Aggregation Scheme Research Articles

CodedPaddedFL and CodedSecAgg: Straggler Mitigation and Secure Aggregation in Federated Learning

We present two novel federated learning (FL) schemes that mitigate the effect of straggling devices by introducing redundancy on the devices’ data across the network. Compared to other schemes in the literature, which deal with stragglers or device dropouts by ignoring their contribution, the proposed schemes do not suffer from the client drift problem. The first scheme, CodedPaddedFL, mitigates the effect of stragglers while retaining the privacy level of conventional FL. It combines one-time padding for user data privacy with gradient codes to yield straggler resiliency. The second scheme, CodedSecAgg, provides straggler resiliency and robustness against model inversion attacks and is based on Shamir’s secret sharing. We apply CodedPaddedFL and CodedSecAgg to a classification problem. For a scenario with 120 devices, CodedPaddedFL achieves a speed-up factor of 18 for an accuracy of 95% on the MNIST dataset compared to conventional FL. Furthermore, it yields similar performance in terms of latency compared to a recently proposed scheme by Prakash et al. without the shortcoming of additional leakage of private data. CodedSecAgg outperforms the state-of-the-art secure aggregation scheme LightSecAgg by a speed-up factor of 6.6–18.7 for the MNIST dataset for an accuracy of 95%.

MAQA: a quantum framework for supervised learning

Quantum machine learning has the potential to improve traditional machine learning methods and overcome some of the main limitations imposed by the classical computing paradigm. However, the practical advantages of using quantum resources to solve pattern recognition tasks are still to be demonstrated. This work proposes a universal, efficient framework that can reproduce the output of a plethora of classical supervised machine learning algorithms exploiting quantum computation’s advantages. The proposed framework is named Multiple Aggregator Quantum Algorithm (MAQA) due to its capability to combine multiple and diverse functions to solve typical supervised learning problems. In its general formulation, MAQA can be potentially adopted as the quantum counterpart of all those models falling into the scheme of aggregation of multiple functions, such as ensemble algorithms and neural networks. From a computational point of view, the proposed framework allows generating an exponentially large number of different transformations of the input at the cost of increasing the depth of the corresponding quantum circuit linearly. Thus, MAQA produces a model with substantial descriptive power to broaden the horizon of possible applications of quantum machine learning with a computational advantage over classical methods. As a second meaningful addition, we discuss the adoption of the proposed framework as hybrid quantum–classical and fault-tolerant quantum algorithm.

Wireless Traffic Prediction Based on a Gradient Similarity Federated Aggregation Algorithm

Wireless traffic prediction is critical to the intelligent operation of cellular networks, such as load balancing, congestion control, value-added service promotion, etc. However, the BTS data in each region has certain differences and privacy, and centralized prediction needs to transmit a large amount of traffic data, which will not only cause bandwidth consumption, but may also cause privacy leakage. Federated learning is a kind of distributed learning method with multi-client joint training and no sharing between clients. Based on existing related research, this paper proposes a gradient similarity-based federated aggregation algorithm for wireless traffic prediction (Gradient Similarity-based Federated Aggregation for Wireless Traffic Prediction) (FedGSA). First of all, this method uses a global sharing enhanced data strategy to overcome the data heterogeneity challenge of multi-client collaborative training in federated learning. Secondly, the sliding window scheme is used to construct the dual channel training data to improve the feature learning ability of the model; In addition, to improve the generalization ability of the final global model, a two-layer aggregation scheme based on gradient similarity is proposed. The personalized model is generated by comparing the gradient similarity of each client model, and the central server aggregates the personalized model to finally generate the global model. Finally, the FedGSA algorithm is applied to wireless network traffic prediction. Experiments are conducted on two real traffic datasets. Compared with the mainstream Federated Averaging (FedAvg) algorithm, FedGSA performs better on both datasets and obtains better prediction results on the premise of ensuring the privacy of client traffic data.

