Wired drone networks have emerged as an alternative to wireless multi-drone systems by eliminating radio interference and enabling stable data transmission. In such networks, drone placement must jointly consider sensor coverage and network connectivity under cable-length constraints caused by drone payload limits. In this paper, we propose a connectivity-aware drone placement method for cable-length-constrained wired drone networks. We formulate the placement problem to maximize the number of relayed sensors while satisfying perdrone cable-length constraints and ensuring full connectivity. The proposed method combines a greedy-based initial placement strategy with a connectivity-aware neighborhood search. Through simulation experiments, we show that our proposed method efficiently finds feasible solutions and achieves higher sensor coverage than simulated annealing method.

In this paper, we propose a robust optimizationbased mobile base station placement method for efficient store-carry-forward data collection using unmanned aerial vehicles (UAVs) in environments with limited or disrupted communication infrastructure. We focus on scenarios in which communication demand is temporally fluctuating and uncertain, with the aim to determine a placement strategy that maintains stable performance under varying future conditions. We formulate the base station placement problem as a robust optimization problem that explicitly accounts for demand uncertainty. We obtain the optimal solution using the column-and-constraint generation algorithm. To evaluate the effectiveness of the proposed approach, we conducted simulation experiments with predefined candidate sites and demand points, and compared the results with those of benchmark methods. The proposed method significantly reduces both the total deployment cost and number of installed base stations while maintaining a high data collection rate and balanced demand coverage, which demonstrates its robustness and efficiency, even under probabilistic demand fluctuations.

Unmanned aerial vehicles (UAVs) have attracted attention as aerial platforms for uplink data collection from ground devices. However, UAV communications are vulnerable to jamming interference, which degrades communication reliability and increases the overall mission time. This paper proposes a trajectory planning method for UAV-assisted data collection with directional antenna control under jamming environments. The proposed method jointly optimizes the UAV hovering positions and antenna directivity using particle swarm optimization in order to balance communication efficiency and mobility cost. By adaptively adjusting the antenna beamwidth, the UAV can suppress interference from jammers while maintaining reliable communication with legitimate ground terminals. In addition, a trajectory refinement strategy is employed to construct an efficient flight path that minimizes unnecessary movement. Through simulation experiments, we show that the proposed method effectively reduces the total mission time while ensuring reliable data collection in the presence of jammers.

This study investigated the effectiveness of machine learning techniques for detecting feint shrew attacks. A shrew attack exploits the retransmission timeout mechanism of the TCP protocol by transmitting intermittent bursts of high-rate packets, which severely degrade communication performance despite its minimal use of traffic, and is regarded as a representative type of low-rate denial-of-service (LDoS) attack [1]. In recent years, the feint shrew attack, a derivative form of the shrew attack, has emerged, making detection even more challenging by distributing traffic across multiple terminals or inserting feint traffic that closely resembles normal communication. To analyze such attacks, we constructed a network traffic dataset using ns-3 simulations and applied two machine learning models, a feedforward neural network (NN) and long short-term memory (LSTM), to classify normal and malicious traffic. The detection performance of these models was compared to evaluate the effect of considering temporal dependencies on feint shrew attack identification. The experimental results show that the LSTM model achieves higher detection accuracy than the NN model, demonstrating that incorporating temporal features is effective in detecting feint shrew attacks and highlighting the potential of machine learning-based analysis for understanding and mitigating LDoS attack behavior.

This paper proposes an unsupervised anomalous network traffic detection method that treats traffic flows as time-series data and integrates a Transformer encoder with DeepSVDD. With the increasing prevalence of encrypted communications, payload-agnostic detection approaches based on flow-level statistics have become increasingly important; however, conventional methods have difficulty adequately representing temporal dependencies among flows. In the proposed method, a sequence of flows is constructed, temporal context within the sequence is learned by the Transformer, and a normality region is subsequently learned by DeepSVDD, where deviations from this region are regarded as anomalies. We validate the effectiveness of the proposed method through experiments using traffic flow data. Although detection performance varies across attack types, the proposed method achieves discriminative performance that is overall comparable to or better than that of baseline methods.

