The concept of managed self-organization is a promising approach for controlling large-scale and complex network systems. However, enhancing the security of such systems remains a challenging task, although the threat of cyber attacks has become serious. In this paper, we focus on a managed potential-based routing mechanism for wireless sensor networks, in which an external controller observes and controls potentials (i.e., routing information), and propose a false injection attack detection and defense mechanism. In false injection attacks, an attacker illegally monitors control inputs from the external controller and injects malicious inputs into nodes, leading to inefficient packet forwarding. To reduce their impact, we introduce the concept of fallback control, in which control inputs are shut off when an attack is detected. Through simulation experiments, we demonstrate that the proposed mechanism can detect attacks and reduce the impact of attacks.

A UxNB is a radio access node to extend service coverage and serve out-vehicle user equipments in a given area, e.g., disaster area or close to stadiums. Since the UxNB is carried in the air by an uncrewed aerial vehicle (UAV), it is also referred to as a flying base station. However, wireless backhaul links are more vulnerable than wired backhaul links, and node failures are more likely to occur at flying base stations than ground base stations. On a multi-hop wireless backhaul, backhaul radio link failure (BH RLF) at the UxNB is caused by either the failure of the wireless link with the UxNB’s parent node or the failure of the parent node. The BH RLF recovery procedures described in the Third Generation Partnership Project (3GPP) specification produce recovery results that are classified into three cases. First, both the new and old parent node of the UxNB are served by the same Distributed Unit (DU) located on the ground. Second, the new parent node is served by a different DU underneath the same Central Unit (CU) located on the ground. Third, the new parent node is served by a different DU underneath a different CU located on the ground. The topology change and the recovery cost increase in the following order: first case, second case, and third case. In this paper, we propose a BH RLF recovery method utilizing a wireless repeater to enable both the new and old parent nodes to be served by the same DU. Evaluation results show that the proposed method reduces the recovery cost, suppresses the topology change, avoids traffic congestion at UxNBs, and inhibits the increase in the number of hops to the DU located on the ground.

With the increasing demand for efficient and reliable data delivery in wide-area transmission scenarios, traditional TCP/IP-based end-to-end architectures encounter inherent limitations, particularly under conditions of packet loss and long transmission paths. Information-Centric Networking (ICN), characterized by content-based naming and in-network caching, offers a promising alternative. However, most existing caching strategies are primarily designed to improve cache hit ratios, with limited attention to leveraging cached replicas for in-transit packet loss recovery. Consequently, they lack mechanisms to jointly optimize both immediate recovery value during ongoing transmissions and future reuse value across subsequent requests. These two value dimensions differ in terms of spatial scope and temporal duration, posing challenges for effective resource coordination. To address this issue, we propose a multi-dimensional value-aware caching strategy that integrates both value dimensions within a unified cache management framework. The strategy estimates the expected retransmission delay of path segments to quantify recovery value and employs a probabilistic placement mechanism to enhance loss resilience. Simultaneously, it evaluates reuse value through sliding-window-based request analysis and places replicas at nodes with low cache utilization to improve cache hit efficiency. Furthermore, a differentiated protection time mechanism is introduced to coordinate cache management. Simulation results demonstrate that the proposed strategy significantly improves transmission reliability, reduces latency, and increases cache hit ratios.

Dynamic Adaptive Streaming over HTTP (DASH) is an adaptive video streaming method over HTTP that achieves high-quality video playback in unstable network environments. SVC-DASH is a method that uses Scalable Video Coding (SVC) as the encoding method for video segments on DASH, which effectively prevents playback stalling due to delays in downloading video segments and improves cache efficiency in P2P video streaming. InterPlanetary File System (IPFS) video streaming is a method of distributing video using an IPFS/HTTP gateway. Video segments are shared between multiple gateways using the IPFS network, and video segments are sent to DASH clients using HTTP to achieve video distribution. This enables video streaming with guaranteed tamper resistance and censorship resistance for video content within IPFS networks. However, SVC uses many files (layers) to represent high-quality video, and this can cause a bottleneck in content exchange in IPFS networks. In this paper, we propose a prefetching method for IPFS video streaming with scalable video coding. In the proposed method, the IPFS/HTTP gateway prefetches the next enhancement layer up to the quality of the previously requested video segments. This reduces the download time of video segments for the client, and improves the quality of the video segments. In the evaluation, we evaluate the video playback bitrate on the Docker network, and show that the playback bitrate of the proposed method is improved compared to the conventional method. We also evaluate the prefetching performance of the proposed method using the hit rate and utilization rate, and show that the proposed method can prefetch video segments while reducing unnecessary traffic in the IPFS network.

With the continuous advancement of intelligent connected vehicle (ICV) technologies, the demand for computing resources by in-vehicle applications has significantly increased. In future zone-oriented centralized electronic/electrical architectures (E/EA), the vehicular computing platform (VCP) equipped with high-performance CPUs/GPUs provides the majority of computing resources. Under this centralized allocation paradigm, the issues arising from the imbalanced distribution of computing resources cannot be overlooked. To efficiently utilize surplus computing resources in VCPs, this paper proposes a multi-protocol integrated task offloading method for heterogeneous in-vehicle networks (IVNs). The approach leverages topic-based communication in Data Distribution Service (DDS) to offload complex tasks from resource-limited embedded devices to the VCP, thereby reducing task execution time. Considering the stringent real-time requirements of safety-critical applications, the Audio Video Transport Protocol (AVTP) is employed instead of DDS for data transmission during offloading when large volumes of pending data need to be transferred, effectively reducing communication latency. A task offloading agent is designed to bridge protocol heterogeneity in IVNs, extending offloading functionality to resource-limited devices. Furthermore, a greedy latency-aware offloading strategy is developed to obtain optimal offloading decisions within nanosecond-level latency. Finally, simulated IVNs based on automotive-grade chips were established, with physical experiment results validating the effectiveness of the proposed methods.

