International Journal of Communication Systems最新文献

Appropriate routing strategies are necessary for mobile ad hoc networks (MANETs) in order to facilitate effective data transfer. In order to counter the prevailing problems, the correct routing schemes will need to be selected as the default configurations are used. In this paper, a special optimal link state routing (OLSR) protocol is proposed to incorporate a deep learning methodology to facilitate efficient video streaming in MANETs. This study presents a new improved variant of the OLSR protocol, which is specially tailored to achieve efficient video streaming in MANETs. It is a radical approach that combines a deep-learning model with blockchain technology to overcome security and reliability issues. It starts with the gathering of video content that is available publicly. In order to detect black-hole nodes, a special twin-attention-based Elman spiking neural network model is applied. The reliability of the neighboring nodes is then measured by means of trust values. The pufferfish optimization algorithm, or the accuracy-aware energy-efficient multipath routing algorithm (AEMRAP), which takes into account node- and link-stability degrees, is used in making routing decisions. Interplanetary file system (IPFS) technology is used to store the data on blockchain and increase its security. The authentication of the blockchain architecture is conducted via the delegated proof-of-stake (DPoS) method that also delivers an extra protection of MANETs against unauthorized access. The study demonstrates superior performance in securing and optimizing video transmission, confirming that the extended OLSR protocol is highly effective for MANET video streaming applications. The proposed model exceeds the current approaches with a throughput of 2100 Kbps, an average end latency of 20.2 s, and a packet-delivery ratio of 92.3%.

The rapid increase in the integration of wireless sensor networks within the Internet of Things (IoT) ecosystem has led to crucial difficulties in providing reliable, energy-efficient, and secure communication. Most of the traditional intrusion detection models face few struggles in mitigating these challenges due to limited scalability and centralized processing. Therefore, this paper proposes a lightweight federated learning–based energy-aware (LF-LEA) model to overcome all the existing issues. The proposed model is an integration of federated learning, bio-inspired optimization, and energy-aware routing for decentralized environments. For local intrusion detection at edge nodes, the proposed model uses a lightweight convolutional neural network to transmit only the model updates rather than raw data, and this ensures the user's data privacy. The integration of the Levy flight and lotus effect mechanisms enhances the exploration and exploitation balance for improved intrusion detection accuracy. Furthermore, the S-LEACH–based routing protocol is incorporated to ensure secure and energy-efficient communication between nodes and base stations. Two benchmark datasets are used to validate the proposed model. The experimental results demonstrated that the proposed model achieves a higher accuracy of 98.60%, a precision of 98.32%, and a packet delivery ratio of 92.7%. In addition, the proposed model achieves a minimum communication delay and false alarm rate. Furthermore, the statistical Wilcoxon rank-sum test is conducted to confirm the effectiveness and consistency of the proposed model across diverse evaluation metrics. The overall result demonstrates that the proposed model ensures privacy preservation, scalability, and energy efficiency, making it a robust model for real-time intrusion detection in IoT-enabled applications, including smart cold storage monitoring systems, industrial automation, and environmental sensing networks.

Severe resource constraints, multi-hop connections, variable topology, and high sensitivity of transmitted data have raised the issue of quality of service (QoS) as one of the serious issues for wireless body area networks (WBANs). Therefore, many papers have been introduced in this field to improve the important aspects of the QoS field. However, studies indicate that some important issues have not been well considered, the most important of which are the lack of measures to consider the dynamic conditions of nodes and the effective support for varying traffic QoS. In addition, the performance of most methods leads to a sharp increase in control overhead. In this paper, a QoS-Temperature-Aware energy-efficient Adaptive Routing (QTAEEAR) based on the development of the Q-learning algorithm has been introduced. QTAEEAR is a three-step method. The routing table is created and updated in the first step. In the second step, routing and selecting the next-hop node are performed based on the Q learning. In the third step, learning is updated in response to new network conditions. Simulation results using NS-2 indicated the optimal performance and superiority of QTAEEAR compared to other similar methods.

