雲端外包儲存(cloud outsourced storage)是一項重要的雲端運算服務,雲端用戶將檔案儲存至雲端伺服器以減少檔案管控與維運所需的成本,而且為了保護檔案的私密性,由雲端用戶自行將檔案進行加密再儲存至雲端伺服器。本論文以kNN的技術為基礎對已加密的檔案可以進行多關鍵字搜尋,可以判斷出每關鍵字與檔案相關程度之高低,使雲端伺服端可以漸進地依序自相關程度較高的檔案開始回傳給雲端用戶,以減少檔案傳輸的通訊成本,並且為了更進一步增加系統的容錯能力與可用性,利用建立相似關鍵字集合的方法達到相似搜尋的功能,甚至雲端用戶輸入錯誤關鍵字,但在錯誤未超過門檻值的前提下,仍舊可以找出相關的檔案。此外本研究也考量在機制運行的過程中也能夠滿足必要的安全需求。 Cloud outsource storage is one of important services in cloud computing. Cloud user uploads data to cloud server in order to reduce data management cost and maintenances cost of hardware and software. To protect data confidentiality, cloud user encrypts files before outsourcing. However, it is difficult for the cloud server to retrieve the corrected encrypted file based on the query of cloud user. This paper proposes the protocol which is based on kNN technology. It can perform multi-keyword search in encrypted cloud data, and it determine relevance of encrypted files and keywords. The cloud server cannot learn the keywords including in the query of cloud user. In order to reduce communication cost of file transport, cloud server can return higher relevance of encrypted files to the cloud user. Moreover, if cloud user type wrong keywords and wrong keyword does not exceed threshold, he still retrieves files from cloud server. In addition, this study also considered that security requirements have to be satisfied.