Adaptive Multi-keyword Ranked Search Over Encrypted Cloud Data

Advances in cloud computing and Internet technologies has changed the way weacquire, store, organize, and process information. More and more data owners are pushedto outsource storage and management services of their data in order to reduce large capitalspending in purchasing and maintaining facilities as well as the complexity in systemmanagement. However, cloud computing also introduces many new security challengesthat demands special attention. This is because data owners lose control over theirsensitive data after outsourcing them to a semi-trusted cloud server.In order to protect data privacy, sensitive data has to be encrypted before outsourcingto the cloud server. This also goes in hand with another challenge regarding effective datautilization since the stored data in encrypted format makes traditional data utilizationbased on plaintext keyword search impractical. Thus, enabling keyword search overencrypted cloud data became a topic of vital importance. Furthermore, data users need tobe provided with a way to rank the search results over the encrypted data so that they canonly obtain the most relevant files based on their search requests.Privacy preserving multi-keyword ranked search over encrypted cloud data (MRSE)is one of the early research publications to address the problem of keyword search overencrypted data using a secured and multi-keyword query. However, in MRSE scheme, thekeyword dictionary is static and the data owner needs to re-run the whole encryption overagain following any changes in the number of keywords. In this dissertation, a newscheme called A-MRSE is proposed which supports all the dynamics in the keyworddictionary with minimum computation and communication overhead.In our newly designed scheme, we introduce new algorithms that can be deployedeach time the data owner makes modifications that affect the size of the keyworddictionary. This implies, the data owner is free either to add more keywords or removesome of them, and is still able to make use of the present secret keys to generate new keysthat corresponds with the modified size of the keyword dictionary. This eliminates hugecomputational overhead posed by earlier schemes that demands to re-run the whole process again.After implementing the A-MRSE scheme and successfully deploy it, we conductmultiple experiments to demonstrate the effectiveness of the newly proposed scheme, andthe results illustrates that the performance of A-MRSE scheme is much better than theMRSE scheme. Typically, for key generation it consumes about32%of total time duringkey expansion and less than1%during key reduction. Remarkable gains are also observedduring index building, trapdoor generation and searching.