Aggregation Protocol Research Articles

VOSA: Verifiable and Oblivious Secure Aggregation for Privacy-Preserving Federated Learning

Federated learning has emerged as a promising paradigm by collaboratively training a global model through sharing local gradients without exposing raw data. However, the shared gradients pose a threat to privacy leakage of local data. The central server may forge the aggregated results. Besides, it is common that resource-constrained devices drop out in federated learning. To solve these problems, the existing solutions consider either only efficiency, or privacy preservation. It is still a challenge to design a verifiable and lightweight secure aggregation with drop-out resilience for large-scale federated learning. In this paper, we propose VOSA, an efficient verifiable and oblivious secure aggregation protocol for privacy-preserving federated learning. We exploit aggregator oblivious encryption to efficiently mask users' local gradients. The central server performs aggregation on the obscured gradients without revealing the privacy of local data. Meanwhile, each user can efficiently verify the correctness of the aggregated results. Moreover, VOSA adopts a dynamic group management mechanism to tolerate users' dropping out with no impact on their participation in future learning process. Security analysis shows that the VOSA can guarantee the security requirements of privacy-preserving federated learning. The extensive experimental evaluations conducted on real-world datasets demonstrate the practical performance of the proposed VOSA with high efficiency.

P4SGD: Programmable Switch Enhanced Model-Parallel Training on Generalized Linear Models on Distributed FPGAs

Generalized linear models (GLMs) are a widely utilized family of machine learning models in real-world applications. As data size increases, it is essential to perform efficient distributed training for these models. However, existing systems for distributed training have a high cost for communication and often use large batch sizes to balance computation and communication, which negatively affects convergence. Therefore, we argue for an efficient distributed GLM training system that strives to achieve linear scalability, while keeping batch size reasonably low. As a start, we propose P4SGD, a distributed heterogeneous training system that efficiently trains GLMs through model parallelism between distributed FPGAs and through forward-communication-backward pipeline parallelism within an FPGA. Moreover, we propose a light-weight, latency-centric in-switch aggregation protocol to minimize the latency of the AllReduce operation between distributed FPGAs, powered by a programmable switch. As such, to our knowledge, P4SGD is the first solution that achieves almost linear scalability between distributed accelerators through model parallelism. We implement P4SGD on eight Xilinx U280 FPGAs and a Tofino P4 switch. Our experiments show P4SGD converges up to 6.5X faster than the state-of-the-art GPU counterpart.

Communication-Efficient and Privacy-Preserving Verifiable Aggregation for Federated Learning.

Federated learning is a distributed machine learning framework, which allows users to save data locally for training without sharing data. Users send the trained local model to the server for aggregation. However, untrusted servers may infer users' private information from the provided data and mistakenly execute aggregation protocols to forge aggregation results. In order to ensure the reliability of the federated learning scheme, we must protect the privacy of users' information and ensure the integrity of the aggregation results. This paper proposes an effective secure aggregation verifiable federated learning scheme, which has both high communication efficiency and privacy protection function. The scheme encrypts the gradients with a single mask technology to securely aggregate gradients, thus ensuring that malicious servers cannot deduce users' private information from the provided data. Then the masked gradients are hashed to verify the aggregation results. The experimental results show that our protocol is more suited for bandwidth-constraint and offline-users scenarios.

Pyas: Pay for Your Aggregation Service—A Privacy Preserving Aggregation Protocol of Time-series Data

Pyas: Pay for Your Aggregation Service—A Privacy Preserving Aggregation Protocol of Time-series Data

TFPA: A traceable federated privacy aggregation protocol

TFPA: A traceable federated privacy aggregation protocol

FedBully: A Cross-Device Federated Approach for Privacy Enabled Cyber Bullying Detection using Sentence Encoders

Cyberbullying has become one of the most pressing concerns for online platforms, putting individuals at risk and raising severe public concerns. Recent studies have shown a significant correlation between declining mental health and cyberbullying. Automated detection offers a great solution to this problem; however, the sensitivity of client-data becomes a concern during data collection, and as such, access may be restricted. This paper demonstrates FedBully, a federated approach for cyberbullying detection using sentence encoders for feature extraction. This paper introduces concepts of secure aggregation to ensure client privacy in a cross-device learning system. Optimal hyper-parameters were studied through comprehensive experiments, and a computationally and communicationally inexpensive network is proposed. Experiments reveal promising results with up to 93% classification AUC (Area Under the Curve) using only dense networks to fine-tune sentence embeddings on IID datasets and 91% AUC on non-IID datasets, where IID refers to Independent and Identically Distributed data. The analysis also shows that data independence profoundly impacts network performance, with AUC decreasing by a mean of 5.1% between Non-IID and IID. A rich and extensive study has also been performed on client network size and secure aggregation protocols, which prove the robustness and practicality of the proposed model. The novel approach presented offers an efficient and practical solution to training a cross-device cyberbullying detector while ensuring client-privacy.

