比特幣的熱門發展讓各界意識到區塊鏈技術之可行性，區塊鏈與其共識演算法提供現今網路中分散式系統新的可能性，而目前的發展階段已從比特幣單純的交易電子貨幣中來到了分散式應用的階段。分散式應用為使用區塊鏈技術作為後端伺服器的應用程式，其相較於以往集中式伺服器有著公開透明、不可逆及可追蹤來源等好處。而目前各界熱烈地討論將其應用於資訊敏感與追蹤來源方面，利用其特性提供相關服務，例如：醫療資料共享、雲端資料隱私保障與產品來源紀錄等等。 本論文利用區塊鏈技術打造物聯網的服務管理平臺，讓物聯網裝置在區塊鏈上分享其資料訊息，並且透過智能合約管理資料之訂閱與通知。實作溫度計與裝置的例子，溫度計在區塊鏈上發布其資料智能合約讓裝置訂閱，資料庫收到訂閱後則依照裝置在其合約上設定的條件進行通知。另一方面，本論文從網路連線的角度，提出節點連線管理的演算法，假設已知網路中各節點的hop數與連結度，利用LMH(Limit Max Hop)演算法產生出的網路拓樸可以降低網路連線數量與各節點連結度，以此減少節點的負擔，利用LML(Limit Max Link)演算法則可產生出網路最大hop數與平均hop數較小之拓樸，以此減少區塊傳遞需要的時間。 最後透過模擬的方式比較兩種演算法與隨機連線方式的差異，從中發現LMH演算法比起隨機方式有著較低的連線數與連結度，但於平均hop數則表現較差，LML演算法則於網路最大hop數與各節點平均hop數皆較隨機方式表現得好。;The rapid growth of Bitcoin has shown the possibility of such Blockchain system. Along the growth of Bitcoin, it’s Blockchain technology and consensus algorithms took the network and distributed system to a new era. At this point, Blockchain technology have expanded to the development of distributed application(DApp). Comparing with the traditional architecture of central server, distributed Blockchain server has the advantage of decentralization, immutable, transaction transparency and trackable. Currently Blockchain system has been applied to medical information sharing, protecting cloud data and tracking the source of product due to the characteristic of it. In this paper, we implement a platform of internet of things service management system through applying Blockchain. The platform allows IoT devices to share their data information over Blockchain and manage the information using smart contractions. In our example, thermometers deploy contracts on the Blockchain letting others to subscribe. After subscriptions, a database will keep track of the data and notice the subscriber. On the other side, we suggest a peering management scheme that improves the topology of Blockchain systems. Assumed given the hop count and degree of every peer in the network. We proposed two algorithms, LMH algorithm: limits network’s maximum hop count and generates topology using fewer connections. LML algorithm: limits the number of connections and generates topology with fewer hop counts. At last, our simulations show that comparing with random scheme, LMH has a better performance on network maximum degree and total link, but relatively poor on average hop count. On the other hand, LML has a better performance on both network maximum hop count and average hop count compared with random scheme.
