Secure Edge Computing Connected Device (SECCD)

The Secure Edge Computing Connected Device project is dedicated to developing an advanced IoT gateway centered around a System-on-Chip (SoC) powered by a RISC-V core. This SoC will incorporate specialized hardware accelerators designed to handle security, privacy-critical functions, and edge Artificial Intelligence (AI) computations. The utilization of Neural Network (NN) models for fast inference addresses the limitations of power consumption that traditional algorithms impose on resource-constrained IoT devices, thereby enhancing performance and energy efficiency. The project aims to strengthen IoT systems against network threats by integrating AI and network attack detection models tailored to specific use cases directly within the SoC. To ensure maximum flexibility and efficiency, the SoC will be implemented on an FPGA, exploiting the Dynamic Partial Reconfiguration (DPR) feature.

Furthermore, the instruction set of RISC-V cores will be extended to manage DPR in a power and time-efficient manner, allowing the system to dynamically adapt to varying operational conditions such as changes in battery levels or required security levels, facilitated by AI. This dynamic capability enables selective activation or deactivation of AI and cryptography accelerators. By implementing a comprehensive approach to security and adaptability, the ARCD project aims to establish new benchmarks for IoT gateway technology, ensuring robust protection and optimal performance across diverse operational environments. As a proof of concept, an IoT network encompassing perception, network, and cloud layers will be constructed to demonstrate resistance against network security threats like spoofing and denial of service. The project will also provide full software support, from drivers and operating systems to application software and network/cloud stacks, creating a complete, design and runtime adaptable, secure, powerefficient, and time-efficient edge computing solution from sensor to cloud.