Energy Efficient Federated Learning Over Wireless Communication Networks

Yang, Zhaohui; Chen, Mingzhe; Saad, Walid; Hong, Choong Seon; Shikh-Bahaei, Mohammad

doi:10.1109/twc.2020.3037554

Cited by 747 publications

(425 citation statements)

References 43 publications

Supporting

Mentioning

424

Contrasting

Order By: Relevance

“…These prior works mostly consider the drones as learners [21]- [24]. However, even with current advances in energy efficient FL [25] and low power computation systems 1 , we consider the concept of drone-mounted ML-computation hardware as heavy and energy inefficient, thus reducing the flight time of the UAV.…”

Section: A State Of the Artmentioning

confidence: 99%

Federated Learning With a Drone Orchestrator: Path Planning for Minimized Staleness

Donevski

Babu

Nielsen

et al. 2021

IEEE Open J. Commun. Soc.

Self Cite

View full text Add to dashboard Cite

In this paper, we investigate the problem of scheduling transmissions for spatially scattered nodes that contribute to a collaborative federated learning (FL) algorithm via wireless links provided by a drone. In the considered system, the drone acts as an orchestrator, coordinating the transmissions and the learning schedule within a predefined deadline. The actual schedule is reflected in a planned path: as the drone traverses it, it controls the distance and thereby the data rate to each node. Hence, the model is structured such that the drone orchestrator uses the path (trajectory) as its only tool to achieve fairness in terms of learning staleness, which reflects the learning time discrepancy among the nodes. Using the number of learning epochs performed at each learner as a performance indicator, we combine the average number of epochs computed and staleness into a balanced optimization criterion that is agnostic to the underlying FL implementation. We consider two methods for solving the complex trajectory planning optimization problem for static nodes: (1) successive convex programming (SCP) and (2) deep reinforcement learning (RL). Considering the proposed criterion, both methods are compared in three specific scenarios with few nodes. The results show that drone-orchestrated FL outperforms an immobile deployment by providing improvements in the range of 57% to 87.7%. Additionally, RL-guided trajectories are generally superior to SCP provided ones for complex node arrangements.

show abstract

Section: A State Of the Artmentioning

confidence: 99%

Federated Learning With a Drone Orchestrator: Path Planning for Minimized Staleness

Donevski

Babu

Nielsen

et al. 2021

IEEE Open J. Commun. Soc.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Joint RRM and training batchsize selection is further investigated in [23] to accelerate the learning speed in FEEL systems. From the perspective of FEEL system performance, there exists a fundamental tradeoff between device energy consumption and learning speed, which is quantified in [24]. In addition, in view of the varying communication-and-computation capacities of different nodes, researchers have also developed a hierarchical network architecture for implementing large-scale FEEL [25].…”

Section: B Federated Edge Learningmentioning

confidence: 99%

Joint Parameter-and-Bandwidth Allocation for Improving the Efficiency of Partitioned Edge Learning

Wen

Bennis

Huang

2020

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

To leverage data and computation capabilities of mobile devices, machine learning algorithms are deployed at the network edge for training artificial intelligence (AI) models, resulting in the new paradigm of edge learning. In this paper, we consider the framework of partitioned edge learning for iteratively training a large-scale model using many resource-constrained devices (called workers). To this end, in each iteration, the model is dynamically partitioned into parametric blocks, which are downloaded to worker groups for updating using data subsets. Then, the local updates are uploaded to and cascaded by the server for updating a global model. To reduce resource usage by minimizing the total learning-and-communication latency, this work focuses on the novel joint design of parameter (computation load) allocation and bandwidth allocation (for downloading and uploading). Two design approaches are adopted. First, a practical sequential approach, called partially integrated parameter-andbandwidth allocation (PABA), yields two schemes, namely bandwidth aware parameter allocation and parameter aware bandwidth allocation. The former minimizes the load for the slowest (in computing) of worker groups, each training a same parametric block. The latter allocates the largest bandwidth to the worker being the latency bottleneck. Second, PABA are jointly optimized. Despite it being a nonconvex problem, an efficient and optimal solution algorithm is derived by intelligently nesting a bisection search and solving a convex problem. Experimental results using real data demonstrate that integrating PABA can substantially improve the performance of partitioned edge learning in terms of latency (by e.g., 46%) and accuracy (by e.g., 4% given the latency of 100 seconds).

show abstract

“…Chen et al [19] formulates the joint learning, wireless resource allocation, and user selection problem as an optimization problem whose goal is to minimize an federated learning (FL) loss function that captures the performance of the FL algorithm. Yang et al [20] investigates the problem of energy efficient transmission and computation resource allocation for FL over wireless communication networks. Wang et al [21] studies the problem of optimizing the deployment of unmanned aerial vehicles (UAVs) equipped with visible light communication (VLC) capabilities.…”

Section: Convolutional Neuron Networkmentioning

confidence: 99%

IDSDL: a sensitive intrusion detection system based on deep learning

Bai

Yang

et al. 2021

J Wireless Com Network

View full text Add to dashboard Cite

Device-free passive (DfP) intrusion detection system is a system that can detect moving entities without attaching any device to the entities. To achieve good performance, the existing algorithms require proper access point (AP) deployment. It limits the applying scenario of those algorithms. We propose an intrusion detection system based on deep learning (IDSDL) with finer-grained channel state information (CSI) to free the AP position. A CSI phase propagation components decomposition algorithm is applied to obtain blurred components of CSI phase on several paths as a more sensitive detection signal. Convolutional neuron network (CNN) of deep learning is used to enable the computer to learn and detect intrusion without extracting numerical features. We prototype IDSDL to verify its performance and the experimental results indicate that IDSDL is effective and reliable.

show abstract

Energy Efficient Federated Learning Over Wireless Communication Networks

Cited by 747 publications

References 43 publications

Federated Learning With a Drone Orchestrator: Path Planning for Minimized Staleness

Federated Learning With a Drone Orchestrator: Path Planning for Minimized Staleness

Joint Parameter-and-Bandwidth Allocation for Improving the Efficiency of Partitioned Edge Learning

IDSDL: a sensitive intrusion detection system based on deep learning

Contact Info

Product

Resources

About