With the continuous addition of an abundant of heterogeneous devices, the limitation of task delay has become an obstacle to the development of the Industrial Internet of Things (IIoT). Task offloading based on edge computing can provide low-latency computing services for these tasks. However, in the actual IIoT scenario, in contrast to cloud computing, edge computing has limited resources and computing capabilities. Resource-constrained edge resources cannot meet the offloading requirements of massive industrial devices. In this article, we propose an optimal joint offloading scheme based on resource occupancy prediction for the problem of computing offloading with limited edge resources. The scheme is divided into two parts, including edge resource occupancy prediction and task offloading. Simultaneously, considering multitask and the limitations of edge resources, gate recurrent unit (GRU) is used to predict the occupancy of edge resources. Formulating an optimal strategy of task offloading by using a reinforcement learning algorithm according to the network state and predicted results. The simulation results show that the scheme can effectively reduce the average delay of tasks, while minimizing the task offloading failure rate.

The Industrial Internet of Things (IIoT) is considered to be one of the most promising revolutionary technologies to increase productivity. With the refined development of manufacturing, the entire manufacturing process is split up into several areas of IoT production. Devices from different domains cooperate to perform the same task, which cause security problems in interacted communication among them. Existing authentication methods cause heavy key management overhead or rely on a trusted third party. It is imperative to protect privacy and ensure the credibility of the device during device interaction. This article proposes a federated hierarchical trust interaction scheme (FHTI) for the cross-domain industrial IoT. It builds a low-privacy network platform through blockchain and protects the data privacy of the IIoT. A hierarchical trust mechanism based on federated detection is designed to realize the unified trust evaluation of cross-domain devices. A trusted cross-domain method based on device trust value is designed to ensure the security and trustworthiness of cross-domain devices. The simulation results show that the FHTI scheme can improve the speed of identity authentication and the detection accuracy of malicious devices.

With the emergence of new services, the complex optical network environment makes it more difficult to predict network anomalies. This paper proposes a multi-index anomaly prediction scheme with hybrid supervised/unsupervised deep learning for elastic optical networks. Aimed at complex optical network indicators, the scheme presents three phases to enhance the abnormal prediction. The scheme first selects the most influential indicators of anomaly label among the mass of network indicators by calculating the Spearman correlation coefficient. Then, considering the timeliness of network data, it predicts time series of different indicators to analyze future network conditions by using long short-term memory neural network. In order to improve the accuracy and efficiency of the anomaly detection model, the scheme further establishes a deep neural network for anomaly classification. We also discuss how to process data without anomaly labels. The feasibility of the proposed scheme is verified on a real network dataset. Experimental results show that the scheme can predict the occurrence of future network anomalies with high accuracy, protect network services from potential abnormalities, and enhance the stability and robustness of the network.

C+L band optical line systems are a promising scenario to meet future network capacity upgrades. While expanding the transmission capacity, this technology has also aggravated the influence of stimulated Raman scattering (SRS) which leads to reduced Signal-to-Noise Ratio (SNR). If the resource allocation is always pursuing the improvement of SNR, the possibility of generating spectrum fragments will increase, thereby reducing the spectrum efficiency. However, higher spectrum efficiency is very likely to be at the expense of SNR, because the spectrum occupation may be too close to decrease SRS. Therefore, there is a balanced relationship between the SNR and spectrum utilization. In this paper, we use C+L optical cross connect (C+L-OXC) based on all-optical wavelength converters to break spectrum contiguity constraints. In view of this, we establish a two-dimensional frequency-time slot model that considers the negative impact of newly arriving services on the SNR of the already carried services, ensuring optimal transmission time when the link load is minimized. Additionally, each service occupies the spectrum resources with the target of the least number of fragments, thereby enhancing spectrum efficiency. This approach effectively resolves the tradeoff between network average SNR and spectrum efficiency by decoupling them from each other. We have established an experimental platform for optical switching, and the experimental results show that the proposed algorithm improves the average SNR in the C+L band, and can effectively optimize the spectrum utilization to improve the spectrum efficiency.

In recent years, the number of devices and terminals connected to the smart city has increased significantly. Edge networks face a greater variety of connected objects and massive services. Considering that a large number of services have different QoS requirements, it has always been a huge challenge for smart city to optimally allocate limited computing resources to all services to obtain satisfactory performance. In particular, delay is intolerable for services in certain applications, such as medical, industrial applications, etc, that such applications require the high priority. Therefore, through flexibly dynamic scheduling, it is crucial to schedule services to the optimal node to ensure user experience. In this paper, we propose a resource allocation scheme for hierarchical edge computing network in smart city based on attention mechanism, for extracting a small number of features that can represent services from a large amount of information collected from edge nodes. The attention mechanism is used to quickly determine the priority of the services. Based on this, task deployment and resource allocation for different task priorities are developed to ensure the quality of service in smart cities by introducing Q-learning. Simulation results show that the proposed scheme can effectively improve the edge network resource utilization, reduce the average delay of task processing, and effectively guarantee the quality of service.

Owing to the strong protection of data privacy, federated learning (FL) has become a key method for achieving intelligent decision making in smart homes. However, under realistic conditions, such as differentiated requirements and heterogeneous service environments, FL in smart homes faces the problem of non-independent and identically distributed (non-IID) data and uneven computing power, which leads to the poor adaptability of global models. To address this issue, this study proposes a cluster FL architecture based on edge-cloud collaboration. First, a Gaussian mixture model-based cluster FL is proposed to improve the model accuracy by clustering features on the FL dataset and ensuring an independent identical distribution of the data. Subsequently, a model training strategy based on edge-cloud collaboration is proposed to achieve the sharing of edge-cloud computing power by split learning, which provides sufficient computing power for model training. The simulation results show that the proposed architecture improves the accuracy of global models while ensuring normal network service provision.