Where academic tradition
meets the exciting future

Deriving Efficient and Dependable Parallel Programs from Simulink Models

Sergey Ostroumov, Pontus Boström, Marina Waldén, Mikko Huova, Deriving Efficient and Dependable Parallel Programs from Simulink Models. TUCS Technical Reports 1111, TUCS, 2014.

Abstract:

Cost effective development of embedded software is becoming an important challenge due to the increased reliance on software in embedded systems. Moreover, real-time constraints and resilience requirements need to be met. Model-based development provides means to address these objectives. Simulink is one such widely used modelling environment that supports a complete design chain starting from modelling and ending in code generation. However, the generated code cannot fully utilize the potential offered by energy and performance efficient many-core platforms such as networks-on-chip (NoC) that can be useful for high performance control and signal processing applications. Furthermore, the failure of a processing core where the derived implementation is mapped lead to the failure of the application. In this paper, we address two primary problems: 1) derivation of efficient parallel implementations from Simulink models, 2) resilient execution of derived implementations on a many-core platform. To address the first issue, we propose an approach in which we map Simulink models to concurrent code suitable for many-core platforms where performance (real-time constraints) and low power consumption are considered. To tackle the second problem, we utilize dynamic reconfiguration and propose a fault-tolerance (FT) mechanism to mask failures of individual cores. Finally, we evaluate the proposed solutions on an industrial case study using a commercially available NoC platform, namely TilePro by Tilera. The evaluation shows only about 1% performance degradation when comparing an FT implementation with a non-FT one.

Files:

Full publication in PDF-format

BibTeX entry:

@TECHREPORT{tOsBoWa14a,
  title = {Deriving Efficient and Dependable Parallel Programs from Simulink Models},
  author = {Ostroumov, Sergey and Boström, Pontus and Waldén, Marina and Huova, Mikko},
  number = {1111},
  series = {TUCS Technical Reports},
  publisher = {TUCS},
  year = {2014},
  keywords = {Simulink; Resilience; Many-Core Platforms; Parallel Programs; Agent-based Management Systems},
  ISBN = {978-952-12-3066-0},
}

Belongs to TUCS Research Unit(s): Distributed Systems Laboratory (DS Lab)

Edit publication