With the advancement of sensing technology, the use of spatial information from LiDAR and similar measurement devices such as a depth camera is rapidly expanding. However, 3D spatial data contains trade secrets such as facility layouts and equipment configurations, making direct sharing a significant business risk. Additionally, from the perspective of data distribution between companies, a mechanism to prove the value of data utilization before purchase is essential. Existing approaches using trusted third parties or conventional encryption require data disclosure for utility verification, failing to achieve both confidentiality and value assessment simultaneously. Therefore, this study proposes a distributed platform that enables secure data exchange between organizations while ensuring confidentiality of 3D spatial information using cryptographic methods. The system operates on a Hyperledger Fabric-based permissioned blockchain to establish trust through immutable proof verification records for data distribution, and enables verification of data utility without disclosing any original data through zero-knowledge proof technology. Specifically, we implement a proprietary algorithm that generates feature values with concealed coordinates while preserving the geometric characteristics of the spatial information. Each participating organization generates feature values from spatial information and records proofs of the validity of this process on the blockchain, allowing other organizations not only to search for useful spatial information based on the feature values but also to verify the reliability of the feature values themselves. This enables previously difficult applications such as collaborative digital twin construction with competitors in manufacturing and logistics industries. Through empirical experiments, we clarify practical processing speeds in a consortium of multiple organizations, confirming the applicability in enterprise environments.

In order to realize beyond visual line of sight (BVLOS) unmanned aerial vehicle (UAV) flights for agricultural support in mountainous areas, it is necessary to estimate the flight range of UAVs where the stable communication with ground control terminals can be maintained for observing the condition of the flight path. Existing research proposes methods for estimating wireless communication qualities using radio wave propagation simulations based on ray tracing technology. However, when incorporating complex structures of the field such as vegetation in the mountainous areas, the computational load of the ray tracing markedly increases, making it difficult to perform frequent simulations in agricultural areas where terrain structures change continuously. Therefore, this study proposes a new wireless communication quality estimation method that integrates lightweight radio wave propagation simulation and machine learning technology. The proposed method first roughly estimates communication quality using the lightweight simulation that considers only coarse terrain structures such as elevation. And then, the estimation results are applied to a machine learning model trained to predict communication quality incorporating accurate 3D terrain structures. Through field experiments, we demonstrate that utilizing machine learning models enables the accurate prediction of radio wave propagation in the field by reflecting detailed terrain features.

A blockchain is an important component in the design of secure distributed file systems, such as cryptocurrencies. One of the key components of the blockchain is the key-value commitment scheme, which constructs a commitment value from two inputs: a key and a value. In a conventional commitment scheme, a single user constructs a commitment value from an input value, whereas in a key-value commitment scheme, multiple users construct a commitment value from their keys and values. Both conventional and key-value commitment schemes must satisfy binding and hiding properties. The key-binding and key-hiding properties guarantee that neither the sender nor the verifier can act maliciously. The concept of a key-value commitment scheme was first proposed by Agrawal et al. in 2020 using a strong RSA assumption. Their scheme satisfies the key-binding but not key-hiding properties. In this paper, we propose two lattice-based key-value commitment schemes, Insert-KVCm/2,n,q,βand KVCm,n,q,β, that satisfy both the key-binding and the key-hiding properties. The key-binding property of both Insert-KVCm/2,n,q,βand KVCm,n,q,βare proven under the short integer solution (SIS∞n,m,q,β) problem. The key-hiding property of both Insert-KVCm/2,n,q,βand KVCm,n,q,βare proven under the Decisional-SIS∞n,m,q,β-form problem, which is newly defined in this paper. We demonstrate the difficulty of the Decisional-SIS∞ n,m,q,β-form problem by showing that the Decisional-SIS∞n,m,q,β-form problem is secure when the SIS∞ n,m,q,β problem is secure. Finally, we analyze the computational costs of Insert-KVCm/2,n,q,βand KVCm,n,q,β. Our method is the first lattice-based key-value commitment scheme with proven the key-binding and the key-hiding properties.

With each new iteration of mobile wireless networks, demands for low latency and real-time processing continue to grow. Traditional cloud-based processing suffers from latency and bandwidth issues, hindering real-time applications. To address this, we propose an architecture that distributes computation to edge servers in order to reduce latency and improve the user experience. Using the OMNeT++ platform with Simu5G and INET libraries, we simulate a detailed 5G network with MEC integration. Central to our architecture is the MEC Orchestrator, which manages edge computing resources and interacts with the MEC Hosts which are small edge-located servers that process data and deliver services closer to the user, minimizing latency and improving the user experience. We then simulate the dynamic interactions between a mobile user's User Equipment (UE) and the MEC infrastructure. To further enhance the efficiency of our proposed architecture, we propose to integrate an algorithm that can predict the UE's future location. This algorithm will enable optimal MEC Host selection for user tasks, ensuring data processing occurs closest to the user's predicted future location.

