Network Operations Research Articles

Guest Editorial: Integrated Sensing and Communications for 6G

6G networks have been envisioned as key enablers for numerous emerging applications, including smart cities and homes, intelligently connected vechiles, smart manufactoring, and industrial Internet-of-Things (IoT). These applications require both extreme wireless connectivity as well as highly accurate and reliable sensing capability. Indeed, among many 6G visions, a common theme is that sensing will play a more significant role than ever before. By equipping wireless systems with the sensing functionality, 6G networks will go beyond classical communication and provide ubiquitous sensing services to measure, or to image, surrounding environments. This sensing functionality and the corresponding ability of the network to collect sensory data from the environment are seen as the foundation for building intelligence in the future smart world. Toward that end, there is a strong need to jointly design sensing and communication operations in 6G networks, which motivates the recent research of Integrated Sensing and Communications (ISAC).

Mechanism Design of Blockage Management Based on Local Flexibility Market under High Proportion of Distributed Energy

Continuous access to distributed energy sources greatly increases the uncertainty and complexity of distribution network operations. The distribution network is prone to congestion phenomena such as voltage or tide overruns and equipment overloads, thus causing a decrease in distribution network power supply reliability. Blockage management in distribution networks based on local flexibility markets is one of the blockage management methods currently adopted in many European regions. In order to clarify the operation mechanism of the local flexibility market, firstly, the basic requirements of the local flexibility market are introduced, including the concept of the local flexibility market, market members, and their rights and responsibilities; secondly, four key issues of local flexibility market construction are analyzed in detail: timing, controllability of resources, operating ownership, and Coordination of market members. Further, the organization and transaction process of the European local flexibility market is sorted out. Finally, two typical cases in the UK and their current market operation are introduced.

Design and performance analysis of optical sequence detector using lithium niobate-based Mach–Zehnder interferometers

Abstract Present article suggested and proposed a new design of 4-bit optical sequence detector using an electro-optic effect by utilizing Lithium niobate (LiNbO3) based Mach–Zehnder interferometer, because of its consistency about production and stability. Suggested sequential detector is designed by OptiBPM (optical beam propagation method) software. This structure uses EO effect to alter the light upon lithium niobate optical waveguide via cross-state to bar-state and replica. Sequence detector is a sequential state machine, that extract an input string of bits and, when the targeted bit has been detected, then it gives an output as 1. In this paper the authors have focused on all-optical 4-bit sequential detector to produce the outcome as 1 when the bit sequence finds as 1101 within the given data. This all-optical device can be utilized in several optical switching and networking operations. The suggested design is replicated with MATLAB along with mathematical descriptions too and has been verified.

Simulating IoT Workflows in DISSECT-CF-Fog.

The modelling of IoT applications utilising the resources of cloud and fog computing is not straightforward because they have to support various trigger-based events that make human life easier. The sequence of tasks, such as performing a service call, receiving a data packet in the form of a message sent by an IoT device, and managing actuators or executing a computational task on a virtual machine, are often associated with and composed of IoT workflows. The development and deployment of such IoT workflows and their management systems in real life, including communication and network operations, can be complicated due to high operation costs and access limitations. Therefore, simulation solutions are often applied for such purposes. In this paper, we introduce a novel simulator extension of the DISSECT-CF-Fog simulator that leverages the workflow scheduling and its execution capabilities to model real-life IoT use cases. We also show that state-of-the-art simulators typically omit the IoT factor in the case of the scientific workflow evaluation. Therefore, we present a scalability study focusing on scientific workflows and on the interoperability of scientific and IoT workflows in DISSECT-CF-Fog.

Adaptive rail transit network operations with a rollout surrogate-approximate dynamic programming approach

This paper presents an adaptive optimization framework for dynamic rail transit network operations with use of a rollout surrogate-approximate dynamic programming method. The optimization algorithm derives coordinated decisions of service schedules and train unit deployment with respect to prevailing passenger demand. Considering the computational effectiveness needed for real-time applications, a state-dependent surrogate function is incorporated to approximate the costs associated with operational decisions over future stages. The surrogate approximation is updated iteratively via a temporal difference learning process with feeding of observations made from the transit network. The proposed framework is implemented and tested on a real-world scenario in Hong Kong Light Rail Transit (LRT) network. The results reveal that the proposed framework is able to reduce significantly the total passengers’ waiting times over existing plans with reasonable computational time via use of the surrogate approximation. This suggests the potential of the proposed optimizer for real time applications in large-scale rail transit networks.

