| 西田 知弘 | 1651083: M, 2回目発表 | 大規模システム管理 笠原 正治, 岡田 実, 笹部 昌弘, 川原 純, 張 元玉 |
|
title: Collaborative Spectrum Sensing Mechanism Based on User Incentives in Cognitive Radio Networks
abstract: In cognitive radio networks, a detection performance of secondary users (SUs) can be improved by cooperatively detecting the signal of primary user (PU). We have proposed a group formation scheme among SUs for CSS in which each SU aims to maximize its own communication opportunity while satisfying the constraint on miss detection probability. In the previous scheme, we have assumed that communication opportunity is equally allocated to each SU. In general, however, the detection performance is different among SUs, and there is an aspect that the incentive to participate in CSS is weak for the SU which has a high contribution to CSS, i.e., high detection performance. In this presentation, we propose a CSS mechanism based on user incentives in cognitive radio networks. Through simulation experiments, we show that each SU obtains the communication opportunities corresponding to its own contribution and the ratio of SUs that can access the channel of PU increases. language of the presentation: Japanese | ||
| 飯塚 翔 | 1651005: M, 2回目発表 | 大規模システム管理 笠原 正治, 池田 和司, 笹部 昌弘, 川原 純, 張 元玉 |
|
title: Coalition formation for cooperative spectrum sensing by estimating detection and false-alarm rate
abstract: Cooperative spectrum sensing is a scheme for reducing sensing errors in cognitive radio. In coalition formations for cooperative spectrum sensing, it is necessary to consider two criteria, detection rate and false-alarm one. In situations where these criteria are unknown, coalition formations need to be done while taking into account the exploration-exploitation trade-off. In this presentation, we propose a technique to estimate the posterior distribution of detection rate and false-alarm one in order to make decisions by bandit algorithm. We show that the proposed method reduces sensing errors through the experiment of coalition formations using estimated posterior distribution. language of the presentation: Japanese | ||
| 河瀬 良亮 | 1651036: M, 2回目発表 | 大規模システム管理 笠原 正治, 池田 和司, 笹部 昌弘, 川原 純, , 丁 明 |
|
title: Application of M/GB/1 priority queue to analyzing transaction-confirmation time in Bitcoin
abstract: In Bitcoin system, a transaction is given a priority value according to its attributes such as the remittance amount and fee, and transactions with high priorities are likely to be confirmed faster than those with low priorities. In this study, we analyze the transaction-confirmation process with a queueing system with batch service, M/GB/1. We consider the joint distribution of numbers of transactions in system and the elapsed service time, deriving the mean transaction-confirmation time. Using the result, we derive the recursice formulae of mean transaction-confirmation times of an M/GB/1 queue with priority service discipline. Numerical examples show the effect of the maximum block size on the mean transaction-confirmation time. We also discuss how the increase in micropayments, which are likely to be given low priorities, affects the transaction-confirmation time. language of the presentation: Japanese | ||
| 森 剛基 | 1651111: M, 2回目発表 | サイバネティクス・リアリティ工学 清川 清, 加藤 博一, 佐藤 智和, 河合 紀彦 |
|
title: Depth Map Inpainting Based on Similar Shape Search Using Normal Vectors
abstract: Depth maps, which are related to length from a camera to objects, are often used for 3D Shape Reconstruction. However, it is difficult to generate ideal 3D models if unnecessary objects exist in target scenes. For this problem, removing unnecessary objects in depth maps and inpainting the depth maps will create 3D models without such objects. Previous methods that inpaint a depth map using similar depth patches in the depth map have been proposed. However, such methods cannot effectively use depth patterns that are not similar in the depth map, but are actually similar in 3D space, which often causes the failure in inpainting. The proposed approach makes good use of these patches by relativizing depth-values to improve the quality of depth map inpainting. I report the quality improvement of depth map inpainting by comparing the method that uses relative depth-values with conventional methods. In addition, I report the proposed method that compensates for changes in depth patterns, which are caused by directions of surfaces, using normal vectors for further improvement of depth map inpainting. language of the presentation: Japanese 発表題目: 法線を用いた類似形状の探索に基づくデプスマップの修復 発表概要: カメラと撮影対象までの距離に関係するデプスマップは，三次元形状復元に利用されるが，シーン中の不要物体によりユーザの望む三次元モデルが生成できない場合がある. このため，デプスマップから不要な物体を除去し，その背景デプスを修復することにより，不要物体のない三次元モデルを生成できる. 従来，デプスマップを修復する際，同一デプスマップ中の類似デプスパッチを用いて修復する手法が提案されているが，三次元空間中では同じ形状であるにも関わらず，撮影位置に依存するデプスマップ上ではデプス値が必ずしも類似しないようなパッチが有効に利用できないため，違和感のない修復が難しいことがある. 提案手法ではそのようなパッチを有効に利用するために，物体のデプス値を相対的な値にすることで修復品質の向上を図る. 本発表では従来のパッチベースの手法と比較した場合の品質の向上を報告する. また，カメラに対する物体の傾きの違いによるデプスパッチのパターン変化を補償するため，デプスマップから得られる法線を利用した修復によりさらなる修復品質の向上を目指しており，本発表ではこれについても報告する. | ||
| 塩竃 実 | 1651058: M, 2回目発表 | ロボティクス 小笠原 司 ☆, 加藤 博一, 多田 充徳(客員）, 村井 昭彦(客員）, 高松 淳 |
|
title: Optimization of Running Shoes for Individuals Using Wearable Devices
abstract: In this research, we develop a motion feature extraction technique using wearable sensors to optimize running shoes that prevent injuries. Many companies develop running shoes with designing its shoe sole stiffness to prevent the increase of population that are suffered from running injuries. These shoes guarantee their beneficial effect when he/she choose ones that are appropriate for his/her motion characteristics, though it is technically difficult to measure one's daily running characteristics precisely and easily. We develop a robust technique to extract the parameters that highly relate to running injuries, and develop a system that chooses running shoes that are appropriate for individual motion characteristics. language of the presentation: Japanese | ||
| 西村 祥吾 | 1661013: D, 中間発表 | 環境知能学 萩田 紀博, 加藤 博一, 神原 誠之 |
|
title: Interaction Designs for Service Robots Increasing Daily-Use Motivation of Users
abstract: It has increased that the introduction of service robots communicating instead of people in our environment, such as street corner stores, home desktops and so on. On the other hand, from the viewpoint of whether the user continues to use these services, the tiredness of the robot due to the monotonous reactions and phsychological barriers to the robot occurs, and we think that it is difficult to realize the long-term robot service. In this research, in order to solve the problem, we propose the design of robot interaction that incleases motivation for daily-use of service robot from user's point of view. 発表題目: ユーザの継続利用意欲を向上するサービスロボットのためのインタラクション設計 概要: 街角の店舗や自宅の卓上など，我々の生活中にはユーザとのコミュニケーションを通してサービスを代行するサービスロボットが増えてきた．一方でこれらのサービスを継続してユーザが利用するかという観点から見ると，ロボットに対する心理的障壁や単調な反応などによるロボットへの飽きが発生し，長期的なロボットサービスの実現には大きな壁があると感じる．本研究ではこの課題を解決するためユーザの視点に立ち，サービスロボットへの継続利用意欲を向上するロボットインタラクションの設計が必要であると考え，これを満たす設計指針を提案する． | ||