OctaiPipe: Accelerating Trustworthy AIoT

cc42033d-437d-4a19-9a9c-c8fa34f13640

Closed

Innovate UK
Innovate UK
Innovate UK
Innovate UK
Innovate UK

£242,570

April 30, 2023

July 31, 2023

In the past 5 years, there has been a dramatic growth and adoption of Internet of Things (IoT) devices. These are found all around us, from smartphones, watches, TVs, smart meters, to industrial machinery, energy infrastructure, and autonomous vehicles. Each device generates data continuously and the volumes of data generated are vast and continue to grow exponentially. At the same time, data usage from deployed IoT devices continues to grow and offers significant economic value potential. As highlighted in 2021 McKinsey report, it is estimated that by 2030 it could enable $5.5 to $12.6 trillion in value globally. However, a key challenge is maintaining the privacy and security of data and algorithms so that users, both individuals and organisations, can trust them. This has profound consequences for modern and future societies, particularly when combined with technologies such as AI.

T-DAB.AI's mission is to help solve this problem of trustworthy AI in IoT. We are doing this by pursuing the development of a secure, auditable, private, and distributed AI for IoT platform targeting deployments to IoT and edge devices called OctaiPipe. OctaiPipe is a first of its kind innovation that brings together privacy preserving machine learning technology, cyber security, and machine learning lifecycle management (MLOps). This will allow IoT enabled businesses to build, deploy, and manage machine learning software that guarantees the privacy and security of device data and its use, allowing the consumer to have a high degree of trust in the AI solutions running on them.

This project is to undertake a feasibility study in assembling a consortium of leading technology organisations, comprising of innovators in AI and IoT hardware, software, and R&D. This feasibility study will not only work to assemble a consortium, but also assess the feasibility of undertaking cutting edge technology development in four key domains of developing trustworthy AI. This will include work to protect algorithms from being poisoned by bad actors, better optimising algorithms to combine privacy and accuracy, make specialised lightweight models more explainable, and make the management systems for on device algorithms more auditable.