A Universal Privacy-Preserving Multi-Blockchain Aggregated Identity Scheme

Cryptocurrencies offer various benefits in terms of privacy protection and cross-border transactions, but they have also been used for illicit activities such as money laundering due to their anonymous nature and the difficulty of cross-border regulation. Additionally, the unethical actions of some virtual asset service providers (VASPs), such as rug pulls and the embezzlement of user funds, have further eroded the trust between users and VASPs. Implementing identity management on blockchains can help restore trust between users and VASPs. However, current solutions have privacy concerns as identity providers have access to the asset balances and transaction records of each user’s wallet account, and no solution can support all public blockchains unconditionally. To address these issues, this paper proposes a multi-chain aggregated identity scheme. In this scheme, the identity provider will issue a non-fungible token (NFT) for users who have undergone verification, and wallet accounts from different blockchains will be added to a cryptographic accumulator. The accumulator value is then bound to the identity NFT through a smart contract by the user. This allows the user to prove to others that the identity of the wallet account owner has been verified. The use of accumulators also allows users to combine proof for multiple wallets into a single proof, which significantly improves the efficiency and provides a way for VASPs such as centralized exchanges to demonstrate Proof of Reserves (PoR) to users. Importantly, this scheme preserves privacy as neither the identity provider nor the VASPs can link the user’s real identity with the wallet accounts. Only regulators can access the user’s identity data held by the identity provider and the user’s wallet account held by the VASP to link real identities with wallet accounts for the purpose of sanctions or criminal investigations. Additionally, the scheme supports all blockchains by allowing wallet accounts from any public blockchain to be added to the accumulator. Furthermore, the NFT implementation in the scheme helps prevent identity loss or theft, as it can only be transferred by the identity provider.

Pairing-Free Certificateless Aggregate Signcryption Scheme for Vehicular Sensor Networks

Smart vehicle applications play an increasingly significant role in constructing intelligent transportation systems. Vehicles with sensors are able to form temporary vehicular sensor networks dynamically, in which a mass of data are collected, shared, and aggregated to the traffic administration for improving road safety and efficiency. In such networks, vehicles exchange urgent or sensitive information with others in open wireless channels. To ensure secure and privacy-preserving communications, we develop a new pairing-free certificateless aggregate signcryption scheme based on the elliptic curve cryptosystem. The certificateless setting prevents the proposed scheme from both certificate management and key escrow issues. Aggregate signcryption without pairings realizes the aggregation and simultaneous authentication of multisource data by efficient cryptographic computing. The proposed scheme is proved to achieve confidentiality and unforgeability with the random oracle model. The security analysis shows that the proposed scheme provides forward secrecy, immunity to various known attacks, and pseudonym-based conditional privacy for vehicles. The performance analysis demonstrates that it satisfies more security properties with higher computation efficiency and lower communication overhead than state-of-the-art schemes.

SparSFA: Towards robust and communication-efficient peer-to-peer federated learning

Federated Learning (FL) has emerged as a powerful paradigm to train collaborative machine learning (ML) models, preserving the privacy of the participants’ datasets. However, standard FL approaches present some limitations that can hinder their applicability in some applications. Thus, the need of a server or aggregator to orchestrate the learning process may not be possible in scenarios with limited connectivity, as in some IoT applications, and offer less flexibility to personalize the ML models for the different participants. To sidestep these limitations, peer-to-peer FL (P2PFL) provides more flexibility, allowing participants to train their own models in collaboration with their neighbors. However, given the huge number of parameters of typical Deep Neural Network architectures, the communication burden can also be very high. On the other side, it has been shown that standard aggregation schemes for FL are very brittle against data and model poisoning attacks. In this paper, we propose SparSFA, an algorithm for P2PFL capable of reducing the communication costs. We show that our method outperforms competing sparsification methods in P2P scenarios, speeding the convergence and enhancing the stability during training. SparSFA also includes a mechanism to mitigate poisoning attacks for each participant in any random network topology. Our empirical evaluation on real datasets for intrusion detection in IoT, considering both balanced and imbalanced-dataset scenarios, shows that SparSFA is robust to different indiscriminate poisoning attacks launched by one or multiple adversaries, outperforming other robust aggregation methods whilst reducing the communication costs through sparsification.