Efforts towards the realization of smart cities are gaining increasing attention to improve the efficiency and comfort of social activities. The smart cities are built on the IoT systems that gather and analyze various sensor data from different parts of the city. The wireless sensor networks are essential for constructing the IoT systems but have various vulnerabilities (e.g., data tampering, eavesdropping) to the integrity of the sensor data due to the openness of wireless communication. The existing study proposes a system for detecting anomalies in the sensor data by verifying the correctness of the time-series characteristics. However, it is difficult for the existing system to detect attacks that tamper with the sensor data by simulating the realistic pattern of the time series. Therefore, in this study, we assume that sensor nodes physically close in proximity can generate sensor data with similar information, and propose a new method to ensure the integrity of the sensor data by enabling nodes participating in different sensor networks to collaborate with the adjacent nodes to verify the data. The proposed system enables collaboration between different wireless sensor networks by temporarily switching communication protocols and forming a temporal network using BLE. Within this network, each sensor node exchanges its data with adjacent nodes belonging to other networks and performs mutual integrity verification using lightweight machine learning techniques, such as autoencoders. To demonstrate the feasibility of the proposed method, we implement a prototype system and show that it can reliably detect data tampering while maintaining low processing and energy overhead.

Despite various countermeasures implemented by railway companies, human accidents related to railways continue to occur. The railway companies are installing platform screen doors (PSDs) on station platforms to prevent accidents in which passengers fall off the platform. However, installing PSDs costs up to several billion yen per station, as well as upgrading station platforms for installation. On the other hand, the existing survey clarifies that there are predictive behaviors for drunken passengers falling from platforms. By detecting the abnormal behavior, the accidents could potentially be prevented. In the existing studies, a system that tracks the movement trajectory of a person by analyzing point-cloud data obtained from 3D LiDAR sensors that can accurately measure the distance, position, and shape of the object, installed at multiple locations is proposed and developed. However, these studies focus on estimating movement trajectories of people and do not address abnormal behavior detection such as the movement of a heavily intoxicated person. Additionally, they do not consider real-time processing capabilities, which is a significant limitation for critical situations. Therefore, in this study, we propose a system that detects the abnormal behaviors of people on the station platform by analyzing 3D point-cloud data taken by the 3D LiDAR. By extracting points corresponding to people from the point-cloud data, observing their movement trajectories, and analyzing them by a combination of rule-based and machine learning-based methods for abnormal detection, the proposed system aims to identify abnormal behaviors. Furthermore, the proposed system reduces processing time by adopting the system structure of edge computing for extracting only points corresponding to people on the edge side and for analyzing only the extracted data on the central part of the system.

This paper proposes novel cooperative relaying named “filter-and-forward cooperative relaying”. The proposed filter-and-forward cooperative relaying finishes signal transmission from a base station to the destination terminal only in two time slots despite of the number of the cooperative relays, which keeps high spectrum efficiency. In the proposed cooperative relaying, the signals received at the relays are filtered and the filtered signals are forwarded for the destination terminal, simultaneously. Moreover, this paper proposes two relay selection techniques for the proposed filter-and-forward cooperative relaying, where the performance degradation due to the Doppler shift in a packet is taken into account. One of the proposed selection techniques named as “MAX-SNR selection” achieves superior transmission performance. The other selection technique named as “distributed selection” can be implemented without any additional resource, which makes it easier to implement the proposed filter-and-forward cooperative relaying. However, the performance of the distributed selection is a little bit inferior to the MAX-SNR selection.

This paper presents a novel approach to synthetic aperture (SA) based ghost image suppression in sparse array imaging for millimeter wave (MMW) forward looking automotive radars. Although MMW radars are promising environmental sensors, such as in self driving systems, their spatial resolution is insufficient for recognizing targets especially in front of vehicles, such as pedestrians or artificial objects, due to their limited aperture sizes or far range investigation (over 10 m). To retain the spatial resolution in far range sensing, the sparse array configuration can be introduced, however it suffers from ghost images, i.e., grating lobes, due to phase uncertainty. This study addresses this issue by presenting a simple array optimization scheme and an image integration approach involving small radar motion (i.e., the SA effect), which can considerably suppress ghost images through a simple procedure. The results of a numerical simulation for a 79 GHz MMW multiple input multiple output radar show that our proposed approach greatly enhances reconstruction accuracy with sufficient spatial resolution.

Synthetic Aperture Radar (SAR) is an all-weather active observation system that enables imaging via microwave remote sensing. However, its inherent speckle noise and complex sea clutter background challenges traditional methods like Constant False Alarm Rate (CFAR) due to blurred small target features and high false alarm rates, such as blurred features of small targets and high false alarm rates. To address these issues, this paper proposes NPSA-Net, a SAR ship detection method based on noise perception and scatter attention. Based on RetinaNet, NPSA-Net introduces two key improvements: First, a Noise Perception Module is designed in the feature preprocessing stage, which acquires the influence weights of speckle noise on image channels by dynamically calculating the spatial variance distribution of feature maps. Second, a Scatter Attention Module is constructed at the output of the feature pyramid network (FPN). It utilizes multi-scale dilated convolutions to capture the scattering characteristics of ships and realizes feature enhancement through a dual-path attention mechanism. Experiments show that on public SAR ship datasets, NPSA-Net achieves significant improvements in precision and recall compared to baseline models, delivering superior detection performance.