GENERIC SYSTEM VERILOG UNIVERSAL VERIFICATION METHODOLOGY BASED REUSABLE VERIFICATION ENVIRONMENT FOR EFFICIENT VERIFICATION OF IMAGE SIGNAL PROCESSING IPS/SOCS

International Journal of VLSI design & Communication Systems (VLSICS) Vol.8, No.6, December 2017
DOI : 10.5121/vlsic.2017.8601 1
AN ULTRA-LOW POWER ROBUST KOGGE-
STONE ADDER AT SUB-THRESHOLD VOLTAGES
FOR IMPLANTABLE BIO-MEDICAL DEVICES
Sruthi Nanduru, Santosh Koppa and Eugene John
Department of Electrical and Computer Engineering
University of Texas at San Antonio, Texas, USA
ABSTRACT
The growing demand for energy constrained applications and portable devices have created a dire need for
ultra-low power circuits. Implantable biomedical devices such as pacemakers need ultra-low power
circuits for a better battery life for uninterrupted biomedical data processing. Circuits operating in sub-
threshold region minimize the energy per operation, thus providing a better platform for energy
constrained implantable biomedical devices. This paper presents 8, 16 and 32-bit ultra-low power robust
Kogge-Stone adders with improved performance. These adders operate at subthreshold supply voltages
which can be used for low power implantable bio-medical devices such as pacemakers. To improve the
performance of these adders in sub-threshold region, forward body bias technique and multi-threshold
transistors are used. The adders are designed using NCSU 45nm bulk CMOS process library and the
simulations were performed using HSPICE circuit simulator. Quantitative power-performance analysis is
performed at slow-slow (SS), typical-typical (TT) and fast-fast (FF) corners clocked at 50 KHz for
temperature ranging from 25̊C to 120̊C. For a supply voltage 0.3V, all the adders had the least PDP. Using
0.3V as the supply voltage, multi threshold voltage and forward body biasing techniques were applied to
further improve the performance of the adders. The PDP obtained using the forward body biasing
technique shows an effective improvement compared to high threshold voltage and multi threshold voltage
techniques. The forward biasing technique maintains a balance between delay reduction and increase in
average power, thus reducing the power delay product when compared to the other two techniques.
KEYWORDS
Kogge-Stone adder, Bio-Medical, Sub-Threshold, Forward body bias, Multi Threshold.
1. INTRODUCTION
The growing market for energy constrained applications and portable devices have created a dire
need for circuits operating at low supply voltages and consume less power. Many portable
biomedical devices require ultra-low power and improved battery life for uninterrupted
biomedical data processing. Low power systems need to operate at low voltages and/or low
currents. A recent explosion in applications that benefit from low energy operation has carved out
a significant niche for sub-threshold circuits. Digital circuits operating with a supply voltage less
than the threshold voltage of the transistors are said to be operating in the sub-threshold region.
Circuits operating in the sub-threshold region consume less active power and dissipate less
leakage power [1]. With the increased demand of the on-board processing of signals in the
implantable devices, designing circuits that consume low power and also perform complex real-
time signal processing is a challenging task. One of the complex signal processing tasks in

2
implantable devices such as pacemakers is the calculation of the Fast Fourier transform (FFT).
Basic building block of these FFTs are adders.
Addition is an indispensable operation in any digital, analog, or control system. An adder is one
of the critical components of a processor, as it can be used in the arithmetic logic units, floating-
point units and in the memory address generation unit. The fast and accurate operation of a digital
system mainly depends on the resident adders in the system. Among all the current adders,
parallel prefix adders are known to have better performance[2]. These parallel prefix adders are
known as carry tree adders.
The carry is generated parallelly in a tree adder and computation speed is improved at the cost of
increased area and power consumption. The main advantage of a carry tree design is that the
number of logic levels is reduced by generating the carries in parallel [2]. Due to the complexity
O (log_2 N) delay through the carry path, the parallel-prefix tree adders are better in terms of
speed. Kogge-Stone adder, Brent-Kung adder, Han-Carlson adder, and Sklansky adder are some
of the parallel prefix adders. Among all the parallel adders, Kogge-stone adder is the fastest
adder. This is due to the reduced number of stages needed to compute the sum and carry in the
Kogge-Stone adders. Kogge-Stone adder implementation is straightforward and also has one of
the shortest critical paths among all the tree adders [2].To utilize this adder for a low power
implantable device, it can be operated in subthreshold region [5]- [8]. In subthreshold region the
performance of the adders is degraded making it unsuitable for signal/data processing. In order to
improve the performance in subthreshold region, forward body bias (FBB) [9]- [12] and multi
threshold transistor (MVT) techniques [3]-[4] are used. This research conducts a quantitative
power-performance analysis of the Kogge-Stone adder using 45nm bulk CMOS technology in
typical-typical (TT), slow-slow (SS) and fast-fast (FF) corners at various operating temperatures.
The rest of this paper is organized as follows. Section 2 describes the steps involved in the
addition operation of the Kogge-Stone adder. Section 3 describes the tools and methodologies
used in simulating the Kogge-Stone adders. Section 4 discusses the simulation results and Section
5 concludes the paper.
2. THE KOGGE-STONE ADDER
The Kogge-Stone adder is a parallel prefix adder form of Carry Look Ahead (CLA) adder. The
Kogge-Stone adder requires more area to implement, but has less fan-out at each stage. In this
research, the Kogge-Stone adder is designed in the sub threshold region to reduce its power
consumption. The architecture of a Kogge-Stone adder is described below.
The functionality of the Kogge-Stone adder can be analysed through three distinct steps as shown
in Figure 1.
Figure 1: The three distinct stages of the Kogge-Stone Adder.