Modelling of ML-Enablers in 5G Radio Access Network-Conceptual Proposal of Computational Framework

The fifth generation (5G) of mobile networks connects people, things, data, applications, transport systems, and cities in smart networked communication environments. With the growth in the amount of generated data, the number of wirelessly connected machines, traffic types, and associated requirements, ensuring high-quality mobile connectivity becomes incredibly difficult for technology suppliers. Mobile operators and network vendors enrolling in 5G face far more rapid demands than any technology before, and at the same time need to ensure efficiency and reliability in the network operations. In fact, intelligent forecasting and decision-making strategies are several of the centerpieces of current artificial intelligence research in various domains. Due to its strong fitting ability, machine learning is seen to have great potential to be employed to solve telecommunication networks’ optimization problems that range from the design of hardware elements to network self-optimization. This paper addresses the question of how to apply artificial intelligence to 5G radio access control and feed ML techniques with radio characteristic-based automatic data collection to achieve ML-based evaluation of 5G performance. The proposed methodology endorses ML tools for the 5G portfolio scenarios requirements assessment and integrates into the mature methods for network performance optimization: self-organizing networks (SON) and minimization of drive tests (MDT). In this context, the proposed treatment guides future network deployments and implementations adopted on a 3GPP standard basis.

中国SKA区域中心跨洲际高速数据传输进展及展望

The Square Kilometre Array (SKA) is the largest radio telescope, and the data generated from its observations are first transmitted from Australia and South Africa to the scientific data processing center 100 km away, and then distributed to various SKA Regional Centres (SRC) with a distance of tens of thousands of kilometers through a high-speed network.In the SKA Phase 1 (SKA1) stage, 710 PB of data needs to be distributed to each SRC through a network of at least 100 Gbps each year. Such high network bandwidth and data scale bring great challenges to data transmission and distribution. This study analyzes a variety of network protocols, such as TCP/UDP/HTTP, and uses different software in the field of radio astronomy for testing and research. Then, the optimal transmission scheme parameters under the current infrastructure of a 10 Gbps network are obtained. In this study, the factors affecting high-speed transmission are discussed, and the corresponding performance optimization strategies are given.This study provides the technical foundation for China's network construction and layout before the official observation of the first stage of SKA and the technology accumulation of global network operations of the SRC in the future.The technical details and methods described in this study are essential for scientific applications that rely on large data cross-international nodes.

Analog nanophotonic computing going practical: silicon photonic deep learning engines for tiled optical matrix multiplication with dynamic precision

Abstract Analog photonic computing comprises a promising candidate for accelerating the linear operations of deep neural networks (DNNs), since it provides ultrahigh bandwidth, low footprint and low power consumption computing capabilities. However, the confined photonic hardware size, along with the limited bit precision of high-speed electro-optical components, impose stringent requirements towards surpassing the performance levels of current digital processors. Herein, we propose and experimentally demonstrate a speed-optimized dynamic precision neural network (NN) inference via tiled matrix multiplication (TMM) on a low-radix silicon photonic processor. We introduce a theoretical model that relates the noise figure of a photonic neuron with the bit precision requirements per neural layer. The inference evaluation of an NN trained for the classification of the IRIS dataset is, then, experimentally performed over a silicon coherent photonic neuron that can support optical TMM up to 50 GHz, allowing, simultaneously, for dynamic-precision calculations. Targeting on a high-accuracy and speed-optimized classification performance, we experimentally applied the model-extracted mixed-precision NN inference scheme via the respective alteration of the operational compute rates per neural layer. This dynamic-precision NN inference revealed a 55% decrease in the execution time of the linear operations compared to a fixed-precision scheme, without degrading its accuracy.

