Quantization Impact on the Accuracy and Communication Efficiency Trade-off in Federated Learning for Aerospace Predictive Maintenance

Abdelkarim Loukili

cs.LGApr 9, 2026

Quantization Impact on the Accuracy and Communication Efficiency Trade-off in Federated Learning for Aerospace Predictive Maintenance

Authors: Abdelkarim Loukili

Abstract

Federated learning (FL) enables privacy-preserving predictive maintenance across distributed aerospace fleets, but gradient communication overhead constrains deployment on bandwidth-limited IoT nodes. This paper investigates the impact of symmetric uniform quantization ( $b \in \{32,8,4,2\}$ bits) on the accuracy--efficiency trade-off of a custom-designed lightweight 1-D convolutional model (AeroConv1D, 9,697 parameters) trained via FL on the NASA C-MAPSS benchmark under a realistic Non-IID client partition. Using a rigorous multi-seed evaluation ( $N=10$ seeds), we show that INT4 achieves accuracy statistically indistinguishable from FP32 on both FD001 ( $p=0.341$ ) and FD002 ( $p=0.264$ MAE, $p=0.534$ NASA score) while delivering an $8\times$ reduction in gradient communication cost (37.88~~KiB $\to$ 4.73~~KiB per round). A key methodological finding is that naïve IID client partitioning artificially suppresses variance; correct Non-IID evaluation reveals the true operational instability of extreme quantization, demonstrated via a direct empirical IID vs.\ Non-IID comparison. INT2 is empirically characterized as unsuitable: while it achieves lower MAE on FD002 through extreme quantization-induced over-regularization, this apparent gain is accompanied by catastrophic NASA score instability (CV,=,45.8% vs.\ 22.3% for FP32), confirming non-reproducibility under heterogeneous operating conditions. Analytical FPGA resource projections on the Xilinx ZCU102 confirm that INT4 fits within hardware constraints (85.5% DSP utilization), potentially enabling a complete FL pipeline on a single SoC. The full simulation codebase and FPGA estimation scripts are publicly available at https://github.com/therealdeadbeef/aerospace-fl-quantization.

Figures & tables

Layer	Type / Configuration	Output Shape	Params
Input	Time-series window	$(B,14,50)$	0
1	Conv1D ( $k=3,s=1,p=1$ ) + ReLU	$(B,32,50)$	1,376
2	MaxPool1D ( $k=2,s=2$ )	$(B,32,25)$	0
3	Conv1D ( $k=3,s=1,p=1$ ) + ReLU	$(B,64,25)$	6,208
4	AdaptiveAvgPool1D	$(B,64,1)$	0
5	Flatten	$(B,64)$	0

Table 1: AeroConv1D architecture specification (9,697 parameters total).

B

: batch size. The parameter count is verified programmatically at simulation startup via an assertion.

	FD001		FD002
Client	Mean RUL	EMD	Mean RUL	EMD
$k=1$	66.5	8.9	72.0	3.4
$k=2$	74.1	1.2	77.6	2.2
$k=3$	68.8	6.5	78.1	2.6
$k=4$	73.3	2.0	74.8	0.6
$k=5$	86.8	11.4	71.5	4.0

Table 2: Per-client mean RUL (cycles) for seed 42, confirming Non-IID label heterogeneity. EMD: Earth Mover’s Distance from the global RUL histogram.

Sub.	Cfg	MAE (cycles)	$p_{\text{MAE}}$	Score $S$ ( $\times 10^{3}$ )	$p_{S}$	CV S
FD001	FP32	$17.52\pm 0.47$	—	$449\pm 123$	—	27.3%
	INT8	$17.51\pm 0.48$	0.520	$447\pm 127$	0.746	28.6%
	INT4	$17.48\pm 0.51$	0.341	$452\pm 115$	0.802	25.3%
	INT2	$19.03\pm 1.62$	0.018	$802\pm 573$	0.064	72.0%
FD002	FP32	$26.99\pm 1.69$	—	$923\pm 206$	—	22.3%
	INT8	$27.24\pm 1.40$	0.265	$951\pm 167$	0.364	16.9%

Table 3: Test-set results: Mean

\pm

Std over 10 random seeds.

p

-values from two-tailed paired

t

-test vs. FP32 (

n=10

\text{df}=9

). Bold :

p<0.05

. NASA score

S

reported as sum over all test windows (Eq. 3 ). CV S : coefficient of variation of

S

across seeds. Cohen’s

\hat{d}

N=10

carries 95% CI

\approx\hat{d}\pm 0.95

; interpret as directional only.

