RAPPER: Robust and APProximate ERror Tolerating Communication
Somayeh Sadeghi-Kohan, Sybille Hellebrand, Hans-Joachim Wunderlich · 2024
Effective error-resilient communication is crucial, especially in autonomous systems. However, some applications can accept a certain level of approximation within a defined range. This opens up opportunities for optimizing system objectives, like minimizing error rates, extending operational lifetimes or reducing power consumption. In this study, we introduce an innovative approach called RAPPER (Robust and Approximate Error Toler-ating Communication) which combines approximation and error correcting codes for optimizing the transmitted data. The under-lying communication scheme exploits an extension of differential encoding to enhance communication performance. Simulation results show an increased interconnect mission time and performance, while simultaneously achieving significant reductions in error rates and power consumption.