3
Figure 2 shows the schematic block diagram of an 8-bit Kogge-Stone adder. It consists of 3 carry
look-ahead network stages. The n-bit Kogge-Stone adder consists of log2n carry look ahead
network stages [13]. Hence, the 16-bit and 32-bit Kogge-Stone adder consists of 4 and 5 carry
look-ahead network stages, respectively.
Figure 2: The Schematic block diagram of an 8-Bit Kogge-Stone Adder.
Pre-Processing stage:In the pre-processing step, propagate and generate signals are computed
corresponding to each pair of A and B bitsas givenby Equations (1) and (2).
G = Ai AND Bi (1)
P = Ai XOR Bi (2)
Carry Generation Network:
In this stage, the carries are computed. All the operations are executed in parallel. They are
segmented after the computation of the carry. It uses carry propagate and generate which are
given by the logic Equations (3) and (4).
G = Gi OR (Pi AND G(iprev )) (3)
P=Pi AND P(iprev )
(4)
Post Processing:This step involves the computation of sum bits and final carry given by
Equations (5) and (6).
Ci = Gi (5)
Si = Pi XOR C(iprev ) (6)

4
3. TOOLS AND METHODOLOGY
In this research, 8, 16 and 32-bitKogge-Stone adders are simulated in the sub-threshold region of
operation using high threshold voltage transistor models for low power consumption. The
minimum operating supply voltage is determined experimentally. Use of multi-threshold voltage
transistors and body biasing techniques are employed in order to improve the performance at the
minimum operating point. All the simulations are performed in HSPICE using NCSU 45nm bulk
CMOS process library at SS, TT and FF corners considering temperature variations from 25̊C to
120̊C operating at a maximum frequency of 50 KHZ.
At first, each Kogge-Stone circuit is simulated using higher threshold voltage cells for low power
consumption and the minimum operating point is obtained. Operating the circuit in sub-threshold
region increases the delay of the circuit exponentially with supply voltage reduction. Whereas, the
power consumption decreases with the decreasing supply voltage.
In order to enhance the performance of the adder at the minimum operating point, forward biasing
technique and MVT (Multi-Threshold Voltage) techniques are applied to the circuit. The
performance improvement at minimum operating point is considered for better energy savings.
To implement the forward biasing technique, the high threshold PMOS transistors present in the
critical path are forward biased to a voltage value lower than Vdd, as shown in Figure 3. For
multi threshold voltage simulation, low threshold cells are inserted in the critical path replacing
the high threshold transistor cells.
Figure 3: Forward Body Biased PMOS Transistor.
The maximum or the worst-case delay of the adder is calculated through the critical path of the
adders. The critical path of the 8-bit, 16bit and 32bit Kogge-Stone adders consists of five gate
stages, six gate stages and seven gate stages, respectively. The delay is measured from the 50% of
the input edge to the 50% of the resultant output edge. The power consumption is evaluated for
random input patterns to achieve an accurate evaluation. Separate tests are conducted to
investigate the dynamic power and the delay. The circuit operation is tested for various input
voltage values ranging from 0.9V to 0.2V. All the outputs are connected to load inverters
powered by the input voltage. The different threshold voltages of the transistors considered is
given in Table 1.
4. RESULTS AND DISCUSSIONS
The 8, 16 and 32 bit Kogge-Stone adders are simulated using NCSU 45nm process library in
typical-typical, slow-slow and fast-fast corners and clocked at a frequency of 50 KHz and the
results are presented in the following sections.