Machine learning enhanced next-generation optical access networks—challenges and emerging solutions [Invited Tutorial

Optical access networks are envisioned to become increasingly complex as they support more and more diverse and immersive services, each with a different capacity, latency, and reliability need. While machine learning has been touted as a silver bullet that will intelligently manage network operations and resources to meet these demands, as it had been anticipated for core and metro networks, there exist various challenges that need to be addressed to progress machine learning models from research to production. In this tutorial, we first aim to motivate the continued push to advance optical access networks and rationalize the use of machine learning in these networks. We then highlight the challenges that are especially amplified due to the traffic dynamicity and heterogeneity, data scarcity, and computation-resource constraints of optical access networks. We discuss emerging machine learning approaches that are being explored to address these challenges. Finally, we consider a fast and self-adaptive machine learning enhanced dynamic bandwidth allocation scheme in an illustrative future use case of supporting immersive human-to-machine communications over the mobile fronthaul of next-generation mobile networks.

DEICA: A differential evolution‐based improved clustering algorithm for IoT‐based heterogeneous wireless sensor networks

SummaryWith the evolution of technology, many modern applications like habitat monitoring, environmental monitoring, disaster prediction and management, and telehealth care have been proposed on wireless sensor networks (WSNs) with Internet of Things (IoT) integration. However, the performance of these networks is restricted because of the various constraints imposed due to the participating sensor nodes, such as nonreplaceable limited power units, constrained computation, and limited storage. Power limitation is the most severe among these restrictions. Hence, the researchers have sought schemes enabling energy‐efficient network operations as the most crucial issue. A metaheuristic clustering scheme is proposed here to address this problem, which employs the differential evolution (DE) technique as a tool. The proposed scheme achieves improved network performance via the formulation of load‐balanced clusters, resulting in a more scalable and adaptable network. The proposed scheme considers multiple parameters such as nodes' energy level, degree, proximity, and population for suitable network partitioning. Through various simulation results and experimentation, it establishes its efficacy over state‐of‐the‐art schemes in respect of load‐balanced cluster formation, improved network lifetime, network resource utilization, and network throughput. The proposed scheme ensures up to 57.69%, 33.16%, and 57.74% gains in network lifetime, energy utilization, and data packet delivery under varying network configurations. Besides providing the quantitative analysis, a detailed statistical analysis has also been performed that describes the acceptability of the proposed scheme under different network configurations.

Balancing Information Preservation and Data Volume Reduction: Adaptive Flow Aggregation in Flow Metering Systems

The critical role of network traffic measurement and analysis extends across a range of network operations, ensuring quality of service, security, and efficient resource management. Despite the ubiquity of flow-level measurement, the escalating size of flow entries presents significant scalability issues. This study explores the implications of adaptive gradual flow aggre- gation, a solution devised to mitigate these challenges, on flow information distortion. The investigation maintains flow records in buffers of varying aggregation levels, iteratively adjusted based on the changing traffic load mirrored in CPU and memory utilization. Findings underscore the efficiency of adaptive gradual flow aggregation, particularly when applied to a specific buffer, yielding an optimal balance between information preservation and memory utilization. The paper highlights the particular significance of this approach in Internet of Things (IoT) and contrasted environments, characterized by stringent resource constraints. Consequently, it casts light on the imperative for adaptability in flow aggregation methods, the impact of these techniques on information distortion, and their influence on network operations. This research offers a foundation for future studies targeting the development of more adaptive and effective flow measurement techniques in diverse and resource-limited network environments.

Reframing Neural Networks: Deep Structure in Overcomplete Representations.

In comparison to classical shallow representation learning techniques, deep neural networks have achieved superior performance in nearly every application benchmark. But despite their clear empirical advantages, it is still not well understood what makes them so effective. To approach this question, we introduce deep frame approximation: a unifying framework for constrained representation learning with structured overcomplete frames. While exact inference requires iterative optimization, it may be approximated by the operations of a feed-forward deep neural network. We indirectly analyze how model capacity relates to frame structures induced by architectural hyperparameters such as depth, width, and skip connections. We quantify these structural differences with the deep frame potential, a data-independent measure of coherence linked to representation uniqueness and stability. As a criterion for model selection, we show correlation with generalization error on a variety of common deep network architectures and datasets. We also demonstrate how recurrent networks implementing iterative optimization algorithms can achieve performance comparable to their feed-forward approximations while improving adversarial robustness. This connection to the established theory of overcomplete representations suggests promising new directions for principled deep network architecture design with less reliance on ad-hoc engineering.

Improving Lightweight AdderNet via Distillation From ℓ2 to ℓ1-norm.

