Signal Processing - March 2017 - 80

for production. Cost is an important consideration for automotive manufacturers. Low cost and high processing performance of the ECUs are needed for mass production. Beyond
the intensive research on various efficient and robust DSP
algorithms, microprocessor manufacturers have focused their
efforts on building computationally powerful automotive oriented microcontrollers with the goal of reducing overall costs
by utilizing various technologies. For example, the Motorola
CPU16 has some very basic DSP instructions, and Freescale
Semiconductor's MPC 5554 Copperhead microcontroller and
MPC5674F Power Architecture microcontroller have a powerful built-in signal processing engine (SPE), which is able to
accelerate signal processing with its vector processing capability that allows two arithmetic operations to be completed
in parallel. The SPE works with either fixed-point or floating-point operations and runs especially efficiently for DSP
algorithms because the specially designed DSP functionality
can accelerate signal processing computations, such as those
associated with FIR and IIR filters and DFTs.
According to a study of software solutions for knock detection by [50], more than 50% reduction of ECU computational
time can be achieved using the Copperhead microcontroller's
SPE functions compared to the conventional C code implementation, and, therefore, the use of the SPE feature makes an ECU
capable of running more applications or extending the life of
one generation of ECU. The microcontroller MPC5674F further provides a special feature to perform frequency-filteringbased knock detection without additional external components.
This avoids use of separate knock detection application-specific
integrated circuits and other components and therefore reduces
overall system cost. The aforementioned example represents a
trend of comprehensive on-chip integration and virtual sensing
capabilities provided in high-performance automotive microcontrollers to help automotive developers accelerate next-generation engine control design and simultaneously achieve both
performance and cost benefits.
From the viewpoint of automotive manufacturers, there
are many vehicle models to be produced every year. For
each feature or application, they wish to have a unified algorithm that can be used in all models of vehicles through
proper configuration and minimal calibration complexity. The development of DSP applications needs to consider
these requirements. Because of the periodic characteristics
of engine signals, DSP has the potential to reduce calibration effort. Some interested frequencies are constant when
the systems are designed to execute in the crankshaft angle
domain and use signals that are sampled in the same domain.
This significantly simplifies the designed DSP systems and
reduces the calibration efforts. As in the previously given
examples of misfire detection and individual cylinder fuel-air
ratio imbalance detection, frequency-domain DSP methods
have a distinct advantage.
Based on the aforementioned criteria and considerations,
it is not difficult to understand why the most popular DSP
methods for engine control and OBD are still frequencyfiltering techniques-particularly because of the techniques'
80

maturity and relatively low implementation cost. The FFT or
DFT methods are becoming feasible when they are implemented in the ECUs with high-performance processors or
ones with additional dedicated DSP features or special DSP
hardware systems. It is still not practical for many advanced
signal processing methods, such as time-frequency analysis
and wavelet transforms, to be deployed for onboard real-time
processing in today's ECU environment because of their
high computational demands. However, these techniques
can possibly be used in off-board analyses during production development stages. We view these challenges as a great
opportunity for signal processing to play a critical role in the
automotive industry. As the title of a January 2014 "Special
Reports" column in IEEE Signal Processing Magazine [51]
indicated, the automotive industry is a key contributor to the
success of the DSP sector. There are many opportunities for
DSP applications in automobiles, including infotainment,
telematics, advanced driver assistance systems, and autonomous driving. DSP applications in powertrain controls seem
not as visible to most automobile consumers as the aforementioned application areas for making smarter cars. However,
one trend is evident: advanced DSP techniques will be essential in various engine control and OBD applications, as well
as in other powertrain controls, especially as ECUs continue
to evolve to have more processing power available for more
mature and sophisticated DSP algorithms that will improve
vehicle fuel economy and emission performance for making
greener cars.

Acknowledgments
We would like to thank Dr. Wade Trappe for his many valuable editing suggestions that were adopted in the manuscript.

Authors
Zhijian James Wu (zhijian.wu@gm.com) received his
M.S.E. and Ph.D. degrees from the University of Michigan,
Ann Arbor, in 1990 and 1991, respectively, and holds bachelor's and master's of engineering degrees in electronic engineering from Shanghai Jiao Tong University, China (1982 and
1984, respectively). He has more than 20 years of R&D experience in engine controls and onboard diagnostics. He joined
General Motors in 2014. He was previously a senior technical
specialist with Fiat Chrysler Automobiles. He has more than
20 patents and technical publications and is a Senior Member
of the IEEE and a member of the Society of Automotive
Engineers International.
Sanjeev M. Naik (sanjeev.m.naik@gm.com) holds a
bachelor's degree from IIT Bombay (1986), an M.S.E.E.
degree from the University of Michigan, Ann Arbor (1988),
and a Ph.D. degree from the University of Illinois, Urbana-
Champaign (1992), all in electrical engineering, and an
M.B.A. degree from the University of Michigan, Ann Arbor
(2000). He is the engineering group manager for controls
advanced engineering at General Motors (GM). Prior to GM,
he worked at Cummins Engine Company. He has several
publications and 50 patents. His technical interests are in the

IEEE SIgnal ProcESSIng MagazInE

March 2017

Table of Contents for the Digital Edition of Signal Processing - March 2017