International Journal of Communication Systems最新文献

In the past few years, due to the massive growth of IoT-related devices in an interconnected ecosystem, serious attacks like distributed denial of service (DDoS), spoofing, sinkhole, and ransomware attacks have been observed. These extend from data breaches and privacy violations to several other types of cyber-attacks. Therefore, this paper proposed a novel type of clustering-based Tree Hierarchical Deep Convolutional Neural Network (TH-DCNN) model with Upgraded Human Evolutionary Optimization Algorithm (UHEOA) as an additional dimension for safeguarding the IoT from such attacks. It utilizes an Improved Soft-K-Means (IS-K-Means) algorithm to effectively cluster the IoT nodes in order to optimize resource utilization. The TH-DCNN guarantees efficient security by way of effective malicious attack recognition, whereas UHEOA adapts model parameters to operate at its best. The proposed TH-DCNN-UHEOA framework is tested in a simulation environment implemented using Python with 500 IoT nodes on a 4000 × 3600 m terrain area for 7 h under random mobility, with broadcast transmission and node restriction. The proposed framework achieves outstanding improvements compared with the state-of-the-art progress, including DNN-CL-IoT, Co-FitDNN-IoT, and CNN-TSODE-IoT. The proposed TH-DCNN-UHEOA achieves a packet delivery ratio (PDR) of up to 25.04%, a network lifetime (NLT) of up to 19.56%, and a detection accuracy of up to 26.76% higher compared with these baselines. All the parameters such as energy consumption, communication cost, throughput, PDR, NLT, energy consumption (EC), number of alive sensor nodes (NAN), accuracy, and number of dead sensor nodes (NDN) determine its efficiency, certifying the framework can repel malicious attacks like DDoS, spoofing, and sinkhole attacks, providing strong security to IoT systems.

A broadband dual-beam dual-polarized antenna array having low sidelobe and controllable beam pointing is proposed. A crossed-dipole antenna element with link lines is introduced to achieve excellent impedance matching, symmetrical and stable radiation pattern, low cross-polarization level, as well as very steady gain at 1.7–2.7 GHz. Based on the element, a new staggered 2 × 2 subarray is proposed to achieve excellent sidelobe suppression over the wide operating band. Then a dual-beam subarray, which consists of two staggered 2 × 2 subarrays and controllable metal fixtures, is designed to introduce dual-beam performance with low sidelobe. The dual-beam pointing can be easily controlled by tuning the metal fixtures. Finally, a dual-beam dual-polarized array is proposed to obtain high gain for practical application. It obtains a wide bandwidth of 45.5% (1.7–2.7 GHz) for reflection coefficient < −14 dB, and good isolation between all the ports larger than 21 dB. The array also has good dual-beam performance, with sidelobe levels below −20 dB and beam-pointing angles that can be varied to ±18°, ±28°, and ±38°.

Millimeter-wave (mmWave) communication plays a crucial role in wireless systems due to its high data rate capabilities and suitability for 5th generation (5G) networks. However, mmWave signals confront significant propagation issues, which include greater path loss, major attenuation from blockages, and sparse multipath propagation, constraining coverage and consistency. Accurate path loss validation is a serious and composite task for successful network planning, optimization, and resource allocations. To overcome these limitations, effective deep learning–based path loss estimation in mmWave communication systems is developed in this research work. Initially, the required data are collected from the standard datasets and given to the preprocessing phase. Once the data are preprocessed, they are given into the deep feature extraction phase, and it is done by applying the Pyramid Multihead Convolutional Cross Attention Network (PMC-CANet). The ability to ensure the efficiency of next-generation wireless networks is what makes it effective in feature extraction tasks. Finally, the path loss estimation process is performed on the extracted deep features through Adaptive Residual Bidirectional Gated Recurrent Unit (AR-BiGRU), where several parameters are tuned using the Updated Random Attribute–based Sculptor Optimization (URA-SO). One of the primary advantages of using AR-BiGRU with USOA for path loss estimation is its ability to process large, high-dimensional datasets, which can include not only geographical and environmental information but also temporal data, such as time-of-day or seasonal variations in path loss. The optimal solution outcome can be achieved by using the developed model. Then, its effectiveness is validated by comparing it with other existing models. This proposed system provides a consistent and best solution for tackling the problems of mmWave signal attenuation, thus enhancing the effectiveness and performance of next-generation wireless networks. The outcomes reveal that the proposed URA-SO-AR-BiGRU obtained an accuracy of 97.12% when taking the batch size as 15, leading to highly reliable and precise path loss estimations.

