This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication.
Novel Battery-Less Wireless Sensor for Traffic Flow Measurement
K. Vijayaraghavan 1 and R. Rajamani 1, *
Abstract— This paper presents a novel battery-less
wireless sensor that can be embedded in the road and used
for measurement of traffic flow rate. Compared to existing
inductive loop based traffic sensors, the new sensor is
expected to provide increased reliability, easy installation
and low cost. The sensor requires no external power source
and has zero idle power loss. Energy to power this sensor is
harvested from the short duration vibrations that results
when an automobile passes over the sensor. Since much of
the earlier work in literature on vibration energy harvesting
has focused on continuous sources of vibration, this paper
focuses on short duration vibrations and on developing low
power control algorithms that can be implemented on the
sensor using an analog circuit. To this effect this paper
develops and compares three control algorithms “Fixed
threshold switching”, “Maximum voltage switching” and
“Switched inductor.” The “Switched inductor” algorithm is
shown to be the most effective at maximizing harvested
energy. Experimental results are presented and validate the
fact that adequate energy can be harvested from each
passing vehicle to enable successful wireless transmission of
traffic data.
Index Terms—Traffic sensor, Batteryless, Wireless,
Energy harvesting.
I. REVIEW OF CURRENT TRAFFIC SENSORS
ransportation agencies all around the country monitor
traffic flow rates on most major highways using
inductive loop detectors (ILDs). The Minnesota
Department of Transportation (MnDOT), for example,
monitors the flow rates at over 6000 points in the
Minneapolis/St. Paul metro area using such ILDs. An ILD
consists of a big loop of metallic coil buried in the lane.
This loop is connected to a road-side station which
provides power to the loop and processes the information
obtained from it to determine if a vehicle is passing over
the sensor. The traffic flow rate information from such
sensors is used to control ramp meters, identify
congestion points, detect incidents and for a number of
other applications.
Inductive loop detectors exhibit high accuracy in
detecting vehicles ([29]). Hence, despite various new non-
intrusive technologies for detecting vehicles such as
image processing based detectors ([6],[9],[15],[17]) and
systems based on audio processing ([2], [3]), inductive
loop detectors remain the most widely used technology.
This work was supported in part by a research grant from the ITS
Institute, University of Minnesota.
1
K. Vijayaraghavan and R. Rajamani are with the Department of
Mechanical Engineering, University of Minnesota, Minneapolis, MN
55455, USA
* rajamani@me.umn.edu, tel : (612) 626-7961, fax : (612) 625-4344.
Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
Despite their popularity, ILDs are not perfect. They
are prone to breakdowns [1], and can only measure traffic
flow rate, not vehicle speed or vehicle classification.
There have been attempts to use estimation algorithms to
do vehicle classification using ILDs and identification
techniques such as fuzzy logic and artificial neural
networks (ANNs) ([13], [27]).
Despite many improvements, the installation of the
ILD involves cutting a large section of the roadway in
each lane and therefore causes considerable traffic
disruption. Owing to its operating principle, the ILD
needs to be continuously powered resulting in
considerable idle power loss. For example, an ILD needs
to be continuously powered at night, even if there is very
little traffic flowing on the highway.
II. NEW BATTERY-LESS WIRELESS TRAFFIC SENSORS
A. Overview
The researchers in this paper have developed a novel
battery-less wireless traffic sensor which requires zero
external energy. The sensor is completely autonomous
and can be embedded in the lane without the need for
power or data cables. In the absence of any automobile,
the sensor stays turned off, consuming no power. Thus,
the sensor has zero idle power loss. When an automobile
passes over the sensor, the sensor turns on and a RF pulse
is transmitted wirelessly to a receiving station. The
receiving station can be as much as 500 feet away from
the sensor. The sensor requires no external power source
as it is powered by harvesting all its required energy from
vibrations that result when a vehicle passes over it.
Further this sensor is significantly smaller compared to an
ILD and can be installed with lower traffic disruptions.
This is especially true because the sensor does not need a
power source and power lines do not need to be run to the
sensor. This new sensor, like the ILD, does not use
complex image processing or audio processing techniques
and hence is expected to provide the same level of high
reliability. Owing to the battery-less and wireless nature
of the sensor, low maintenance costs can also be
expected. Further the sensor can measure the number of
axles (vehicle classification) and can be modified to also
measure the weight of the passing vehicle in addition to
the traffic flow rate. It is also possible to configure several
sensors to transmit to a single receiving station.
B. Principle
The proposed sensor is based on the principle of
vibration energy harvesting (VEH) to enable wireless
transmission of signals. Reference [26] provides a good
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

