*Result*: Benchmarking Fully Homomorphic Encryption in Trusted Execution Environments : A comparison between Intel TDX, Virtual Machine and native execution ; Prestandajämförelse av fullständig homomorf kryptering i betrodda exekveringsmiljöer : En jämförelse mellan Intel TDX, virtuell maskin och körning utan virtualisering

Title:
Benchmarking Fully Homomorphic Encryption in Trusted Execution Environments : A comparison between Intel TDX, Virtual Machine and native execution ; Prestandajämförelse av fullständig homomorf kryptering i betrodda exekveringsmiljöer : En jämförelse mellan Intel TDX, virtuell maskin och körning utan virtualisering
Publisher Information:
Högskolan Dalarna, Institutionen för information och teknik
Publication Year:
2025
Collection:
Dalarna University: Publikationer
Subject Terms:
Data security and privacy have become more important than ever. The need for data integrity is also becoming more relevant. Supporting these requirements while also outsourcing computing to cloud services is challenging, but there are different ways to meet these requirements such as data encryption. Combining Trusted Execution Environments (TEE) with Homomorphic Encryption (HE) can provide enhanced data confidentiality as well as data integrity. Intel TDX is a newly released TEE, and little is known about combining Fully Homomorphic Encryption with Intel TDX. This study examines the performance of Intel TDX compared to a regular virtual machine (VM) implementations and examines what trade-offs must be made when prioritizing between execution rate and security. To investigate this, experiments were conducted using the different environments. A Fully Homomorphic Encryption (FHE) benchmark program with two different parameter sets was used to measure its execution time. The analysis involved plots, regression analysis and a Pareto front plot. The sample times ranged between approximately 9 ― 89 seconds. TDX and VM environments were observed to be similar in execution time, with no environment consistently being faster. A linear relationship was found, predicting execution time vs lattice dimension. The regression analysis showed an impressive R2 value of 0.99, while the Pareto front analysis showed the optimal parameter set when considering execution rate and estimated security level for our data, Computer Sciences, Datavetenskap (datalogi)
Document Type:
*Dissertation/ Thesis* bachelor thesis
File Description:
application/pdf
Language:
English
Rights:
info:eu-repo/semantics/openAccess
Accession Number:
edsbas.52F67D32
Database:
BASE

*Further Information*

*Data security and privacy have become more important than ever. The need for data integrity is also becoming more relevant. Supporting these requirements while also outsourcing computing to cloud services is challenging, but there are different ways to meet these requirements such as data encryption. Combining Trusted Execution Environments (TEE) with Homomorphic Encryption (HE) can provide enhanced data confidentiality as well as data integrity. Intel TDX is a newly released TEE, and little is known about combining Fully Homomorphic Encryption with Intel TDX. This study examines the performance of Intel TDX compared to a regular virtual machine (VM) implementations and examines what trade-offs must be made when prioritizing between execution rate and security. To investigate this, experiments were conducted using the different environments. A Fully Homomorphic Encryption (FHE) benchmark program with two different parameter sets was used to measure its execution time. The analysis involved plots, regression analysis and a Pareto front plot. The sample times ranged between approximately 9 ― 89 seconds. TDX and VM environments were observed tobe similar in execution time, with no environment consistently being faster. A linear relationship was found, predicting execution time vs lattice dimension. The regression analysis showed an impressive R2 value of 0.99, while the Pareto front analysis showed the optimal parameter set when considering execution rate and estimated security level for our data.*