Redesigning the Internet to overcome the limitations of the current one has been a topic of discussion for some years. These discussions aim to provide solutions for optimizing resource consumption and providing more flexibility, which are key features in the modern world of interconnected devices and applications. Today, the applications created adhere to the norms of the current Internet, which is based on well-established protocols and tools. However, it is also restricted by its limitations, which are expected to be addressed in content-oriented future Internet architecture (FIA) solutions. These approaches aim to provide a network environment in which the data itself is the key in the communication, opposite to the current solution, which relies on hosts' address and location. The transition between different architectures in FIAs is anticipated to be challenging, as existing solutions must be modified to meet FIA protocols and patterns. Therefore, to ensure a seamless evolution of the network, regardless of the environment in which it is used, it is essential to provide a means of enabling this transition. For instance, solutions like NDN.JS and COIN, which were developed for NDN, are intended to facilitate communication across different architectures. This survey aims to provide a comprehensive overview of the potential benefits of interoperability between the current Internet/web and FIAs, as well as how existing applications can benefit from it.

Active distribution network (ADN) plays a vital role in the smart power grid implementation process. Congestion prediction and avoidance are essential in ADN to prevent overloads, improve efficiency, and ensure system stability. Various existing approaches are developed to alleviate congestion in ADN, but they are unable to predict the accurate congestion and require more time to generate appropriate control commands for power dispatch. Therefore, more advanced and adaptive methods are needed for accurate congestion management in modern power systems. This work implements a deep learning algorithm combined with an efficient cryptography model to ensure secure data transmission in ADN. Initially, network characteristics are collected during data transmission from the distributed network for attack detection and preprocessed using Two-Step Sparse Switchable Normalization (TSSN-Net) to normalize the data and Self-Attention-based Imputation for Time Series (SAITS) based missing value imputation to improve quality by filling in the blanks. The preprocessed data were used to select the features using redundancy analysis and interaction weight (RAIW). The Gaussian Process Reference-based General Regression Neural Network (GPR-GRNN) is then used to anticipate congestion using the selected features. Once the congestion is predicted, it is avoided using the Density-Based Congestion Control Protocol (DBCCP), which reduces the loss of packets to avoid the network congestion. After that, the gathered data are securely stored in the cloud server through linear feedback shift register encryption (LFSRE) based encryption algorithm. The proposed approach achieves an accuracy of 97.60%, PPV of 96.70%, selectivity of 97.20%, and NPV of 96.10%. The proposed approach enables significant advancement in modern power systems focusing on intelligent forecasting and uncertainty-aware congestion management for long-term success.

Utilizing licensed assisted access (LAA) for cellular long-term evolution (LTE) presents a viable solution to address the growing issue of limited wireless spectrum. Nevertheless, realizing the advantages of LTE-LAA requires a fair coexistence framework to ensure harmonious operation alongside existing WiFi networks. This study explores how collaborative and coexistent strategies between WiFi and cellular technologies in unlicensed bands can enhance the capacity of heterogeneous wireless networks. To this end, we focus on optimizing spectrum allocation to boost the performance of systems where LAA-LTE and WiFi networks operate together. By using the carrier aggregation technology, the unlicensed bands can be appropriately distributed to individual mobile devices. Our approach integrates a distributional reinforcement learning algorithm and three distinct one-to-many bargaining solutions, enabling adaptive responses to various wireless environments. Based on the learning and bargaining methodologies, cellular and WiFi access points act cooperatively with each other to enhance conflicting performance criteria. The core innovation of our method lies in leveraging hybrid optimization strategies to their fullest potential while simultaneously achieving mutual agreement among different network entities through collaborative mechanisms. The simulation results confirm the efficiency of the proposed hybrid control strategy and validate its overall effectiveness.

The preservation of high data rates and reliable operation of reconfigurable intelligent surface (RIS)–assisted wireless communication systems requires accurate channel state information (CSI), especially within the high-frequency bands of millimeter wave (mmWave). But the problem is that CSI acquisition in the direction of large-scale antenna arrays and RIS elements as the passive elements is highly fickle because of high dimensionality and the extra need for vast amounts of pilot overhead and a subsequent decision complexity. To address this challenge, we propose a novel coherence-optimized training pilot–based channel estimation with a deep compressed sensing–enabled ResUNet encoder–decoder architecture having skip connections. We take advantage of having the double structure of the sparsity inherent in angular cascaded channels in multiuser MIMO networks so as to recover the sparse signal efficiently with minimal pilot overhead. In contrast to traditional compressed sensing or classical deep learning architecture, the proposed ResUNet architecture achieves better convergence and estimation performance due to the talents of residual learning and attention blocks that allow the preservation of essential spatial information structure across the network layers. Moreover, we look at an optimum sensing matrix with a mutual coherence criterion to improve the recovery in a greater measure through a wide range of sparsity diligence and signal-to-noise ratio (SNR). Large-scale simulations show that our approach generates a better normalized mean squared error (NMSE) compared with state-of-the-art schemes like double-structured orthogonal matching pursuit (DS-OMP), JDCNet, and classic CNN-based estimators. It has been shown that performance is validated at different SNRs, pilot lengths, and different sparsity ratios, which indicate robustness and scalability of the proposed architecture. All these findings reinforce the fact that the combination of pilot optimization and deep compressed sensing can be used to deliver impactful and efficient CSI acquisitions in RIS-aided mmWave MIMO systems.