review of many of these VEH techniques. Some of the be made bigger than the size of the sensor and a flexible
earlier work has also focused on developing control sealant used in the gap around the sensor. This will
algorithms to optimize the amount of energy harvested ensure that expansion and contraction of the sensor due to
([16], [22]). However, the VEH techniques in literature temperature variations can be accommodated. This
focus predominantly on harvesting energy from a accommodation is necessary in order to prevent
continuous source of vibration. When a vehicle passes mechanical failure and also to prevent unloaded
over the sensor, the resulting mechanical vibrations are of mechanical displacement of the sensor beam in the
short duration. Hence, although the concept of VEH is not absence of vehicle loads on it. A photograph of the
new, it has never before been used to power a traffic sensor is shown in figure 1(b).
sensor. Further the optimal algorithms that have been
D. Controller constraint
proposed earlier cannot be implemented in a stand-alone
sensor as they require an external control input (and In this paper, we develop a controller to extract
possibly an external power source). Hence new energy harvested from short duration vibration loads
algorithms have been developed and implemented in this similar to impact loading. The control systems that
paper. enables Vibration Energy Harvesting (VEH) itself needs
to be completely powered from the energy that is
C. Hardware harvested. The design has been restricted to controllers
The proposed sensor, as shown in figure 1(a), that can be implemented using simple onboard analog
consists of a structure made up of three beams. Beam #1 electronics.
is 6ft long (1.8 m) and is called the “main beam”. The
sensor will be embedded in the road such that III. SYSTEM MODEL FOR CONTROL
automobiles pas directly over this main beam. The load
A. Mechanical sub-system
acting on the main beam is transmitted to two “support
beams” located at either ends. The support beams have a The VEH sensor consists of a mechanical sub-system
gross length of 10” (254 mm) (with an effective length of and an electrical sub-system. When the piezo is open (the
8” or 200 mm), and the ends of each support beam are terminals are unconnected), the mechanical sub-system is
rigidly fixed. A total of eight piezo elements (four piezos simply modeled as a vibrating beam structure. For a
for each of the support beams) are bonded to the support simple beam structure in vibration, the various modes of
beams. The location of the piezos were chosen as close to vibration can be calculated using equation (1). More
the ends of the support beam as possible. The piezos are complicated structures require a FEM model solution.
connected electrically in parallel. ANSYS simulations  2  x, t    x, t   4  x, t 
A b  EI  0 (1)
reveal that the average of the strain over the area of all the t 2
t x 4
piezos depends only on the total load acting on the main where
beam and is independent of the lateral location of the load A is the linear density of the beam
on the main beam. Hence this configuration was chosen b is the damping in the beam
in order to make the average voltage developed by the
EI is the stiffness of the beam
piezos independent of the location of the load. It should
be further noted that the speed of the passing vehicle can
be measured by measuring the time difference in the 4.25
loading between two consecutive sensors placed a short
4.2
longitudinal distance apart. 5 4.15

4.1
4
4.05
Main beam
3
102.1035 2. 37
2.34 2.35 2.36 2.37

Support beam
10 10 1010
2

1
(a)

-1

-2

-3
-60

1 2 3 4
10 1 10 2 10 3 10
(b) 10 10 10 104
Figure 1: Sensor schematic and photograph Frequency (rad/sec)
Regarding installation of the sensor in the road, the sensor Figure 2: Bode magnitude plot of impulse response of the sensor
can be placed in a slot made in the road. The slot should

