COVERAGE DRIVEN FUNCTIONAL TESTING ARCHITECTURE FOR PROTOTYPING SYSTEM USING SYNTHESIZABLE ACTIVE AGENT

International Journal of VLSI design & Communication Systems (VLSICS) Vol.9, No.3, June 2018
DOI : 10.5121/vlsic.2018.9304 41
COVERAGE DRIVEN FUNCTIONAL TESTING
ARCHITECTURE FOR PROTOTYPING SYSTEM USING
SYNTHESIZABLE ACTIVE AGENT
Dipakkumar Modi and Usha Mehta
EC Department, Institute of Technology, Nirma University, Ahmedabad, India
ABSTRACT
Time and efforts for functional testing of digital logic is big chunk of overall project cycle in VLSI industry.
Progress of functional testing is measured by functional coverage where test-plan defines what needs to be
covered, and test-results indicates quality of stimulus. Claiming closer of functional testing requires that
functional coverage hits 100% of original test-plan. Depending on the complexity of the design, availability
of resources and budget, various methods are used for functional testing. Software simulations using
various logic simulators, available from Electronic Design Automation (EDA) companies, is primary
method for functional testing. The next level in functional testing is pre-silicon verification using Field
Programmable Gate Array (FPGA) prototype and/or emulation platforms for stress testing the Design
Under Test (DUT). With all the efforts, the purpose is to gain confidence on maturity of DUT to ensures
first time silicon success that meets time to market needs of the industry. For any test-environment the
bottleneck, in achieving verification closer, is controllability and observability that is quality of stimulus to
unearth issues at early stage and coverage calculation. Software simulation, FPGA prototype, or
emulation, each method has its own limitations, be it test-time, ease of use, or cost of software, tools and
hardware-platform. Compared to software simulation, FPGA prototyping and emulation methods pose
greater challenges in quality stimulus generation and coverage calculation. Many researchers have
identified the problems of bug-detection / localization, but very few have touched the concept of quality
stimulus generation that leads to better functional coverage and thereby uncover hidden bugs in FPGA
prototype verification setup. This paper presents a novel approach to address above-mentioned issues by
embedding synthesizable active-agent and coverage collector into FPGA prototype. The proposed
architecture has been experimented for functional and stress testing of Universal Serial Bus (USB) Link
Training and Status State Machine (LTSSM) logic module as DUT in FPGA prototype. The proposed
solution is fully synthesizable and hence can be used in both software simulation as well as in prototype
system. The biggest advantage is plug and play nature of this active-agent component, that allows its
reusability in any USB3.0 LTSSM digital core.
KEYWORDS
Testing, Functional Coverage, Synthesizable Active Agent, Universal Serial Bus (USB), Link Training and
Status State Machine (LTSSM)
1. INTRODUCTION
In recent times the complexity of the VLSI designs has increased rapidly with advancements in
nanotechnology and gate capacity from million gates to billion gates in a chip. High-speed serial
communication interfaces such as USB and PCIe provides high data transfer capabilities which is
a need of modern era applications. In cycle accurate and time-based software simulations, the
testing-time becomes bottleneck while running iterative simulations to achieve functional
coverage and detect bugs in time, and overall it affects project schedule

