在物聯網與霧網路環境中的分散式探索網路資源

Abstract

隨著物聯網時代的到來，我們見證智慧裝置與服務成為商務解決方案中的關鍵部分，然而，為了使物聯網變得真正普及，一個廣域網路規模的分散式物聯網服務索引是必要的。此方法將與現行封閉物聯網資源於密封環境中且限制遍佈資源之存取與協調的專屬雲端垂直解決方案比較。 本論文提升了相連物聯網資源的視野，每項資源或資源集合都表示為允許分散資源發現、選擇、與存取的點對點覆蓋網路中的一個節點（代理人），我們分析物聯網對點對點覆蓋的特定環境需求，並與現行的解決方案做比較。而後提出基於三維德洛涅三角剖分的新型覆蓋設計，其支援空間查詢、節點移動性、發布/訂閱與要求/應答的發現模式，再對不同架構情況下之覆蓋的訊息負擔進行效能分析。最後，描述一個基於第五代移動通訊技術霧運算生態系統的部署情境，並強調本論文提出的解決方案之適當性。

At the dawn of the IoT era we are witnessing smart devices and services becoming a pivotal part of business solutions. However, in order for IoT to become truly pervasive, a WAN scale distributed IoT service indexing is required. It will contrast with current IoT discovery mechanisms realized through the proprietary, cloud based vertical solutions which enclose IoT resources into hermetic environments and limit ubiquitous resource access and orchestration. This dissertation promotes the vision of connected IoT resources. Every resource or federation of resources is represented as a node (agent) in a P2P overlay which allows distributed resource discovery, selection and access. IoT environment-specific requirements towards construction of a P2P overlay are analyzed and compared against existing solutions. Later a proposition of novel overlay design is presented that is based on a 3-dimensional Delaunay triangulation which supports spatial query, node mobility as well as publish/subscribe and request/reply discovery modes. A performance analysis in terms of messaging overhead with different architectures cases of the overlay is performed. Finally, a deployment scenario based on Fog computing ecosystem for 5G communication technology is described, with emphasis on aptitude of the proposed solution.