Attribute Based Homomorphic Encryption (ABHE) Scheme For Outsourced Big Data Computation
Loading...
Date
2017-09
Authors
Tan, Soo Fun
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
Outsourced Big Data computation raises several security and privacy issues
such as data breaches, identity theft, as well as malicious insider threats, from
espionage to sabotage. In this study, Attribute Based Homomorphic Encryption
(ABHE) scheme is proposed as a mechanism to protect confidentiality and privacy
issues of outsourced Big Data computations. The Homomorphic Encryption scheme
is a cryptosystem that is capable of computing on encrypted data, whereas the
Attribute Based Encryption scheme is a specialization of Public Key Encryption that
realises the implicit access control on encrypted data without the engagement of
conventional trusted server. Therefore, the proposed ABHE scheme is a promising
tool that is capable of performing computations on encrypted data, meanwhile
providing access control on these private data directly, thus making end-to-end data
protection for outsourced Big Data computation possible. However, the construction
of ABHE scheme always affecting the computation capabilities of Homomorphic
Encryption scheme. Moreover, existing schemes still suffering from practical
deployment issues such as slow running speed, and huge ciphertext size, as well as
their applications are only considered the single–data owner scenario. To bridge
these gaps, this study is subsequently divided into six phases, includes problem
formulation and preliminary investigation, algorithms design I, algorithms design II,
algorithms design III, security analysis and lastly, experimental development and
performance assessment. For supporting a multi-user environment of outsourced Big
Data computation, the algorithm design I focus on incorporating Attribute Based
Encryption (ABE) scheme onto Homomorphic Encryption (HE) scheme, with the
aim i.e. without affecting the capability of homomorphic computation. To support
the high volume of outsourced Big Data processing, the algorithm design II aimed to
reduce generated ciphetext size in the proposed ABHE scheme. Subsequently, the
proposed ABHE scheme is further extended into non-circuit based approach in
algorithm design III in order to improve its computation time for supporting high
velocity of outsourced Big Data computations. Theoretical and experimental results
shown that the proposed non-circuit based ABHE scheme has greatly reduced the
computation time and ciphertext size as compared to circuit based approach. For
instance, the proposed non-circuit based ABHE scheme took approximately 8.07
milliseconds and 1.64049 seconds, as compared to recent HE scheme, which took
approximately 8.14 milliseconds and 1.67247 seconds, for performing a single
additive and multiplicative homomorphism respectively. Subsequently, the
encryption of 4 megabytes’ of data that generates a ciphertext more than 280
gigabytes in recent study is further reduced to 130 gigabytes in the proposed ABHE
scheme, indeed, with the added capability of controlling access on these encrypted
data implicitly. Subsequently, the proposed ABHE scheme was proven semantically
secure under Indistinguishable under Non-Adaptive Chosen Ciphertext attack (INDCCA1)
and collusion attack with the hardness of Decision Ring-LWEd,q,𝓍 problem.
Description
Keywords
Attribute Based Homomorphic Encryption as a mechanism , to protect confidentiality outsourced Big Data computations