42
The next level of verification with prototyping and emulation systems provide accelerated
verification platforms to run time-sensitive scenarios. For high-speed serial protocols, there are
two time-sensitive test areas; a) link-training and low power phase b) data-transfer phase.
Verification in FPGA prototyping system greatly help stressing out the DUT with heavy data
traffic stimulus and reduces the test-time for long and iterative simulations. The DUT can be
simulated by injecting heavy data-traffic with different types of data, and number of data-packets
with varying length to uncover more corner scenarios. The data transfer test-scenarios are driven
from higher-level in the prototyping system, where user has better controllability and
observability for closed-loop testing and hence proper functional coverage can be realized.
However, for link-training related test scenarios there is very limited access where generating
right stimulus and achieving functional coverage is a big challenge.
The USB3.0 LTSSM, many other standard protocols, has timers ranging from 10 µsec to 300
msec for link-training or low power state transitions. Stress testing and coverage for such huge
time sensitive scenarios is herculean task in software simulation, and its unreachable in standard
prototype system. This paper proposes architecture where time-sensitive link-training and low
power states logic is targeted with an embedded active-agent in prototype system that uses
advanced error injection. In a proposed architecture, the active agent controls the stimulus and
coverage collector provides observability for closed-loop functional testing of such time-sensitive
scenarios. The technique used in this architecture focuses on stress testing time sensitive LTSSM
transition arcs and mutation coverage where injected error is observed for predictable behaviour
of DUT. This approach is feasible for many such prototyping systems where lower layers of DUT
can be stress tested using synthesizable active-agent such as proposed in this paper.
2. BACKGROUND AND MOTIVATION OF RESEARCH
In a typical test-environment of prototyping system, traffic injection to DUT is done in two ways;
a) through a test-component b) using a standard real device.
For USB designs when the DUT is USB target device, the test-component at other end could be a
proprietary hardware from Electronic Design Automation (EDA) company, that mimic real USB-
host component. The DUT in such emulated system may run at slower operating clock and sees
the test-component as protocol compliant device at the other end. Such test-components provide
greater flexibility in generating desired stimulus. However, these high-end proprietary test-
components are costly, and interface requirements with DUT are complex. Moreover, if multiple
USB components exist in a system, multiple such test-components would be required to stress test
the system concurrently.
Prototyping system with a real device against a DUT is comparatively easy to build and cost
effective. Such test setup also helps early software development and testing with real components.
The advantage is that all the testing happens in real world and at-speed. The drawback is
controllability of stimulus as the other end of the DUT is actual device where it behaves as per the
protocol defined. In such prototyping system heavy data-traffic can be injected with upper layer
application control. For USB link, when the DUT is USB target device and other end of the link
is real USB-Host, host side application can inject various types of transfers and data traffic.
However, there is very minimum control over how the lower layer of the design is stimulated.
Lower level module for USB-target device is link layer logic with LTSSM. Also, negative testing
options are very limited. Now, here is where the opportunity lies for improvement in the
prototyping system that became motivation of our research work.
The architecture presented here provides controllability of stimulus with synthesizable active-
agent and observability with coverage collector for link-layer LTSSM of the USB3.0 target

43
device DUT. The prototype of the proposed architecture is operational and presents promising
results with better stimulus generation. The scope of improvement and future work is to reduce
resource overhead in the prototyping system to provide competitive edge to this approach. The
same approach is feasible for many such prototyping systems where lower layers of design can be
stress tested using synthesizable active-agent such as proposed in this paper.
3. RELATED WORK
Authors from academia and industry have researched in area of pre-silicon verification,
synthesizable test-components, controlled stimulus, issue detection, and coverage collection. All
the work relates to improving effectiveness of testing and its suitability to specific application.
F. Moraes et al. in the paper [1], “A generic FPGA emulation framework” perfectly highlights
challenges associated with software-simulation based testing and proposes generic emulation
framework to improve controllability and observability of design under verification (DUV). The
presented structure is combination of software and hardware components. It uses host-PC for
higher level control of stimuli and the tests are driven to DUV via standard interfaces like PCIe or
Ethernet.
S. S. Shankar et al. in paper [2], “Synthesizable Verification IP to Stress Test System On-Chip
Emulation and Prototyping Platforms”, authors have highlighted the challenges associated with
pre-silicon verification and proposed synthesizable verification architecture compliant with
Standard Co-Emulation Modelling Interface (SCE-MI) infrastructure through which the protocol
specific traffic is injected at industry standard interfaces. This approach is based on combination
of high-level test sequences and protocol specific transactors implemented with synthesizable
models. The paper presents the approach with USB2 logic with Transceiver Macrocell Interface
(UTMI) interface as design under test. This approach highly relies on directed testing and the
stimulus is controlled from higher layer test-sequences.
H. Krrikyan et al. in paper [3], “Prototyping system for USB3.0 link layer using synthesizable
assertions and partial reconfiguration”, presents problems related to bug detection/localization
and coverage calculation. The proposed solution is based on embedding synthesizable assertions
into prototype and the experiment was done on USB3.0 link layer FPGA prototype. The paper
focuses on observability aspect of verification with assertion-based testing in prototype system.
Synthesizable assertions allow its usability in both software-based simulation test-environment
and FPGA prototype /emulation platforms. The experiment utilizes FPGA partial configuration
feature for Xilinx architecture-based FPGA and tool set.
O. Amin et al. proposes an architecture in paper [4], “System Verilog Assertions Synthesis Based
Compiler”, helps assertion-based verification idea for prototyping systems. The non-synthesizable
code written as SystemVerilog Assertions (SVA) in software-simulation based test-environment
can be converted to synthesizable code that can be ported on FPGA. Again, this approach focuses
on observability part of DUT verification and is used to watch how the design performs in testing
setup.
A. Yehia et al. in paper [5], "Faster coverage closure: Runtime guidance of Constrained Random
stimuli by collected coverage” proposes closed-loop verification concept to improve the coverage
and make verification more efficient. This approach is targeted for software simulation based
functional testing where stimuli based on constrained random verification uses data from
coverage collector to optimize the testing run time. It shows how collected coverage can
dynamically guide random stimuli generators during simulation to generate uncovered scenarios