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
It is possible to construct a reduced order model of
the system, using only the most dominant vibration mode.
To determine this dominant frequency, the impact
response of the sensor was experimentally obtained.
Using the impact response data shown here in figure 2,
the frequency of the first mode of vibration is determined
to be 38 Hz .
For purposes of developing the control system, this
dominant frequency is used to model the mechanical
system as a spring mass damper system
F F 2k 2m n 2
2
u  2 n u   n u  (2)
m Hence, it can be seen that if the resonant frequency
It must be noted that the system model deviates from
equation (2) when the piezo sources current to any
connected electronics, and voltage difference across the
piezo decreases. This change in piezo voltage will induce
a proportional force, Felec on the mechanical system and
hence equation (2) is modified to
2 F F (3)
u  2 n u   n u   elec
m m
where Felec is the force applied on the mechanical
system by the piezo and is proportional the voltage across
the capacitor of the piezo VCpiezo .
The strain,  , at the location of the piezo can be
represented in terms of a length scale l as
 u l (4)
From equations (3) and (4), the mechanical sub-
system dynamics can be rewritten in terms of strain as
2 2 2
  2 n  n    n F   elec n VCpiezo (5)
where  is the amount of static strain produced for a unit
applied force and  elec is the amount of static strain
produced for a unit change in piezo voltage.
In order to determine the value of  , let us consider a
unit static load acting at the center of the main beam of
the traffic sensor. This load would results in an average
load of 1 2 units being transmitted to each support beam.
It can be inferred from Ketchum et. al. ([10]) that the
resulting bending moment at the location of the piezo
( x0 ) is given by
 Ls x s  1
M piezo    0     (6)
 8 2  2
For the support beam, Ls  0.2m , x 0s  0.01875m ,
b s  0.025m , h s  0.00625m and E s  200GPa . Hence
I s  5.0863 10 -10 mm 4 and
 s  2.4  10 7 N 1 (7)
Thus
  2.4  10 7 N 1 (8)
The value of  elec is determined in section III.C.A
quick brief analysis is presented here on parameter
variations and how these variations would affect beam
displacement. The ratio of beam vibration amplitude X
to load magnitude F is given as a function of frequency
 by
Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
Xk 1 (9)

F 
2 2
2    
1     2 
  2   
 n   n 
Since the load in this case is approximately that of an
impulse, the vibration of the beam will largely occur at
the resonant frequency    n . With    n , the ratio of
vibration amplitude to force becomes
X 1 1 (10)
 
 n changes by 20%, the vibration amplitude can decrease
to 0.7 times its nominal value or increase to 1.56 times the
nominal value. Likewise, if the damping ratio changes by
20%, the vibration amplitude can decrease to 0.85 times
the nominal amplitude or increase to 1.25 times its
nominal value. In the design of the energy harvesting
system, it is important to ensure that the power needed by
the wireless transceiver and associated electronics is a
factor of 2 less than the power provided by the nominal
system design. This would ensure that the system would
continue to function reliably even if there were changes in
mechanical parameters in the sensor.
B. Electrical model
At low frequencies, the piezo electric material which
is the critical part of the VEH system, is modeled as a
voltage source in series with a capacitance using
equations (11-13) ([5], [20]). A more sophisticated model
can be found in Weinbert et. al.([32]).
V piezo  Vstrain  VCpiezo (11)
1 (12)
C piezo 
VCpiezo  i p dt
(13)
V strain  g
The piezo electric material consists of T107-A4E-602
piezo sheets that were purchased from Piezo Systems, Inc.
The piezo sheets were cut to rectangular strips of size
37.5mm  25mm . The piezos were bonded to the sensor
such that they are subject to strain in the “1” direction.
Hence the “31” parameters of the piezo are used in the
calculation. For A4E piezo, using the modulus of
elasticity at constant electric field E and thickness  , the
open circuit voltage per unit strain is
g  1.4623 10 5 V (14)
The equivalent capacitance was estimated to be
C piezo  0.14333 F .
In order to calculate the overall electrical dynamics,
other components making up the electrical subsystem
need to be considered. The electrical system shown in
figure 3 also contains a bridge rectifier (denoted by
“Diode Bridge”) connected to the piezo element. The
diode bridge is connected to a storage capacitor C s ,
which in turn is connected to transmitter via a load switch
(SWL). Each diode making up the bridge is modeled by a
piecewise linear model ([24])
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

 V piezo  VCs  2Vd 6 EVpiezo

 signV piezo   elec  d 31VCpiezo (21)

 Rd E shs
 (15) From equation (5) we can deduce that for a static
ip   if V piezo  VCs  2Vd
 VCpiezo
  elec   elecVCpiezo (22)
 0 otherwise
Hence
The transmitter is modeled as a load resistor RL in
6 EVpiezo (23)
the circuit. The control circuit is not included in the  elec  d 31
E shs
dynamics owing to its extremely small current
Thus the coupling coefficient is
consumption. The piezo current i p can then be written as
 elec  6.0192  10 -8 V 1 (24)
equation (15) and the dynamics of the capacitor voltage
can be given by equation (16) as
IV. PROPOSED CONTROL SYSTEMS
C sVCs  1 R L VCs  i p if SW L is closed (16) Of the total energy generated in the piezo, only the

