Yifei Wei’s research while affiliated with Beijing University of Posts and Telecommunications and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (28)


System architecture
Example of ABF
Performance comparison with different number of candidate clients
Clients selection distribution
The execution time of different algorithms

+3

Two-Stage Client Selection Scheme for Blockchain-Enabled Federated Learning in IoT
  • Article
  • Full-text available

November 2024

·

9 Reads

Xiaojun Jin

·

Chao Ma

·

Song Luo

·

[...]

·

Yifei Wei

Federated learning enables data owners in the Internet of Things (IoT) to collaborate in training models without sharing private data, creating new business opportunities for building a data market. However, in practical operation, there are still some problems with federated learning applications. Blockchain has the characteristics of decentralization, distribution, and security. The blockchain-enabled federated learning further improve the security and performance of model training, while also expanding the application scope of federated learning. Blockchain has natural financial attributes that help establish a federated learning data market. However, the data of federated learning tasks may be distributed across a large number of resource-constrained IoT devices, which have different computing, communication, and storage resources, and the data quality of each device may also vary. Therefore, how to effectively select the clients with the data required for federated learning task is a research hotspot. In this paper, a two-stage client selection scheme for blockchain-enabled federated learning is proposed, which first selects clients that satisfy federated learning task through attribute-based encryption, protecting the attribute privacy of clients. Then blockchain nodes select some clients for local model aggregation by proximal policy optimization algorithm. Experiments show that the model performance of our two-stage client selection scheme is higher than that of other client selection algorithms when some clients are offline and the data quality is poor.

Download

Cloud-edge collaboration-based task offloading strategy in railway IoT for intelligent detection

September 2024

·

6 Reads

Wireless Networks

Driven by technologies such as deep learning, online detection equipment can perform comprehensive and continuous monitoring of high-speed railways (HSR). However, these detection tasks in the railway Internet of Things (IoT) are typically computation-intensive and delay-sensitive, that makes task processing challenging. Meanwhile, the dynamic and resource-constrained nature of HSR scenarios poses significant challenges for effective resource allocation. In this paper, we propose a cloud-edge collaboration architecture for deep learning-based detection tasks in railway IoT. Within this system model, we introduce a distributed inference mode that partitions tasks into two parts, offloading task processing to the edge side. Then we jointly optimize the computing offloading strategy and model partitioning strategy to minimize the average delay while ensuring accuracy requirements. However, this optimization problem is a complex mixed-integer nonlinear programming (MINLP) issue. We divide it into two sub-problems: computing offloading decisions and model partitioning decisions. For model partitioning, we propose a Partition Point Selection (PPS) algorithm; for computing offloading decisions, we formulate it as a Markov Decision Process (MDP) and solve it using DDPG. Simulation results demonstrate that PPS can rapidly select the globally optimal partition points, and combined with DDPG, it can better adapt to the offloading challenges of detection tasks in HSR scenarios.


Figure 2. The architecture of the DDPG algorithm.
Figure 6. Simulation results of penalty. (a) Average delay ratio against absolute value of penalty; (b) success rate against absolute value of penalty.
Figure 7. Simulation results of different algorithms. (a) Comparison of algorithm (number of devices is 10); (b) comparison of algorithm (number of devices is 20).
Simulation parameters.
Computation Offloading Strategy for Detection Task in Railway IoT with Integrated Sensing, Storage, and Computing