DECTBEEP: A Novel Differential Evolution Clustering Tree based Energy-Efficient Protocol for Data Aggregation in Context of WSN

DECTBEEP: A Novel Differential Evolution Clustering Tree based Energy-Efficient Protocol for Data Aggregation in Context of WSN

Lattice based distributed threshold additive homomorphic encryption with application in federated learning

In federated learning (FL), a parameter server needs to aggregate user gradients and a user needs to protect the value of their gradients. Among all the possible solutions to the problem, those based on additive homomorphic encryption (AHE) are natural. As users may drop out in FL and an adversary could corrupt some users and the parameter server, we require a dropout-resilient AHE scheme with a distributed key generation algorithm. In this paper, we aim to provide a lattice based distributed threshold AHE (DTAHE) scheme and to show their applications in FL. The main merit of the DTAHE scheme is to save communication bandwidth compared with other latticed based DTAHE schemes. Embedding the scheme into FL, we get two secure aggregation protocols. One is secure against a semi-honest adversary and the other is secure against an active adversary. The latter exploits a smart contract in a ledger. Finally, we provide security proofs and performance analysis for the scheme and protocols.

Development and Implementation of an Internal Quality Control Sample to Standardize Oligomer-Based Diagnostics of Alzheimer's Disease.

Protein misfolding and aggregation are pathological hallmarks of various neurodegenerative diseases. In Alzheimer's disease (AD), soluble and toxic amyloid-β (Aβ) oligomers are biomarker candidates for diagnostics and drug development. However, accurate quantification of Aβ oligomers in bodily fluids is challenging because extreme sensitivity and specificity are required. We previously introduced surface-based fluorescence intensity distribution analysis (sFIDA) with single-particle sensitivity. In this report, a preparation protocol for a synthetic Aβ oligomer sample was developed. This sample was used for internal quality control (IQC) to improve standardization, quality assurance, and routine application of oligomer-based diagnostic methods. We established an aggregation protocol for Aβ1-42, characterized the oligomers by atomic force microscopy (AFM), and assessed their application in sFIDA. Globular-shaped oligomers with a median size of 2.67 nm were detected by AFM, and sFIDA analysis of the Aβ1-42 oligomers yielded a femtomolar detection limit with high assay selectivity and dilution linearity over 5 log units. Lastly, we implemented a Shewhart chart for monitoring IQC performance over time, which is another important step toward quality assurance of oligomer-based diagnostic methods.

A security-friendly privacy-preserving solution for federated learning

Federated learning is a privacy-aware collaborative machine learning method where the clients collaborate on constructing a global model by performing local model training using their training data and sending the local model updates to the server. Although it enhances privacy by letting the clients collaborate without sharing their training data, it is still prone to sophisticated privacy attacks because of possible information leakage from the local model updates sent to the server. To prevent such attacks, generally secure aggregation protocols are proposed so that the server will not be able to access the individual local model updates but the aggregated result. However, such secure aggregation approaches may not allow the execution of security mechanisms against some security attacks to model training, such as poisoning and backdoor attacks, because the server cannot access the individual local model updates and; therefore, cannot analyze them to detect anomalies resulting from these attacks. Thus, solutions that satisfy privacy and security at the same time or new privacy-preserving solutions that allow the server to execute some analysis on the local model updates without violating privacy are needed for federated learning. In this paper, we introduce a novel security-friendly privacy solution for federated learning based on multi-hop communication to hide clients’ identities. Our solution ensures that the forwardee clients in the path between the source client and the server cannot execute malicious activities by altering model updates and contributing to the global model construction with more than one local model update in one FL round. We then propose two different approaches to make the solution also robust against possible malicious packet drop behaviors by the forwardee clients.

