一個在行動感測網路中的品質導向資料蒐集機制

Abstract

行動無線感測網路已經廣泛地使用在許多方面，如動物追蹤和環境監測。有別於傳統靜態的無線感測網路，由於行動無線感測網路受到它本身的行動性和無線存取點的佈署範圍限制，它會有間斷性連結的問題。另一方面，一般在無線感測網路中，對於資料蒐集的架構是採取集中式架構，再加上大量的資料以及有限的無線頻寬，所以在這樣的環境下要蒐集所有的資料是一件很困難的任務。因此，降低資料上傳量也是一個很重要的議題。 在本篇論文中，我們提出了一個在行動感測網路中的品質導向資料蒐集機制，簡稱MobiQC。我們針對如何從行動感測裝置蒐集資料的問題來做設計，使得蒐集到的資料可以符合給定的服務品質(QoS)規範，並且降低傳輸到伺服器的資料量。在MobiQC裡，我們使用了線性回歸模組以及核回歸模組的技術；同時，我們也使用了在網路中的集成法。透過以上的技術，我們可以降低資料傳輸量並保證資料的品質，且我們的實驗顯示MobiQC是有效用的。

Mobile sensor networks have been widely used in many applications, such as animal tracking and environment monitoring. Different with the static one, it suffers from the problem of intermittent connection due to its mobility and the coverage of wireless network. On the other hand, since the data collection performed in a mobile sensor network usually follows a centralized infrastructure, collecting all data from mobile sensor nodes is a hard task due to the large quantities of data and limited wireless bandwidth. Therefore, decreasing the amount of data uploaded is also an important issue. Moreover, in many applications, it is sufficient to support approximate data collection by tolerant errors. In this paper, we proposed a propose a QoS-aware data collection framework in mobile sensor networks (MobiQC). We focus on the problem of data collection from mobile sensors such that the quality of data satisfies given QoS specifications and the amount of data transmitted to the server is reduced. In the MobiQC, we use both linear and kernel regression model to reduce total transmitted data and in-network aggregation is employed to further reduce the amount of data while guaranteeing the quality of data. Our experiments verify the effectiveness and the efficiency of our proposed MobiQC.