5
4.1. Typical-Typical corner
All the adder architectures are simulated at TT corner using high threshold transistors for supply
voltage varying from 0.9V to 0.2V. The PDP is measured and plotted for 8, 16 and 32 bit Kogge-
Stone adder as shown in Figure 4, Figure 5 and Figure 6 respectively.
Table 1: Threshold voltage values of 45nm bulk transistor models.
Figure 4: PDP of 8-bit Kogge-Stone adder using 45nm Bulk models in TT corner.
Figure 5: PDP of16-bit Kogge-Stone adder using 45nm Bulk models in TT corner.

6
From the plots it can be observed that, at a supply voltage 0.3V, all the adders had the least PDP.
Using 0.3V as the supply voltage, MVT and FBB techniques are applied to further improve the
performance of the adders. The PDP (fJ) is measured at different temperature points and are
presented in Table 2- Table 4 for 8, 16 and 32 bit K-S adders respectively.
Figure6: PDP of32-bit Kogge-Stone adder using 45nm Bulk models in TT corner.
Table 2: PDP (fJ) of 8 bitK-S adder for TT corner.
Temp(C) 120 100 75 50 25
HVT 0.9978 0.756 0.5451 0.411 0.35
FBB 0.98 0.74 0.5315 0.399 0.334
MTV 7.826 6.95 5.989 5.05 4.33
Temp(C) 120 100 75 50 25
HVT 2.794 2.008 1.514 1.138 0.96
FBB 2.679 2.019 1.441 1.077 0.899
MTV 9.553 8.202 6.716 5.349 4.301
Temp(C) 120 100 75 50 25
HVT 7.64 5.771 4.117 3.076 2.574
FBB 7.229 5.437 3.868 2.873 2.383
MTV 14.052 11.844 9.485 7.403 5.879
Figure 7 shows the graphical analysis of the Kogge-Stone adder performance for the three applied
techniques i.e. high threshold voltage (HVT), forward body biasing and multi threshold voltage in
TT corner for a temperature of 25̊C and at supply voltage of 0.3V.

7
Figure 7: Kogge-Stone adder performance for various bit widths for TT process corner for HVT, FBB and
MVT techniques.
From the Figure 7, it can be observed that the PDP obtained using the FBB technique shows an
effective improvement compared to HVT and MVT techniques. Even though both the techniques
i.e. FBB and MVT help in improving the performance when compared to HVT technique, they
simultaneously increase the average power. The forward biasing technique maintains a balance
between delay reduction and increase in average power, thus reducing the power delay product
when compared to other two techniques. The percentage of energy savings by FBB is
significantly higher than the MVT technique.
4.2. Slow-Slow corner
All the adder architectures are also simulated at SS corners using high threshold transistors for
supply voltage varying from 0.9V to 0.2V. The PDP is measured and plotted for 8, 16 and 32 bit
Kogge-Stone adders and is given in Figure 8, Figure 9, and Figure 10 respectively.
Figure 8: PDP of 8-bit Kogge-Stone adder using 45nm Bulk models in SS corner.

8
Figure 9: PDP of16-bit Kogge-Stone adder using 45nm Bulk models in SS corner.
Figure 10: PDP of32-bit Kogge-Stone adder using 45nm Bulk models in SS corner.
From the results presented, it can be observed that approximately at 0.3V of supply voltage, all
the adders exhibit the lowest PDP. Using 0.3V as the supply voltage, MVT and FBB techniques
are applied to further improve the performance of the adders. The PDP (fJ) is measured at
different temperature points and given in Table 5- Table 7.
Table 5: PDP (fJ) of 8 bit adder for SS corner.
Temp(C) 120 100 75 50 25
HVT 0.738 0.541 0.364 0.249 0.192
FBB 0.732 0.526 0.347 0.239 0.1896
MTV 2.81 2.23 1.628 1.158 0.79
Temp(C) 120 100 75 50 25
HVT 2.092 1.532 1.027 0.7028 0.537
FBB 2.027 1.455 0.957 0.66 0.524
MTV 7.441 6.446 4.853 3.657 2.619

