| Abstract: | 網路在大多數人的日常生活中扮演著非常重要的角色，並且也成為繼報紙、廣播、電視之後的第四大的資訊傳播媒介。現在人們在任何時間、任何地點透過行動裝置來存取及分享任何訊息。舉例來說，Google Search、Bing、Yahoo! Search等搜尋引擎提供搜尋資料的服務。然而，這些搜尋服務屬於集中式的系統，若系統管理員是不可信賴的，它們可能會竄改原有的搜尋結果或是提供有偏見的資訊。因此本研究使用分散式網路的架構，並且使用隨機化(Randomization)的方式來搜尋資料，提供搜尋者有極高的搜尋成功率，並且確保所有的節點?能夠分享資源而不需要依賴集中式架構的管理者來發布訊息，但當網路架構在大規模及移動式的動態網路中，會有以下問題: 1) 網路節點數量多且分佈範圍廣泛; 2) 網路節點隨時都會移動到不同的區域; 3)傳輸範圍也會受到訊號強弱的限制; 4) 網路節點隨時都會新的節點離開或是加入。因此，如何在大規模及移動式的動態網路中節省發佈資料、查詢資料、更新會員資訊及位置、所需的時間，並且提供有效率的搜尋及檢索資料，即是本研究欲解決的問題。本研究提出了一個具有信賴機制的資訊發佈暨檢索系統，名為TIPSS (Trustworthy Information Publication and Search System)。TIPSS系統先把整個網路分成多個小區域，並且允許每個節點都可在任何時間加入、移動、以及離開區域。當來源節點(Source node)或搜尋節點(Requesting node)需要發佈或查詢資料時，它會先把原始資料切割成多個詮釋資料(Metadata)，再使用區域敏感函數(Locality-Sensitive Hashing, LSH)來計算出所屬的區域，最後再讓所屬區域的中繼節點(Relay node)在該區域中隨機選擇2√n個節點來發佈詮釋或搜尋資料。從實驗結果中，本研究證明了TIPSS不僅能提供使用者降低搜尋成本，同時也可維持級高的搜尋成功率以降低整個網路的負荷及節省搜尋成本。
As Internet plays an important role on most people's life, it also becomes one of the biggest information media after newspapers, radio, and television. In nowadays, people can access and share any messages through their mobile devices at any time and any place. For example, Google Search, Bing, Yahoo! Search are all search engines that information retrieval services. However, currently these services are centralized search system. If the system administrator cannot act as benign, the overall search results might be tampered.Therefore, our study is based on a decentralized search engine and use randomization to search for information, to provide high retrieval rate, and to ensure that all nodes are able to share resources without relying on the centralized administrator to publish messages. But in large-scale network architecture and dynamic mobile network, we might encounter the following problems: 1) the number of massive nodes are distributed over a wide range; 2) network node can move to different region at any time; 3) transmission range of the signal strength varies; 4) nodes might leave or join at any time. Thus, how we can efficiently distribute data, search data, or update member information in large-scale and dynamic mobile network, is the aim of this study.In this study, we propose a Trustworthy Information Publication and Search System called TIPSS. In TIPSS, we divide the whole network into numerous small areas, and allows every node to join, leave, or move to other regions. When Source node or requesting node need to publish or retrieve information, we first construct its metadata, and then use the Locality-Sensitive Hashing (LSH) to calculate its belonging region, and finally inform relay nodes to randomly selected 2√n nodes in the calculated region to publish metadata or requests. From our experiment, we have demonstrated that our TIPSS can provide maintain minimum message costs while maintaining a high retrieval rate, and ultimately can reduce the overall network load traffic costs. |