44
and avoid generating redundant ones. Similar concept is applied in our architecture for the
prototyping system.
4. PROPOSED ARCHITECTURE
Presented in Figure 1. is use case showing active agent placement in overall FPGA prototype
system. The prototyping system on Xilinx Kintex®-7 FPGA, constitutes USB3.0 Device
controller as DUT, the application layer of DUT is supported by standard mass storage device
driver running on local processor with Advance Microcontroller Bus Architecture (AMBA) -
Advanced eXtensible Interface (AXI) based system-bus. The link layer of DUT connects to
USB3.0 PHY via intel PHY Interface for PCIe and USB (PIPE). The other side of the USB link is
standard host with USB3.0 host port that generates stimulus while running real bulk-transfer
application. Building such prototyping system is easy and cost effective compared to protocol
specific high-end emulators or test-components. The objective of our prototyping system is to
address time sensitive stimulus generation and stress test the DUT. Here data intensive traffic is
injected from standard USB host and the application can inject desired stimulus. To address
stimulus generation for lower layer logic area, LTSSM in this case, the synthesizable active agent
plays a role.
The synthesizable active-agent design is based on Universal Serial Bus (USB) specification [6]
and PIPE [7] specification. The configuration space access of the USB device IP core is accessed
through AXI-Lite interface that follows AMBA-AXI specification [8]. The active agent intercepts
the PIPE interface signalling of USB3.0 link layer and injects error to create specific test scenario
that otherwise wouldn’t have occurred in normal prototype test-environment setup.
USB
PHY
USB3.0 Device IP Core (DUT)
LTSSM
Link Layer
Protocol
Layer
Application
Layer
System with
USB Device-
driver
PIPE interface
Xilinx Zynq FPGA platform
USB3cable
Standard
USB3.0 Host
Active
AgentConfig
Space
Figure 1. Active-agent in prototyping system for USB3 LTSSM
The link Layer interface with PHY is Intel PIPE interface having unidirectional data-flow that
allows adding pipe-line stages to insert our active-agent, without affecting specification timing
requirements. The active-agent is placed on the receive path at upstream port (USP) PIPE
interface where USB3.0 Device IP core connects with standard USB3.0-PHY.
This active agent is a plug-and-play component and work with any USB3.0 controller DUT with
just knowledge about the core’s LTSSM current state value that is available as part of debug port
in typical FPGA prototype system. The active-agent maintains scenario coverage information and
accordingly injects anomaly at available opportunity in an ongoing test. For simplicity This paper
focuses on specific portion of LTSSM and the same would be extended to cover more scenarios
for stress testing of complete LTSSM in an automated way.
In traditional simulation-based verification architecture there are passive agents acting as protocol
checker, scoreboard data collector, or functionality monitors that are vital components in overall

45
closed-loop verification. Also, call-back mechanism is used to inject error in the transfer by
manipulating the information and thus, creating specific directed scenario. Similarly, active-agent
manipulates information, within specification boundaries, at standard interface thereby deviating
from standard state-transition of DUT. It is obvious that this active-agent needs some extra
information, to be more effective, rather than purely taking the DUT as black-box. This
information helps find appropriate opportunity to inject variation in the received information from
other end of the USB-link and execute certain directed scenarios.
Active-agent architecture incorporates most part of decode logic in USB link layer that includes
descrambler, CRC, LFPS detection, and de-framing. It adds pipe-line stages to the receive
datapath to cover for decode latency and allow time to inject variation at right time. The active-
agent also implements configuration space with information on masking specific scenario
injection and collecting coverage statistics.
Major functional blocks are shown in the architectural diagram Figure 2. of active agent.
Figure 2. USB3.0-LTSSM Active Agent Architecture
4.1. DESCRAMBLER
Before the information can be identified to act upon by the agent it must be decoded at the receive
path of PIPE interface, and the first stage is descrambling. It’s based on free running linear
feedback shift register (LFSR) and if enabled the logic follows USB3.0 specification with
polynomial;
G(x) = x^16 + x^5 + x^4 + x^3 + 1
Control (K codes) symbols are bypassed and the LFSR is reset to FFFF at every COM symbol. If
the PHY contains this logic, it can be easily bypassed from the agent logic.
4.2. FRAMING DETECTION
After descrambling the decode logic performs detection of framing pattern and special symbols.
The sliding pattern match mechanism is used to detect symbol position in aligned or unaligned
packet. Framing detection is performed for Link Commands (LC), Header Packet (HP), Data
Packet (DPP, END/ENDB), and Ordered Sets (OS). The packets are checked for CRC and after
anomaly injection the CRC is recalculated to re-validate the packet.
4.3. LFPS DETECT
Low Frequency Periodic Signalling (LFPS) detection is performed as per USB3.0 specification
and is primarily used for low power mode exit and warm-reset. The details of normative LFPS