Vehicular ad hoc networks (VANETs) play a central role in intelligent transportation systems (ITSs), yet their performance is often constrained by high mobility, limited scalability, privacy risks, and increasing security threats, including quantum-enabled attacks. Existing approaches typically integrate software-defined networking (SDN), federated learning (FL), blockchain, or cryptographic techniques in isolation, resulting in fragmented control and inconsistent security guarantees. To overcome these limitations, this work proposes FLEDGE-SDVN, a unified FL-enabled, blockchain-secured, and postquantum-resilient SDN framework for secure and energy-efficient VANET operation. The framework employs dynamic hierarchical clustering for stable topology management and incorporates an FL-based trust model to detect malicious vehicles without sharing raw data. A lightweight blockchain ensures tamper-resistant maintenance of trust records, while Kyber- and Dilithium-based postquantum cryptography provides long-term secure authentication. SDN controllers enforce global routing policies, optimize traffic flow, and support adaptive decision-making. Extensive NS-3 and SUMO simulations demonstrate that FLEDGE-SDVN significantly enhances network performance compared to CRAS-FL, EECT, and DistB-VNET. Specifically, the framework improves packet delivery ratio by 10%–22%, reduces end-to-end delay by 13%–28%, enhances energy efficiency by 17%–31%, and achieves trust accuracy up to 96%. These results confirm that FLEDGE-SDVN offers a scalable, secure, and future-ready solution for next-generation vehicular communication systems.

Clustered wireless sensor networks (WSNs) comprise many small, battery-powered nodes that collect and transmit data to a central base station. Choosing the right cluster heads (CHs) is a crucial task in WSNs, as it involves a multi-objective problem. It requires balancing conflicting goals such as minimizing energy consumption and delay while maximizing packet delivery and network lifetime. Current multi-objective algorithms, such as MPSO and MFDO, try to balance these objectives. Nevertheless, they are computationally expensive to run, sensitive to parameter changes, and cannot easily adapt to large or dynamic WSNs. To address these limitations, this paper proposes a lightweight multi-objective sloth (MSloth) algorithm that introduces a novelty in addressing the research gap of unstable and energy-inefficient CH selection in WSNs. This bio-inspired dynamic weighting selects CHs based on node energy, density, and distance to the base station, while a sloth-inspired refractory mechanism and dynamic weighting schedule reduce CH rotation, ensuring adaptive and energy-efficient network performance. We compared MSloth against three other multi-objective optimization algorithms, MPSO, MFDO, and MOANA, using both simulated networks with 100–300 nodes and real WSN datasets with samples of 1000 nodes; the numerical results over 1000 and 10,000 rounds show that MSloth used 20%–27% less energy, demonstrated better network lifetime, and maintained low delays with 0.9 s for the synthetic dataset and 0.65 s for the real dataset. In tests involving 300 nodes, MSloth saved 20.26 J of energy out of a total of 50 J, kept 190 nodes active, achieved an 83.4% packet delivery rate, and recorded only 0.55 s of delay; statistical analysis confirmed that these improvements were significant across performance metrics. These findings suggest that MSloth is an effective and lightweight protocol suitable for real-world applications such as environmental monitoring, warning systems, and smart farming.

The increasing demand for high-speed, low-latency communication has led to significant advancements in hybrid optical-wireless access networks. Among these, time and wavelength division multiplexing passive optical networks (TWDM-PONs) have received considerable attention for their high capacity and efficiency. In this study, a reconfigurable fronthaul for cloud radio access networks (C-RANs) is developed using deep belief networks (DBNs) and a novel dynamic bandwidth allocation (NoDBA) algorithm. The primary objective is to design a low-latency, high-efficiency mobile fronthaul that optimally manages bandwidth allocation while ensuring seamless connectivity. The proposed DBN-NoDBA model, integrated with the XGS-PON standard, is evaluated through simulations, demonstrating superior performance compared with existing XGS-PON algorithms for handling mobile fronthaul and backhaul traffic. Additionally, a fast-converging genetic algorithm is introduced to enhance energy efficiency by dynamically activating and deactivating fronthaul units based on real-time traffic variations. The results confirm that DBN-NoDBA effectively reduces latency, optimizes bandwidth utilization, and significantly develops energy savings, making it a promising solution for next-generation 5G and beyond optical-wireless networks.