A Data-Synchronous Event Model for GNU Radio

In this paper, we present a synchronous event stream model overlaid on the existing GNU Radio streaming dataflow model. GNU Radio has long utilized a traditional static streaming dataflow model to interconnect modular signal processing blocks. While this model fits many radio and signal processing applications well, GNU Radio and the applied signal processing community face several pressing needs which require extensions to this model. The event stream model exploits the inherent data parallelism present in many multi-access wireless protocols, greatly enhancing performance on modern concurrent hardware. It also greatly reduces the complexity necessary to implement modern communications protocols and greatly improves code and component re-use by removing the need to co-locate state machines and in-band signaling within primitive GNU Radio blocks. These features ultimately contribute to the agility and rapid development cycle targeted by GNU Radio. We describe the basic components of the event stream model. Through this discussion, we show how the model works as a system to provide additional opportunities for parallelism, creates simplified complex waveform design, promotes rapidly reconfigurable behavior, and encourages good modular design practices. We discuss a number of critical performance issues and considerations to take into account when the creating of this system. We further describe how it interoperates with many existing GNU Radio subsystems, including existing streaming blocks, hierarchical blocks, stream tags, and asynchronous message handling.