Figure 1: MAE convergence over 20 FL rounds on C-MAPSS FD001. Shaded bands:

\pm

1 std over 10 seeds. FP32, INT8, and INT4 converge to indistinguishable final MAE; INT2 exhibits slower convergence and higher variance.

Partition	Config	MAE (cycles)	Score $S$ ( $\times 10^{3}$ )
IID	FP32	$17.34\pm 0.36$	$421\pm 41$
IID	INT4	$17.28\pm 0.26$	$440\pm 31$
Non-IID	FP32	$17.52\pm 0.47$	$449\pm 123$
Non-IID	INT4	$17.48\pm 0.51$	$452\pm 115$

Table 4: Methodological bias: IID vs. Non-IID partitioning on FD001 (mean

\pm

std over 10 seeds). IID suppresses variance, making INT4 appear to outperform FP32; Non-IID reveals statistical parity.

Figure 2: MAE by subset and quantization level. Hatching indicates FD002 (6 operating conditions). Asterisk (*):

p<0.05

vs. FP32 on MAE. Warning: The lower MAE of INT2 on FD002 is an over-regularization artefact, not a genuine accuracy improvement (see Section 5.4 ).

Figure 3: NASA score

S

convergence on C-MAPSS FD001. Lower is better. INT2 early-round values reach

10^{9}

and are off-scale; the y-axis is clipped for readability. Negative values arise when systematic under-prediction dominates; INT2 oscillates between extreme positive and negative scores, illustrating non-reproducibility.

Figure 4: Gradient-distortion privacy proxy

\mathcal{L}_{\mathrm{priv}}

on FD001 (log scale) over 20 FL rounds. Higher values indicate greater gradient distortion and higher gradient-inversion attack cost [ 20 , 4 ] . FP32 is omitted (

\mathcal{L}_{\mathrm{priv}}=0

by definition).

\mathcal{L}_{\mathrm{priv}}

is not a formal DP bound; see Section 3.5 .

Figure 5: Accuracy–communication trade-off on FD001. Error bars:

\pm

1 std over 10 seeds. Arrow: FP32

\to

INT4 operating point. Annotated

p

-value from paired

t

-test on NASA score (INT4 vs. FP32).

Cfg	LUT	%LUT	DSP	%DSP	Lat.	Fit
FP32	51 717	18.9%	17 239	684.1%	16 $\mu$ s	$\times$
INT8	12 929	4.7%	4 309	171.0%	4 $\mu$ s	$\times$
INT4	6 464	2.4%	2 154	85.5%	2 $\mu$ s	$\checkmark$
INT2	3 232	1.2%	1 077	42.7%	1 $\mu$ s	$\checkmark$

