Federated unlearning (FU) algorithms allow clients in federated settings to exercise their "right to be forgotten" by removing the influence of their data from a collaboratively trained model.Existing FU methods maintain data privacy by performing unlearning locally on the client-side and sending targeted updates to the server without exposing forgotten data; yet they often rely on server-side cooperation, revealing the client's intent and identity without enforcement guarantees -compromising autonomy and unlearning privacy.In this work, we propose EFU (Enforced Federated Unlearning), a cryptographically enforced FU framework that enables clients to initiate unlearning while concealing its occurrence from the server.Specifically, EFU leverages functional encryption to bind encrypted updates to specific aggregation functions, ensuring the server can neither perform unauthorized computations nor detect or skip unlearning requests.To further mask behavioral and parameter shifts in the aggregated model, we incorporate auxiliary unlearning losses based on adversarial examples and parameter importance regularization.Extensive experiments show that EFU achieves nearrandom accuracy on forgotten data while maintaining performance comparable to full retraining across datasets and neural architectures -all while concealing unlearning intent from the server.Furthermore, we demonstrate that EFU is agnostic to the underlying unlearning algorithm, enabling secure, function-hiding, and verifiable unlearning for any client-side FU mechanism that issues targeted updates.

Device heterogeneity poses major challenges in Federated Learning (FL), where resource-constrained clients slow down synchronous schemes that wait for all updates before aggregation. Asynchronous FL addresses this by incorporating updates as they arrive, substantially improving efficiency. While its efficiency gains are well recognized, its privacy costs remain largely unexplored—particularly for high-end devices that contribute updates more frequently, increasing their cumulative privacy exposure. This paper presents the first comprehensive analysis of the efficiency-fairness-privacy trade-off in synchronous vs. asynchronous FL under realistic device heterogeneity. We empirically compareFedAvgand staleness-aware FedAsync using a physical testbed of five edge devices spanning diverse hardware tiers, integrating Local Differential Privacy (LDP) and the Moments Accountant to quantify per-client privacy loss. Using Speech Emotion Recognition (SER) as a privacy-critical benchmark, we show that FedAsync achieves up to 10× faster convergence but exacerbates fairness and privacy disparities: high-end devices contribute 6–10× more updates and incur up to 5× higher privacy loss, while low-end devices suffer amplified accuracy degradation due to infrequent, stale, and noise-perturbed updates. These findings motivate the need for adaptive FL protocols that jointly optimize aggregation and privacy mechanisms based on client capacity and participation dynamics, moving beyond static, one-size-fits-all solutions.

Federated Learning (FL) enables model training across decentralized devices by communicating solely local model updates to an aggregation server. Although such limited data sharing makes FL more secure than centralized approached, FL remains vulnerable to inference attacks during model update transmissions. Existing secure aggregation approaches rely on differential privacy or cryptographic schemes like Functional Encryption (FE) to safeguard individual client data. However, such strategies can reduce performance or introduce unacceptable computational and communication overheads on clients running on edge devices with limited resources. In this work, we present EncCluster , a novel method that integrates model compression through weight clustering with recent decentralized FE and privacy-enhancing data encoding using probabilistic filters to deliver strong privacy guarantees in FL without affecting model performance or adding unnecessary burdens to clients. We performed a comprehensive evaluation, spanning various datasets and architectures, to demonstrate EncCluster scalability across encryption levels. Our findings reveal that EncCluster significantly reduces communication costs — below even conventional FedAvg — and accelerates encryption by more than four times over all baselines; at the same time, it maintains high model accuracy and enhanced privacy assurances.

Federated learning (FL) as a novel paradigm in Artificial Intelligence (AI), ensures enhanced privacy by eliminating data centralization and brings learning directly to the edge of the user's device. Nevertheless, new privacy issues have been raised particularly during training and the exchange of parameters between servers and clients. While several privacy-preserving FL solutions have been developed to mitigate potential breaches in FL architectures, their integration poses its own set of challenges. Incorporating these privacy-preserving mechanisms into FL at the edge computing level can increase both communication and computational overheads, which may, in turn, compromise data utility and learning performance metrics. This paper provides a systematic literature review on essential methods and metrics to support the most appropriate trade-offs between FL privacy and other performance-related application requirements such as accuracy, loss, convergence time, utility, communication, and computation overhead. We aim to provide an extensive overview of recent privacy-preserving mechanisms in FL used across various applications, placing a particular focus on quantitative privacy assessment approaches in FL and the necessity of achieving a balance between privacy and the other requirements of real-world FL applications. This review collects, classifies, and discusses relevant papers in a structured manner, emphasizing challenges, open issues, and promising research directions.

Context: Federated Learning is an approach to distributed machine learning that enables collaborative model training on end devices. FL enhances privacy as devices only share local model parameters instead of raw data with a central server. However, the central server or eavesdroppers could extract sensitive information from these shared parameters. This issue is crucial in applications like speech emotion recognition (SER) that deal with personal voice data. To address this, we propose Optimized Paillier Homomorphic Encryption (OPHE) for SER applications in FL. Paillier homomorphic encryption enables computations on ciphertext, preserving privacy but with high computation and communication overhead. The proposed OPHE method can reduce this overhead by combing Paillier homomorphic encryption with pruning. So, we employ OPHE in one of the use cases of a large research project (DAIS) funded by the European Commission using a public SER dataset.

Context: Speech Emotion Recognition (SER) is a valuable technology that identifies human emotions from spoken language, enabling the development of context-aware and personalized intelligent systems.To protect user privacy, Federated Learning (FL) has been introduced, enabling local training of models on user devices.However, FL raises concerns about the potential exposure of sensitive information from local model parameters, which is especially critical in applications like SER that involve personal voice data.Local Differential Privacy (LDP) has prevented privacy leaks in image and video data.However, it encounters notable accuracy degradation when applied to speech data, especially in the presence of high noise levels.In this paper, we propose an approach called LDP-FL with CSS, which combines LDP with a novel client selection strategy (CSS).By leveraging CSS, we aim to improve the representatives of updates and mitigate the adverse effects of noise on SER accuracy while ensuring client privacy through LDP.Furthermore, we conducted model inversion attacks to evaluate the robustness of LDP-FL in preserving privacy.These attacks involved an adversary attempting to reconstruct individuals' voice samples using the output labels provided by the SER model.The evaluation results reveal that LDP-FL with CSS achieved an accuracy of 65-70%, which is 4% lower than the initial SER model accuracy.Furthermore, LDP-FL demonstrated exceptional resilience against model inversion attacks, outperforming the non-LDP method by a factor of 10.Overall, our analysis emphasizes the importance of achieving a balance between privacy and accuracy in accordance with the requirements of the SER application.