Enhancing quality of assertion generation: methods for automatic assertion generation and evaluation

Hertz, Samuel

Permalink

https://hdl.handle.net/2142/55156 Copy

Description

Title

Enhancing quality of assertion generation: methods for automatic assertion generation and evaluation

Author(s)

Hertz, Samuel

Issue Date

2013-06-11

Director of Research (if dissertation) or Advisor (if thesis)

Vasudevan, Shobha

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• assertion
• assertion generation
• assertion evaluation
• assertion ranking
• GoldMine
• static analysis
• data mining
• evaluation
• ranking
• verification
• validation

Abstract

We present methods for automatically generating and evaluating register transfer level (RTL) assertions. We detail the GoldMine methodology and each of its data mining algorithms. We introduce the Best-Gain Decision Forest algorithm to mine concise RTL assertions. We develop an assertion ranking methodology. We define assertion importance, complexity, rank and ideality and we detail methods to compute each of them. We present a case study and experimental results to demonstrate the effectiveness of assertion rank. We develop an assertion rank aggregation methodology. We define assertion coverage and expectedness. We aggregate rankings for assertion importance, complexity, coverage and expectedness. We present experimental results to demonstrate the value of these metrics and the rank aggregation methodology. We rigorously analyze the performance of each data mining algorithm in GoldMine. We present experimental results that demonstrate each algorithm's performance with respect to various metrics.

Graduation Semester

2013-05

Permalink

http://hdl.handle.net/2142/55156

Copyright and License Information

Copyright 2013 Samuel Hertz

Owning Collections