Secure Cloud Computing

We are in the age of big data. Every industry has massive data sets that need to be stored, managed, and computed. It is infeasible for users to analyze large-scale data-sets on traditional computer hardware due to its limited computing resources. Although cloud computing holds the ability to compute these large scale data sets, outsourcing data to a cloud server poses a threat to the confidentiality and integrity of the data.

At Wichita State University, professor Dr. Sergio Salinas has found a new way to securely compute big data at the cloud. This new method allows for massive data sets to be securely computed at the cloud, while taking advantage of the cloud’s vast computing infrastructure.


 Cloud computing offers the following benefits: 

  • Massive decrease in computation times
  • Decreased in all hardware costs:
    • Maintenance & upkeep
    • Physical infrastructure

Our solution allows all of the advantages that cloud computing offers, but in a secure manner that is faster and cheaper than existing methods.


As massive data collection becomes prevalent, various industries now have a need for big data calculations. Industries that have a current need for big data calculations include:

  • biomedicine
  • power systems
  • finance
  • simulations
  • social networks

This model for cloud security proposes methods fo perform safe computations on big data of various forms, including quadratic programming and solving linear systems of equations (with special consideration for sparse linear systems of equations). By transforming the data through an efficient and effective procedure at the client, the data can be sent and computed faster at the cloud without risking the integrity or security of the data. All of this is done through the use of traditional linear algebra software, avoiding the costly requirements of ciphertext-based operations.

When performing operations on linear systems of equations, the matrix to be calculated is transformed by the client by adding a sparse matrix with random values and randomly permuting its rows and columns. This new matrix is proven to be computationally indistinguishable under a chosen-plaintext attack (CPA). Then, with the help of the cloud, the client finds the solution vector iteratively using the conjugate gradient method. Assuming the matrix is sparsely populated, both the client and the cloud are able to exploit the sparsity of the transformed matrix for computational savings.

When performing operations on quadratic programs, the algorithm requires the user to perform computations with only O(max{n2,mn}) complexity. To accelerate the computations, the cloud runs the algorithm in parallel. The user only communicates with the cloud to transform its QP and to receive the solution, resulting in a constant number of data exchanges between the user and the cloud. It is proven that the transformed QP problem can protect the user’s data privacy. In particular, the transformed QP has the property of computational indistinguishably under a CPA.


Dr. Sergio Salinas is an assistant professor in the Dept. of Electrical Engineering and Computer Science. Dr. Salinas leads the Cyber-Physical System Security Lab (CPSSL) at WSU. His primary research interests include privacy and security, cyberphysical systems (e.g. smart grids), big data, bioinformatics, cloud computing, and social networks.

Dr. Salinas is a contributing member to professional societies such as the Association of Computing Machinery (ACM) and the Institute of Electrical and Electronic Engineers (IEEE), having published 18 peer reviewed research papers for various journals and conferences. He is a regular reviewer of IEEE Journals, and commonly referees conferences, such as the International Conference on Computer Communications (INFOCOM) 2017.



Download a Printable PDF


Rob Gerlach
Director of Technology Transfer

- Technology Patent Pending -