Federated Learning in the Sky: Joint Power Allocation and Scheduling with UAV Swarms

Abstract

Unmanned aerial vehicle (UAV) swarms must exploit machine learning (ML) in order to execute various tasks ranging from coordinated trajectory planning to cooperative target recognition. However, due to the lack of continuous connections between the UAV swarm and ground base stations (BSs), using centralized ML will be challenging, particularly when dealing with a large volume of data. In this paper, a novel framework is proposed to implement distributed federated learning (FL) algorithms within a UAV swarm that consists of a leading UAV and several following UAVs. Each following UAV trains a local FL model based on its collected data and then sends this trained local model to the leading UAV who will aggregate the received models, generate a global FL model, and transmit it to followers over the intra-swarm network. To identify how wireless factors, like fading, transmission delay, and UAV antenna angle deviations resulting from wind and mechanical vibrations, impact the performance of FL, a rigorous convergence analysis for FL is performed. Then, a joint power allocation and scheduling design is proposed to optimize the convergence rate of FL while taking into account the energy consumption during convergence and the delay requirement imposed by the swarm's control system. Simulation results validate the effectiveness of the FL convergence analysis and show that the joint design strategy can reduce the number of communication rounds needed for convergence by as much as 35% compared with the baseline design.

Machine Learning Methods for Management UAV Flocks -- a Survey

“…Zeng et al [128] used a distributed swarm approach for the power control and scheduling of UAV flocks. In particular, they used distributed federated learning (FL) algorithms within a UAV swarm that consisted of a leading UAV and several following UAVs.…”

UAV-assisted Online Machine Learning over Multi-Tiered Networks: A Hierarchical Nested Personalized Federated Learning Approach

“…Distributed ML over wireless networks: Recent literature concerning ML by wireless networks has shifted towards federated learning [23], [24], and is mostly focused on studying the convergence and behavior of federated learning over wireless networks [9], [10], [12], [25], [26], [27], [28], [29], [30], [31], [32], [33]. Conventional federated learning assumes training a single ML model for all the engaged devices.…”

“…Each x ∈ D i (t) is a data sample containing model features and (possibly) a target variable. Different from most of the existing literature in distributed learning where the data distributions of the devices/workers are assumed to be static [9], [10], [12], [25], [26], [27], [28], [29], [30], [31], [33], we consider online model training and deployment (captured via the notion of concept/model drift in Sec. 5), where the goal is to obtain a personalized ML model for each cluster c to use for real-time inference.…”

“…While ( 5),( 6),( 9) are only realized by HN-PFL when t = t L s (k), i.e., at a local aggregation, and while ( 7),( 8), (10) are only realized when t = t G s (k ), i.e., at a global aggregation, defining them for each t ∈ T s will be useful in our theoretical analysis as we model their evolution over time.…”

“…Current literature has taken some initial steps toward integrating UAVs into distributed ML, e.g., [8], [9], [10], [11], [12], which mostly focused on implementing the classic federated learning architecture using UAVs where either the data is assumed to already be at the UAVs [8], [9], [10], [11] or the UAVs are merely used for parameter aggregation instead of a cellular base station [12]. Specifically, current literature is missing a holistic ML paradigm that develops a UAVspecific learning architecture for UAV-assisted networks, where the UAVs are utilized to train an ML model via data collection from energy-constrained devices.…”