Model-Based Integrated Simulation (MILAN)

摘要

The motivation for the Model-based Integrated Simulation (MILAN) project is to develop an extensible modeling, simulation, and design space exploration framework for the design of latency and energy efficient embedded systems for signal processing applications. Design of embedded systems requires minimization of energy dissipation (to maximize battery life) while meeting a given latency constraint (typically real-time constraints). While until now Application Specific Integrated Circuits (ASICs) were considered the primary choice for low power high performance embedded systems, the recent advances in the design of general purpose processors (GPP), digital signal processors (DSP), field programmable gate arrays (FPGAs), and memories have provided viable commercial-off-the-self (COTS) alternatives to ASICs. These devices are designed using low-leakage process and support a number of low power operating and standby states, dynamic voltage and frequency scaling, among others to support energy optimization. In the MILAN project, we focus on signal processing applications that process a stream of input frames while meeting a given latency constraint for the processing of a single frame. MILAN is a joint effort by the University of Southern California and Vanderbilt University and is supported by the DARPA Power Aware Computing and Communication Program.