46
implementation can be found in the USB specification. The latency increases by few clock cycles
while the decode logic is active before it acts for scenario injection based on constraints. This
does not impact the functionality of the DUT neither it violates USB3.0 or PIPE specification.
4.4. CONFIGURATION SPACE
The active agent also has configuration space that is accessible through simple register space
access interface that is memory mapped in local processor. The configuration spaces access is
used to fetch coverage statistic and to enable or disable specific scenario injection logic and to
configure and modify constraints of scenario injection driver.
4.5. ERROR INJECTION AND PIPELINE STAGES
This block is part of pipeline stage and performs error injection in receive path with manipulated
information. The required information for manipulation and content update is received from other
modules. Scenario injection logic acts based on the information from coverage statistics that help
update the constraints of scenario injection logic for uncovered state transitions during future test
iterations. Coverage constraints applicability is restricted to state-transitions where occurrence of
transition conditions can be varied in time. Manipulating receive path information is tricky and
requires extra care while exercising this approach staying within boundaries of USB3.0
specification. The implementation is simple once the concept is understood and the efforts put in
developing such active-agent are worth as the results seen are promising for coverage closer on
prototype system.
Table 1. Xilinx-Kintex7 FPGA series resource utilization
Resource Slice LUTs Slice Registers RAMB36
USB-Device 21764 12766 16
Active agent
% of DUT
2840 810 0
13% 6% 0%
5. VERIFICATION AND RESULTS
As per USB3.0 specification there are total of 20 LTSSM state and sub-states for USP with more
than 85 possible transition arcs where 20 transitions are associated with ‘directed’ from upper
layer through warm-reset. From rest of the transition arcs, less than 20 transition arcs are usually
covered in normal prototype environment. There is scope of error injection, within specification
boundaries, to stress test the DUT and the active agent plays effective role. All the LTSSM states
and sub-state transitions are tracked and coverage information is collected at state entry event.
During the research work specific set of test conditions were targeted for this experiment.
A simple event of cable connect-disconnect can exercise more test-condition and increase the
DUT coverage. However, performing this connect-disconnect test manually and expecting it to
occur at specific point in time at all possible opportunities, is practically impossible in typical
FPGA prototype platform. With active agent this is made possible. USB device can be connected
or disconnected from host port at any time and cover all reachable scenarios on existing prototype
platform. At PIPE interface the connect-disconnect event is stimulated using relevant input
signals as listed in table 2.

