TUT14: Optimal Workflow for 5G Edge Intelligence

Edge computing has proliferated many applications involving real-time interactions. The upcoming Metaverse requires support for AR VR content that needs to be streamed interactively in real-time. It requires the content to be stored hierarchically in the Edge and transferred over the 5G network to meet the SLAs. In addition, the Edge server stores the AI models for decision-making over the captured data. It calls for the support of the underlying 5G network for the high-speed data transactions required for in-time decision-making. This tutorial provides a detailed description of the expectations from the different components of the 5G network to support the Edge intelligence. It spans Edge capture, Edge training, Edge inferencing, and Edge offloading. The working of each of these modules will be detailed, emphasizing the importance of the supporting 5G network. Devices have become intelligent nowadays and acquire data only when it is required. For example, the security camera can acquire frames at a higher rate only when some event is happening in front of it. To optimize the workflow, a mobile Edge device has to decide when to acquire the data, how much to acquire, how much to process in place and how much to transfer to the Edge server or the other Edge devices. The tutorial will discuss these problems in detail, providing possible solutions.

Short Bio: Dr. Manjunath Ramachandra is working at CTO office of Wipro Limited, Bangalore as Principal consultant. He has filed about 90 patents, conducted 26 tutorials, and delivered 37 keynotes, authored 210 papers and two books. He was a key member in the TDLS task force of Wi-Fi Alliance, served as the editor for the regional profiles standard in Digital living network alliance (DLNA) and as the industrial liaison officer for the CE-Linux Forum. He is a member of Telecom standards development society of India and actively contributes to 5G Edge intelligence standards aligned with 3GPP. He is a member of the inter-ministerial advisory group for quantum communication. He is the founding member for the model Verification and validation group in IEEE Industrial AI standards. He has chaired several international conferences and workshops. His research interests include AI, communications, Quantum computing, signal processing applications for networking etc.