Professional Documents
Culture Documents
TESTABILITY
Heera Narayanan Dr. HelenMary M C Anjana G
Dept.of Electronics and Dept.of Electronics and Dept.of Electronics and
Communication Engineering Communication Engineering Communication Engineering
Rajiv Gandhi Institute of Technology Rajiv Gandhi Institute of Technology Rajiv Gandhi Institute of Technology
Kottayam,Kerala, India Kottayam,Kerala, India Kottayam,Kerala, India
heeranarayanan007@gmail.com helenmarymc@rit.ac.in anjanageethaanoor@gmail.com
Abstract—Concurrent error detection (CED) means detecting detection includes detecting a fault before consecutive error
errors parallel with normal operation of the circuit. Proposed occurs. Otherwise it affects the reliability of the system. As
adder comprises of a carry select adder which can perform reliability is a major factor, this condition is to be satisfied.
error detection and testability. Error can be detected by pure
comparison. Comparing includes parity of the sum and predicted So the proposed adder should have the property of concurrent
one and also carry outputs. Parity of the sum is generated by error detection along with simple testability ,in that case,
using XOR operation. Input patterns are produced for testing. fault in it can be easily found at maintenance or at any
Test patterns for different blocks have been designed. For future stage[4],[8],[10],[11]. These limitations motivated the research
intelligent autonomous systems, this system is very important for a system with concurrent error detection and simple
for finding accidents. Coding is done in Verilog HDL, simulated
using ISIM Simulator and implemented using Xilinx ISE 14.7. testability. Main aim of this work is to establish such a system.
Index Terms—Concurrent error detectability,parity, testability, Arithmetic and Logic Unit (ALU) is a fast electronic circuit
test patterns which can carry out both arithmetic and logic operations.
ALU has direct input output access to input output devices,
I. I NTRODUCTION memory and processor controller [5].ALU can be implemented
Nowadays, in the upcoming era of integrated circuits, using our proposed system. ALU is a very crucial element for
system reliability is a major problem. Concurrent error performing operations in all systems. The rest of the paper
detection is very crucial in reliable systems. A lot of is organized as follows. Related works are given in section
arithmetic operations are used in VLSI, out of which addition II. Proposed system with its error detection and testability are
is the most important operation [1].Reliability is a major explained in section III. Experimental results are presented in
factor of concern. A lot of concurrent error detectable adders section IV. Concluding remarks are given in section V.
were proposed earlier [3], [6], [7] and [9].Parity prediction
II. RELATED WORKS
is used for detecting errors. In this method, errors can be
detected by comparison. Here, predicted parity and parity of A. A C-testable multi-block carry select adder[4]
the result i.e. the XORed value of the sum bits are compared. A C-Testable multiple-block carry select adder is proposed.
It uses cell fault model where adders and multiplexers are
Proposed system is an adder with concurrent error detection. taken as cells. In order to obtain C-testability, some inputs are
Lot of easy testable adders were proposed, but the peculiarity added. Modifications are done on LSB. This circuit uses 16
lies in the concurrent error detection property.Testability is patterns for testing. If n combinations are present, 2n patterns
considered as most powerful property in those adders. It is are formed, neglecting the redundancy.
useful for testing parallel with operation. The proposed system
mainly consists of a multi-block carry select adder(CSA).This B. On line detection of faults in carry select adders[7]
system assumes two carry inputs to each block, either 0 or Delay is an important factor for all digital circuits. Most
1.Final sum output is calculated by selecting one carry by important arithmetic operation is addition. A carry select
using a multiplexer[13].This is the case of parity prediction. adder with self-checking property is implemented. Transistors
Predicted parity and parity of sum outputs are compared for determine the overhead of the adders. Various error detection
detecting faults. It also compares its duplicated carry outputs. schemes have been discussed in this session, like coding
schemes. Any deviation from the expected format is consid-
In the previously proposed systems, single error faults ered erroneous. If any change occurs, it is the indication of
can be detected but not multiple outputs. Concurrent error error. Performance comparison of various schemes are also
Authorized licensed use limited to: East Carolina University. Downloaded on June 20,2021 at 23:56:51 UTC from IEEE Xplore. Restrictions apply.
explained in this session. This method is applicable for single two carry outputs. Parity of carry signals are also produced. It
faults only. Checking can be done using a compact header. has 2 operands Xk and Yk and sum result Sk . Gate level design
of INC and MUX are shown in the Fig.3.Concurrent error
C. Carry checking/Parity prediction adders and ALUs[9] detectability is obtained by special design of half adder and
Self-checking system generation is highly challenging incrementer. It consist of 2 carry outputs, where one is used
nowadays. Arithmetic codes, which is a type of self-checking for addition purpose and the other is used for parity prediction.
code, is used in arithmetic circuits for checking arithmetic In each block, XOR gates are used .The functionality of XOR
operations. Low hardware overhead is a matter of concern.