ESE: Efficient Security Enhancement Method for the Secure Aggregation Protocol in Federated Learning

ESE: Efficient Security Enhancement Method for the Secure Aggregation Protocol in Federated Learning

Anonymous Homomorphic IBE with Application to Anonymous Aggregation

All anonymous identity-based encryption (IBE) schemes that are group homomorphic (to the best of our knowledge) require knowledge of the identity to compute the homomorphic operation. This paper is motivated by this open problem, namely to construct an anonymous group-homomorphic IBE scheme that does not sacrifice anonymity to perform homomorphic operations. Note that even when strong assumptions, such as indistinguishability obfuscation (iO), are permitted, no schemes are known. We succeed in solving this open problem by assuming iO and the hardness of the DBDH problem over rings (specifically, ZN2 for RSA modulus N). We then use the existence of such a scheme to construct an IBE scheme with re-randomizable anonymous encryption keys, which we prove to be IND-ID-RCCA secure. Finally, we use our results to construct identity-based anonymous aggregation protocols.

A Reliable Protocol for Data Aggregation and Optimized Routing in IoT WSNs based on Machine Learning

A Reliable Protocol for Data Aggregation and Optimized Routing in IoT WSNs based on Machine Learning

Privacy-preserving data aggregation scheme against deletion and tampering attacks from aggregators

Because privacy-preserving data aggregation protocols provide data privacy and data compression, they have been extensively studied in smart grids. However, most of the existing data aggregation protocols are based on honest-but-curious aggregators, or adopt computationally intensive asymmetric homomorphic encryption, making these protocols only provably secure in weak security models, or resulting in high computational complexity for terminal users. To address the aforementioned issues, in this work, we propose a novel lightweight privacy-preserving data aggregation scheme against malicious aggregators based on our symmetric homomorphic encryption. The proposed scheme can reduce the computational cost of smart meters while also resisting tampering and deletion attacks from malicious aggregators. One of the highlights of the proposed scheme is its ability to withstand deletion attacks from malicious aggregators, which makes it unnecessary for the malicious aggregator to aggregate the readings of all smart meters instead of only a part of them. Therefore, our scheme is more secure than other schemes. It is the first privacy-preserving data aggregation scheme based on a stronger security model without multiple rounds of interaction. Finally, the detailed security analysis shows that the proposed scheme satisfies desirable security properties. Moreover, experimental results demonstrate that the proposed scheme is superior to the other four schemes in terms of computational complexity and communication overhead.

SCDAP –secured cluster based data aggregation protocol for energy efficient communication in wireless sensor networks

A network of real time devices that can sense and transmit the information from the deployed environment by using multi hop communication is called as Wireless Sensor Network (WSNs). Despite the rapid advancement of WSN, where an increasing number of physical devices so called as sensors nodes are connected with each other, providing the improved security with optimized energy consumption during data transmission, communication and computation remains huge challenge. In wireless sensor networks, numerous sensor nodes are deployed in the physical environment to sense and collect the required information from the given environment. The sensed information is needed to be transmitted from the nodes to the control station in an energy efficient manner. Data aggregation is one kind of techniques which will optimize the energy usage in wireless sensor networks during the data transmission. In data aggregation, the unnecessary data is removed which will significantly reduce energy of the nodes during data transmission. However, collected data during the data aggregation should be completely protected and there are various threats that can be launched by the intruders to carry out unauthorised data access and can cause threat to the integrity of the network. Therefore, ensuring data security during the data aggregation process is very important and essential for the security of the network. In this paper, a Secure Cluster based Data Aggregation Protocol (SCDAP) have been proposed to provide better security through secure authentication and verification process, and to reduce overall energy consumption of the network by implementing secure clustering process to eliminate the redundant data in the network. Moreover, the proposed system is more efficient in generating public and private keys for effective and secure data transmission and verification process. The proposed system is experimentally tested in NS-3 tool and proves that the proposed system reduces high energy consumption, computational and communicational cost, end-to-end delay and improves the packet delivery ratio. Moreover, the proposed system provides better security in the network when compared to other existing systems during the data aggregation.