To achieve efficient inference with a hardware-friendly design, Adder Neural Networks (ANNs) are proposed to replace expensive multiplication operations in Convolutional Neural Networks (CNNs) with cheap additions through utilizing l1 -norm for similarity measurement instead of cosine distance. However, we observe that there exists an increasing gap between CNNs and ANNs with reducing parameters, which cannot be eliminated by existing algorithms. In this paper, we present a simple yet effective Norm-Guided Distillation (NGD) method for l1 -norm ANNs to learn superior performance from l2 -norm ANNs. Although CNNs achieve similar accuracy with l2 -norm ANNs, the clustering performance based on l2 -distance can be easily learned by l1 -norm ANNs compared with cross correlation in CNNs. The features in l2 -norm ANNs are encouraged to achieve intra-class centralization and inter-class decentralization to amplify this advantage. Furthermore, the roughly estimated gradients in vanilla ANNs are modified to a progressive approximation from l2 -norm to l1 -norm so that a more accurate optimization can be achieved. Extensive evaluations on several benchmarks demonstrate the effectiveness of NGD on lightweight networks. For example, our method improves ANN by 10.43% with 0.25× GhostNet on CIFAR-100 and 3.1% with 1.0× GhostNet on ImageNet.

Fast Algorithms for Deep Octonion Networks.

This brief presents the results of a study of the possibilities of reducing the arithmetic complexity of computing basic operations in octonionic neural networks and also proposes new algorithmic solutions for efficiently performing these operations. Here, we primarily mean the operation of multiplying octonions, the operation of computing the dot product of two octonion-valued vectors, and the operation of multiple multiplications of an octonion by several other octonions. In order to reduce the computational complexity of these operations, it is proposed to use the fast Walsh-Hadamard transform, which is well known in digital signal processing. Using this transform reduces the number of multiplications and additions of real numbers required to perform computations. Thus, the use of the proposed algorithms will speed up computations in octonion-valued neural networks.

Spectral–Spatial Morphological Attention Transformer for Hyperspectral Image Classification

In recent years, convolutional neural networks (CNNs) have drawn significant attention for classification of hyperspectral images (HSIs). Due to their self-attention mechanism, the vision transformer (ViT) provides promising classification performance compared to CNNs. Many researchers have incorported ViT for HSI classification purposes. However, its performance can be further improved because the current version does not use spatial-spectral features. In this paper, we present a new morphological transformer (morphFormer) that implements a learnable spectral and spatial morphological network, where spectral and spatial morphological convolution operations are used (in conjunction with the attention mechanism) to improve the interaction between the structural and shape information of the HSI token and the CLS token. Experiments conducted on widely used HSIs demonstrate the superiority of proposed morphFormer over the classical CNN models and state-of-the-art transformer models. The source will be made available publicly at https://github.com/mhaut/morphFormer.

Development of a neural network model for the mechanical properties prediction of aluminum alloys during die casting with crystallization under pressure

A specialized software based on a neural network model is developed, which makes it possible to predict the hardness of aluminum alloys, the crystallization of which proceeds under conditions of comprehensive compression at a pressure of 500 MPa and with a change in the percentage of chemical elements. The object of control is the crystallizing metal. Theoretical provisions that allow correctly training a neural network are considered. A schematic circuit of the casting process with crystallization under pressure is developed, the initial data for neural network modeling and the activation function of neurons are selected. A model and an algorithm for searching for a neural network are developed, a listing of the program code, which includes the functions of normalizing values, sampling the neural network, visualization, and graph comparison operations is adduced. Verification of the program operation, which allows predicting the mechanical properties of products from aluminum alloys, is carried out by an experimental method. Is revealed that the developed neural network is able to predict the hardness of parts from aluminum alloys with a minimum error. The training sample, supplemented with new experimental data, went through the process of information treatment on the constructed neural network model. Keywords neural network model, automation, control, specialized software, aluminum alloys, pressure, crystallization, hardness

Securing wireless sensor network from node capture attack using an efficient optimization defense strategy model

