Seamless Dynamic Runtime Reconfiguration in a Software-Defined Radio

Michael Dickens, J. Nicholas Laneman, Brian P Dunn
RFware, LLC, South Bend, IN
SDR WInnComm Europe, 2011


   title={Seamless Dynamic Runtime Reconfiguration in a Software-Defined Radio},

   author={Dickens, Michael and Laneman, J.N. and Dunn, B.P.},



Download Download (PDF)   View View   Source Source   



We discuss implementation aspects of a software-defined radio system that allows for dynamic waveform reconfiguration during runtime without interrupting dataflow processing. Traditional software-defined radio systems execute a waveform statically, exactly as it is programmed. Reconfiguration is provided by executing a different waveform, which requires the system to stop processing data while reconfiguration occurs, and also may incur an unacceptable delay for some applications. Recent research has demonstrated basic reconfiguration by programming multiple branches into a waveform and dynamically switching between branches. This technique requires redundant resources and in general cannot be expanded to encompass all possible waveforms of interest, but, if implemented carefully, could be made to seamlessly process data. We propose a system that allows for dynamic insertion and removal of entire waveforms, individual constituent blocks, and block algorithm implementations tailored to specific processors. Our system performs this reconfiguration while maintaining processing state, seamlessly without interrupting data-processing, and with only the resources necessary for the given waveform and processors. In order to leverage this new level of reconfigurability, we created a new system component: a supervisor. This system supervisor monitors the state of each processor and waveform execution, and moves computations among available processors as their loads, capabilities, and block algorithm implementations allow. An example using a simple supervisor is provided to demonstrate the effectiveness of our system.
Rating: 2.5/5. From 1 vote.
Please wait...

* * *

* * *

HGPU group © 2010-2021 hgpu.org

All rights belong to the respective authors

Contact us: