2014 Fiscal Year Research-status Report
ネットワークとユーザの状況に対応する自律的なIoT・M2Mプラットフォーム構築
| Project/Area Number |
25730061
|
|---|
| Research Institution | The University of Aizu |
|---|
Principal Investigator |
王 軍波 会津大学, 産学イノベーション センター, 准教授 (40646882)
|
|---|
| Project Period (FY) |
2013-04-01 – 2016-03-31
|
| Keywords | Disaster / Abnormal detection / Situation recognition / Network modeling / Network optimization / Deployment / Mesh Network |
|---|
| Outline of Annual Research Achievements |
In this year, we have focused on
(1) Optimized task allocation methods among local cloud and remote cloud: We investigate how to efficiently offload codes from wearable devices to computation resources in the mobile and cloud resources, to guarantee a certain level of user experience, called a just-in-time objective for code offloading, i.e., maximizing the number of tasks that must be executed within a given delay from their direct previous ones.
(2) Network situation understanding and optimization in a disaster scenario: We have proposed optimized method to deploy mobile mesh routers and control effeiciently data flows in a disaster scenario, by considering situation and communication demands of users, through analysis of communication network. The optimization objectives are to increase QoS and minimize the congested communication demand in the whole communication network when forwarding data to gateways.
and (3) abnormal situations detection in a smart home: , we propose a situation-aware abnormality detection system based on support vector data description (SVDD) for elderly people. First, a sensing system is proposed to detect the details of a person's situation. Then, we propose various features that are analyzed and designed for each situation. Then, a method to detect abnormal activities in a situation based on SVDD is proposed.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Basically, the projects goes smoothly. Till now,
we have studied how to organize situations happened around a user and proposed a mechanism called situator to implement the proposed functions in 2013, related with challenge (1) "how to manage IoT resources and realize realtime communication" in the original proposal. Also we have studied network modeling methods and optimization methods considering disaster situations/scenarios in 2014.
Related with challenge (2) "task allocation among local Cloud and remote Cloud" in the original proposal, we have studied how to efficiently offload codes from wearable devices to computation resources in the mobile and cloud resources, to guarantee a certain level of user experience, called a just-in-time objective for code offloading in 2014.
Related with challenge (3) "situation recognition" in the original proposal, we have propose a situation-aware abnormality detection system based on support vector data description (SVDD) for elderly people. We also implemented a real system to evaluate the system.
|
| Strategy for Future Research Activity |
In 2015, we are going to further study how to know situations from big data analysis and then support IoT/M2M system. Especially, we are going to study the following subproblems as follows,
(1) How to know situations from spatial big data from social medias, e.g., twitters. Here we will study theoretically and proposed lots of original solutions. Furthermore, to let the proposed methods be able to easily implemented in a real system in the future, we are going to consider a specific scenario, i.e., disaster.
(2) How to efficiently construct a communication network in a disaster scenario. In 2013 and 2014, we have studied mobile ad hoc network constructed based on smart phones and mobile mesh network due to movable access points in a disaster scenario. In 2015, we will further study the research problem, .e.g, outage probability analysis and optimization of a communication network based on small cell technique in disaster scenario.
(3) Further we will consider how to join the big data analysis, situation recognition and network planing together to propose original solutions.
|
