A Random Linear Network Coding Platform MPSoC Designed in 22nm FDSOI
2022 IEEE COMPUTER SOCIETY ANNUAL SYMPOSIUM ON VLSI (ISVLSI 2022)(2022)
摘要
Random linear network coding (RLNC) has great potential to improve security, reliability, energy efficiency and throughput of many applications in networking and storage applications. The high computation costs and power consumption caused a reduction of interest in RLNC research more than ten years ago. We present a distributed parallel computation platform aiming at making RLNC affordable and scalable enough to be deployed in real-life-sized applications. As key component of this platform, an MPSoC was developed, produced and measured in our lab. The design aims at high energy efficiency and utilizes a hierarchical communication system for scalability to reach data rates needed by real-life applications with a reasonable power budget. For example, our platform would suffice to equip a 36 Gb/s backplane, 20W Ethernet switch with an RLNC accelerator on a power budget of 2.4W, showing an energy efficiency of 37 pJ/b.
更多查看译文
关键词
MPSoC,RLNC,NoC,memory management,low power,network coding
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要