SDAFA: Secure Data Aggregation in Fog-Assisted Smart Grid Environment

The tremendous growth of about 8 billion devices connected to each other in various domains of Internet of Things (IoT)-based applications have attracted researchers from both industry and academia. IoT is a network of several devices connected with each other to provide sensing capabilities, particularly in smart grid (SG) environment. Various challenges such as the efficient handling of massive IoT data can be addressed with advances in fog computing. The secure data aggregation challenge is one such issue in IoT-based smart grid systems, which include millions of smart meters. Typical SG-based data aggregation approaches have high computation and communication costs, however, many efforts have been made to overcome these limitations while leveraging fog computing but no satisfactory results have been obtained. Moreover, existing solutions also suffer from high storage requirements. The traditional data aggregation schemes such as GCEDA (Grouping of Clusters for Efficient Data Aggregation) and SPPDA (Secure Privacy-Preserving Data Aggregation) also suffer from a few shortcomings. SPPDA follows a mixed aggregation architecture that includes trees and clusters which can lead to some performance complexities and is not energy-efficient, whereas GCEDA does not support heterogeneity. To overcome these problems, this research provides a fog-assisted strategy for secure and efficient data aggregation in smart grid. The concept of smart grid is implemented in fog environment, which was not the case in previous schemes. We used communication between smart meters (SMs) and fog nodes (FNs) to transmit confidential data in compressed form towards FN. The FN further aggregates the received data which can then be updated in cloud repositories later. We presented two algorithms—data aggregation and data extraction at FN and cloud, respectively, to achieve secure communication. The performance of the proposed strategy has been evaluated against existing data aggregation techniques GCEDA and SPPDA for various performance parameters such as storage, communication cost and transmission cost. The proposed scheme overcomes the limitation of heterogeneity and mixed aggregation which was faced in GCEDA and SPPDA and the results revealed outstanding performance in comparison with both, so the proposed solution can be used in a smart grid environment for efficient and secure data transmission.

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.

Reliable cluster based data aggregation scheme for IoT network using hybrid deep learning techniques

BackgroundSeveral IoT nodes are deployed in the monitoring environment to ensure reliability. In both sensor and sink nodes, the same data is sensed and forwarded. While redundant data maintains reliability, sink nodes waste energy processing the redundant data. ObjectiveIn order to upholdthe compromiseamong energy ingesting and reliability, necessary to eliminate the redundancies in sensed data up to an appropriate level. Data aggregation algorithms currently assign time slots based on data sensing period and program rate, disregarding packet loss and latency. MethodologyIn this paper, we suggest a cluster based reliable data aggregation (CRDA) scheme for IoTnetwork which ensures data collection and aggregation in energy efficient manner and transfer to another end very effectively. We first introduce a monarch and sine-cosine (MSC) algorithm to form clusters by grouping the IoT sensors which ensures the effective data transferring. In data aggregation phase, we utilize the multiple design metrics to compute the trust degree of each IoT sensors and design an improved sunflower optimization (ISFO) algorithm to optimize the design constraints. The highest trust degree owned swelling is act as cluster head (CH) of the cluster which ensures data aggregation. A reformative optimal–learning-based deep neural network (ROL-DNN) is then used to compute routes between IoT sensors which ensures reliabledata aggregation and transferring. Resultsand analysis: Finally, we validate our proposed routing with the different simulation scenario and their results are compared with the existing routing protocols to prove the effectiveness.

PCAS: Cryptanalysis and improvement of pairing-free certificateless aggregate signature scheme with conditional privacy-preserving for VANETs

Certificateless aggregate signature (CLAS) is an effective approach for the multiple authentications in multi-users and multi-information environments like the vehicular ad hoc networks (VANETs), which can address the certificate management issue in the traditional public key cryptography (PKC) and solve the key escrow problem in identity-based cryptography (IBC). In recent years, a new CLAS mechanism (LICLAS) is presented for the authentication of sensors in healthcare wireless medical sensor networks (HWMSNs). In this paper, we make an analysis and prove that LICLAS is not able to against forgery attacks. Thus, we present an improved CLAS scheme (PCAS) for vehicle authentication in VANETs, which satisfies the privacy and security requirements. PCAS constructs the signature algorithm based on Elliptic Curve Cryptography (ECC) to avoid bilinear pairing and Map-to-Hash function operations. In order to further improve security and efficiency of PCAS, the vehicle key generation and signature algorithm is revised. The pseudonym mechanism is also adopted to achieve conditional privacy preservation. Besides, PCAS allows to execute a batch verification of messages, where RSUs may verify the aggregated signatures from vehicles to reduce time overhead. Under the assumption that the elliptic curve discrete logarithm problem (ECDLP) is hard, PCAS is proved to be existentially unforgeable against adaptive chosen message attacks in the random oracle model. Through the performance analysis, under the same time complexity, PCAS is shown to more effective. Compared with LICLAS, for a single message and 2000 messages, the transmission overhead of PCAS is reduced by 25% and 25% respectively, and the computation overhead is reduced by 16.56% and 25.34% respectively.