July 2024

·

12 Reads

·

1 Citation

Electronics

Online detection devices, powered by artificial intelligence technologies, enable the comprehensive and continuous detection of high-speed railways (HSRs). However, the computation-intensive and latency-sensitive nature of these detection tasks often exceeds local processing capabilities. Mobile Edge Computing (MEC) emerges as a key solution in the railway Internet of Things (IoT) scenario to address these challenges. Nevertheless, the rapidly varying channel conditions in HSR scenarios pose significant challenges for efficient resource allocation. In this paper, a computation offloading system model for detection tasks in the railway IoT scenario is proposed. This system includes direct and relay transmission models, incorporating Non-Orthogonal Multiple Access (NOMA) technology. This paper focuses on the offloading strategy for subcarrier assignment, mode selection, relay power allocation, and computing resource management within this system to minimize the average delay ratio (the ratio of delay to the maximum tolerable delay). However, this optimization problem is a complex Mixed-Integer Non-Linear Programming (MINLP) problem. To address this, we present a low-complexity subcarrier allocation algorithm to reduce the dimensionality of decision-making actions. Furthermore, we propose an improved Deep Deterministic Policy Gradient (DDPG) algorithm that represents discrete variables using selection probabilities to handle the hybrid action space problem. Our results indicate that the proposed system model adapts well to the offloading issues of detection tasks in HSR scenarios, and the improved DDPG algorithm efficiently identifies optimal computation offloading strategies.



Fig. 1. System architecture.
Fig. 2. Model accuracy and loss with σ 2 S 2 = 0.1.
Fig. 3. Model accuracy and loss with σ 2 S 2 = 0.3.
Fig. 5. Model accuracy and loss with σ 2 S 2 = 0.3 use BP neural network.
Fig. 6. Model accuracy and loss with different size of sliding window. (a) is the variation of LR model accuracy with the size of sliding window; (b) is the variation of BP model with the size of sliding window.
Distributed IIoT anomaly detection scheme based on blockchain and federated learning

April 2024

·

12 Reads

·

3 Citations

Journal of Communications and Networks

Anomaly detection in the industrial internet of things (IIoT) devices is significant due to its fundamental role in protecting modern critical infrastructure. In the IIoT, anomaly detection can be carried out by training machine learning models. Data sharing between factories can expand the data from which the model is trained, thus improving the performance of the model. However, due to the sensitivity and privacy of IIoT data, it is also difficult to build a high-performance anomaly detection model between factories. To address this problem, we design an anomaly detection method for IIoT devices combined blockchain of main-side structure and federated learning. We store the global model on the main-chain while the side-chain records the hash value of the global models and local models, which updated by participating nodes, controlling nodes access to the global model through the main-side blockchain and the smart contracts. Only the nodes participating in the current federated learning training can get the latest global model, so as to encourage the nodes to take part in the training of the global model. We designed a proof of accuracy consensus algorithm, and select the nodes to participate in training according to the accuracy of the local model on the test dataset to resist the poisoning attack of the models. We also use the local differential privacy (LDP) algorithm to protect user data privacy from model inference attacks by adding noise to the local model. Finally, we propose a new algorithm named Fed_Acc to keep the accuracy of the global model stable when the users add a lot of noise to their local models.


Joint Task Offloading and Resource Allocation for Intelligent Reflecting Surface-Aided Integrated Sensing and Communication Systems Using Deep Reinforcement Learning Algorithm

December 2023

·

38 Reads

Sensors

This paper investigates an intelligent reflecting surface (IRS)-aided integrated sensing and communication (ISAC) framework to cope with the problem of spectrum scarcity and poor wireless environment. The main goal of the proposed framework in this work is to optimize the overall performance of the system, including sensing, communication, and computational offloading. We aim to achieve the trade-off between system performance and overhead by optimizing spectrum and computing resource allocation. On the one hand, the joint design of transmit beamforming and phase shift matrices can enhance the radar sensing quality and increase the communication data rate. On the other hand, task offloading and computation resource allocation optimize energy consumption and delay. Due to the coupled and high dimension optimization variables, the optimization problem is non-convex and NP-hard. Meanwhile, given the dynamic wireless channel condition, we formulate the optimization design as a Markov decision process. To tackle this complex optimization problem, we proposed two innovative deep reinforcement learning (DRL)-based schemes. Specifically, a deep deterministic policy gradient (DDPG) method is proposed to address the continuous high-dimensional action space, and the prioritized experience replay is adopted to speed up the convergence process. Then, a twin delayed DDPG algorithm is designed based on this DRL framework. Numerical results confirm the effectiveness of proposed schemes compared with the benchmark methods.



