How do we ensure that our personal data can be utilised to its fullest potential without compromising our privacy? Homomorphic encryption, which allows computation outsourced to the cloud to be done directly on encrypted data, is a potentially powerful way of tackling the challenges involved.

Explaining the science

The challenges involved in privacy-preserving data analysis involve finding secure ways of providing public access to private datasets, securely decentralising services that rely on private data from individuals, enabling joint analysis on private data held by several organisations, and securely outsourcing computations on private data.

Fully homomorphic encryption has potential application in all these challenges. This technology enables secure, outsourced computation, where data owners can upload their privately encrypted data to a cloud computing provider. Computations in the server are then done directly on the data without prior decryption, and the results are sent encrypted to the data owner who can decrypt them themselves.

Fully homomorphic encryption schemes are still a long way from being applicable in practice for outsourced secure computation. A more practical alternative is so-called ‘somewhat homomorphic’ encryption schemes, which support only a limited, fixed number of nested multiplications, but are faster and more compact. Many useful data analysis tasks can be implemented using this more limited form of encryption.

Project aims

Several libraries for somewhat homomorphic encryption are available based on different cryptographic assumptions, with different features and usability constraints.

The key questions for this project are:

  • What is the current state of the implementations of homomorphic encryption?
  • What homomorphic encryption scheme (and implementation) should someone use in their application, for a given set of requirements?

The aim of the project is not to give a complete answer to these questions, but to set up an evaluation platform called SHEEP (SHEEP is a Homomorphic Encryption Evaluation Platform). This platform will be used to assess several implementations of encryption schemes, with a variety of homomorphic properties.


Researchers and collaborators

Contact info

[email protected]

Research Engineering

View the Research Engineering page

Members of the Research Engineering Group at the Turing are contributing their expertise to this project.

They are producing a software package that allows users to express their calculation in a common language, then perform them using a range of homomorphic encryption libraries in order to compare performance. In addition to the interfaces to the different libraries, the package provides an API and a web front-end to allow user interaction, plus a range of example notebooks to illustrate usage of the package.