Thresholding for Optimal Data Processing in a Software Defined Radio Kernel

The authors discuss a developing software-defined radio framework and kernel, Surfer, which, among other changes compared with known current frameworks, moves control of when to process a block and how much data to process into the blocks themselves via buffer thresholds. Varying threshold values shows a trade-off between data processing latency and throughput, with respect to overhead time. Processing and overhead timing statistics are collected during runtime and can be retrieved by a user's constraint function in order to allow for user-based optimal data processing. A simple waveform example is included to demonstrate system functionality as well as provide insights into the latency-throughput tradeoff.