Edge-AI Addresses the Big Data Challenges of IoT in the Cloud Edge

Academic research is appearing in a new disciplinary field through the amalgamation of Cloud Edge and Artificial Intelligence named Edge-AI. This amalgamation of technologies is promising to accelerate insights to end-user devices and remove the reliance on the Cloud for central data collection and analysis. IoT devices and their data will exponentially grow to 75 billion devices by 2025, which is 2.5 times the amount of IoT data processed in 2020. With this expected growth, the ability to transmit and process IoT sensor data in a time-sensitive and meaningful manner will challenge the current central Cloud architecture. This research aims to provide an overview of how Edge-AI deployed in the Cloud Edge will elevate IoT life-critical systems and improve privacy and security in analytics data. This research uses the Systematic Literature Review methodology to discuss collaborative learning frameworks, Machine Learning distribution techniques, data consensus models and security and privacy in the context of Edge-AI. A keyword search was conducted, and the material was selected for review. The research concludes that Edge-AI is a promising architectural concept encompassing many subcomponents. Edge-AI will be required as IoT devices increase in numbers to support the efficient use of life-critical systems like autonomous vehicles and health monitors. There is significant fragmentation in research into the major subcomponents of Edge-AI, which hamper the ability to adopt widely. The Efficient Edge-AI Framework proposed in this research aims to progress this field by presenting a holistic framework encompassing the gaps in previous research.