Embedded AI and Computation Offloading for 6G Green Communication
Naila Bouchemal,Amar Ramdane-Cherif,Nicolas Chollet
DOI: https://doi.org/10.1109/6GNet58894.2023.10317778
2023-10-18
Abstract:While 5G systems are still not yet fully deployed, 6G Key Performance Indicators (KPIs) are already defined and qualified in terms of the true benefits that the 5G system brings: Ultra high capacity (10x of 5G), User data rate in scale of Gb/s (10x of 5G), Support for millions of connected devices (10x of 5G), Energy efficiency (300x of 5G), Ultra-low latency (10x of 5G), Ultra-high positioning accuracy (30x of 5G). Key Value Indicators (KVIs) such as Trustworthiness, Sustainability, and Fairness; are also targets of future 6G networks. Nevertheless, among these indicators, sustainability is one of the primary goals. We must continue to ensure good network performance regarding Quality of Service (QoS), security, flexibility, and intelligence while reducing the environmental impact. This problem mainly occurs in Massive Machine Type Communications (MTC), where the number of devices increases exponentially, and most of these devices are battery-constrained and challenging to be charged. Several 6G technologies are candidates, such as Green Communications, to address these challenges. It aims to reduce energy consumption and serve the network to a vast number of servers cost-effectively. Green communication integrates energy saving in wireless links, hardware devices, and Infrastructure. In other words, to ensure sustainability requirements, we need to rethink the design of every component of the 6G network: energy consumption in Wireless Links, energy consumption in Infrastructure, and energy consumption in Tx/Rx Hardware. Embedded Artificial Intelligence (AI) is an effective solution. It allows constrained hardware to run complex AI algorithms, avoiding energy-costly transmission processes. However, Embedded AI devices still need to offload some computation tasks, such as training, and current MTC communication protocols struggle to meet the requirements. Thus, we propose a new communication design based on embedded AI and computational offloading in Massive MTC communication. By using these two solutions, we show how future 6G networks could integrate green communication energy saving in both Tx/Rx hardware devices and Infrastructure.
Computer Science,Engineering,Environmental Science