Internet of Things (IoT) systems using real world data are making our lives more convenient and efficient by using environmental information. The IoT system consists of an IoT device that obtains real-world data, a server that processes and stores the data obtained by the IoT device, and actuators that adjust environmental information. If the server is compromised by an attacker, the data stored on the server may be tampered, and cause the actuator to operate incorrectly based on the tampered data. Existing studies focus on the possibility that multiple IoT devices may also exist near the in a proximity to detect the occurence of the data tampering by the malicious device or attackers. In this existing study, unique data is shared among the multiple IoT devices in close proximity using BLE (Bluetooth Low Energy) and sent to a server along with the obtained real-world data. The server compares and analyzes the data sent from the IoT devices to verify whether the real-world data is tampered or not by the malicious IoT device or attackers. However, the unique data may be recorded by an attacker and shared via BLE to a distant IoT device, which misleads the IoT system into believing the proximity of an IoT device that is not originally in proximity, leading to a false sense of data integrity. Therefore, in this study, we propose a method to verify the proximity of an IoT device to other IoT devices based on the similarity of environmental information to ensure the integrity of data before it is stored in the blockchain. To solve the existing problem, this study proposes the development of a data management infrastructure equipped with the capability to identify devices in physical proximity by recording and verifying environmental information dependent on time and location, thereby enhancing security against such attacks. The approach further addresses attacks that attempt to tamper with the data before it is stored on the blockchain, enhancing security measures against these threats.

With the spread of ICT technology, it has become common for people to spend long hours at their desks in the office, which hurts their health. In addition, this effect leads to stress, which in turn reduces concentration and makes it difficult to produce good results at work. Consequently, it is necessary to observe regular stress and encourage breaks to prevent prolonged sitting at the desk. To solve this problem, sensing systems have been researched and developed to measure heartbeats by mounting sensors on desk chairs and user interfaces of the computer and to analyze the measurement results. The existing methods require many sensors to be installed on various devices and thus cause large costs. On the other hand, the existing methods measure heartbeats without contacting the device. For example, existing research has constructed a fatigue estimation system using a non-contact Doppler sensor that can measure heartbeats and breathing from a distance. However, existing Doppler sensors are directional, meaning that they can only observe a specific direction, making it impossible to observe a large number of people at the same time with only a limited number of sensors. Therefore, in this study, we combine an omnidirectional camera and a Doppler sensor to develop a system that analyzes and records the stress values of each individual by estimating the position of each individual from the images captured by the camera and adjusting the orientation of the Doppler sensor in that direction. In the proposed system, by combining an omnidirectional directional camera and a directional sensor, a large number of people can be placed in the observation range with a small number of sensors.

In recent years, Japan has a steady increase in fitness enthusiasts. However, it is accompanied by a rise in injury rate due to improper fitness knowledge. To address this issue, IoT-based fitness assessment systems are gaining attention. However, many existing systems can not simultaneously assess fitness movements and vital signals, and they have challenges in monitoring people in large spaces. Therefore, in this study, we propose a remote fitness monitoring system that employs a collaborative approach with various sensors, including fisheye camera, RGB camera, infrared sensor, and millimeter wave radar. This system enables non-contact monitoring of users' physiological information during fitness, while also evaluating the quality of their movements. Through location detection via image analysis and actuation by servo motor, this system achieves wide- range and continuous monitoring of trainees using a small number of devices. Furthermore, we conduct experiments to validate the system's effectiveness in terms of scalability of the monitoring range and the effectiveness of vital signs monitoring.

As society ages, the workload of desk workers is increasing, leading many to experience overwork and heightened job-related stress. However, due to busy schedules, they often neglect their health and cannot regularly undergo health checks. Consequently, many desk workers suffer from various health issues, including the risk of overwork-related death. To address these issues, existing research has explored real-time health monitoring using medical devices and cameras, but this approach is costly and raises privacy concerns. Therefore, we propose a new system that utilizes low-cost, non-contact multi-sensors to monitor the vital signs of office workers in real-time. The system employs infrared sensors to monitor body and environmental temperature, analyzes temperature images to identify sitting posture, and uses millimeter-wave sensors to track heart and respiration rates. Additionally, a multi-modal deep learning model with an attention mechanism extracts key features from these vital signs to assess individual health status and psychological stress levels. The system delivers real-time monitoring results to desk workers' devices and provides advice when potential health issues are detected, such as suggesting rest or seeking medical attention.