http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A light-weight dynamic Ontology for Internet of Things using Machine Learning Technique
Hafizur Rahman,Md. Iftekhar Hussain 한국통신학회 2021 ICT Express Vol.7 No.3
Ensuring semantic interoperability in the future Internet of Things can be a challenging task due to their heterogeneous nature and increasing scale. Ontologies are widely used to achieve semantic interoperability among IoT applications and services. But, available ontologies are very complex, static or unable to fulfill the requirements of IoT. To address this concern, we proposed a light-weight dynamic ontology using only the most important concepts and clustering technique. It provides dynamic semantics automatically to include additional concepts using machine learning technique. Compared to the existing ontology, the proposed model reduces query response time and memory consumption to some extent.
Dew-based offline computing architecture for healthcare IoT
Kishore Medhi,Nurzaman Ahmed,Md. Iftekhar Hussain 한국통신학회 2022 ICT Express Vol.8 No.3
Due to the resource-constraint nature and lack of lightweight computing solutions for diagnostic devices in healthcare IoT, provisioning time-critical responses is still challenging. In this paper, we propose DC-Health, a Dew Computing enabled IoT healthcare solution for offline and ultra-low latency decisions. The proposed solution connects a large number of healthcare devices and provisions user-specific services even when Internet connectivity is not available. The computation module is placed at the extreme edge rather than the cloud to reduce the complexity and to improve the user-specific services. In addition to the other computing facilities provided by the cloud, fog, and edge, our solution performs with a negligible dependency on the Internet.We develop a prototype of DC-Health, which monitors the heart condition using the ECG sensors with end-mile services, flexibility in terms of user-control, and mobility feature. The experimental implementations show that the proposed architecture minimizes the network response time by at least 92% and 98%, compared to the fog and cloud-based approaches, respectively. Along with this, the proposed technique also reduces the CPU and memory usages, and response time by around 30% compared to the conventional method.