47
By overriding inputs signals to DUT at PIPE interface device connect-disconnect event is tested
for each all LTSSM states and sub-states. The active-agent has helped significantly in uncovering
multiple corner conditions that did not get caught during normal testing on the prototype system.
Table 2. PIPE interface signals for active-agent
Signal Direction Description
RxDataValid Input Instruct the Link layer DUT logic to ignore the data
interface for one clock cycle. A value of one indicates the
upper layer logic will use the data, a value of zero
indicates it will not use the data.
RxValid Input Indicates symbol lock and valid data on RxData and
RxDataK
PhyStatus Input Used to communicate completion of several PHY
functions including stable PCLKafter Reset# de-assertion,
power management state transitions, rate change, and
receiver detection. When this signal transitions during
entry and exit from any PHY state where PCLK is not
provided, then the signalling is asynchronous.
RxElecIdle Input Indicates receiver detection of an electrical idle. While
de-asserted with the PHY in P0, P1, P2, or P3, indicates
detection of LFPS
RxStatus[2:0] Input Encodes receiver status and error codes for the received
data stream when receiving data.
3'b000 : Received data OK
3'b001 : 1 skip OS added
3'b010 : 1 skip OS removed
3’b011 : Receive detected
3'b100 : 8b10b decode error
3'b101 : EB overflow
3'b110 : EB underflow
3'b111 : Receiver disparity error
PowerPresent Input Indicates the presence of VBUS
Scenario injection is simpler and more effective while exercising negative conditions for state
transition. Let’s understand this with one of the core state of the ‘LTSSM: Polling’ state and sub-
states. Polling.LFPS is a sub-state designed to establish the PHY’s DC operating point for LFPS
operation, and to synchronize the operation between the two link partners after exiting from
Rx.Detect. This is also a sub-state for a port to identify itself based on various Polling.LFPS
signatures.
Figure 3. Shows polling sub-states of USB LTSSM. Let’s understand how the active agent
generates desired stimulus and help generate quality stimulus. In normal operation, with standard
prototype system, the Polling state moves straight forward through Polling.LFPS ->
Polling.RxEQ -> Polling.Active-> Polling.Configuration-> Polling.Idle-> U0. There are multiple
other functionally possible test conditions that are not reachable in standard prototype system.
When the scenario injection is enabled the active agent finds opportunity of injecting specific
condition when the LTSSM enters Polling.LFPS. So apart from its normal route that is
Polling.LFPS Polling.RxEQ, the LTSSM is stress tested for error conditions in Polling.LFPS
state and for transitions to SS.Disabled or Compliance. There is a provision to enable or disable
scenario injection by masking condition via configuration space access interface.

48
Rx.Detect
Polling.LFPS
Polling.RxEQ
Polling.Active
Polling.Configuration
Polling.Idle
Compliance
SS.Disabled
U0
Hot Reset
Loopback
Rx.Detect
Figure 3. USB3.0-LTSSM – Polling sub-states
Table 3. lists multiple functional testing scenarios for Polling.LFPS sub state in UpStream Port
component. It is evident from the list of possible scenarios that just state transition coverage is not
sufficient to gain confidence before claiming closer in pre-silicon verification with FPGA
prototype systems.
Table 3. Polling.LFPS Test scenarios and Active agent
The real benefit of active agent is not just covering all the state transition of DUT on FPGA
prototype, rather its ability to inject error without violating the specification to stress test specific
state of LTSSM.
Table 4. presents improved results with active-agent generated stimulus for the DUT.

49
Table 4. Polling.LFPS coverage
Transition arc Test Coverage
Current State Exit State # Default With Active agent
Polling.LFPS Polling.RxEQ 1 1 1
Polling.LFPS Polling.RxEQ 1a 0 1
Polling.LFPS Polling.RxEQ 1b 0 1
Polling.LFPS Compliance 2 0 1
Polling.LFPS SS.Disabled 3 0 1
Irrespective of what method is chosen for offline simulation, pre-silicon verification, or post-
silicon testing, important aspects of any verification idea are controllability, and observability. In
proposed method, this is covered to some extent with control of scenario injection and coverage
collector. However major concern with such non-trivial testing methodology for FPGA prototype
is trouble shooting and bug localization. To manage this concern, a simple debug mechanism was
created. Explained in brief; local memory was added, with configurable depth, that logs state
transitions and the contents rollover when the test is run for longer period beyond its storage
capacity. Interrupt is asserted to local processor for initiating the memory-read when entry to
Rx.Detect occurs for warm-reset and was not intended as per scenario injection. The content of
memory gives transaction log that helps understand state transitions. In our experiment we’ve
debug port from DUT that provides all important error flags due to which the state transitions
deviate from its expected route. This information helps in bug-localization.
6. CONCLUSION
The stress testing and coverage improvement using proposed architecture is demonstrated in this
paper with sample experiment. The proposed approach is very efficient in uncovering hidden
bugs without inviting much in hardware or software infrastructure resources. The experiment was
done on limited scenarios of LTSSM, however more scenarios would be added in future to
exercise further stress testing with varying state exit and error injection. This active agent is
reusable component and can be safely used with any USB3.0 USP DUT. The same methodology
can be applied to USB3.0 DSP, and for that matter, any standard protocol interface for the
purpose of stress testing and finding corner scenarios at earliest with less effort on FPGA
prototype systems.
ACKNOWLEDGEMENTS
The authors would like to thank Softnautics LLP. for their supports and providing FPGA
prototype platform along with all the associated infrastructure to test active-agent. The authors
would also like to thank Dr. D. J. Shah for his guidance.
REFERENCES
[1] F. Moraes et al., (2012) "A generic FPGA emulation framework", 19th IEEE International
Conference on Electronics, Circuits, and Systems, pp. 233-236.
[2] S. S. Shankar & J. S. Shankar, (2011) "Synthesizable verification IP to stress test system-on-chip
emulation and prototyping platforms", International Symposium on Integrated Circuits, Singapore.
pp. 609-612.
[3] H. Krrikyan, T. Hovhannisyan and S. Manukyan, (2015) "Prototyping system for USB3.0 link layer
using synthesizable assertions and partial reconfiguration", Computer Science and Information
Technologies (CSIT), pp. 19-22.