software-defined networking (SDN) offers flexible traffic management but remains vulnerable to sophisticated cyberattacks, necessitating accurate and efficient network traffic detection. Existing SDN-based intrusion detection systems often suffer from high computational cost, poor scalability, and reduced accuracy in high-throughput or encrypted environments. To address these limitations, the NTD-SDN-RICCNN framework is proposed, which integrates fast robust iterative filtering (FRIF) for noise removal with spectral graph fast Fourier transform (SGFFT) for discriminative feature extraction. Rotation-invariant coordinate convolutional neural network (RICCNN) optimized with weighted velocity-guided gray wolf optimizer (WVGWO) for parameter tuning. The proposed method reduces redundant feature processing while improving detection accuracy and inference speed. Experiments on the SDN intrusion detection dataset show that NTD-SDN-RICCNN attains 99.7% accuracy, 99.6% precision, 99.5% recall, and reduces computational time by up to 32.5% compared to the state-of-the-art baselines. These results demonstrate the method's effectiveness and scalability for real-time SDN intrusion detection in diverse network conditions.

In wireless sensor networks (WSN), the sensor nodes are deployed randomly where the sensor nodes are not positioned away from each other. The intersection of sensing ranges creates an overlapping area. Every sharing node perceives the same event and generates redundant and associated data if it takes place inside the overlapping area. In this paper, a multimetric-based optimization of cluster selection and data redundancy elimination through copula variational LSTM in wireless sensor networks (CS-DRE-CVLSTM-WSN) is proposed. Initially, cluster formation using semantic invariant multiview clustering (SIMVC) for data aggregation in WSN by leveraging data characteristics and connectivity patterns is discussed. Then, the clusters formed are given to the Wader Hunt Optimization Algorithm (WHOA) for cluster head selection. For this, a novel method to enhance data redundancy elimination efficiency, considering factors, like trust degree computation, energy efficiency, link quality, path loss, distance from the target node to the base station, and aggregation delay is proposed. The selected clusters are fed into copula variational LSTM (CV-LSTM) to optimize cluster selection and predict temporal trends in key network metrics. The offline meta RL (OMRL) framework is proposed to suppress redundant data streams and decide which data to transmit or aggregate. The reward system assigns positive values for efficient aggregation and reduced communication costs, while penalizing data loss or unnecessary transmissions. The output is a set of policies for effective data aggregation and redundancy elimination. The performance of the proposed CS-DRE-CVLSTM-WSN method is evaluated with existing methods like the reliable cluster dependent data aggregation scheme for IoT with hybrid deep learning methods (RC-DAS-IoT-HDL), new machine learning-driven data aggregation for predicting data redundancy in IoT connected WSN (ML-PDR-IoT-WSN), and two vector data prediction techniques for energy-efficient data aggregation in WSN (TVDP-EEDA-WSN), respectively.

This paper investigates an energy-limited multi-user mixed radio frequency/free space optical (RF/FSO) system, where the RF links and FSO link undergo Rayleigh fading and Gamma-Gamma turbulence models incorporating the effect of pointing errors, respectively. Simultaneous wireless information and power transfer (SWIPT) based on the power splitting protocol is employed to maintain communication continuity of the system operating in an energy-limited scenario. Moreover, an energy-based user selection strategy is adopted to enhance the overall performance. Closed-form expressions for outage probability (OP) and approximate bit error rate (BER) are provided by using Meijer-G functions. The accuracy of theoretical derivations is validated by Monte-Carlo simulations. Results demonstrate that, under the user selection strategy, merely increasing the number of users cannot significantly boost system performance. Furthermore, influenced by multiple factors, the prudent design of the power splitting proportion enables system performance optimization.

The quick propagation of Internet of Things (IoT) devices in 6th Generation (6G) networks intensifies security challenges due to high-dimensional and diverse nature of IoT data, which complicates feature selection and increases computational overhead. Dynamic and evolving attack patterns, including various intrusion types, malware, denial-of-service attempts, and coordinated botnet attacks, further reduce detection reliability. To address these challenges, Optimized Periodic Implicit Generative Adversarial networks with advanced Transformer (OPIGAT) proposes an end-to-end framework for robust and scalable IoT security. The framework begins with a preprocessing stage that handles missing values, outliers, and normalization for clean and consistent data. Feature optimization occurs through a Hybrid Tuna Particle Swarm Optimization Algorithm (HTPSO), which Merges Particle Swarm Optimization (PSO) global search capability with Tuna Swarm Optimization (TSO) spiral foraging-inspired local refinement, enabling precise selection of compact and highly discriminative features while reducing dimensionality. OPIGAT classifier detects diverse attacks, with the generative component synthesizing realistic traffic patterns and the transformer module capturing contextual relationships, enhancing anomaly detection and reducing false positives. Finally, a lightweight blockchain integrated with InterPlanetary File System (IPFS) ensures secure and scalable data management, employing proof of authority, Elliptic Curve Cryptography (ECC) with Ring signature encryption, and smart contract–based revocation to maintain privacy, integrity, and efficiency. Extensive experiments on CICIoT-2023 and ROUT-4-2023 datasets demonstrate superior accuracy (98.78% and 97%), high detection efficiency (98.21%), and a low false alarm rate (20%), while IPFS-enabled storage supports seamless scalability. These results establish OPIGAT as a secure, efficient, and highly effective solution for 6G IoT intrusion detection.

