Paper 2019/1437

Reverse Outsourcing: Reduce the Cloud's Workload in Outsourced Attribute-Based Encryption Scheme

Fei Meng and Mingqiang Wang

Abstract

Attribute-based encryption (ABE) is a cryptographic technique known for ensuring fine-grained access control on encrypted data. One of the main drawbacks of ABE is the time required to decrypt the ciphertext is considerably expensive, since it grows with the complexity of access policy. Green et al. [USENIX, 2011] provided the outsourced ABE scheme, in which most computational overhead of ciphertext decryption is outsourced from end user to the cloud. However, their method inevitably increases the computational burden of the cloud. While millions of users are enjoying cloud computing services simultaneously, it may cause huge congestion and latency. In this paper, we propose a heuristic primitive called reverse outsourcing to reduce the cloud's workload. Specifically, the cloud is allowed to transform the ciphertext decryption outsourced by the end user into several computing tasks and dispatches them to idle users, who have some smart devices connected to the internet but not in use. These devices can provide computing resources for the cloud, just like the cloud hires many employees to complete the computing work. Besides, the computing results returned by the idle users should be verified by the cloud. We propose a reverse outsourced CP-ABE scheme in the rational idle user model, where idle users will be rewarded by the cloud after returning the correct computing results and they prefer to get rewards instead of saving resources. According to the Nash equilibrium, we prove that the best strategy for idle users is to follow our protocol honestly, because the probability of deceiving the cloud with incorrect computing results is negligible. Therefore, in our scheme, most computational overhead of ciphertext decryption is shifted from the cloud to idle users, leaving a constant number of operations for the cloud.