Compressing Microcontroller Execution Traces to Assist System Analysis

Recent technological advances have made possible the retrieval of execution traces on microcontrollers. However, the huge amount of data in the collected trace makes the trace analysis extremely difficult and time-consuming. In this paper, by leveraging both cycles and repetitions present in an execution trace, we present an approach which offers a compact and accurate trace compression. This compression may be used during the trace analysis without decompression, notably for identifying repeated cycles or comparing different cycles. The evaluation demonstrates that our approach reaches high compression ratios on microcontroller execution traces.