Combining dictionary coding and LFSR reseeding for test data compression

Abstract

In this paper we describe a method to combine dictionary coding and partial LFSR reseeding to improve the compression efficiency for test data compression. We also present a fast matrix calculation method which significantly reduces the computation time to find a solution for partial LFSR reseeding. Experimental results on ISCAS89 benchmark circuits show that our approach is better than either dictionary coding or LFSR reseeding, and outperforms several test data compression methods proposed recently.

Correlation-Based Rectangular Encoding

“…As shown in Table V, the proposed scheme achieves the highest compression ratio in all the circuits, which is enabled by reduction in the number of specified bits. Compared with [23], more compression ratio is observed in the proposed study. Another important advantage over [23] is that the decoder in the proposed scheme is independent of circuit or test data, while [24] requires a dictionary that is dependent on circuit.…”

“…The same test set is employed in [22], [24], and the proposed study, while [23] does not ( [23] employs Mintest test set). As shown in Table V, the proposed scheme achieves the highest compression ratio in all the circuits, which is enabled by reduction in the number of specified bits.…”

“…Another important advantage over [23] is that the decoder in the proposed scheme is independent of circuit or test data, while [24] requires a dictionary that is dependent on circuit. Note that the Mintest test set in [23] has a smaller number of test cubes and more inconsistent in terms of the number of specified bits than the test set used in the proposed study.…”

“…Whereas the method in [22] is only applicable for LFSR reseeding, where the seed is periodically loaded, and the proposed scheme is applicable for any linear decompressor including combinational and sequential continuous-flow decompressors (for which the method in [22] cannot be used). Dictionary coding and LFSR reseeding are combined in [23] such that either one or the other is used to load each scan slice. In the proposed method, rectangular coding is combined with a linear decompressor and both are used together for all scan bit slices, enabling a continuous-flow decompression with greater efficiency.…”

Periodic Scan-In States to Reduce the Input Test Data Volume for Partially Functional Broadside Tests

“…The fully specified scan-in states of functional broadside tests are not compatible with the most popular test data compression methods [Barnhart et al 2001;Chandra et al 2011;Koenemann 1991;Krishna et al 2004;Lin and Rajski 2012;Pomeranz 2016;Rajski et al 2002;Sun et al 2004;Tenentes et al 2008]. These methods compress tests into initial states (seeds) for a linear-feedback shift register (LFSR).…”