論文

The increasing amount of data running in cloud-computing environments has started inflating networks. To solve the problems caused by network inflation (e.g., latency and privacy), new types of computing environments with multiple layers have been proposed. However, service placement inside these multilayer computing environments has not been proposed. Nodes inside multilayer computing environments have different preferences, and the services deployed also have restrictions on deployment. Therefore, services must be placed carefully inside the computing environment. To place these services, we introduce a service classification method according to their properties and restrictions. However, when accommodating dynamic placement, rapid classification is needed to avoid serious damage caused by restriction changes. Therefore, we propose a classifying method using k-Nearest Neighbor Classification (k-NN). In addition, to accelerate the process, we use a dimension reduction method called Self-Organizing Maps (SOM) to preprocess the data. The proposed classification method is expected to be used as the primary step in service placement. The method will supply service placers with the identification of which layer services should be deployed.

Smart community is the extended concept of the smart grid. Many services, including, but not limited to infrastructure services are deployed in the smart community. Some of the services in the smart community are time critical and require low latency between requests and responses. However, commonlyused cloud computing architecture is not designed with an emphasis on the latency. Therefore, the network architecture for low latency required services are necessary. This paper proposes a serviceproviding network for smart communities. In the proposed network, service-oriented router is utilized to provide the service in intermediate nodes between a cloud server and an end node. By responding to therequest at the intermediate node instead of the cloud where is close to the end devices, the latency is reduced. To demonstrate the feasibility of the proposed network, a simulation of the proposed network with two time-critical services were carried out in the network simulator ns-3. The ancillary service and intersection accident avoidance service are chosen to be considered in the simulation since they are attention-grabbing time-critical services for the smart community. The reduction of both latency and variation of the latency in both services was observed in the simulation. For the ancillary service, the proposed method has reduced the average latency to 10.85 ms that is low enough to provide stable service. The result shows the feasibility of application task allocation between intermediate SoR nodes based on a latency requirement in the smart community.

Depending on the location of the air conditioner and the shape of a room, air-conditioning control may be inefficient resulting in temperature imbalance. When attempting to solve this problem, it is vital to understand the spatial structure of a room (including its size and shape) and the location of air conditioners and then automatically control the airflow and direction according to the structure. However, such a method for recognizing spatial structures has not yet been established. In this paper, we propose a spatial recognition method using stereo infrared array sensors (SIRA sensors) installed in an air conditioner. Our system detects objects in the obtained thermal images and estimates their distances using triangulation. In addition, the room's size and shape are estimated based on the assumption that the room size lies within the detection range. The distances to the front and left/right walls were estimated in one-meter-wide classes. The estimation accuracy was compared using two types of IRA sensors: thermopile array sensors and thermal diode infrared sensors. Regarding the distance estimation of persons from the captured stereo thermal images, the average error rate was 12.5% for both types. The distance to each wall was estimated within a 1 m error range for the thermal diode infrared sensor. Moreover, applications of the proposed spatial recognition to air-conditioning control were demonstrated. Specifically, we propose a method to control the airflow direction and volume by considering the room's geometry. An L-shaped room was modeled and simulated. From the results, the spatial recognition reduced the unevenness in temperature by adjusting the airflow based on the room shape. These results indicate that the proposed method can be practically used for spatial recognition to efficiently improve user comfort by controlling air-conditioning based on the spatial structure and eliminating uneven temperature.

Achieving a uniform comfort within an indoor housing environment is important for health and productivity while saving the energy consumption of a Heating, Ventilation, and Air Conditioners (HVAC) device. The optimal control of an HVAC system is a well-studied area. While many works explore the optimal temperature set-point, a few works consider effective airflow direction control. This work proposes an airflow direction control method that aims uniform comfort of the indoor environment using a deep reinforcement learning (DRL) approach. We implemented our proposed DRL framework using computational fluid dynamics (CFD) simulation software. Our proposed method was evaluated for comfort and energy consumption. The experimental results show the improvements for our proposed method in comfort by 21.3 % while reducing energy consumption by 34.5 % for the average than the baseline method.

