Network Protocol Acceleration
Cyber-Physical Systems (CPS) and other next-generation Embedded Systems in the Internet of Things (IoT), for example, require special network protocol acceleration technology to meet the cost / performance targets. In general, networking technology today is driven by fast Ethernet connectivity for 10GbE, 40GbE, 100GbE deployed in today's datacenters, and Embedded Systems are a fast-follower to this.
TCP/IP and UDP/IP are well established network protocols from the Internet. Because of their maturity and their ubiquituous deployment they offer many benefits for Embedded Systems. However, the layered communication architecture can put a processing burdon on an Embedded Systems CPU: A generally accepted rule of thumb is that 1 hertz of CPU processing is required to send, or receive, 1 bit of TCP/IP per second. TCP offload engines were a first step to offload, from the CPU, processing of TCP/IP, especially when high-speed network interfaces are used, and the processing overhead of the network stack becomes even more significant.
MLE and the German Fraunhofer Heinrich-Hertz-Institute have partnered to market a TCP/IP and UDP/IP Network Protocol Acceleration Platform that goes beyond state-of-the-art TCP offload engines. Please read more how this customizable solution can accelerate your next 1GbE or 10GbE FPGA or ASIC system.