Occupancy detection is crucial for efficiently managing energy in building. However, purchasing occupancy sensors incurs additional cost, and their accuracy is typically low. This paper proposes AI-based occupancy prediction model using Channel State Information (CSI) from WiFi APs. Two models have been implemented based on logistic regression and deep neural network. By synergistically combining their respective advantages, its accuracy can be improved. This enables high-accuracy occupancy prediction using existing WiFi APs without deploying new sensors. Experimental results show that proposed model achieves a high accuracy exceeding 94%, which is an acceptable rate for real-world applications.

For the real-time object detection and tracking, CNN-based YOLO(You Only Look Once) is widely used. However, employing YOLO in a multi-streaming environment poses a challenge due to the increasing demand for computing resources as the number of streams increases. This paper proposes a system that minimizes FPS degradation without necessitating additional computing resources when using YOLO in a multi-streaming environment. Experimental results show that the proposed system can handle multi-streaming with 8 CCTV cameras without encountering FPS degradation, while maintaining delays at an acceptable level.

Recently, the amount of power generation from new and renewable energy sources is increasing compared to electricity demand, which is harming the stability of the system. To solve these problems, a microgrid-type power source configuration is being selected instead of expanding power infrastructure, and this paper proposes a microgrid operation strategy through optimization-based ESS scheduling. To verify the effectiveness of the proposed method, the algorithm was applied and analyzed by simulating a virtual microgrid, and energy savings of 7.1% were verified.

Demand-side energy management is becoming increasingly important owing to concerns related to global warming and energy shortages. In particular, as the development of Internet of Things (IoT) enables the precise control of home appliances, the demand for home energy management systems (HEMSs) is expected to increase. This article proposes an artificial intelligence-based HEMS (AI-HEMS) that provides both energy efficiency and resident satisfaction. To this end, we implemented three prediction mechanisms: 1) derivation of comfort temperature; 2) device-free sleep prediction; and 3) occupancy-probability-based outing prediction. Based on these mechanisms, we present four intelligent heater control strategies: 1) outing; 2) occupancy; 3) comfort; and 4) sleep-based control. For evaluation, an experimental testbed was constructed and measurements were taken. To increase the reliability of the evaluation by excluding natural energy fluctuation factors, a treatment group (100 households) and a control group (2281 households) were recruited to measure the energy savings of the treatment group compared to the control group. In a 48-day evaluation, AI-HEMS was found to provide an energy-saving rate of approximately 14% and resident satisfaction of approximately 91%. These results imply that the proposed system can save energy while maintaining a high level of satisfaction.

Since the concept of Cognitive Radio (CR) was first introduced, it has been considered as a key technology of future wireless devices to better utilize radio spectrum. A number of CR-MAC protocols have been studied for Cognitive Radio Networks (CRNs), but most state-of-the-art frequencyhopping based protocols focus mainly on channel mobility and spectrum resource allocation. They lack integration of other essential features such as time synchronization and cooperative spectrum sensing, which are practically crucial, into a full-blown CR-MAC protocol. In this paper, we propose MSHCS-MAC: a mac protocol for Multi-hop CRNs based on Slow Hopping and Cooperative Sensing approach which is the cutting edge frequency hopping scheme. The contributions of MSHCS-MAC are threefold: (1) Support of multi-hop communication without dedicated control channel and multiple transceivers, (2) Integration of essential CR-MAC features such as bootstrapping, multi-channel operation, cooperative spectrum sensing and time synchronization, (3) Practical implementation and evaluation on commercial devices. The evaluation results show that MSHCS-MAC provides reasonable performance in the experimental testbed with supporting multihop communication and essential CR-MAC features.

In CR networks, to share the conditions whether the channel is occupied or not, many existing researches generally adopt common control channel(CCC) and multiple radio transceivers due to easy design and implementation for multi-hop communications. However, they are resource wasteful because one channel should be always occupied for CCC and the radio transceiver is additionally required. In this paper, we propose a new time synchronization method for multi-hop communication by extending the SHCS-MAC protocol. We implemented the proposed time synchronization method on IEEE 802.15.4 Zigbee PHY based USRP SDR devices using GNU Radio and conduct the experiment to confirm it works properly.