Checkers are used for finding this factor. It does not require
any translators for codes and used with memories as well.
It is valid for any schemes, any adders and so on. It also
utilizes comparison and two rail checking. But it results in high
hardware cost, so some techniques are introduced to reduce the
cost. It includes avoiding the duplication of complex blocks,
partial carry duplication etc.
Authorized licensed use limited to: East Carolina University. Downloaded on June 20,2021 at 23:56:51 UTC from IEEE Xplore. Restrictions apply.
there are three possibilities. In first case, carry signals of the possible combinations are provided, out of which two patterns
INC is affected. As there are two carry signals, one is used for are enough for testing, like 110 and 111.Repeating patterns are
addition and the other is used for parity prediction. As carry neglected. So three patterns for HA and INC, two patterns for
signal for addition is affected, the carry signal ck,j creates MUX. If any fault occurs, it will be visible on output through
an error in the sum signal sk,j . If the erroneous value is not MUX.
propagated ,it is detected as the carry is not used for parity
prediction. If it is propagated, then error is also propagated. TABLE I
Error in q carry signals produce q+1 erroneous sum results. I NPUT TEST PATTERNS FOR blockk (k > 1)
In second case, sum signal is affected. If the sum signal is Pattern xk,j , yk,j Cink
erroneous, then the sum bit will be changed. So the parity of j≥1 j=0
0
Cink
the sum bit is also changed. While comparing predicted parity t0 01 11 0
with parity of sum, fault can be found. Predicted parity will t1 11 11 0
t2 10 10 0
be non-erroneous because carry signals are not affected. Third
t3 00 11 0
case is a combination of previous two cases. Both sum signal t4 01 01 1
and carry are affected. Sum bits and carry bits are consistent. t5 11 11 1
In top position of INC, carry is affected but sum is correct t6 10 00 1
t7 00 01 1
and vice-versa for the bottom positions of the INC.
Evaluation of the proposed system is done. Proposed adder
with error detection unit is given in the Fig.4. Proposed
adder in the system generates sum output and parity. A parity C. ALU using proposed adder
generator is present, which consist of XOR gates, it is used for
generating parity of the sum. A two rail checker compares the ALU has an important role in executing arithmetic and
parity of sum and predicted parity along with carry outputs. logic operations. When large numbers of bits are present,
Inputs provided to the checker are complementary inputs ,then cascading of adders are required. These can lead to carry
the output should also be complementary. If the output is propagation delay which can be reduced by carry select adder
erroneous, then it won’t be complementing to each other. configuration. Block diagram of ALU is given in Fig 5.
Hence error is detected. ALU can perform several arithmetic and logic operations. In
B. Easy testability order to perform all these operations, three lines are provided
Testability is done with input test patterns. For each bit posi- P[2],P[1],P[0].Our proposed adder is rebuild with ALU block.
tion, bit values for j ≥ 1 .INC and HAs are designed carefully. AND operation is done between P[2] and carry output from
All possible combinations of the patterns are provided, out of each stage. Output of the AND gate acts as select line for
which recurring patterns are discarded. Only three patterns are multiplexer[2]. ALU with proposed adder is given in Fig. 6.
enough for testing, like (x,y,cin) = 100,011 and 111 through Output from ALU block is fed as input to our proposed adder.
MUX. Test patterns for blockk are given in TABLE I. Same Some of the operations performed are increment, add with
test pattern generation can be used for a MUX as well. All carry, OR etc
Authorized licensed use limited to: East Carolina University. Downloaded on June 20,2021 at 23:56:51 UTC from IEEE Xplore. Restrictions apply.
Fig. 6. ALU using proposed adder.
Fig. 8. ALU Output waveform 1.