C sVCs  i p if SW L is open fraction transferred to the storage capacitor C s is
where Vd is the forward voltage drop across the diode available to drive the electrical load. The peak voltage
(normally about 1.1 V). across the storage capacitor, VCsmax , is a measure of this
energy. The peak power at the load, given by VCsmax  RL ,
2

is also a function of VCsmax . Hence in the following sub-

Figure 3: Energy harvesting circuit
C. Electro-mechanical coupling
The VEH system consists of piezo electric elements
bonded to the support beams as shown in figure 1. The
piezos are bonded to the top and bottom surfaces of the
support beam. When a vehicle passes over the sensor, the
piezo experiences a strain from the loading and thus
generates a voltage. When the piezo sources current to the
circuit, VCpiezo increases and V piezo decreases. Since the
sections, the available storage capacitor voltage, VCsmax , is
determined for each of three different control algorithms.
For the purpose of simulation, we consider the sensor
model with the parameter values calculated in section III.
The input load resulting from the front and the rear tires,
is modeled as two short duration pulses with a magnitude
of 3937.5 N as shown in figure 4. The tire passing over
the sensor would damp out the mechanical vibrations in
the sensor. Hence the impulse response in figure 2 could
not be directly used to find the damping ratio in section
III. Instead a value of   0.7 was found to be
appropriate for the simulations. The components making
up the electrical system are modeled by an equivalent
1 K load resistor.
4000
3500

piezo is bonded to the mechanical system, this decrease in 3000

piezo voltage translates to a steady state stress 2500
 Cpiezo  EVpiezo d 31 VCpiezo  (17)
Force(N)

2000
The force developed by the piezo as a result of this
stress is given by 1500
piezo
FVCpiezo  b s   Cpiezo (18)
1000
The pair of piezos that were considered, bonded to
500
the top and bottom of the support beam, are separated by
the height of the support beam h s . The moment 0
0 1 2 3 4 5
developed by them is Time(s)
piezo
M VCpiezo  b s  h s  Cpiezo (19) Figure 4: Force Input used for simulation
This moment acting on the mechanical system would A. Fixed Threshold Switching
result in a static strain This algorithm is an adaptation of traditional
piezo
M VCpiezo h s 2 (20) algorithms used for harvesting energy from sustained

EsI s oscillations. In this algorithm, the load is connected to C s ,
Noting that I s  1 12 b s h s  we get
3
by setting the control input to logic high (1) when the

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

value of VCs crosses a predetermined high threshold Vhigh . 3

The control is turned off (0), if VCs falls below a low

2.5
threshold Vlow . The control signal to SWL can be given by

Load Current (mA)

the control law state transition diagram shown in figure 5. 2
VCs  Vhigh
1.5

Vcontrol = 0 Vcontrol = 1 1

0.5
VCs  Vhigh VCs  Vlow
VCs  Vlow
0
Figure 5: State transition diagram for “Fixed Threshold Switching” 0 1 2 3 4 5
Time(s)
Once the load switch SWL is closed, the voltage Figure 7: Load current for 1K load with “Fixed Threshold
across the storage capacitor VCs does not rise any further. switching” algorithm
The maximum value VCsmax is equal to the on-threshold

Cumulatinve energy transfered to the load (uJ)

200

Vhigh . Thus
  C piezo  
min 

  V strain

max
 2Vd , Vhigh    150


  C s  C piezo   (25)
VCs  
max
if V strain  2Vd
max
100


 0 otherwise

50
For a sufficiently large Vstrain ,
max

VCsmax  Vhigh (26)