IoT devices possess a radio-frequency fingerprint (RFF) due to inherent variations in manufacturing, making it device-specific and difficult to alter. This unique RFF, derived from RF signals, is often employed for physical layer authentication (PLA) by matching it against a pre-existing accesslist. However, such schemes struggle to accommodate a massive number of devices. This paper presents a novel new RFF-tag-based approach for device authentication. Essentially, the device sends its binary RRF feature within its message, and the receiver authenticates the device by matching this feature with the one extracted from the received RF signal. To prevent direct transmission of the raw binary feature, the RFF-tag, which is a parity check segment from an error-correcting codeword encoding the feature, is utilized. In the proposed scheme, all the signal processing is confined within the physical layer. The absence of an accesslist within the access point enables scalability to accommodate a massive number of IoT devices. In a simulation involving 500 randomly assigned legitimate devices and 4500 illegitimate devices, it is shown that the average false alarm rate and average misdetection rate are 1.60% and 1.38%, respectively, when the SNR is 22 dB.

We study irregular repetition slotted ALOHA (IRSA) with multi-antenna reception over Rayleigh block fading channels. An exact closed-form expression for the average decoding error probability ${\bar \varepsilon _{m,L}}$ is derived for collision sizes m=1 and m=2 by applying the inclusion–exclusion principle, which generalizes known single-antenna results and remains valid for any finite number of antennas. Using this result, we develop a density-evolution-based analysis of multi-antenna IR-SA systems and characterize belief-propagation (BP) thresholds. Numerical results for the corresponding maximum a posteriori (MAP) decoding thresholds and converse bounds are also presented, demonstrating threshold saturation with spatial coupling.

Machine learning-based malware detection has achieved high performance, but it remains vulnerable to adversarial malware. This paper proposes an adversarial malware detection method based on the stability of explanations generated by LIME. By applying LIME multiple times and measuring explanation stability using Spearman's rank correlation, the proposed method identifies adversarial malware without relying on specific attack patterns. Experiments on the AndroZoo dataset show that the proposed method detects various adversarial attacks with high accuracy while reducing computational cost compared to Kernel SHAP-based approaches.

This paper presents an analytical framework for Buffered-SIC (successive interference cancellation) slotted ALOHA (BSIC-SA) systems with real-time feedback under composite fading channels. The BSIC-SA employs buffered inter-slot SIC to utilize the residual signals from previous slots and intra-slot SIC under random fading for enhanced packet recovery. The received signal-to-noise ratio (SNR) is modeled by a mixture gamma (MG) distribution, which generalizes the exponential SNR model of Rayleigh fading by incorporating both exponential and power terms in its probability density function (PDF). By exploiting the series representations of the lower and upper incomplete gamma functions, closed-form expressions for the state transition probabilities in the Markov process of a two-device system are derived. The transmission probability and coding rate are jointly optimized to maximize the achievable sum rate. Our proposed analysis accommodates various fading environments without requiring separate derivations for individual channel models. Numerical analyses and simulation results confirm the validity of the proposed framework and demonstrate the achievable maximum sum rate for multi-device BSIC-SA systems.

In recent years, unmanned aerial vehicles (UAVs) have been extensively studied as aerial base stations for wireless communication systems. In most previous studies, researchers have focused on optimizing UAV trajectories to maximize communication efficiency; however, the impact of intentional interference such as jamming has received comparatively little attention. In this paper, we propose a directional antenna-assisted UAV path planning method that simultaneously satisfies legitimate users' communication demands and mitigates the impact of jamming attacks. The UAV is equipped with a directional antenna whose coverage pattern varies with altitude and antenna gain, introducing a trade-off between communication efficiency and robustness against interference. To address this, we jointly optimize the UAV's hovering positions and antenna parameters using particle swarm optimization within user clusters formed by spatial clustering. Then we connect the obtained hovering points via a traveling salesman problem-based path construction to minimize the total flight distance. Through simulation experiments, we show that the proposed method effectively avoids jamming signals while ensuring successful data transmission for all users.

