Analysis of Augmentation Methods for RF Fingerprinting under Impaired Channels

Dedicated fieldbuses were developed to provide temporal determinisms for industrial distributed real-time systems. In the early stages, communication systems were dedicated to a single protocol and generally supported a single service. Industrial Ethernet, which is used today, supports many concurrent services, but usually only one real-time protocol at a time. However, shop-floor communication must support a range of different traffic from messages with strict real-time requirements such as time-driven messages with process data and event-driven security messages to diagnostic messages that have more relaxed temporal requirements. Thus, it is necessary to combine different real-time protocols into one communication network. This raises many challenges, especially when the goal is to use wireless communication. There is no research work on that area and this paper attempts to fill in that gap. It is a result of some experiments that were conducted while connecting a Collaborative Robot CoBotAGV with a production station for which two real-time protocols, Profinet and OPC UA, had to be combined into one wireless network interface. The first protocol was for the exchange of processing data, while the latter integrated the vehicle with Manufacturing Execution System (MES) and Transport Management System (TMS). The paper presents the real-time capabilities of such a combination—an achievable communication cycle and jitter.

Section: Wireless Communicationmentioning

confidence: 99%

Industrial Shared Wireless Communication Systems—Use Case of Autonomous Guided Vehicles with Collaborative Robot

Stój

Fojcik

Cupek

et al. 2022

“…The reference [ 40 ] investigated the effect of environmental changes on the effectiveness of RF fingerprint identification and the problem that it is not easy to collect samples from different environments; the tapped delay line and clustered delay line (TDL/CDL) models were trained to improve the accuracy of recognizing transmitters significantly from 74% to 87.94% on the unobserved data. The reference [ 41 ] optimized the solution for the reference [ 40 ] and proposed a fine-grained augmentation approach to improve the learning performance of the deep learning model, resulting in an accuracy of 96.61% for recognition.…”

Section: Related Workmentioning

confidence: 99%

“…According to the research in references [ 33 , 34 , 35 , 36 , 37 , 38 , 39 ], edge computing has been more frequently employed for equipment monitoring. The references [ 40 , 41 ] show that deep learning augmentation algorithms have better recognition results in the case of insufficient training data. It is critical to developing a current-based edge monitoring approach with solid adaptability, cheap costs, and high reliability to suit the real-time monitoring needs of traditional processing equipment.…”

Section: Related Workmentioning

confidence: 99%

Monitoring the Production Information of Conventional Machining Equipment Based on Edge Computing

Wang

Shen

Zhu

et al. 2022

A production status monitoring method based on edge computing is proposed for traditional machining offline equipment to address the deficiencies that traditional machining offline equipment have, which cannot automatically count the number of parts produced, obtain part processing time information, and discern anomalous operation status. Firstly, the total current signal of the collected equipment was filtered to extract the processing segment data. The processing segment data were then used to manually calibrate the feature vector of the equipment for specific parts and processes, and the feature vector was used as a reference to match with the real-time electric current data on the edge device to identify and obtain the processing start time, processing end time, and anomalous marks for each part. Finally, the information was uploaded to further obtain the part processing time, loading and unloading standby time, and the cause of the anomaly. To verify the reliability of the method, a prototype system was built, and extensive experiments were conducted on many different types of equipment in an auto parts manufacturer. The experimental results show that the proposed monitoring algorithm based on the calibration vector can stably and effectively identify the production information of each part on an independently developed edge device.

“…With the explosive development of mobile Internet and deep learning (DL), intelligent edge computing services based on collaborative learning are widely used in various application scenarios [ 1 , 2 , 3 , 4 , 5 ]. For example, autonomous vehicles and face recognition cameras; these intelligent services put forward higher requirements on the privacy and security of models [ 6 , 7 , 8 ].…”

Section: Introductionmentioning

confidence: 99%

Edge-Cloud Collaborative Defense against Backdoor Attacks in Federated Learning

Yang

Zheng

Wang

et al. 2023

Federated learning has a distributed collaborative training mode, widely used in IoT scenarios of edge computing intelligent services. However, federated learning is vulnerable to malicious attacks, mainly backdoor attacks. Once an edge node implements a backdoor attack, the embedded backdoor mode will rapidly expand to all relevant edge nodes, which poses a considerable challenge to security-sensitive edge computing intelligent services. In the traditional edge collaborative backdoor defense method, only the cloud server is trusted by default. However, edge computing intelligent services have limited bandwidth and unstable network connections, which make it impossible for edge devices to retrain their models or update the global model. Therefore, it is crucial to detect whether the data of edge nodes are polluted in time. This paper proposes a layered defense framework for edge-computing intelligent services. At the edge, we combine the gradient rising strategy and attention self-distillation mechanism to maximize the correlation between edge device data and edge object categories and train a clean model as much as possible. On the server side, we first implement a two-layer backdoor detection mechanism to eliminate backdoor updates and use the attention self-distillation mechanism to restore the model performance. Our results show that the two-stage defense mode is more suitable for the security protection of edge computing intelligent services. It can not only weaken the effectiveness of the backdoor at the edge end but also conduct this defense at the server end, making the model more secure. The precision of our model on the main task is almost the same as that of the clean model.