The deployment of Information and Communication Technology in existing infrastructure has increased interest in smart communities. Smart communities provide various services such as ancillary services, road traffic information management, and health care management. Each service has distinctive characteristics such as processing delay, the need for anonymity, computational cost, and the amount of data. Therefore, to optimize the system, locations to provide these services should be organized. In addition, secondary use of personal data is assumed in smart communities. Thus, measures to provide privacy protection should be implemented. In this paper, a new information platform is proposed, which allows management of a user's privacy policy, management of data usage, smooth data sharing, and adequate service location organization by introducing the concept of 'Relationship.' Response time associated with organization was evaluated on the implemented platform to ensure the feasibility of the proposed method, and it was about 15 minutes when the number of Relationships is 12,800.

In the last decade, the significant development of smart building technologies has led to the formation of various energy sensing and monitoring applications. Energy monitoring of appliances relies on techniques such as Intrusive Load Monitoring (ILM) and Non-Intrusive Load Monitoring (NILM). I L M is referred to a technique that a sensor installed for each load. In N I L M method, disaggregation of measured energy of all appliances at utility service entry is the main goal to provide a simple and cost-effective method of monitoring the appliances like sequence time domain reflectometry (STDR). This manuscript provides an overview of developments in energy consumption sensing and monitoring in three key areas of Internet of things (IoT), WSN and STDR for the advancement of smart building technology. This paper also provides some research directions for smart home of the future.

The development of the Internet of Things (IoT) has created an environment in which numerous sensors and actuators are connected to the Internet. Machines and management systems in factories use data from such sensors and actuators to improve their work efficiency, and are essential parts of today's smart factories. The vision of a smart factory is based on the concept of Industry 4.0 (I4.0), a subset of the fourth industrial revolution, in which smart factories support the operator and maintenance processes of the factory from an I4.0 perspective. The analysis of big data gathered by IoT devices in factories, particularly for the use of anomaly detection, can aid in achieving product quality stabilization. For example, if a large refrigerator in a warehouse breaks down, the quality of stock food will deteriorate, and food loss may become significant. In the case of anomaly detection, machine status monitoring and accident prediction are required to reduce the operation and maintenance costs. Furthermore, the introduction cost of such systems can be reduced by generalizing them (the systems). However, the data types as well as the sensor and actuator types, differ between factories. Therefore, nonparametric statistical methods are required for anomaly detection. By contrast, factor analysis requires a costless method, one that does not require an overhaul of machinery. Consequently, it is necessary to adopt a machine learning-based method using sampled data. In this study, we proposed a method of anomaly detection and factor analysis for cooling systems in smart factories using appropriate methodologies for detection and analysis. The proposed method consists of two phases: anomaly detection and factor analysis. In the anomaly detection stage, Gaussian kernel density estimation was used to calculate the occurrence distribution. Two types of anomaly scores, cumulative density value and KL divergence, were defined. The probability distribution was estimated with a constant window frame to reflect a tendency to increase. In the factor analysis stage, target values were predicted using LightGBM. The factor of abnormalities was detected by comparing the results of two predictions: one using all the features, and the other using the data, which excluded a factor to detect the contribution of the factor.

In Internet of Things (IoT) networks, devices use various application protocols, such as MQ telemetry transport (MQTT), constrained application protocol (CoAP), and extensible messaging and presence protocol (XMPP). However, because of the various protocols used, devices adopting different protocols cannot communicate mutually; thus, service interoperability issues arise. To enhance service interoperability, protocol conversion is required. Because different protocols are used in IoT networks, they are required to correspond to various protocols and have high extensibility for protocol conversion. In this paper, a protocol conversion method that satisfies these requirements is proposed. The proposed method converts packets into a middle format before converting them into target protocols. Conversion rules are described for each protocol; thus, the proposed method has high extensibility. We confirmed that clients that use the proposed conversion method could successfully connect to other devices using different protocols and communicate mutually through servers. The protocols included MQTT, CoAP, XMPP, and SMTP. Furthermore, the throughput degradation caused by the conversion process is small.

自由に桁数を選択可能なハッシュ関数を利用し，そのORを取ることでグルーピングを行うとともに，アドホックネットワークにおけるルーティングテーブルをHELLOメッセージ中の有効ノード情報を削減し，軽量な分散ルーティングテーブル構築手法を提案し評価している．