We address the challenge of selecting a drone flight path that protects delivery destination privacy. A straight-line path makes the destination easily identifiable, compromising privacy. Existing methods, DLT (Dummy Location-based Tracking) and RLT (Random Location-based Tracking), focus on either increasing randomness or minimizing delivery time but fail to balance both. DLT introduces intermediate destinations, while RLT prioritizes efficiency by selecting waypoints within a defined region. To overcome this limitation, we propose a probabilistic method that selects DLT with probability p and RLT with 1 − p, effectively balancing delivery time and route randomness. Through simulation experiments, we show that our proposed method enhances privacy protection while maintaining efficient delivery times.

This paper proposes a flight path planning method for unmanned aerial vehicles (UAVs) adopting simulated annealing to maintain line of sight (LoS) between UAVs while avoiding obstacles. Existing approaches often assume obstacle-free environments, limiting the applicability of their methodologies in urban and post-disaster scenarios. The proposed method initializes feasible UAV paths and iteratively refines them adopting simulated annealing to optimize LoS maintenance and inter-UAV spacing. The proposed method incorporates a scoring function that balances the total path length, obstacle avoidance, LoS constraints, and UAV separation. Simulation results obtained in a three-dimensional environment demonstrate that the proposed method achieves a higher LoS maintenance rate and better UAV spacing than conventional approaches. This confirms the method’s effectiveness in ensuring stable UAV communication and efficient data collection in complex environments.

In this paper, we propose a two-stage detection method for detecting malware and adversarial malware. In the first stage, we use a general-purpose detector that targets several types of adversarial malware to determine whether the input data are malware. In the second stage, we use attack-specific detectors to detect malware that was determined to be benign in the first stage. The second-stage detector is composed of multiple detectors. We train each detector using data generated by a specific adversarial malware generation method. Through experiments on the Android malware dataset, we demonstrated that our proposed method improved accuracy compared with existing methods.

With the rapid increase in cyber threats, detecting unknown malware has become a critical challenge. Traditional deep learning-based malware detection methods struggle to identify new malware types because of their reliance on large labeled datasets. To address this issue, we propose a malware detection framework that integrates one-class support vector machine (SVM) and model-agnostic meta-learning (MAML). Our approach consists of two stages: (1) we use one-class SVM to distinguish between known and unknown malware classes, thereby identifying potential novel threats, and (2) we apply MAML to the extracted unknown samples to rapidly adapt and classify new malware types with minimal training data. Experiments conducted on a malware image dataset demonstrated that one-class SVM successfully identified the presence of unknown malware. Furthermore, our proposed MAML-based adaptation method significantly improved accuracy compared with conventional CNN-based models, which suggests that the combination of anomaly detection and meta-learning can enhance malware classification performance.

In this paper, we consider the optimal placement problem in seafloor optical wireless networks. In a seafloor optical wireless network, data collected by sensors are forwarded to relay nodes and delivered to a sink node via multiple relay nodes. The performance of data delivery to the sink node depends greatly on the placement of relay nodes. Therefore, this paper formulates an optimization problem in which the objective function is to minimize information freshness, i.e., the age of information. Since this optimization problem is a mixed integer linear problem once the relay node locations are determined, we propose a relay node placement method using a genetic algorithm. In the proposed method, the placement position of the relay node is assumed to be a gene, and the fitness value is obtained by solving the mixed integer programming problem. Through the simulation experiments, we show that efficient data transfer can be achieved by using the proposed method for relay node placement.