Figure 1 Overall System Model width (Ɓ) as a resource. As all the ß , are equipped with MEC, therefore, hereafter, we denoted base stations as ß ! ( providing Č, Ș, and Ɓ. The total computation, storage, and bandwidth resources are denoted as Č -.-/& (cycles), Ș -.-/& (bytes), and Ɓ -.-/& (Hz). The resources mentioned earlier at each ß , are virtualized by different T ! ' in the form of a set of slices Ǒ ! = 3Ǒ # , Ǒ2, Ǒ % , … Ǒ ) 5 where n is the total number of slices owned by numerous S ! '
Simulation Parameter Table
AN INTELLIGENT NETWORK SLICING FRAMEWORK FOR DYNAMIC RESOURCE SHARING IN MULTI-ACCESS EDGE COMPUTING ENABLED NETWORKS

June 2023

·

47 Reads

Latin American Applied Research - An international journal

Dynamic resource sharing in multi-access edge computing (MEC) enabled networks has gained tremendous interest in the recent past, paving the way for the realization of beyond fifth generation (B5G) communication networks. To enable efficient and dynamic resource sharing, Network Slicing has appeared as a promising solution, virtualizing the network resources in the form of multiple slices employed by the end-users requiring strict latency, proximate computations, and storage demands. In literature, network slicing is primarily studied in the context of communication resource slicing, and little research has been devoted to jointly slicing communication, energy, and MEC resources. In this paper, we, therefore, proposed a joint network-slicing framework that considers 1) communication resources, 2) compute resources, 3) storage resources, and 4) energy resources, and intelligently and dynamically shares the resources between different slices, aiming to improve tenants' overall utility. To this end, we formulated a utility maximization problem as Markov-chain Decision Process. We utilized a tenant's manager that employs a deep reinforcement learning technique named "deep deterministic policy gradient" to enable dynamic resource sharing. Simulation results reveal the effectiveness of the proposed scheme.



Fig. 1. The system model.
Fig. 7. (a) Total utility of MVNO v.s. computation charge, (b) total utility of MVNO v.s. caching charge, and (c) total utility of MVNO v.s. bandwidth charge.
Slicing-based resource optimization in multi-access edge network using ensemble learning aided DDPG algorithm

February 2023

·

34 Reads

·

7 Citations

Journal of Communications and Networks

Recently, the technological development in edge computing and content caching can provide high-quality services for users in the wireless communication networks. As a promising technology, multi-access edge computing (MEC) can offload tasks to the nearby edge servers, which alleviates the pressure of users. However, various services and dynamic wireless channel conditions make effective resource allocation challenging. In addition, network slicing can create a logical virtual network and allocate resources flexibly among multiple tenants. In this paper, we construct an integrated architecture of communication, computing and caching to solve the joint optimization problem of task scheduling and resource allocation. In order to coordinate network functions and dynamically allocate limited resources, this paper adopts an improved deep reinforcement learning (DRL) method, which fully jointly considers the diversity of user request services and the dynamic wireless channel conditions to obtain the mobile virtual network operator (MVNO) maximal profit function. Considering the slow convergence speed of the DRL algorithm, this paper combines DRL and ensemble learning. The simulation result shows that the resource allocation scheme inspired by DRL is significantly better than the other compared strategies. The output of the result of DRL algorithm combined with ensemble learning is faster and more cost-effective.


Citations (12)


... The proposed HSR MEC system integrates communication relays installed on HSTs to facilitate task communication with Base Stations (BSs) [21]. As illustrated in Figure 1, each HST is equipped with a device-detection sensor that identifies devices generating tasks. ...

Reference:

Deep Reinforcement Learning-Based Task Partitioning Ratio Decision Mechanism in High-Speed Rail Environments with Mobile Edge Computing Server
Computation Offloading Strategy for Detection Task in Railway IoT with Integrated Sensing, Storage, and Computing

Electronics

... The model incorporates a linear sigmoid singleton deep neural network (LS2DNN) for classification and utilizes the chebyshev chaotic mapping adapted jaya optimization algorithm (C2MJOA) for feature selection. Jin et al. [37] proposed an anomaly detection method for IIoT devices that combines blockchain with FL to enhance privacy and model performance. They store the global model on the main-chain, use side chain for model hash recording and apply smart contracts to control access to the global model. ...