The fixed threshold switching is the simplest 0
0 1 2 3 4 5
algorithm, and would serve as a baseline for evaluating Time(s)
the performance of other control algorithms. Simulation Figure 8: Cumulative energy transferred to a 1K Load with “Fixed
results obtained using this baseline control law are shown Threshold switching algorithm”
below. The voltage in the storage capacitor for For benchmark numbers on power required for wireless
Vhigh  2.75V and Vlow  1.5V is seen in figure 6. Figure 7 transmission, the Linx transceiver TXM-418-LC requires
shows the current through the 1K (1 Kohm) load 2.5 mW while the CC2500 transceiver from Texas
resistance and figure 8 shows the cumulative energy Instruments requires 300 micro-watts for its operation.
transferred to the resistor. The peak power is 7.56 mW Hence both of these transceivers can be reliably powered
and the total energy transferred to the load is 30 µJ. from the energy harvested in this traffic sensor.
B. Max Voltage Switching
10
In this algorithm, the load is connected to the storage
capacitor C s , when the voltage VCs reaches a maximum
Storage capacitor voltage (V)

8
value. The control SWL is turned off, if VCs falls below
6 the off-threshold Vlow . The control law can be given by the
state transition diagram shown in figure 9.
4 VCs is max imum

Vcontrol = 0 Vcontrol = 1
0
0 1 2 3 4 5
Time(s)
|VCs | VCs ≥ Vlow
Figure 6: Voltage Output for “Fixed Threshold switching” or VCs < Vlow
algorithm
VCs < Vlow
Figure 9: State transition diagram for “Max Voltage switching”

The occurrence of maximum can be determined using

analog electronics. For instance, the max-detector can be

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

realized using a high pass RC filter given by equation threshold. If VCs 0   VCsi and VCpiezo 0   VCpiezo
i
, equation
(27). A maximum is declared when the output of (28) is modified to
this filter falls below a threshold. The value of this
Veffective  Vstrain
max
 VCpiezo
i
 VCsi  2Vd (30a)
threshold is small and determines how close to zero the
derivative must become for the voltage to be recognized   C piezo 
    Veffective  VCs
i
as maximum.   C s  C piezo 
RCs max 
 (30b)
V filter  VCs (27) VCs  if Veffective  0
RCs  1 

As shown in Appendix A, if the displacement of the V i
 Cs otherwise
beam has only one extremum value u max (with a
corresponding V strain  V strain
max
), then VCsmax can be calculated
 C  sign V max  V i
 s strain 
Cpiezo 

V
 i
 VCpiezo
 C s  C piezo  effective
using max  
VCpiezo  if Veffective  0
 C piezo 
   V strain
 C s  C piezo 
max

 2V d if V strain
max
 2V d  

 Vi
VCs  
max
 Cpiezo otherwise
 (30c)
 0 otherwise For the particular load acting on the sensor
(28a) (consisting of four extremums), VCsmax is approximately
 Cs  four times the value of VCsmax obtained from just the first
    Vstrain  2Vd
max

  C s  C piezo 
pulse. From figure 11, VCsmax is 3.9V. Figure 12 shows the
 current through the 1K (1 Kohm ) load resistance and
max
VCpiezo  max
if Vstrain  2Vd figure 13 shows the cumulative energy transferred to the
 resistor.
 10
0 otherwise

Storage capacitor voltage (V)

(28b) 8
Thus for a sufficiently large strain voltage, the
difference between Vstrain
max
and 2  V d is distributed 6
between C s and C piezo in the inverse ratio of their
capacitance and 4
 C piezo 
VCsmax  
C C
  V strain

max
 2V d (29)   2
 s piezo 
Figure 10 shows simulation results for the max
voltage algorithm. 0
0 1 2 3 4 5
10 Time(s)
Figure 11: Voltage Output for modified “Max Voltage switching
algorithm”
Storage capacitor voltage (V)

8
4

6 3.5

3
4
Load Current (mA)

2.5

2 2

1.5
0
30 1
4 2
5 1
Time(s)
Figure 10: Voltage Output for “Max Voltage switching algorithm” 0.5

Since we have apriori knowledge about the nature of 0

0 1 2 3 4 5
the loading, it is possible to modify the switch SWL to Time(s)
turn on only at the end of the second pulse, and not turn Figure 12: Load current for 1K load with modified “Max Voltage
on after the first pulse. One way of achieving this would switching algorithm”
involve checking for a maximum larger than a low

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

Cumulatinve energy transfered to the load (uJ)