This paper explores the optimization of transmission probability and code rate in multi-device slotted ALOHA systems employing successive interference cancellation (SIC) and feedback over Nakagami- m fading channels. Although previous research has derived the optimal probability and the code rate to maximize the sum rate in a two-device scenario analytically using Markov process, its extension to configurations with more than two devices remains challenging. The complexity of Markov modeling arises from two main challenges: 1) an increase in the number of devices greatly expands the state space, precluding the analytical determination of system throughput; 2) deriving transition probabilities in the Nakagami- m channel model is difficult. In this study, we use computer simulations to evaluate the sum rate and to search for the optimal transmission probability and code rate to maximize the sum rate of the systems. Our simulation findings reveal that at an average SNR the optimal code rate is independent of the number of devices and transmission probability. In a 30-device slotted ALOHA system with an SNR of 15 dB, the maximum sum rate of 2.9235 is almost achieved at a transmission probability of 0.0665 and an optimal code rate of approximately 4.02, regardless of the number of devices.

We examine a neural network-based system designed to identify $K$ devices, each with device-specific I/Q imbalances in the RF modulator, as they transmit signals to an access point using time-division multiple access. Traditional methods begin training with randomly chosen initial parameters and need numerous pilot symbols, which can be impractical in scenarios such as when a unmanned aerial vehicle (UAV) serves as the access point, collecting data from massive device sensor networks. In this work, we focus on adaptive learning with minimal pilot symbols to train the neural network to classify devices by their I/Q imbalances. Reptile is an efficient meta-learning algorithm designed to help models rapidly adjust to new tasks with minimal training data. By employing Reptile, we pre-train neural network model parameters offline, facilitating rapid online adaptation and fine-tuning for effective identification of new devices. Simulations indicate that Reptile surpasses conventional methods in device identification with fewer pilot symbols.

This paper proposes an assembly-based data processing method using multiple computing Unmanned Aerial Vehicles (UAVs). While a computing UAV offers computational capabilities, its limited resources make it impractical for managing all processing tasks independently. Instead, each computing UAV is assigned a specific task, similar to service chaining in wired networks, ensuring sequential task execution. However, when UAVs sequentially visit sensors to process stored data, efficiency is reduced because UAVs cannot process tasks while flying. To address this issue, the proposed assembly method allows UAVs to collect data and exchange it among themselves at the assembly point. This approach reduces travel distance and improves processing efficiency. Through simulations, this study demonstrated that the assembly-based data processing method leads to significant enhancement in data processing efficiency.

In recent years, interest in drone-based logistics has grown due to the increasing demand for efficient and sustainable package transportation, driven by the expansion of e-commerce and rising environmental awareness. In this study, we focus on flight scheduling for the efficient transportation of packages between logistics bases, rather than on last-mile delivery. In scenarios where the number of packages handled at each base varies, efficient transportation can be achieved by having drones visit high-demand bases more frequently. To this end, we consider a system with two types of drones: local drones that visit all bases, and express drones that visit only selected high-demand bases. We formulate this problem as a mixed integer linear programming (MILP) model that minimizes the total transportation time. This model simultaneously determines which bases should be visited frequently and computes flight schedules that enable efficient package delivery. Unlike existing transportation models that assume fixed linear routes, our model allows for flexible routing, including direct flights and loop-based paths between bases. To ensure scalability, we also propose an approximation method that significantly reduces the computational cost. As the number of logistics bases increases, the exact solution of the MILP becomes intractable. Therefore, we pre-select candidate high-demand bases based on package volume and spatial layout, thereby reducing the number of decision variables. This makes it possible to compute high-quality solutions even in large-scale environments. Through numerical experiments, we show the effectiveness of our proposed methods for the transportation of packages between logistics bases.

Complex-valued neural network (CvNN)-based device identification is proposed to enhance the security of communications. Devices have in-phase/quadrature (I/Q) impairments that arise from variations in the manufacturing process. An access point (AP) receives I/Q symbols from a device and forms a constellation pattern, which has a unique feature of the device’s inherent I/Q impairments. The CvNN is able to recognize the device by categorizing the pattern. Simulations demonstrate that the CvNN has a lower error rate in identification than a realvalued neural network. This is because the CvNN captures the inherent correlation in complex-valued I/Q symbols caused by I/Q imbalances.