19-09-2014, 03:42 PM
An Efficient Test Data Compression Using
Viterbi Algorithm
An Efficient Test Data.pdf (Size: 302.26 KB / Downloads: 38)
Abstract-
This paper presents An Efficient Test Data
Compression Using Viterbi Algorithm that provides high
encoding efficiency and scalability with respect to the number
of test channels. Proposed numerous test vector compression
technique is linear decompression is used for simplicity high
encoding efficiency. The proposed viterbi decompressor goal is
produced large amount of output using small volume of inputs.
Viterbi encoder is a good decompressor its used for increase
the no of outputs. Experimental results are observed on Xilinx
based simulator “Isim” wave forms and compared to previous
test data compression architectures. Using Verilog and tool
used is Xilinx ISE12.1 software.
. INTRODUCTION
Proposed numerous test vector compression technique is
linear decompression is used for simplicity high encoding
efficiency. Compression architectures are classified into
three types are namely
1. Code based
2. Linear decompression
3. Broad cast scan
Linear compression is popular for simplicity and high
encoding efficiency. Linear decompressor produced output of
set of compressed test vectors in random. Effects on the
compressed, requirements.
Reducing high power consumption
Not providing the systematic methodology to trade
off compressions
This technique is used successful linear decompressor
mainly combinational linear expansion or use linear
feedback shift register (LFSR) implemented by XOR gates.
Linear decompressed text vectors can be obtained by ATPG
having solving linear equations. Output response comparator
can be designed by using lossy algorithm which having
multiple input signatures registers (MISR).
Viterbi algorithm is a dynamic programming technique and
has time complexity O (l . 2F
).
l = length of input sequence, F= number of flip flops
l is a linear function proposed algorithm generates
compressed text vector quickly even for large circuits. F is
TEST STIMULUS COMPRESSION
A test cube is defined as a deterministic test vector in
which the bits that are not assigned value by the ATPG
procedure are left as “don’t cares” (X’s). Normally, ATPG
procedures perform random fill, in which all the X’s in the
test cubes are filled randomly with 1’s and 0’s to create fully
specified test vectors. However, for test stimulus
compression, random fill is not performed during ATPG so
the resulting test set consists of incompletely specified test
cubes. The X’s make the test cubes much easier to compress
than fully specified test vectors. Test stimulus compression
should be an information lossless procedure with respect to
the specified (care) bits in order to preserve the fault coverage
of the original test cubes.
After deompression the resulting test patterns shifted into the
scan chains should match the original test cubes in all
specified (care) bits. Many schemes for compressing test
cubes have been proposed
. CONCLUSION
In this paper, we proposed new and with the number of channels,
linear compression scheme which is based on the viterbi algorithm
defining proper branch metrics, path metric the proposed technique
can include and optimise different test constraints, and selects, a set
of compressed vectors that has minimum cost of function among all
possible states. Limitations of proposed schemes