The nonlinear power amplifier (NPA) in a cognitive radio (CR) network can degrade sensing performance. While spectrum sensing primarily focuses on the receiver, modeling the transmitter's NPA is crucial to understand its impact on detection. This paper investigates spectrum sensing performance in CR networks assisted by intelligent reconfigurable surfaces (IRS), considering the impact of NPA distortions. A novel multiple-input single-output (MISO) IRS-aided CR system, termed MICR, is proposed. The sensing performance of the proposed MICR system is analyzed in terms of the probabilities of false alarm and detection, considering the effect of IRS elements. The results are presented using receiver operating characteristic (ROC) curves. Furthermore, the impact of various system parameters such as the number of samples, number of IRS elements, IRS position, number of transmitting antennas, and phase shifts is analyzed. Additionally, the effects of line-of-sight (LoS) and non–line-of-sight (NLoS) propagation are evaluated. The ROC performance using different signal detectors is analyzed. The impact of interference, IRS overhead, and sensing energy on the proposed MICR system is also investigated. Extensive simulations demonstrate that IRS significantly improves the sensing performance of the CR network, both with and without NPA.

Performance issues with spectrum sensing are caused by multipath fading and shadowing, and obstacle blocking. This paper presents a novel Hybrid spectrum sensing framework that cooperatively combines energy detection (ED), matched filter detection (MFD), and cyclostationary feature detection (CFD) to detection reliability in cognitive radio networks. To mitigate fading and shadowing effects, cooperative spectrum sensing (CSS) is incorporated, leveraging multiple secondary users (SUs) for spatial diversity, achieving P_d = 0.80 at SNR = −20 dB using the OR fusion rule The key novelty lies in a dynamic switching mechanism that adaptively selects among ED, MFD, and CFD based on real-time signal and PU information availability, enabling operation under full, limited, or absent PU knowledge. Monte Carlo simulations in the low SNR region (−20 to −5 dB) demonstrate that the proposed hybrid detector achieves P_d values of 1, 0.98, and 0.95 for CFD, ED, and MFD, respectively, at P_f = 0.1. Moreover, it achieves energy efficiencies of 1.86 bits/J (non-CSS) and 2.05 bits/J (CSS), emphasizing the importance of energy-efficient spectrum access in power-constrained CR-IoT applications. The framework also minimizes average detection delay to 1.16 ms, ensuring faster PU identification. Owing to its adaptability and energy-aware operation, the proposed method is suitable for TV white spaces, CR-IoT, 5G-and-beyond networks, VANETs, and disaster-recovery communications.

Efficient resource allocation is vital for ensuring quality of service (QoS) in Long-Term Evolution (LTE) networks, especially for call admission control (CAC). This study introduces a multilevel bandwidth (BW) allocation model that dynamically assigns BW based on call requirements and computes the minimum necessary BW when availability is limited. Calls are classified as new calls (NC) or handoff calls (HC) based on network conditions. To enhance CAC decision-making, a cascade feedforward network (Cascade FFN), integration of a cascade neuro-fuzzy network and a deep feedforward neural network (DFNN), is employed. The proposed model achieves high BW utilization of 0.911, power efficiency of 65.098 dBm, and mean throughput of 516,830.584 bps. Also, it minimizes call blocking probabilities for NC, which is 0.591, and for HC, which is 0.605; the call dropping probabilities for NC, which is 0.606, and for HC, which is 0.592; the mean delay of 0.066 s; and the number of dropped users of 820.620. These results demonstrate significant improvements in LTE performance through intelligent BW allocation and CAC.