Wireless Sensor Networks (WSNs) are a promising evolving technology for recent real-world applications in collaborative settings. These applications ranging from household monitoring to crucial military applications transmit critical data and thus more challenging approaches are required for providing security. In WSNs, a node capture attack (NCA) is a more serious attack among various attacks which supports an impostor to perform different network (NW) operations, and thereby the security of the entire NW is compromised easily. For conquering the surpassing issues of sensitivity-related techniques and for enhancing the NCA’s AE, we establish the NCA technique amalgamating several objectives like speed and location to lessen comprehensive intricacy. For securing the NW out of the NCA, this study proffers Improved Flower Pollination Optimization Defense Strategy Algorithm (IFPODSA) intending to compute the multiple objective functions for effectual chief administration procedure employing Hankel matrix formation. This proffered IFPODSA is correlated with two prevailing methodologies like Fruit Fly Optimization Algorithm (FFOA) and Finding Robust Assailant Optimization-Particle Swarm Optimization and Genetic Algorithm (FRAOPSOGA) concerning diverse criteria. Consequently, it has been observed that the proffered IFPODSA attains 458.7 kbps of throughput, 81.56% of PDR, 18.2% of energy consumption, and 64.36% of overhead.

A COVID-19 Detection Model Based on Convolutional Neural Network and Residual Learning

A model that can obtain rapid and accurate detection of coronavirus disease 2019 (COVID-19) plays a significant role in treating and preventing the spread of disease transmission. However, designing such a model that can balance the detection accuracy and weight parameters of memory well to deploy a mobile device is challenging. Taking this point into account, this paper fuses the convolutional neural network and residual learning operations to build a multi-class classification model, which improves COVID-19 pneumonia detection performance and keeps a trade-off between the weight parameters and accuracy. The convolutional neural network can extract the COVID-19 feature information by repeated convolutional operations. The residual learning operations alleviate the gradient problems caused by stacking convolutional layers and enhance the ability of feature extraction. The ability further enables the proposed model to acquire effective feature information at a low cost, which can make our model keep small weight parameters. Extensive validation and comparison with other models of COVID-19 pneumonia detection on the well-known COVIDx dataset show that (1) the sensitivity of COVID-19 pneumonia detection is improved from 88.2% (non-COVID-19) and 77.5% (COVID-19) to 95.3% (non-COVID-19) and 96.5% (COVID-19), respectively. The positive predictive value is also respectively increased from 72.8% (non-COVID-19) and 89.0% (COVID-19) to 88.8% (non-COVID-19) and 95.1% (COVID-19). (2) Compared with the weight parameters of the COVIDNet-small network, the value of the proposed model is 13 M, which is slightly higher than that (11.37 M) of the COVIDNet-small network. But, the corresponding accuracy is improved from 85.2% to 93.0%. The above results illustrate the proposed model can gain an efficient balance between accuracy and weight parameters.

An investigation of machine learning-based intrusion detection system in mobile ad hoc network

Building stable networks is one of the most demanding issues in the current era, as the world is increasingly reliant on computers and technology. The standard MANET protocols, software, and facilities presume a collaborative and networking atmosphere that does not consider protection. Intrusion detection systems (IDS) that track centralised network operations and detect malicious nodes are often used to supplement certain security because mitigation strategies are never sufficient. This study describes ML techniques for distributing valuable properties to IDS for green smart transportation on MANET. The performance of ML-IDSs and a review of their adequacy in MANETs help the users determine intrusion when learning about the MANET context. ML optimised KDD IDS. Ensemble learning in this IDS process gave anomaly scores to controlled packets. Our solution to actual MANET dataset shortages is this ML technique. ML techniques, simulation, and a functioning prototype had created a more resilient IDS for green smart transportation. ML-enhanced IDS detected and reduced MANET harmful activity. This research expanded IDS knowledge through ubiquitous learning.

Effects of fluorescent pair on the kinetics of DNA strand displacement reaction

Fluorescent labels are widely used in the characterizations of DNA-based reaction network operations. We systematically studied the effects of commonly used fluorescent pairs on thermal stabilities of signal-substrate duplex and the strand displacement kinetics. It is demonstrated that the modifications of duplex with fluorescent pairs stabilize DNA duplex by up to 3.5 °C, and the kinetics of DNA strand displacement circuit is also evidently slowed down. These results highlight the importance of fluorescent pairs towards the kinetic modulation in designing nucleic acid probes and complex DNA dynamic circuits.