INTRODUCTION
THE PROBLEM—PROTECTING BUSINESS-CRITICAL INTELLECTUAL PROPERTY BY SECURING AI/ML APPLICATIONS
Machine Vision products follow the process of
- acquiring an image of a product or process,
- processing the data delivered by the image, and
- making critical decisions (such as quality or acceptance of a product or process) based on the data.
The algorithms that deliver this functionality represent critical intellectual property (IP), and create significant value for the products. It is absolutely business critical that these algorithms are protected, and can only be seen and used by developers of the products themselves. When designing, manufacturing, and provisioning MV products, application protection must be provided.
THE EmSPARK™ SOLUTION
The industry-leading Machine Vision vendor selected SecEdge’s EmSPARK™ Security Suite in order to protect its critical IP, which would be required during product design, manufacturing, and deployment in the field. The following EmSPARK™ features and functions were implemented:
- EmSPARK™’s CoreTEE™ operating system creates a secure enclave, separating secure applications and data from traditional “rich” applications (such as Linux and OpenSSL).
- EmSPARK™ provides trusted applications (TA’s) for key and certificate management, secure storage, and cryptography to ensure protection of critical data. An end-to-end secure boot, update and failure recovery process was implemented to protect the product’s firmware integrity in manufacturing and deployment.
- Using EmSPARK™’s Software Developer’s Kit (SDK), the vendor was able to develop applications that would house their algorithms in the CoreTEE’s Secure Enclave.
- EmSPARK™ API’s enabled secure communication between the secure applications and the rich applications during device operation.
- By providing encrypted code, housed in the EmSPARK™ secure environment, the vendor was able to ensure that the algorithms would be protected during the manufacturing, software provisioning, field deployment, and firmware upgrade stages of the Machine Vision product’s life cycle.
Header Image provided by Arseny Togulev on Unsplash.