Integrated Event-Based SoC: Revolutionizing Sensor and AI Processor Performance with Low-Latency, Energy-Efficient Neuromorphic Computing

基于事件的集成 SoC：通过低延迟、节能的神经拟态计算彻底改变传感器和 AI 处理器的性能

• 批准号: 10072308
• 金额: $ 6.37万
• 依托单位: RIGPA LTD
• 依托单位国家: 英国
• 项目类别: Grant for R&D
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=10072308
• 关键词: Integrated; Event; Based; SoC; Revolutionizing

Our project focuses on creating an innovative System-on-Chip (SoC) design that integrates an event-based photodetector sensor with a neuromorphic processor. This approach tackles the latency, energy inefficiency, and complexity issues commonly found in traditional sensor and processor architectures. The SoC design employs advanced chiplet technology for scalable and cost-effective fabrication, optimizing performance and cost.The integrated SoC solution offers several key advantages:1. Enhanced Latency and Energy Efficiency: By merging the sensor and processor into one module, our design reduces latency and energy consumption. The event-based sensor captures data only when significant events occur, further decreasing energy usage and data processing requirements.2. Advanced Neuromorphic Processing: The processor is designed to imitate biological neural networks, allowing efficient, low-power processing of sensor data. This architecture is particularly suitable for edge computing applications, where low latency and energy efficiency are crucial.3. Scalable Chiplet Technology: Utilizing cutting-edge chiplet packaging technology enables us to fabricate the SoC design at different process nodes, striking a balance between performance, energy efficiency, and affordability.Our SoC design targets a wide range of applications, including autonomous vehicles, robotics, IoT devices, and smart city infrastructure, addressing the growing demand for high-performance, energy-efficient sensors and processors. With the edge computing market expected to exceed $15 billion by 2025, our innovative SoC solution is well-positioned to capture a significant share of this expanding market.Offering a unique, affordable, and energy-efficient solution, our project aims to reshape the sensor and processor landscape, providing a cutting-edge alternative to traditional architectures. Upon completion, our integrated SoC design will contribute to the development of more efficient, responsive, and sustainable technologies, driving advancements in various sectors and improving the overall quality of life.

我们的项目专注于创建创新的片上系统 (SoC) 设计，将基于事件的光电探测器传感器与神经形态处理器集成在一起。这种方法解决了传统传感器和处理器架构中常见的延迟、能源效率低下和复杂性问题。 SoC 设计采用先进的小芯片技术，可实现可扩展且经济高效的制造，优化性能和成本。集成 SoC 解决方案具有以下几个关键优势： 1．增强的延迟和能源效率：通过将传感器和处理器合并到一个模块中，我们的设计减少了延迟和能源消耗。基于事件的传感器仅在重大事件发生时才捕获数据，进一步降低能源使用和数据处理要求。2．先进的神经形态处理：该处理器旨在模仿生物神经网络，从而能够高效、低功耗地处理传感器数据。该架构特别适合低延迟和能源效率至关重要的边缘计算应用。 3．可扩展的 Chiplet 技术：利用尖端的 Chiplet 封装技术，我们能够在不同的工艺节点制造 SoC 设计，在性能、能源效率和经济性之间取得平衡。我们的 SoC 设计面向广泛的应用，包括自动驾驶汽车、机器人、物联网设备和智能城市基础设施，满足对高性能、节能传感器和处理器不断增长的需求。到 2025 年，边缘计算市场预计将超过 150 亿美元，我们的创新 SoC 解决方案已做好充分准备，在这个不断扩大的市场中占据重要份额。我们的项目旨在提供独特、经济实惠且节能的解决方案，旨在重塑传感器和处理器领域，提供传统架构的尖端替代方案。完成后，我们的集成 SoC 设计将有助于开发更高效、响应更快和可持续的技术，推动各个领域的进步并提高整体生活质量。