Scheduling In-Band Network Telemetry With Convergence-Preserving Federated Learning
- 14 March 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Networking
- Vol. 31 (5) , 2313-2328
Abstract
Conducting federated learning across distributed sites with In-Band Network Telemetry (INT) based data collection faces critical challenges, including control decisions of different frequencies, convergence of the models being trained, and resource provisioning coupled over time. To study this problem, we formulate a non-linear mixed-integer program to optimize the long-term INT overhead, resource cost, and federated learning cost. We then design polynomial-time online algorithms to solve this problem with only observable inputs on the fly, featuring laziness-aware resource adaption, online-learning-based INT flow selection and model aggregation control, as well as expectation-preserving randomized dependent rounding. We rigorously prove the parameterized-constant competitive ratio of our approach against the offline optimum, and the time-averaged constraint violation that vanishes in the long run. With extensive trace-driven evaluations, we confirm the superiority of our approach over other alternative approaches for reducing total cost and the efficacy of our trained models for solving real machine learning problems, reducing the real-time cost by $34\%$ on average.Keywords
Funding Information
- National Natural Science Foundation of China (61832005, 61872175, 61832008)
- China University Industry Research Innovation Foundation (2021FNA04005)
- Fundamental Research Funds for the Central Universities (2022300297)
- Collaborative Innovation Center of Novel Software Technology and Industrialization
- U.S. National Science Foundation (CNS-2047719, CNS-2225949)
This publication has 33 references indexed in Scilit:
- Resource Central2017
- An Online Convex Optimization Approach to Proactive Network Resource AllocationIEEE Transactions on Signal Processing, 2017
- Engineering Egress with Edge Fabric2017
- Application Bandwidth and Flow Rates from 3 Trillion Flows Across 45 Carrier Networks2017
- Pingmesh2015
- A Saddle Point Algorithm for Networked Online Convex OptimizationIEEE Transactions on Signal Processing, 2015
- On-Line Multicast Scheduling with Bounded Congestion in Fat-Tree Data Center NetworksIEEE Journal on Selected Areas in Communications, 2013
- Moving Big Data to The Cloud: An Online Cost-Minimizing ApproachIEEE Journal on Selected Areas in Communications, 2013
- Mapping the expansion of Google's serving infrastructure2013
- Dependent rounding and its applications to approximation algorithmsJournal of the ACM, 2006