System-on-Chip Overlay Methodologies for Software Defined Radio

在过去的五年中，基于现场可编程门阵列（fpga）的可编程片上系统（SoC）技术获得了相当大的发展势头。这些soc将FPGA的高度并行、确定性计算与一个或多个传统处理核心相结合，能够运行操作系统（os）和基于软件的处理。随着这些设备的发展，它们变得越来越复杂，现在将许多不同类型的处理元素组合到单个芯片中（FPGA资源，应用处理器，实时处理器，图形处理器，视频编码器等），因此被供应商Xilinx采用了多处理器SoC （MPSoC）的名称，最近，他们的专业无线电版本，射频SoC (RFSoC)，于2017年发布。虽然能够访问所有这些优化资源是受欢迎的，但它确实在开发时间、精力和完成完整系统设计所需的技能组合方面提出了挑战。针对这一设计挑战的一个新兴解决方案是“overlay”范式，Xilinx以其Python Productivity for Zynq（“Pynq”）方法的形式对其进行了最好的捕获和支持。Pynq寻求通过提供：(i)一组开源工具来加速系统的实现；（ii）一个用生产力语言Python开发软件的框架；（iii）引入硬件“覆盖”理念，其中硬件系统设计具有主题但相对灵活的功能，旨在重用和共享。随着系统复杂性和上市时间的压力不可避免地增加，Pynq方法和其他类似的方法预计将变得越来越重要。overlay / Pynq方法非常适合软件定义无线电（SDR）技术，其中无线电的一个或多个方面的功能是通过软件控制的。由于其提供的采样率和系统集成，新的RFSoC提供了突破性的SDR功能。因此，将Pynq应用于SDR，特别是通过RFSoC平台，提供了一个非常令人兴奋的研究机会，这几乎没有被探索过。基于Pynq覆盖层的SDR系统的开发将为下一代系统提供高度灵活和快速开发的无线电应用，包括加速对5G候选技术的评估，以及实现动态频谱接入（DSA）无线电方案（其中无线电频谱的使用是偶然实现的，而不是通过传统的固定频谱分配方法）。提出的研究项目将考虑SDR覆盖方法的以下几个方面：RFSoC的使用，包括对能力、局限性的评估，以及对其资源的最佳利用，以开发无线电发射机和接收机架构。2. 开发用于SDR应用的硬件覆盖系统，重点关注无线数据网络的现有和新兴标准。研究完全优化的SDR硬件和通用功能/灵活性之间的权衡。开发了一个Pynq SDR框架，采用多个覆盖来实现无线电功能的充分灵活性。评估和描述这些系统，考虑如何在网络场景中部署它们，并评估它们的性能。考虑可能的业务/社区重用模型。

In the last five years, programmable System on Chip (SoC) technology based on Field Programmable Gate Arrays (FPGAs) has gained considerable momentum. These SoCs combine the highly parallel, deterministic computation of an FPGA, with one or more traditional processing cores, capable of running Operating Systems (OSs) and software-based processing. As these devices have developed, they have gained in complexity, with a number of different types of processing elements now combined into a single chip (FPGA resources, applications processors, real-time processors, graphics processors, video encoders, and so on), hence the name Multi-Processor SoC (MPSoC) adopted by vendor Xilinx, and more recently still, their specialist radio version, the Radio Frequency SoC (RFSoC), which was released in 2017. While having access to all of these optimised resources is welcome, it does present a challenge in terms of development time, effort, and the combination of skills required to complete a full system design. An emerging solution to this design challenge is the 'overlay' paradigm, best captured and supported by Xilinx in the form of their Python Productivity for Zynq ('Pynq') methodology. Pynq seeks to accelerate the realisation of a system by providing: (i) a set of open-source tools; (ii) a framework for developing software in the productivity language, Python; and (iii) the introduction of a hardware 'overlay' philosophy, where hardware systems are designed with themed but relatively flexible functionality, intended for reuse and sharing. The Pynq approach, and others like it, are expected to grow in significance as pressures on system complexity and time-to-market inevitably grow. The overlay / Pynq approach is highly suited to Software Defined Radio (SDR) technology, wherein the functionality of one or more aspects of a radio is controlled via software. The new RFSoC provides a breakthrough SDR capabilities, due to the sampling rates and system integration that it offers. The application of Pynq to SDR, specifically via the RFSoC platform, therefore provides a highly exciting research opportunity which has barely been explored. The development of SDR systems based on Pynq overlays would enable highly flexible and rapidly developed radio applications for next generation systems, including accelerating the evaluation of candidate technologies for 5G, and the realisation of Dynamic Spectrum Access (DSA) radio schemes (wherein use of the radio spectrum is achieved opportunistically, rather than via the traditional fixed spectrum allocation approach). The proposed research project will consider the following aspects of overlay methodologies for SDR:1. Use of the RFSoC, including an assessment of capabilities, limitations, and best use of its resources for developing radio transmitter and receiver architectures. 2. The development of hardware overlay systems for SDR applications, focussing on existing and emerging standards for wireless data networks.3. Research into the trade-off between fully optimised SDR hardware, and aspects of generic functionality / flexibility.4. Development of a Pynq SDR framework employing multiple overlays to achieve full flexibility of radio functionality. Assessment and characterisation of these systems, consideration of how to deploy them in a network scenario, and evaluation of their performance.5. Consideration of possible business / community reuse models.