50
[4] O. Amin, Y. Ramzy, O. Ibrahem, A. Fouad, K. Mohamed and M. Abdelsalam, (2016) "System
Verilog Assertions Synthesis Based Compiler," 17th International Workshop on Microprocessor and
SOC Test and Verification (MTV), pp. 65-70.
[5] A. Yehia, (2013) "Faster coverage closure: Runtime guidance of Constrained Random stimuli by
collected" Saudi International Electronics, Communications and Photonics Conference, pp. 1-6.
[6] Universal Serial Bus Rev3.2 Specification, September 2017. Available: https://www.usb.org
[7] PHY Interface For the PCI Express, SATA, and USB3.1 Architectures Specification v5.0. Available:
https://www.intel.com
[8] AMBA AXI-4 Specification. Available: https://www.arm.com
[9] Abhishek Jain, Piyushkumar Gupta, Hima Gupta, Sachin Dhar, (2013) “Accelerating SystemVerilog
UVM based VIP to improve Methodology for Verification of Image Processing Designs using HW
emulator” International Journal of VLSI design & Communication Systems (VLSICS), Vol. 4, No.6,
pp. 13-25.
[10] Sateesh Kumar H.C., Sayantam Sankar, Satish S Bhairannawar, Raja K.B. Venugopal K.R., (2015)
“FPGA Implementation of Moving Object and Face Detection using Adaptive Threshold”
International Journal of VLSI design & Communication Systems (VLSICS), Vol. 6, No.5, pp. 15-35.
[11] N. Mohamad, C. Y. Ooi, N. Ismail and J. Teh, (2016) "SVA checker generator for FPGA-based
verification platform" IEEE International Symposium on Circuits and Systems (ISCAS), pp. 1750-
1753.
[12] B. Vermeulen and S. K. Goel, (2002) "Design for debug: catching design errors in digital chips" IEEE
Design & Test of Computers, vol. 19, no. 3, pp. 35-43.
[13] A. M. Gharehbaghi, M. Fujita, (2011) “Transaction-based debugging of system-on- chips with
patterns”, Proc.of IEEE International Symposium on Quality Electronic Design, pp. 1 -6.
[14] C. H. Chen, J. C. Ju and I. J. Huang, (2010) "A synthesizable AXI protocol checker for SoC
integration," International SoC Design Conference, pp. 103-106.
[15] Tan J, Fresse V, Rousseau F, (2011) “Generation of emulation platforms for NoC exploration on
FPGA” 22nd IEEE International Symposium on Rapid System Prototyping (RSP), pp. 186-192.
AUTHORS
Dipakkumar Modi received the BE degree in instrumentation and control engineering
from L. D. College of Engineering, Ahmedabad, Gujarat in 2001. He is currently
working with Softnautics LLP as technical director IP group. He has worked in VLSI
industry for more than 16 years, his research interests are in field of SoC design,
hardware accelerated verification, and emulation.
Prof. Usha Mehta has done her Ph. D. in the area “Testing of VLSI Design” and is
currently working as Professor at Nirma University, Ahmedabad. She has more than
22 years of academic and industrial experience. She has one patent, has published a
book and more than 50 research papers in international journals and conferences.
She is a senior member of IEEE and currently holds the position of chair, IEEE WIE
Affinity group, Gujarat section. Her research interests are in testing VLSI Design,
Digital VLSI Design, and programmable Logic.

COVERAGE DRIVEN FUNCTIONAL TESTING ARCHITECTURE FOR PROTOTYPING SYSTEM USING SYNTHESIZABLE ACTIVE AGENT

More Related Content

What's hot (18)

Similar to COVERAGE DRIVEN FUNCTIONAL TESTING ARCHITECTURE FOR PROTOTYPING SYSTEM USING SYNTHESIZABLE ACTIVE AGENT (20)

Recently uploaded (20)

COVERAGE DRIVEN FUNCTIONAL TESTING ARCHITECTURE FOR PROTOTYPING SYSTEM USING SYNTHESIZABLE ACTIVE AGENT