Side-channel attack resilient implementation of homomorphic encryption using elliptic curve cryptography for secure cloud computing

In recent times, the amount of data exchanged between the cloud storage and the users has proliferated. The security of that data is also critical. To secure that data and to enhance its integrity, it should be encrypted before being uploaded into the cloud. In this work, a side-channel attack-secured additive homomorphic encryption is implemented using elliptic curve cryptography on an FPGA platform. An elliptic curve scalar multiplication, which is the critical component of elliptic curve cryptography, is designed in the general prime field using standard projective coordinate representation and implemented for 192, 224, and 256 bits as per the left-to-right double-and-add algorithm using radix-4 Booth-encoded modular multipliers in both FPGA devices and the ASIC platform. A minimum of 8242 slices is required to implement the proposed 256-bit elliptic curve scalar multiplication in the Virtex-6 FPGA device. The area of the proposed 192, 224, and 256-bit elliptic curve scalar multiplication is estimated as 149.225K, 208.178K, and 266.981 KGE in the ASIC using Cadence gpdk-45 nm technology libraries. A correlation power analysis attack is mounted on the FPGA implementation of the proposed elliptic curve scalar multiplication with an 8-bit data size to determine the value of scalar ‘n’. The attack is successful with a minimum of 2301 traces, and a high correlation coefficient value is obtained. Scalar randomization is proposed and integrated with the design as a countermeasure part to thwart the correlation power analysis attack, which is successful, and hence the left-to-right double-and-add algorithm used to determine elliptic curve scalar multiplication is made secure against side-channel attacks. This secured hardware implementation of elliptic curve cryptography is utilized to encrypt the data uploaded to the cloud, where additive homomorphic encryption is employed to process the data. Hence, additive homomorphic encryption becomes side-channel attack resilient, and cloud computations are secured. • The hardware implementation of 192, 224, and 256-bit elliptic curve scalar multiplication over the general prime field in standard projective coordinate representation is optimized to achieve minimum area compared to the existing works in the literature. • An 8-bit ECSM of the proposed method is implemented on the SAKURA-G FPGA board, and the power traces are collected during the corresponding execution. A correlation power analysis attack is mounted on the SAKURA-G FPGA board, and the scalar value is extracted successfully, utilizing the hypothetical power values of 8-bit elliptic curve scalar multiplication and the captured power traces to demonstrate the vulnerability present in the left-to-right double-and-add algorithm used to determine elliptic curve scalar multiplication. • Scalar randomization using the number of points in the subgroup formed by the chosen base point of the curve is implemented as a novel countermeasure to overcome the correlation power analysis attack and also validated. • Hence, the left-to-right double-and-add algorithm used to determine elliptic curve scalar multiplication is made secure against correlation power analysis attacks, simple power analysis attacks, and timing attacks. Then a secured EC-ElGamal scheme is implemented to perform elliptic curve cryptography-based encryption before the data is uploaded to cloud storage. • Hence, the additive homomorphic encryption employed to process the cloud data, which is encrypted using secured elliptic curve cryptography, also becomes side-channel attack-tolerant.