Distributed IIoT anomaly detection scheme based on blockchain and federated learning

Journal of Communications and Networks

... Howeve method is slow in handling large-scale problem instances and can only be used in scenarios. Due to the excellent performance of the Deep Deterministic Policy Gr (DDPG) algorithm in solving problems within continuous action spaces, the auth [25][26][27] use the DDPG algorithm for dynamic resource allocation to optimize netwo formance. However, these studies have some limitations when using heuristic and forcement learning algorithms to address the separation between slice requests a source allocation. ...

Slicing-based resource optimization in multi-access edge network using ensemble learning aided DDPG algorithm

Journal of Communications and Networks

... The aim of this work is to enhance collaboration among distributed RL agents and improve efficiency. Authors [91] proposed an Energy Efficient Deep Deterministic Policy Gradient Resource Allocation (EE-DDPG-RA) method for RAN. The aim of this work is to improve long-term throughput while satisfying the QoS requirement. ...

Dynamically Resource Allocation in Beyond 5G (B5G) Network RAN Slicing Using Deep Deterministic Policy Gradient

Wireless Communications and Mobile Computing

... In [43], Xu et al. proposed a mobile-compatible offloading and resource allocation scheme based on the DRL method (i.e., Deep Q-Network (DQN)) aiming to minimize system cost. In [44], Yu et al. proposed an optimization strategy for task offloading and resource allocation for static devices based on the DRL method (i.e., Deep Deterministic Policy Gradient (DDPG)), aiming to minimize system energy 3 The source code has been released at: https://github.com/qiongwu86/Federated-SSL-task-offloading-and-resourceallocation consumption and resource overhead. ...

Blockchain-enabled RCS Task Offloading and Resource Allocation Policy Using DRL Approach
  • Citing Conference Paper
  • October 2022

... At the same time, we gained inspiration from the work of Li J et al. [16,17] and applied it to the image stitching step [18]. The keypoints are detected and the attention information is added to the keypoints. ...

Multi-Level Feature Aggregation-Based Joint Keypoint Detection and Description

... While the schedules produced were near-optimal, the high computational runtime rendered the model unsuitable for real-time scenarios, particularly in automotive environments. In contrast, ref. [24] proposed a DRL-based model to solve the routing problem in TSN networks. A key innovation of their work was the integration of graph convolutional networks within the DRL agent. ...

Joint Routing and Scheduling Optimization in Time-Sensitive Networks Using Graph-Convolutional-Network-Based Deep Reinforcement Learning
  • Citing Article
  • December 2022

IEEE Internet of Things Journal

... However, although detection can theoretically detect the condition of the pantograph, such as completeness or defect, there is no unified standard and it completely depends on the label specified by the annotator, which is not conducive to the update of the data set or the change in the standard in practical work, and the annotated file may be invalid. We investigated some popular object detection models YOLO [3,4] and found that segmentation is relatively more suitable for pantograph detection tasks. ...

Online Rail Fastener Detection Based on YOLO Network

... The works in [17]- [19] are DQN-based cache allocation approaches, but they simply choose a combination of content to store every step, resulting in huge computational overhead. To alleviate the overhead, the authors fixed the content size uniformly and deployed DQN agents on all nodes with storage. ...

A DQN-Based Cache Strategy for Mobile Edge Networks

... Edge computing can also support decentralized decision-making for faster and more reliable energy management [133]. Likewise, blockchain technology can offer secure and transparent energy transactions, empowering homeowners to trade excess energy [134]. Embracing these technological advancements can create a sustainable, efficient, and intelligent residential energy landscape with fully utilized SHEMSs and the integration of ESs. ...

Deep Reinforcement Learning for Edge Computing Resource Allocation in Blockchain Network Slicing Broker Framework
  • Citing Conference Paper
  • April 2021