DSP design tool requirements for embedded systems: A telecommunications industrial perspective

Abstract

Abstract. This paper describes the trends in DSP (Digital Signal Processing) for telecommunications design at Bell Northern Research (BNR) 1 and the tools needed to address them. The paper is in three parts: First, we present the results of a three month survey of DSP design practices at BNR. We briefly describe the characteristics of the designs, as well as the DSP design tools used. However, the emphasis is on the main bottlenecks in the design process, and the tools required to address them in the future. Then, we present a proposal for a next generation DSP design environment for telecommunication applications, based on the survey results. Particular emphasis will be given to code generation, system-level simulation, and behavioral synthesis, the three most requested design tools. Finally, we provide a description of FlexWare, an embedded software development system which is being developed internally. This system addresses one important aspect of this next generation environment, namely design tools for application-specific instruction-set processors (ASIP). FlexWare is composed of two main components: CodeSyn, a retargetable microcode synthesis system; and Insulin, a VHDL-based instruction set simulation system.

Instruction Selection

“…According to the literature, Liem et al [237,276,277] appear to have been the first to have done so.…”

Design of Embedded Systems: Formal Models, Validation, and Synthesis

“…Since the designers of general-purpose CPUs face different problems than the designers of embedded systems, we will only consider those authors who synthesize an Application-Specific Instruction Processor (ASIP, [Pau95]) and the micro-code that runs on it. The designer of a general-purpose CPU must worry about backward compatibility, compiler support, and optimal performance for a wide variety of applications, whereas the embedded system designer must worry about addition of new functionality in the future, user interaction, and satisfaction of a specific set of timing constraints.…”

Hardware/Software Cosynthesis of DSP Systems

“…Thus, we do not explore techniques for fine-grain cosynthesis [21], including synthesis of applicationspecific instruction processors (ASIPs) [43], nor do we explore cosynthesis for control-dominant systems, such as those based on procedural language specifications [22], communicating sequential processes [50], and finite state machine models [6]. All of these are important directions within cosynthesis research, but they do not fit centrally within the DSP-oriented scope of this chapter.…”