Coded Federated Computing in Wireless Networks with Straggling Devices and Imperfect CSI

By Sukjong Ha, Jingjing Zhang, Osvaldo Simeone, Joonhyuk Kang. 2019

41 citations recorded by the source corpus.

Distributed computing platforms typically assume the availability of reliable and dedicated connections among the processors. This work considers an alternative scenario, relevant for wireless data centers and federated learning, in which the distributed processors, operating on generally distinct coded data, are connected via shared wireless channels accessed via full-duplex transmission. The study accounts for both wireless and computing impairments, including interference, imperfect Channel State Information, and straggling processors, and it assumes a Map-Shuffle-Reduce coded computing paradigm. The total latency of the system, obtained as the sum of computing and communication delays, is studied for different shuffling strategies revealing the interplay between distributed computing, coding, and cooperative or coordinated transmission.

Subfields: Artificial Intelligence, Computer Networks and Communications, Electrical and Electronic Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. H. Vincent Poor
2. Tony Q. S. Quek
3. Zuozhu Liu
4. Howard H. Yang
5. Mingzhe Chen
6. Walid Saad
7. Shuguang Cui
8. Zhaohui Yang

In Computer Networks and Communications

1. H. Vincent Poor
2. Howard H. Yang
3. Tony Q. S. Quek
4. Zuozhu Liu

In Electrical and Electronic Engineering

1. H. Vincent Poor
2. Howard H. Yang
3. Tony Q. S. Quek
4. Zuozhu Liu
5. Zhiguo Ding
6. Mingzhe Chen
7. Walid Saad
8. Yuanwei Liu

Citations in the corpus

• Age-Based Scheduling Policy for Federated Learning in Mobile Edge Networks (2019, background): “To this end, the stragglers issue is more crucial for FL than for conventional trainings run in data centers [3,4,12].”
• Scheduling Policies for Federated Learning in Wireless Networks (2019, methodology): “The requirement for such decomposition method to work is that during each global aggregation, the changes in the local variables { a t [ k ] } Kk =1 on each UE and that in the global vector v ( a t ) are kept consistent [19].”
• A Survey on Federated Learning for Resource-Constrained IoT Devices (2021, unknown): “If the number of such straggler clients is high, then the overall model convergence would be at stake [99], [100].”
• Edge Learning for B5G Networks With Distributed Signal Processing: Semantic Communication, Edge Computing, and Wireless Sensing (2022, unknown): “dition, to alleviate the “straggler effect’ where the slowest edge device acts as a bottleneck of learning performance [114], a new protocol for FL was advocated in [60] through joint optimization”
• Machine Learning in the Air (2019, unknown): “A similar observation is also made in [80] by considering coded wireless computation in the map-shuffle-reduce framework, where physical layer techniques are leveraged to provide robustness against worker and channel uncertainties.”