Improve the Compression Ratio by Compacting Bit-streams and Using Modified
Hadamard Transform
Jishun Kuang, Chenghua Wang, Chao Yin
College of Computer Science and Electronic Engineering
Hunan University
Changsha, Hunan
e-mail: jshkuang@hnu.edu.cn, wch466@qq.com, superyin_91@outlook.com
Abstract—A modified Hadamard matrix is proposed and the
compression ratio is improved by using the compatible
preconditioning method on the basis of transforming and
decomposing compression method. With the modified matrix
to transform the energy more focused, to better match test sets,
and the hardware implementation is relatively simple. With
the idea of test vector compaction, input bit-streams of the
circuit under test are converted to compatible before the test
set is decompose, so that a number of continuously compatible
bit-streams are compacted into a single bit-stream.
Experiments on the Mintest test set of the ISCAS89 circuit
show that both of these methods can improve the compression
ratio and the hardware cost is small. The average compression
rate of the VIHC coding method can reach 79.69%.
Keywords- test data compression; transformation; hadamard
transformation; coding
I. INTRODUCTION
As the size of the integrated circuit is increasing, the
amount of test data increases, resulting in increased cost of
test data storage and extended application time, resulting in
an increase in chip test costs. Test data compression can
effectively reduce the amount of test data, shorten the test
application time and reduce test power consumption, has
become a very important part of the test area.
The test data compression is divided into test excitation
compression and test response compression. To ensure that
the fault coverage is constant, test excitation compression is
generally lossless, while test response compression can use
lossy compression. Test excitation compression can be
divided into: Test Set Compaction, Built-in Self Test (BIST)
and Test Data Compression (TDC). Test vector compression
is divided into three major categories [1]: based on coding
compression, compression based on linear decompression
structure, and compression based on broadcast scanning.
This paper discusses the test excitation compression, on
the basis of decomposing test sets compression [2], two
methods to improve the compression ratio are proposed. In
decomposing test sets compression, the test set is
decomposed into a principal component set and a residual
set. The principal component is simple, can be placed on the
measured chip, generated by the on-chip circuit; the residue
component set can be compressed effectively and stored in
the tester.
When we are decomposing test sets, the first problem
that needs to be solved is what kind of transformation is
used. This transformation requires both energy concentration
and hope that hardware is relatively simple to implement.
We have modified the Hadamard matrix used for
transformation. The second problem to be solved in
decomposing test sets is how to encode the transformed data.
For this reason, we use the idea of test set compaction to
propose a method to compacting the input terminal data of
the circuit and improve the compression ratio. The specific
approach is to decompose the original test set before the use
of modified Hadamard matrix transformation, to remove the
original bit-stream set of compatible data, and then once
again with the modified Hadamard matrix on the
incompatible data to decompose, the principal component set
and the residual component set are obtained, and finally the
residual component set is encoded and compressed. The
experimental results show that both removing the compatible
data and the modified Hadamard matrix can improve the
compression rate.
Section 2 of this article describes the related work,
including Hadamard transformation, commonly used test
excitation coding compression method and transform
decompose compression technology. Section 3 presents a
specific method for removing compatible data and modifies
the Hadamard matrix. Section 4 describes the hardware
implementation of the decompression circuit. Section 5 is the
experimental result. Finally summarize the full text.
II. R
ELATED WORKS
A. Hadamard Transform
There are many signals in nature, such as images, sounds,
video, etc. After transform coding can be high-rate
compression, and there are a variety of transforming and
coding methods.
=YHX
(1)
Equation 1 is a Hadamard transform, in which X and Y
represent a vector, and H is a Hadamard matrix. Although
the energy concentration of Hadamard transform is poor, but
its has less computation, hardware is easy and low cost, and
has been used to compress test excitation [3-4].
Hadamard transform, also known as Walsh-Hadamard
Transforms, is a generalized Fourier transform, often used in
video coding compression. Hadamard transform uses the
Hadamard matrix, which is a matrix compose of +1 and -1.
2017 9th International Conference on Intelligent Human-Machine Systems and Cybernetics
978-1-5386-3022-8/17 $31.00 © 2017 IEEE
DOI 10.1109/IHMSC.2017.154
174