Privacy-enhanced and non-interactive linear regression with dropout-resilience

Linear regression, the most basic and widely used machine learning algorithm, has played an essential role in many areas such as healthcare, economics, and weather prediction. However, in practice, regression training requires a large dataset, usually stored in a distributed form from multiple data owners. Moreover, it is difficult to federate data among users to train a global model because of privacy issues and communication limitations. To address these challenges, we propose a non-interactive privacy-enhanced training scheme for linear regression based on functional encryption - FELR. In particular, based on the secure aggregation scheme and Paillier inner-product functional encryption, two cloud servers can train the model on the ciphertext. The solution does not require any interaction between the cloud servers and data owners. Throughout the training period, the cloud servers do not know any local private data. A detailed security analysis demonstrates that FELR can provide high security for data owners. The experimental conclusions also demonstrate that the performance of the training scheme on the data owners' side is approximately 25% higher than that of the existing solution, which significantly decreases the computational costs.

AAIA: an efficient aggregation scheme against inverting attack for federated learning

AAIA: an efficient aggregation scheme against inverting attack for federated learning

FSDA: Flexible Subset Data Aggregation for Smart Grid

Data aggregation is a practical method to protect individual privacy while using data utility, which aggregates the real-time electricity consumption data in an area into a single value in smart grids. Based on traditional data aggregation, multisubset data aggregation can obtain data distribution characteristics rather than just the sum of data. However, almost all existing multisubset aggregation schemes do not pay attention to adjusting the size of subsets according to the data distribution, which limits the data utility. In this article, a flexible subset data aggregation (FSDA) scheme based on Paillier homomorphic encryption for smart grids is proposed. The proposed FSDA scheme can dynamically adjust the size of each subset and aggregate the electricity consumption data into the corresponding subsets while protecting individual privacy. It is proved that FSDA scheme can obtain the distribution of consumption data and resist various internal attacks aiming to expose the individual or aggregated data. Experimental results show that FSDA scheme has lower overhead compared with related schemes, which is practical for resource-constrained smart meter devices.

Analysis and Improvement of an Efficient Certificateless Aggregate Signature With Conditional Privacy Preservation in VANETs

Certificateless aggregate signature reduces the trust in key generation center and eases the inconvenience of certificate management at a low cost of communication and computation. Because of these merits, a lot of new certificateless aggregate signature schemes have been proposed to provide enough security and privacy in vehicular ad hoc networks (VANETs). However, many of these schemes are insecure against various attacks. In this article, we analyze the security of an efficient certificateless aggregate signature with conditional privacy preservation in Internet of vehicles. The results of the analysis show that it is insecure against forgery attack. In order to overcome the security vulnerability, we put forth an improved scheme. Then, the formal and informal security analysis of our improved scheme both are provided, which show that the improved scheme meets the expected security requirements and is secure against several known attacks in VANETs. Finally, performance evaluation shows that our improved scheme is feasible for the VANETs in terms of computation and communication overhead.

A Robust and Optional Privacy Data Aggregation Scheme for Fog-Enhanced IoT Network

With the advances in fog computing, various users’ data, collected by smart devices in the Internet of Things (IoT), are published to facilitate services and efficiency. This leads to the users’ worry about security and privacy issues in fog-enhanced IoT. Although privacy protection can ease such worry, users need to pay some extra price, such as more computation costs. Besides, there are some users who want convenience rather than privacy. They prefer to publish their raw data to exchange for better convenience or benefit. In this article, an optional privacy-preserving data aggregation scheme based on BGN homomorphic encryption is proposed, without any trusted third party. In the proposed scheme, each user can choose either no privacy encryption or privacy encryption to upload its own data according to its own privacy sensitivity. With the help of the fog node, the control center can get the aggregated data of all smart devices and the single data of smart devices that choose the no privacy option. In addition, the proposed scheme is analyzed to achieve correctness, privacy preservation, and robustness. Experimental results and comparisons show the proposed scheme is satisfactory in terms of computation cost and communication overhead.

