Modeling dynamic partial reconfiguration in the dataflow paradigm

An important number of studies have shown the benefit of dynamic partial reconfiguration in reconfigurable computing. Signal processing applications can make use of this technology in such a way that it allows greater flexibility, performances and cost reduction. However several points still need to be addressed and represent critical challenges. One of them concerns architectures modeling as abstraction is strongly required to help designers in building efficient designs. Dataflow is a well adopted modeling paradigm for signal processing application to allow early stage system properties evaluation. This paper describes a first attempt to model dynamic partial reconfiguration in the dataflow paradigm. Our proposal leads to an efficient and simple approach suitable for signal processing systems.