

AI Summary

This work addresses the challenge of incomplete traffic matrices in data center networks, which are often difficult to fully observe and for which existing completion methods lack interpretability and uncertainty quantification. The authors propose a process-oriented completion paradigm that formulates the problem as parameter inference: under a local stationarity assumption, traffic is decomposed in the log domain into a low-rank component and a sparse deviation term. Missing entries are recovered by jointly inferring shared parameters across multiple partially observed snapshots. To circumvent the intractability of marginal likelihood integration and boundary degeneracy, a regularized surrogate objective is designed and optimized via block coordinate descent. Experiments on real-world data center datasets demonstrate that the proposed method significantly outperforms baseline approaches, particularly under conditions of extreme sparsity and bursty traffic.

Abstract

Traffic matrix measurement is fundamental for datacenter operations, but obtaining complete traffic matrices at scale remains challenging due to the prohibitive cost of global fine-grained measurement and partial observations resulting from network faults. Although existing matrix completion methods (reduce cost) achieve satisfactory performance in specific scenarios, their reliance on restrictive assumptions or black-box mappings results in a lack of interpretability and an inability to characterize uncertainty. In this paper, we propose Utimac, an uncertainty-aware traffic matrix completion for data center networks. Our analysis shows that, within a locally stationary window, log-domain traffic can be decomposed into a principal statistical component and a sparse deviation component. Based on this insight, we formulate traffic matrix completion as a parameter inference problem: multiple partially observed frames within a window are used to infer shared parameters and recover missing entries. To avoid the intractability and boundary degeneracy of the original integral-form marginal likelihood, we construct a regularized surrogate objective and solve the resulting joint optimization problem with block coordinate descent. Utimac consistently outperforms all baselines on data center networks datasets in both overall and burst scenarios, with its advantage becoming more pronounced as observations grow sparser. All code is publicly available in an anonymous repository: https://anonymous.4open.science/r/Utimac-0551/

Problem

• traffic matrix completion
• data center networks
• partial observations
• uncertainty quantification
• matrix completion

Innovation

• traffic matrix completion
• uncertainty-aware
• parameter inference
• log-domain decomposition
• block coordinate descent