Analisis Jaringan LAN Menggunakan Teknologi EtherChannel untuk Meningkatkan Performa Jaringan pada SMK Kartika XX-I Makassar

We often encounter network quality decreasing over time due to the increasing number of users using the same amount of bandwidth. The decrease in network quality is due to the increase in the number of network users, with this happening a solution is also provided using the concept of utilizing the technology found in EtherChannel. There are also two options for using this technology, namely the PagP (Port Aggregation Protocol) and LACP (Line Aggregation Control Protocol) protocols. By raising this problem, it aims to test the level of network quality using the PagP and LACP protocols. This test was carried out using the client-server concept, which uses an FTP server application as a container for data transfer or data exchange, while the VLC software application is used as a streaming service. The value generated by the PagP protocol is much better than LACP when compared to using FTP services.

An Efficient Certificate-Based Aggregate Signature Scheme With Provable Security for Industrial Internet of Things

In Industrial Internet of Things (IIoT), the certificate-based aggregate signature (CBAS) scheme was used to ensure privacy and security of IIoT, thus, the concrete construction of CBAS scheme without bilinear pairing has proposed, which was employed as an underlying building block to create an efficient data aggregation protocol. In this article, we first point out that the previous CBAS scheme cannot keep its claimed security. Particularly, a malicious KGC can compute a valid forgery signature, because it has the knowledge of the master secret key. After that, we discussed how specific forgery attacks could be made to against the existing CBAS schemes, which demonstrates that the previous related constructions cannot achieve its claimed security. In order to further provide a secure CBAS scheme for IIoT, we suggest a method to improve the CBAS scheme and propose a concrete construction of it. In the random oracle model, based on the hardness of discrete logarithm problem, the security of our new proposal can be proved. Compared with the previous CBAS schemes, our construction has better security, which is efficient with respect to computation and communication.

Secure multiparty quantum aggregating protocol

Secure multiparty quantum computation is an important and essential paradigm of quantum computing. All the existing aggregating protocols are $(n, n)$ threshold approaches, where $n$ represents the total number of players. If one player is dishonest, the aggregation protocols cannot aggregate efficiently. In this paper, we propose a $(t, n)$ threshold-based aggregating protocol, where $t$ represents the threshold number of players. This protocol uses Shamir's secret sharing, quantum state, SUM gate, quantum Fourier transform, blind matrix, and Pauli operator. This protocol can perform the aggregation securely and efficiently. In this protocol, we simulate this aggregating protocol using the IBM quantum processor to verify the correctness and feasibility.

Blockchain-Oriented Privacy Protection of Sensitive Data in the Internet of Vehicles

The Internet of Vehicles is the specific instantiation of the Internet of Things in the field of transportation. Vehicle and driving data are often used to mine information about people's lifestyles and preferences. Without advanced data encryption and strict authorization measures, personal privacy and security are at great risk. To solve the problem that the relatively low security of vehicles connected to the Internet may threaten the privacy of users, a blockchain-based sensitive data privacy-protection scheme for vehicles connected to the Internet is proposed. First, association rules are used to mine big data in the Internet of Vehicles. Then, a data security aggregation protocol is established. Finally, an end-to-end encryption mechanism is created. The privacy protection of big data in the Internet of Vehicles is achieved through static and dynamic strategies. The experimental results show that the scheme can improve the hiding rate of sensitive data, the key generation time and encryption time are shorter than those of typical technologies, and the encryption ability is better. Therefore, in protecting the privacy of big data in the Internet of Vehicles, the blockchain proposed in this paper has better privacy protection and encryption abilities and is applicable to the protection of sensitive data in the Internet of Vehicles.

Efficient Verifiable Protocol for Privacy-Preserving Aggregation in Federated Learning

Federated learning has gained extensive interest in recent years owing to its ability to update model parameters without obtaining raw data from users, which makes it a viable privacy-preserving machine learning model for collaborative distributed learning among various devices. However, due to the fact that adversaries can track and deduce private information about users from shared gradients, federated learning is vulnerable to numerous security and privacy threats. In this work, a communication-efficient protocol for secure aggregation of model parameters in a federated learning setting is proposed where training is done on user devices while the aggregated trained model could be constructed on the server side without revealing the raw data of users. The proposed protocol is robust against users’ dropouts, and it enables each user to independently validate the aggregated result supplied by the server. The suggested protocol is secure in an honest-but-curious environment, and privacy is maintained even if the majority of parties are in collusion. A practical scenario for the proposed setting is discussed. Additionally, a simulation of the protocol is evaluated, and results demonstrate that it outperforms one of the state-of-art protocols, especially when the number of dropouts increases.