TABLE II
ALU OPERATIONS IN SPECIFIC TIME INTERVAL
Output waveform 1
1-2s x2 [31:0]
2-3s x2 [31:0]+1
3-4s x2 [31 : 0] + y2 [31 : 0]
4-5s x2 [31 : 0] + y2 [31 : 0] + 1
5-6s x2 [31 : 0] − y2 [31 : 0] − 1
Output waveform 2
1-2s x2 [31 : 0] − y2 [31 : 0]
2-3s x2 [31 : 0] − 1
3-4s x2 [31 : 0]
4-5s x2 [31 : 0]or y2 [31 : 0]
—
Authorized licensed use limited to: East Carolina University. Downloaded on June 20,2021 at 23:56:51 UTC from IEEE Xplore. Restrictions apply.
of other CED techniques are available, out of which parity
prediction is most effective. Its area overhead is very low. It
is very easy to detect error in this circuit as it involves simple
comparison. It involves duplication, so it is highly reliable.
The proposed adder is used for the construction of ALU.A
32 bit ALU is implemented using proposed adder. Different
arithmetic and logic operations can be performed.
R EFERENCES
[1] Nobutaka Kito, Member, IEEE, and Naofumi Takagi, Senior Member,
IEEE,“Concurrent Error Detectable Carry Select Adder with Easy Testa-
bility,”IEEE Transactions on computers,2018.
[2] Priyanka Nautiyal, Pitchaiah Madduri, Sonam Negi,“Implementation
of an ALU Using Modified Carry Select Adder for Low Power and
Area-Efficient Applications,“International Conference on Computer and
Computational Sciences (ICCCS),June 2015.
[3] N. Kito and N. Takagi, “Low-overhead fault-secure parallel prefixadder
by carry-bit duplication,“IEICE Transactions on Information and Sys-
tems, vol. E96-D, no. 9, pp. 1962-1970,September 2013.
[4] N. Kito, S. Fujii, and N. Takagi, “A C-testable multiple-block carry select
adder,”IEICE Transactions on Information and Systems, vol. E95-D, no.
4, pp. 1084-1092, Apr. 2012.
[5] Mahmut Yilmaz1, Albert Meixner2, Sule Ozev1, and Daniel
J. Sorin1,“Lazy Error Detection for Microprocessor Functional
Units,“22nd IEEE International Symposium on Defect and Fault Tol-
erance in VLSI Systems,May 2007.
[6] D.Vasudevan and P.Pala,“A technique for modular design of self-
checking carry select adder,“Proc.20th IEEE International Sympiosium
on defect and fault tolerance in VLSI Systems (DFT ’05) , pp. 325-333
Oct 2005.
[7] B. Kumar and P. Lala, “On-line detection of faults in carry-select
adders,”Proc. International Test Conference (ITC ’03), vol. 1, pp. 912-
918, Sep. 2003.
[8] D. Gizopoulos, M. Psarakis, A. Paschalis, and Y. Zorian, “Easily testable
cellular carry lookahead adders,“Journal of Electronic Testing, vol. 19,
pp. 285298, June 2003.
[9] M. Nicolaidis, “Carry checking/parity prediction adders and
ALUs,”IEEE Transactions on Very Large Scale Integration (VLSI)
Systems, vol. 11, no. 1, pp. 121-128, Feb. 2003.
[10] R. D. Blanton and J. P. Hayes, “On the design of fast, easily testable
ALUs,“IEEE Transactions on Very Large Scale Integration (VLSI)
Systems, vol. 8, pp. 220223, Apr. 2000.
[11] S. Kajihara and T. Sasao, “On the adders with minimum tests,“Proc.
Sixth Asian Test Symposium (ATS ’97), pp. 1015, Nov. 1997.
[12] J.Li and E.E.Swatrzlander Jr, “Concurrent error detection in ALUs by
recomputing with rotated operands,”International Workshop on Defect
and Fault Tolerance in VLSI Systems,May 1992.
[13] O. Bedrij, “Carry-select adder,”IRE Transactions on Electronic Comput-
ers, vol. EC-11, no. 3, pp. 340-346, June 1962.
Authorized licensed use limited to: East Carolina University. Downloaded on June 20,2021 at 23:56:51 UTC from IEEE Xplore. Restrictions apply.