9
Temp(C) 120 100 75 50 25
HVT 5.786 4.277 2.83 1.926 1.463
FBB 5.651 3.969 2.604 1.791 1.491
MTV 11.53 8.991 6.431 4.549 3.141
Figure 11: Kogge-Stone adder performance for various bit widths for SS process corner for HVT, FBB and
MVT techniques.
From the Figure 11, it can be observed that the PDP in FBB technique shows an effective
improvement compared to HVT and MVT techniques. Even though both the techniques i.e. FBB
and MVT, help in improving the performance when compared to HVT technique, they
simultaneously increase the average power. The FBB technique maintains a balance between
delay reduction and increase in average power, thus reducing the PDP when compared to other
two techniques. The percentage of energy savings by FBB is significantly higher than the MVT
technique.
4.3. Fast-Fast corner
All the adders are simulated using the same procedure described earlier for the TT and SS
corners.
Figure 12: PDP of 8-bit Kogge-Stone adder using 45nm Bulk models in FF corner.

10
Figure 13: PDP of16-bit Kogge-Stone adder using 45nm Bulk models in FF corner.
Figure 14: PDP of 32-bit Kogge-Stone adder using 45nm Bulk models in FF corner.
Table 8: PDP (fJ) of 8 bit adder for FF corner.
Temp(C) 120 100 75 50 25
HVT 13.07 9.67 6.53 4.457 3.211
FBB 12.82 9.62 6.38 4.37 3.12
MTV 136.42 119.15 94.69 67.93 54.77
Temp(C) 120 100 75 50 25
HVT 36.57 26.98 18.15 12.31 8.8
FBB 35.02 26.2 17.31 11.77 8.39
MTV 157.6 143.13 108.7 83.2 65.1

11
Temp(C) 120 100 75 50 25
HVT 100.74 74.23 49.79 33.64 23.91
FBB 95.13 71.24 46.65 3.66 22.43
MTV 233.51 191.65 148.94 115.56 88.93
Figure 15: Kogge-Stone adder performance for various bit widths for FF process corner for HVT, FBB and
MVT techniques.
From the Figure 15, it can be observed that the PDP in FBB technique shows an effective
improvement than HVT and MVT techniques. The FBB technique maintains a balance between
delay reduction and increase in average power, thus reducing the PDP when compared to other
two techniques. The percentage of energy savings by FBB is significantly higher than the MVT
technique.
5. CONCLUSIONS
The Kogge-Stone adder presented in this research work is efficient enough to operate in the sub-
threshold region. Initially, the Kogge-Stone adder using HVT is efficiently employed for energy
constrained operations using NCSU 45nm bulk CMOS process at SS, TT and FF corners
operating in subthreshold region at a frequency of 50 KHz to obtain the minimum energy
operating point.
For better energy savings at the minimum operating point, performance improvement techniques
such as forward body biasing and the use of multi threshold voltage transistors are explored. In
forward body biasing technique, a body bias voltage is utilized which lowers the threshold
voltage of the transistor cells which in turn improves the performance of the Kogge-Stone adder
circuit.
In the multi threshold voltage technique, in an adder circuit consisting of high threshold voltage
cells, the low threshold voltage cells are placed in the critical path of the circuit to improve the
performance of the Kogge-Stone adder circuit. Based on the comparative analysis performed,
among all the three techniques discussed in this research work, the working of the Kogge-Stone
adder in forward body bias technique proved to be both efficient and robust with a lower delay
and power-delay product values compared to the other two techniques. In the subthreshold region
of operation for a supply voltage of 0.3V, the adder had the least PDP. Even though the FBB
technique exhibited the lower PDP values, HVT technique has marginally lower PDP. The PDP

