On the Applicability of Secret Share Algorithms for Saving Data on IoT, Edge and Cloud Devices

A common practice to store data is to use remote Cloud-based storage systems. However, storing files in remote services can arise privacy and security issues, for example, they can be attacked or even discontinued. A possible solution to solve this problem is to split files into chunks and add redundancy by means of Secret Share techniques. When it comes to Internet of Things (IoT), Edge and Cloud environments, these techniques have not been evaluated for the purpose of storing files. This work aims to address this issue by evaluating two of the most common Secret Share algorithms in order to identify their suitability for different environments, while considering the size of the file and the availability of resources. In particular, we analysed Shamir's Secret Share schema and the Redundant Residue Number System (RRNS) to gauge their efficiency regarding storage requirement and execution time. We made our experiments for different file sizes (from 1kB up to 500MB), number of parallel threads (1 to 4) and data redundancy (0 to 7) in all aforementioned environments. Results were promising and showed that, for example, to have seven degrees of redundancy, Shamir uses eight times more storage than RRNS; or, Shamir is faster than RRNS for small files (up to 20 kB). We also discovered that the environment on which the computation should be performed depends on both file size and algorithm. For instance, when employing RRNS, files up to 500kB can be processed on the IoT, up to 50MB on the Edge, and beyond that on the Cloud; whereas, in Shamir's schema, the threshold to move the computation from the IoT to the Edge is about 50kB, and from the Edge to the Cloud is about 500kB.