Table 5: Analytical FPGA resource projections — Xilinx ZCU102 (274 080 LUT

|

2 520 DSP

|

912 BRAM36). Latency at 500 MHz, sequence length 50. Comm. cost excludes per-layer scale overhead (

8\times 4

\approx

0.03 KiB). Projections derived from hls4ml scaling model; not validated on physical silicon.

Explore similar work

Jun 25, 2026

Quantization in Federated Learning: Methods, Challenges and Future Directions

Federated Learning (FL) has become a foundational paradigm for privacy-preserving distributed intelligence, yet its scalability remains fundamentally constrained by communication bottlenecks, device heterogeneity, and the challenges of training under statistically non-IID data. Quantization is one of the most effective mechanisms for mitigating these limitations, reducing both uplink/downlink payloads and on-device computation. This paper provides the first FL-centric systematic review of quantization, introducing a novel taxonomy organized around FL-specific dimensions, including client heterogeneity, aggregation consistency, communication-scheduling adaptation, non-IID robustness, privacy/security integration, and hardware/energy co-optimization. Beyond cataloging existing methods, we analyze how quantization interacts with core FL behaviors such as client drift, partial participation, convergence stability, secure aggregation, and differential privacy. We further identify cross-method insights, open research gaps, and design guidelines for practitioners deploying quantized FL on mobile, IoT, and edge platforms. This survey thus establishes quantization not merely as a compression technique, but as a fundamental systems component shaping the performance, robustness, and practicality of modern FL.

Farwa Ikram, Dipanwita Thakur, Antonella Guzzo +1

DIMES - University of Calabria, Italy

Neural Network Quantization Federated Learning

Ask PDF

Mar 18, 2026

QuantFL: Sustainable Federated Learning for Edge IoT via Pre-Trained Model Quantisation

Federated Learning (FL) enables privacy-preserving intelligence on Internet of Things (IoT) devices but incurs a significant carbon footprint due to the high energy cost of frequent uplink transmission. While pre-trained models are increasingly available on edge devices, their potential to reduce the energy overhead of fine-tuning remains underexplored. In this work, we propose QuantFL, a sustainable FL framework that leverages pre-trained initialisation to enable aggressive, computationally lightweight quantisation. We demonstrate that pre-training naturally concentrates update statistics, allowing us to use memory-efficient bucket quantisation without the energy-intensive overhead of complex error-feedback mechanisms. On MNIST and CIFAR-100, QuantFL reduces total communication by 40% (

\simeq40\%

total-bit reduction with full-precision downlink;

\geq80\%

on uplink or when downlink is quantised) while matching or exceeding uncompressed baselines under strict bandwidth budgets; BU attains 89.00% (MNIST) and 66.89% (CIFAR-100) test accuracy with orders of magnitude fewer bits. We also account for uplink and downlink costs and provide ablations on quantisation levels and initialisation. QuantFL delivers a practical, "green" recipe for scalable training on battery-constrained IoT networks.

Charuka Herath, Yogachandran Rahulamathavan, Varuna De Silva +1

Institute of Digital Technologies, Loughborough University, UK

Federated Learning Neural Network Quantization Internet of Things

Ask PDF

Apr 25, 2026

Enhanced Privacy and Communication Efficiency in Non-IID Federated Learning with Adaptive Quantization and Differential Privacy

Federated learning (FL) is a distributed machine learning method where multiple devices collaboratively train a model under the management of a central server without sharing underlying data. One of the key challenges of FL is the communication bottleneck caused by variations in connection speed and bandwidth across devices. Therefore, it is essential to reduce the size of transmitted data during training. Additionally, there is a potential risk of exposing sensitive information through the model or gradient analysis during training. To address both privacy and communication efficiency, we combine differential privacy (DP) and adaptive quantization methods. We use Laplacian-based DP to preserve privacy, which is relatively underexplored in FL and offers tighter privacy guarantees than Gaussian-based DP. We propose a simple and efficient global bit-length scheduler using round-based cosine annealing, along with a client-based scheduler that dynamically adapts based on client contribution estimated through dataset entropy analysis. We evaluate our approach through extensive experiments on CIFAR10, MNIST, and medical imaging datasets, using non-IID data distributions across varying client counts, bit-length schedulers, and privacy budgets. The results show that our adaptive quantization methods reduce total communicated data by up to 52.64% for MNIST, 45.06% for CIFAR10, and 31% to 37% for medical imaging datasets compared to 32-bit float training while maintaining competitive model accuracy and ensuring robust privacy through differential privacy.

Emre Ardıç, Yakup Genç

Department of Computer Engineering, Gebze Technical University, Gebze, Kocaeli 41400, Türkiye

Non-IID Federated Learning Differentially Private Federated Learning Neural Network Quantization Federated Learning

Ask PDF