For most of the existing key-policy attribute-based encryption schemes

there are some problems in the decryption

such as the high cost of the client computing

long decryption time and high resource consumption. Some schemes propose outsourcing decryption to cloud servers. However

these schemes do not give the specific parallelization method of outsourcing decryption in cloud server

and there are problems of low efficiency of cloud decryption. To solve these problems

this paper presents a KP-ABE scheme for decryption outsourcing. In this scheme

most of decryption computation is outsourced to Spark platform; and according to the decryption characteristics of KP-ABE

a decryption parallelization algorithm is designed to complete the parallel decryption of leaf nodes and root nodes. The performance analysis shows that most of decryption computing is done by cloud servers and the client can decrypt the shared data by shared access tree with only once exponential operation

and the parallel design can effectively improve the cloud decryption rate.