Federated Learning under Channel Uncertainty: Joint Client Scheduling and Resource Allocation

Wadu, Madhusanka Manimel; Samarakoon, Sumudu; Bennis, Mehdi

doi:10.48550/arxiv.2002.00802

Cited by 6 publications

(8 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The optimization of learning time was studied in [8], [27]- [33], and joint optimization for learning time and energy consumption was considered in [34]- [37]. These works considered resource (e.g., transmission power, communication bandwidth, and CPU frequency) allocation (e.g., [27], [28], [34]- [37]), cost-aware client selection (e.g., [29], [30]), client scheduling (e.g., [31]- [33]), and model pruning [8] for prespecified (i.e., non-optimized) design parameters (K and E in our case) of the FL algorithm.…”

Section: Related Workmentioning

confidence: 99%

Cost-Effective Federated Learning in Mobile Edge Networks

Luo¹,

Li²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Federated learning (FL) is a distributed learning paradigm that enables a large number of mobile devices to collaboratively learn a model under the coordination of a central server without sharing their raw data. Despite its practical efficiency and effectiveness, the iterative on-device learning process (e.g., local computations and global communications with the server) incurs a considerable cost in terms of learning time and energy consumption, which depends crucially on the number of selected clients and the number of local iterations in each training round. In this paper, we analyze how to design adaptive FL in mobile edge networks that optimally chooses these essential control variables to minimize the total cost while ensuring convergence. We establish the analytical relationship between the total cost and the control variables with the convergence upper bound. To efficiently solve the cost minimization problem, we develop a low-cost sampling-based algorithm to learn the convergence related unknown parameters. We derive important solution properties that effectively identify the design principles for different optimization metrics. Practically, we evaluate our theoretical results both in a simulated environment and on a hardware prototype. Experimental evidence verifies our derived properties and demonstrates that our proposed solution achieves near-optimal performance for different optimization metrics for various datasets and heterogeneous system and statistical settings.

show abstract

Section: Related Workmentioning

confidence: 99%

Cost-Effective Federated Learning in Mobile Edge Networks

Luo¹,

Li²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Further, a joint user selection and resource allocation policy towards minimizing the FEEL loss function was proposed in [20] by taking into account the packet error over wireless links. In addition, [21] presented an effective client scheduling and resource allocation policy for FEEL over wireless links with imperfect channel state information (CSI). Besides, the trade-off between learning time and energy consumption in FEEL was investigated in [22] and [23], where [22] developed a closed-form communication and computation resource allocation in a synchronous manner, and [23] designed an iterative algorithm for joint power control and resource allocation.…”

Section: A Prior Workmentioning

confidence: 99%

Scheduling for Cellular Federated Edge Learning with Importance and Channel Awareness

Ren

Wen

et al. 2020

Preprint

View full text Add to dashboard Cite

In cellular federated edge learning (FEEL), multiple edge devices holding local data jointly train a learning algorithm by communicating learning updates with an access point without exchanging their data samples. With limited communication resources, it is beneficial to schedule the most informative local learning update. In this paper, a novel scheduling policy is proposed to exploit both diversity in multiuser channels and diversity in the "importance" of the edge devices' learning updates. First, a new probabilistic scheduling framework is developed to yield unbiased update aggregation in FEEL. The importance of a local learning update is measured by gradient divergence. If one edge device is scheduled in each communication round, the scheduling policy is derived in closed form to achieve the optimal trade-off between channel quality and update importance. The probabilistic scheduling framework is then extended to allow scheduling multiple edge devices in each communication round. Numerical results obtained using popular models and learning datasets demonstrate that the proposed scheduling policy can achieve faster model convergence and higher learning accuracy than conventional scheduling policies that only exploit a single type of diversity.

show abstract

“…Although FL is a very efficient way of designing an ML framework for wireless communication, it is mostly considered for the applications of wireless sensor networks, e.g., UAV (unmanned aerial vehicle) networks [7], [8], vehicular networks [9]. In [7], the authors applies FL to the trajectory planning problem where a UAV swarm is employed and the data collected by each UAV are processed for gradient computation and a global NN at the "leading" UAV is trained.…”

Section: Introductionmentioning

confidence: 99%

“…In [7], the authors applies FL to the trajectory planning problem where a UAV swarm is employed and the data collected by each UAV are processed for gradient computation and a global NN at the "leading" UAV is trained. In [8] and [9], client scheduling and power allocation problems are investigated for FL framework respectively. Different from [7]- [9], [10] considers a more realistic scenario where the gradients are transmitted to the BS through a noisy wireless channel and a classification model is trained for image classification.…”

Section: Introductionmentioning

confidence: 99%

Federated Learning for Hybrid Beamforming in mm-Wave Massive MIMO

Elbir

Coleri

2020

IEEE Commun. Lett.

View full text Add to dashboard Cite

Machine learning (ML) for wireless communications requires the training of a global model with a large dataset collected from the users. However, the transmission of a whole dataset between the users and the base station (BS) is computationally prohibitive. In this work, we introduce a federated learning (FL) based framework where the model training is performed at the BS by collecting only the gradients from the users. In particular, we design a convolutional neural network (CNN), whose input is the channel data and it yields the analog beamformers at the output. We have evaluated the performance of the proposed framework via numerical simulations and shown that FL is more tolerant than ML to the imperfections and corruptions in the channel data as well as having less complexity.

show abstract

Federated Learning under Channel Uncertainty: Joint Client Scheduling and Resource Allocation

Cited by 6 publications

References 8 publications

Cost-Effective Federated Learning in Mobile Edge Networks

Cost-Effective Federated Learning in Mobile Edge Networks

Scheduling for Cellular Federated Edge Learning with Importance and Channel Awareness

Federated Learning for Hybrid Beamforming in mm-Wave Massive MIMO

Contact Info

Product

Resources

About