Secure Mobile Intensive Tasks Offloading to the Cloud

Abstract

Mobile devices with internet access are switching towards the Cloud for resource intensive tasks. Cloud Computing has emerged as an attractive platform for mobile users to offload their resource intensive computations (e.g. consider a large image that has to be compressed and placed as wallpaper) due to limited processing and power capabilities of mobile devices. However, individuals are still reluctant to offload their tasks to the cloud due to a number of security concerns like loss of control and maintaining confidentiality and privacy of data. In this thesis, we address Mobile and Cloud Computing security concerns by encompassing Trusted Computing concepts. Trusted Computing incorporates tamper resistant hardware commodity Trusted Platform Module (TPM) and enables parties to verify what code is running on the remote computer through Remote Attestation process. We propose different scenarios 1) data computation on the mobile device 2) data computation on cloud node 3) data computation on cloud node with encryption 4) data computation on cloud node with encryption and Remote Attestation 5) data computation on cloud node with two-way Remote Attestation and encryption, each scenario more secure than the previous at the cost of some additional performance overheard. Since, there is always tradeoff between security and performance thus users can adopt any scenario matching their requirements based on threshold. We took Android as a client of cloud services and image compression as workload to analyze the performance of proposed scenarios. We performed image compression experiments using both android emulator and a real phone on different image sizes. For emulator, offloading a 244KB image to the Cloud (Scenario -2) saves 34% of mobile device resources and reduces overall execution time to 25%. Offloading the same in Scenario-3 and 4 caused an increase of 20% in mobile device resource utilization with increase of 1.3 and 4.1 times respectively in overall execution time. However, offloading large sized images like 3MB in Scenario-3, 4 and 5 saved mobile device resources up to 88% for Scenarios-3 and 4 and 76% for Scnenario-5 with reduction of 89%, 82%, 69%, and 58% in overall execution times for Scenrios-2, 3, 4 and 5 respectively. Moreover, we performed experiments on Android Nexus One Phone and experiments show that offloading 2.1MB image to the Cloud (Scenario-3, Scenario-4) saves 79% of mobile device resources.