In the EPCglobal standards for RFID architecture frameworks and interfaces, the Electronic Product Code Information System (EPCIS) acts as a standard repository storing event and master data that are well suited to Supply Chain Management (SCM) applications. Oliot-EPCIS broadens its scope to a wider range of IoT applications in a scalable and flexible way to store a large amount of heterogeneous data from a variety of sources. However, this expansion poses data security challenge for IoT applications including patients' ownership of events generated in mobile healthcare services. Thus, in this paper we propose Secure-EPCIS to deal with security issues of EPCIS for IoT applications. We have analyzed the requirements for Secure-EPCIS based on real-world scenarios and designed access control model accordingly. Moreover, we have conducted extensive performance comparisons between EPCIS and Secure-EPCIS in terms of response time and throughput, and provide the solution for performance degradation problem in Secure-EPCIS.

In the Internet of Things, tremendous data is generated by huge number of things and stored in globally distributed data repositories. In this respect, many IoT applications would need to find and get the data indirectly from the data repositories or directly from things. GS1, global leading standard organization, firstly presents the concept of Discovery Service (DS) to find the desired EPC Information Services (EPCIS) like data repositories for things. To realize GS1 DS, many works are proposed with considerations of performance and security issues. However, most works are only applied to the Inter-DS finding one of distributed Intra-DSes; thus deterioration in the Intra-DS actually finding desired data repositories may adversely affect entire DS. In this paper, we propose Oliot-Discovery Service (DS) which deals with performance as well as security issues by especially focusing on Intra-DS aspect. For this, we have designed and implemented the access control pursuing the fine-grained access control model and the two-layered storage architecture comprised of cache and main databases. Moreover, it supports not only indirect access to things' data through the data repositories but also direct access to things which would be also required for many IoT applications. In order to show feasibility of Oliot-DS, we have constructed experimental test-bed and evaluated the performance. The experiment results show that Oliot-DS provides durability against many user transactions, and the reasonable write and read performances by preventing unauthorized access.

Tree based hierarchical routing is widely used in 6LoWPAN due to advantages of low control overhead and low memory resources. However, it generates a triangular detour path considering point-to-point (P2P) traffic. Reactive routing provides a reasonable P2P path by conducting flooding-based route discovery, but the flooding increases packet interference due to huge control overhead. Multipath routing can improve the reliability by preparing for the alternative path. Nevertheless, it needs an extra table for storing the alternative path and it is not appropriate to diverse traffic patterns. In this paper, we propose a traffic-aware multipath routing (TAMR). By managing an overlaid topology comprised of two different topologies without additional overhead, TAMR can dynamically choose a proper routing protocol according to traffic patterns and it can improve network reliability. Besides, TAMR is suitable for 6LoWPAN, because it does not need to store global routing state and generates low control overhead. In order to evaluate the performance, we have conducted the simulation by comparing with representative routing protocols in 6LoWPAN. The results show the overall performance of TAMR surpasses others in terms of hop count, packet delivery ratio, number of control packets, and memory usage.

One of the important challenges in the Internet of Things (IoT) is how to acquire the physical context of things. IP-based wireless sensor networks (IP-WSNs) could be a promising approach to collecting the physical context of things and to integrating WSNs to the Internet. However, realizing IP-WSNs in IoT exposes two major challenges. One is how to embed the Internet Protocol (IP) in resource-constrained sensor nodes. The other is how to achieve real-world deployment of WSNs and its integration with the Internet on the fly and on the cheap. In this paper, we present the SNAIL (Sensor Networks for All-IP World) project and introduce a new type of IP-WSN gateway, which supports dual wireless access points for WiFi and IP-WSN, enabling deployment of SNAIL nodes in an easy and rapid manner as for the solution. To show the proof-of-concept, we implement a new SNAIL platform from tiny sensor nodes to a gateway.