12
obtained using MVT was at least twice that of the PDP obtained using the FBB technique for all
the adders.
REFERENCES
[1] Alice Wang , Benton Highsmith Calhoun , Chandrakasan. “Sub-threshold Design for Ultra-Low
Power Systems”, Integrated Circuits and Systems, 2006th Edition.
[2] Priya Gupta, Anu Gupta and Abhijit Asati, “Power-Aware Design of Logarithmic Prefix Adders in
Sub-threshold Regime: A comparative Analysis”, International Conference on Information and
Communication Technologies, 2015, pp. 1401-1408.
[3] T. McPherson et al., “760 MHz G6 S/390 Microprocessor Exploiting Multiple V_t and Copper
Interconnects,” Proceedings of the IEEE International Solid-State Circuits Conference, pp. 96-97,
February 2000.
[4] J.T. Kao and A. Chandrakasan, “Dual-Threshold Voltage Techniques for Low Power Digital
Circuits,” IEEE Journal of Solid-State Circuits, Vol. 35, No. 7, pp. 1009-1018, July 2000.
[5] A. Raychowdhury, B. C. Paul, S. Bhunia, and K. Roy, “Computing with sub-threshold leakage:
Device/circuit/architecture co-design for ultralow power sub-threshold operation”, IEEE Trans. Very
Large Scale Integr. (VLSI) Syst., vol. 13, no. 11, pp. 1213–1224, Nov. 2005.
[6] J. P. Campbell, L. C. Yu, K. P. Cheung, J. Qin, J. S. Suehle, A. Oates, and K. Sheng, “Large random
telegraph noise in sub-threshold operation of nano-scale n MOSFETs”, in Proc. IEEE Int. Conf. IC
Design Technol., May 2009, pp. 17–20.
[7] Priya Gupta, Anu Gupta and Abhijit Asati, “A Review on Ultra Low Power Design Technique:
Subthreshold Logic”, International Journal of Computer Science and Technology, April-June 2013,
pp. 64-71.
[8] D.Bol, R.Ambroise, D.Flandre and J.D.Legat. “Interests and limitations of technology scaling for
subthreshold logic,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 17, no.
10, pp. 1508–1519, 2009.
[9] S. Narendra, D. Antoniadis, and V. De, “Impact of Using Adaptive Body Bias to Compensate Die-to-
Die V_t Variation on Within-die V_t variation,” Proceedings of the IEEE International Symposium
on Low Power Electronics and Design, pp. 229-232, August 1999.
[10] A. Keshavarzi et al., “Technology Scaling Behavior of Optimum Reverse Body Bias for Standby
Leakage Power Reduction in CMOS IC’s,” Proceedings of the IEEE International Symposium on
Low Power Electronics and Design, pp. 252-254, July 1999.
[11] C. Wann et al., “CMOS with Active Well Bias for Low Power and RF/Analog Applications,”
Proceedings of the IEEE International Symposium on VLSI Technology, pp. 158-159, June 2000.
[12] S. –F. Huang et al., “Scalability and Biasing Strategy for CMOS with Active Well Bias,” Proceedings
of the IEEE International Symposium on VLSI Technology, pp. 107-108, June 2001.
[13] Raghav, Himadri Singh, Sachin Maheswari, and B. Prasad Singh. "Performance Analysis of
Subthreshold 32-Bit Kogge-Stone Adder for Worst-Case-Delay and Power in Sub-micron
Technology." Communications in Computer and Information Science VLSI Design and Test (2013):
100-07.
[14] Gupta, S. K., A. Raychowdhury and K. Roy. "Digital Computation in Subthreshold Region for
Ultralow-Power Operation: A Device-Circuit-Architecture Codesign Perspective." Proceedings of the
IEEE 98, no. 2 (2010): 160-190.

13
AUTHORS
Sruthi Nanduru received her M.S in Computer Engineering from the University of Texas at San Antonio
in August 2016. She is currently working in Intel. Her research interests include, ultra low power
computing, low power VLSI systems, computer architecture and performance evaluation.
Santosh Koppa received his Ph. D. in Electrical Engineering from the University of Texas at San
Antonio. He is currently working in Global Foundries US Inc. His research interests include, ultra low
power analog and digital circuits, Internet of Things, low power VLSI systems.
Eugene John received his Ph. D. in Electrical Engineering from the Pennsylvanian State University. He is
currently a professor in the department of electrical and computer engineering. His research interests
include, ultra-low power computing, low power VLSI systems, integrated circuit IP security and trust and
computer architecture and performance evaluation

GENERIC SYSTEM VERILOG UNIVERSAL VERIFICATION METHODOLOGY BASED REUSABLE VERIFICATION ENVIRONMENT FOR EFFICIENT VERIFICATION OF IMAGE SIGNAL PROCESSING IPS/SOCS

More Related Content

What's hot (18)

Similar to GENERIC SYSTEM VERILOG UNIVERSAL VERIFICATION METHODOLOGY BASED REUSABLE VERIFICATION ENVIRONMENT FOR EFFICIENT VERIFICATION OF IMAGE SIGNAL PROCESSING IPS/SOCS (20)

Recently uploaded (20)

GENERIC SYSTEM VERILOG UNIVERSAL VERIFICATION METHODOLOGY BASED REUSABLE VERIFICATION ENVIRONMENT FOR EFFICIENT VERIFICATION OF IMAGE SIGNAL PROCESSING IPS/SOCS