A Concept of Vector Clock Utilization in an Iterative Tracing Approach for Distributed Embedded Systems

Abstract

Tracing is an inevitable concept for assessing system behavior and optimizing resource utilization in parallel embedded real-time architectures. In the last decades, tracing and trace-recording analysis gained increasing importance due to more complex, computation intense and especially parallel applications spread across a variety of domains. Safety requirements, load balancing, effective memory utilization, product line engineering, heterogeneous targets or demands of common standards are just some of the occurring challenges. Our work addresses iterative system augmentation among models, performance-optimizing features and tracing in order to increase system performance and allow advanced analysis processes. Along with this automatic approach, development results benefit from lower costs and efforts, optimized parallelization, more effective resource utilization and reduced time-to-markets. By replacing a chain of different tools and corresponding manual data transformation and data adaptation with our iterative and comprehensive concept, we provide an advanced, integrated and highly adaptable solution for embedded system development. Furthermore, we illustrate the utilization of vector clocks instead of timestamps for process synchronization and trace analysis. Causal order determination as well as constraint derivation are described among examples and benefits are outlined correspondingly.