AI-augmented intelligent RFICs for transmitter predistortion for 5G and 6G wireless and space communication applications

从2G开始，信号带宽在每一代都增长了五倍，在5G达到了100 MHz。数字预失真(DPD)技术是在3G中引入的，自那以来一直用于减少射频前端的损伤，允许使用更节能的组件和操作模式，但需要在基带中进行额外的计算。在目前的DPD技术中，信号处理和发射机组件以几倍于信号带宽的速度工作，将纠错信息和信号信息一起传输到射频前端。随着信号带宽的扩大，5G DPD处理复杂性和能耗已增加到抵消射频前端能效收益的程度。此外，还需要一个完整的接收器来监控射频前端并适应其行为变化。这种基带-射频-基带环路复杂、昂贵，并且跟随系统变化的速度很慢。随着波束形成技术的应用，5G+前端的工作环境变化很快，现有的DPD技术无法跟踪这些快速变化，本项目旨在设计一种具有快速适应射频前端环境和工作条件变化能力的人工智能(AI)增强的RFIC预失真前端。通过包括监控射频前端环境和操作条件变化的传感器，以及使预失真器适应这些变化的人工智能引擎，建议的解决方案将降低信号处理的复杂性、能耗和延迟。该解决方案将是第一个真正智能的射频系统，能够感知其环境并能够适应变化。概念验证RFIC将使用最先进的氮化镓(GaN)纳米技术设计，并在世界领先的NanoFab设施中制造。

Signal bandwidths have been increasing five times in every generation from 2G, reaching 100 MHz in 5G. Digital Predistortion (DPD) techniques were introduced in 3G and have been used since to reduce the impairments in RF front-ends, allowing the use of more energy efficient components and modes of operation at the cost of additional computation in baseband. As a result, the energy consumption of 3G base stations were reduced by up to 10%.In current DPD technology, the signal processing and transmitter components operate at few times the signal bandwidth to carry the correction information along with the signal information to the RF front-end. With wider signal bandwidths, 5G DPD processing complexity and energy consumption have increased to a point that offsets the energy efficiency gains in the RF frontend. Moreover, a full receiver is needed to monitor the RF front-end and adapt to its behavioural changes. This baseband-RF-baseband loop is complex, expensive, and slow in following the system changes. With the use of beamforming techniques, 5G+ Front-ends operating conditions vary rapidly and current DPD technology is unable to track these fast changes.This project aims to design an artificial intelligence (AI)-augmented RFIC predistorted front-end with the ability to quickly adapt to changes in the RF front-end environment and operating conditions. By including sensors to monitor the RF front-end environmental and operating conditions changes, and an AI engine to adapt the predistorter to these changes, the proposed solution will reduce the signal processing complexity, energy consumption, and latency. This solution would be the first truly intelligent RF system that is aware of its environment and able to adapt to changes. A proof-of-concept RFIC will be designed using state-of-the-art Gallium Nitride (GaN) nano technology and fabricated in world-leading nanofab facilities.