immediate — Innovative Methods, MEthodologies and tools for DIAgnosis and TEst

As the complexity of systems increases, test and diagnosis assume even a more relevant role for the maintenance of the device, in all its components. The aim of this research is to analyze the two mentioned issues to improve diagnostic resolution capability for faulty complex systems.

Initially, the attention has been focused on a novel diagnosis approach, based on an incremental and adaptive testing process, for reducing testing costs and efforts, by selecting — at each step of the process — the tests whose execution and outcome would provide the most significant information for identifying the faulty candidate. The incremental strategy allows us to perform only a subset of the all planned tests, and, based on the resulting partial syndrome, either the faulty candidate is found, or the next test(s) to be executed is selected, to refine the search.

The proposed methodology, initially presented in [DFT2009], is called incremental Automatic Fault Detective, iAFD, and uses a Bayesian Belief Network (BBN) as the probabilistic reasoning engine and defines a diagnosis strategy to incrementally select, given a partial syndrome, the next test to be executed to identify, in a possibly reduced number of steps, the faulty component.

Once the general framework had been defined, we have further refined the approach, by studying the various aspects driving the incremental process, from the characterization of the condition to interrupt the search — presented in [DSD2010] –, since no additional test outcomes can improve the diagnosis, to the test set modification ([DFT2010,ITC2010,DFT2011]), for improving the resolution capability of the methodology.

In the last year, the research has focused on the use of a different reasoning engine for identifying the most promising test to be executed, based on data mining. Preliminary results (presented in [DFT2013]) show relevant improvements in the number of tests (reduced w.r.t. the BBN-engine) and a less empirical termination condition. Different machine-learning techniques have also been taken into account for implementing the incremental engine and the results of the comparative analysis are reported in [DFT2014B].

The research is still active, tackling other aspects for improving results and making the methodology and framework suitable for different working scenarios.

Recent publications:

• [DFT2014B]
  C. Bolchini, L. Cassano, “Machine Learning-based Techniques for Incremental Functional Diagnosis: a Comparative Analysis,” in Proc. IEEE Intl Symp. Defect and Fault Tolerance in VLSI and Nanotechnology Systems, DFT, pp. 245-250, Oct 2014.
• [DFT2013]
  C. Bolchini, E. Quintarelli, F. Salice, P. Garza, “A data mining approach to incremental adaptive functional diagnosis,” in Proc. IEEE Intl Symp. Defect and Fault Tolerance in VLSI and Nanotechnology Systems, DFT, pp. 13-18, Oct 2013, doi: http://dx.doi.org/10.1109/DFT.2013.6653576
• [DFT2011]
  L. Amati, C. Bolchini and F. Salice, “Optimal Test Set Selection for Fault Diagnosis Improvement,” in Proc. IEEE Intl Symp. Defect and Fault Tolerance in VLSI and Nanotechnology Systems, DFT, pp. 93-99, Oct 2011, doi: http://dx.doi.org/10.1109/DFT.2011.57
• [ITC2010]
  L. Amati, C. Bolchini, F. Salice and F. Franzoso, “Improving Fault Diagnosis Accuracy by Automatic Test Set Modification,” in Proc. IEEE Int. Test Conference, Nov 2010, doi: http://dx.doi.org/10.1109/TEST.2010.5699250
• [DFT2010]
  L. Amati, C. Bolchini and F. Salice, “Test Selection Policies for Faster Incremental Fault Detection+“, in Proc. IEEE Intl Symp on Defect and Fault Tolerance of of VLSI Systems, pp. 392-400, Oct 2010, doi: http://dx.doi.org/10.1109/DFT.2010.45
• [DSD2010]
  L. Amati, C. Bolchini, F. Salice and F. Franzoso, “A Formal Condition to Stop an Incremental Automatic Functional Diagnosis,” in Proc. 13th EUROMICRO Conf. on Digital System Design – Architectures, Methods and Tools, Sep 2010, doi: http://dx.doi.org/10.1109/DSD.2010.98
• [DFT2009]
  L. Amati, C. Bolchini, L. Frigerio, F. Salice, B. Eklow, A. Suvatne, E. Brambilla, F. Franzoso and M. Martin, “An incremental approach to functional diagnosis,” in Proc. IEEE Intl Symp on Defect and Fault Tolerance of of VLSI Systems, pp. 392-400, Oct 2009, doi: http://dx.doi.org/10.1109/DFT.2009.29

This work has been/is partially supported by the Cisco and by the Cisco University Research Program Fund, an advised fund of Silicon Valley Community Foundation.

---

+ The title is erroneous, it refers to Fault Diagnosis.