Privacy-preserving knowledge sharing for few-shot building energy prediction: A federated learning approach

The data-driven method is a promising way to predict the energy consumption of buildings, however suffering from the data shortage problem in various scenarios. Even though transfer learning can improve the few-shot prediction performance by utilizing other buildings’ data, the centralized approach poses potential privacy risks. To tackle this issue, the paper proposes a privacy-preserving knowledge sharing framework to facilitate the few-shot building energy prediction based on federated learning. First, a private data aggregation scheme is established to encrypt the sensitive data with shared random masks and guarantee the privacy of the data preprocessing and model optimization. Then, to alleviate the intrinsic data heterogeneity, a dynamical clustering federated learning algorithm is proposed to implement the intra-cluster and inter-cluster knowledge sharing along with the iterative clustering process for participating buildings. Finally, the network-based transfer learning approach is incorporated into the distributed framework to establish the customized model based on trained cluster models and further boost the prediction performance for each building. Extensive experiments on the Building Data Genome Project 2 (BDGP2) dataset indicate that the federated approach witnesses a desirable prediction performance while preserving the privacy of building occupants.

A Digital Twin-Based Drone-Assisted Secure Data Aggregation Scheme with Federated Learning in Artificial Intelligence of Things

A Digital Twin-Based Drone-Assisted Secure Data Aggregation Scheme with Federated Learning in Artificial Intelligence of Things

All-optical aggregation and de-aggregation between OOK, QPSK and 8QAM signals based on nonlinear effects

An all-optical aggregation and de-aggregation scheme between on–off keying (OOK) signals, quadrature phase shift keying (QPSK) signals and 8-ary quadrature amplitude modulation (8QAM) signals is proposed and numerically simulated based on nonlinear effects in the high nonlinear fiber (HNLF). In the de-aggregation processing, the input 30 Gbit/s 8QAM signal is firstly de-aggregated into one 20 Gbit/s QPSK signal and one 10 Gbit/s binary phase shift keying (BPSK) signal by deploying self-phase modulation (SPM), cross-phase modulation (XPM), four-wave mixing (FWM) and parameter amplification (PA) effects. One continuous optical carrier can be extracted from the converted BPSK signal by designing a feed-forward based modulation stripping configuration. The converted OOK signal can be obtained through vector superposition between the extracted optical carrier and the converted BPSK signal. In the aggregation processing, the converted 20 Gbit/s QPSK signal and the 10 Gbit/s OOK signal are fed to another HNLF and aggregated into one 30 Gbit/s 8QAM signal again by utilizing XPM and PA effects. Moreover, the aggregated 8QAM signal keeps the same wavelength and information integrity as the input 8QAM signal. the Error vector magnitude (EVM) and the bit error ratio (BER) for the 8QAM signal before and after aggregation and de-aggregation are measured to evaluate the scheme performance. The proposed all-optical aggregation and de-aggregation scheme for the advanced and simple modulation formats can be applied at the intermediate network node (INN) to solve the network traffic aggregation and de-aggregation with different modulation formats.

Task-Similarity-Based VNF Aggregation for Air-Ground Integrated Networks.

In a harsh environment, function aggregation of air-ground integrated network service function chaining (SFC) deployment can easily cause network load imbalance, which affects the network security and reliability. In this study, a task-similarity-based virtual network function (VNF) aggregation scheme was proposed. It considered air-ground network resource consumption and load balance before SFC mapping. A model for selecting VNFs to be aggregated based on task similarity was built. The tasks were classified based on their similarity. Furthermore, the VNFs to be aggregated were selected within the class under the constraints of the underlying physical resources. Load balancing was achieved by adjusting the similarity threshold. Moreover, an SFC mapping selection scheme based on network resource awareness was used to obtain the most suitable physical nodes for single-chain and multi-chain mapping according to various attributes of physical network nodes. The simulation results indicated that the proposed scheme with a better load balance design outperformed existing works on VNF aggregation. We also demonstrated that the task-similarity-based scheme was resource-consumption efficient and effective.

Point pattern analysis and classification on compact two-point homogeneous spaces evolving time.

This paper introduces a new modeling framework for the statistical analysis of point patterns on a manifold defined by a connected and compact two-point homogeneous space, including the special case of the sphere. The presented approach is based on temporal Cox processes driven by a -valued log-intensity. Different aggregation schemes on the manifold of the spatiotemporal point-referenced data are implemented in terms of the time-varying discrete Jacobi polynomial transform of the log-risk process. The n-dimensional microscale point pattern evolution in time at different manifold spatial scales is then characterized from such a transform. The simulation study undertaken illustrates the construction of spherical point process models displaying aggregation at low Legendre polynomial transform frequencies (large scale), while regularity is observed at high frequencies (small scale). K-function analysis supports these results under temporal short, intermediate and long range dependence of the log-risk process.