200 VCs is maximum

150 Vcontrol = 0 Vcontrol = 1

VCs | VCs ≥ Vlow

100 VCs < Vlow
or
VCs < Vlow
Figure 15: State transition diagram for SWL (“Switched Inductor”)
50
If switch SWP is closed at first local extremum of the
displacement of the beam, the voltages across capacitors
0 C s and C piezo at the end of first half-LC oscillation ( VCs1
0 1 2 3 4 5
Time(s) and VCpiezo
1
respectively) are given by
Figure 13: Cumulative energy transferred to a 1K Load with
  C piezo 
modified “Max Voltage switching algorithm”

 C  C piezo 

  V strain
max

 2V d if V strain
max
 2V d
Since the storage capacitor is allowed to charge to a VCs1   s 
higher voltage, this algorithm will deliver a larger peak 
power in comparison to the “Fixed Threshold switching”.  0 otherwise
Hence this algorithm is always more efficient at (31a)
harvesting vibration energy than the simple fixed   C s 
threshold algorithm described in the previous subsection.  
  V strain

max
 
 2V d  sign V strain
max

Indeed, the value of peak power (15.4 mW) and the  C s  C piezo 

amount of total energy transferred to the load (70 µJ) are 1
VCpiezo  max
if V strain  2V d
larger than those that were obtained for the “Fixed 
Threshold Switching” algorithm. 
 0 otherwise

C. Switched Inductor
(31b)
This section proposes a third algorithm that would where
further enhance VCsmax . This algorithm uses the circuit
  1  exp    1 2 , 1    2 (32a)
shown in figure 16. The new circuit uses an inductor ( L )
and an additional switch SWP in addition to the 1 C (32b)
  Rtotal
components shown in figure 3. The voltage drop across L 2 L
is given by VL . SWP is turned on when V piezo reaches a C pCs (32c)
C
maximum and SWL is turned on when VCs reaches a C p  Cs

maximum. The switches SWP and SWL are turned off

when the respective voltages V piezo  V L and VCs drops This equation has been derived in appendix B. Thus when
max
V strain  2V d
below an off-threshold Vlow . As discussed in the previous
 C piezo 
subsection, the occurrence of maximum can be VCsmax  
C C 

  Vstrain
max
 2Vd  (33)
determined using analog electronics. The control law for  s piezo 
SWP is given by the state transition diagram shown in
figure 14 and the control law for SWL is given by the state
transition diagram state transition diagram shown in
figure 15.
|Vpiezo| is maximum
VCs
Cs
Vcontrol = 0 Vcontrol = 1

|Vpiezo| |VL + Vpiezo|

or |VL + Vpiezo| < Vlow
≥ Vlow
|VL + Vpiezo| < Vlow
Figure 14: State transition diagram for SWP (“Switched Inductor” ) Figure 16: Energy harvesting with Inductor

By comparing equations (33) with (29), we notice

that VCsmax is greater by a factor of at least  , this factor

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

being equal to the peak overshoot of the LCR circuit. The 10

“Switched Inductor” algorithm yields a higher VCsmax due
to the presence of the inductor. In the absence of the diode

Storage capacitor voltage (V)

8
bridge rectifier, the second order dynamics of the LCR
circuit will exhibit an oscillatory behavior for an
6
extremely small time period and the dynamics would
eventually converge to its steady state value. The bridge
rectifier in the circuit would however clamp VCs to the 4

peak overshoot voltage of the transient response, resulting

in higher available voltage for the storage capacitor. 2
Using Rtotal  327.2 (this is the estimated total
resistance in the circuit) in equations (31), we get 0
0 1 2 3 4 5
1
VCs  1.25 and VCpiezo
1
 187.42 . Now VCpiezo
1
 V strain
max
Time(s)
 VCs1  2Vd . Hence the piezo would drive the circuit in the Figure 18: Voltage Output for modified “Switched Inductor”
algorithm
reverse direction. VCs2  0.27 and VCpiezo 2
 40.53 .
7
Hence VCsmax  VCs2  1.52 . The estimated value for VCs is
seen to be in close agreement with the simulation result 6
from figure 17.
5

Load Current (mA)

10
4
Storage capacitor voltage (V)

8
3

6 2

1
4
0
0 1 2 3 4 5
2 Time(s)
Figure 19: Load current for 1K load with modified “Switched
Inductor” algorithm
0
0 1 2 3 4 5
Time(s)
Cumulatinve energy transfered to the load (uJ)

Figure 17: Voltage Output for “Switched Inductor” algorithm 200

As mentioned in the earlier section, owing to apriori

knowledge about the nature of the loading, it is possible 150
to modify the SWL to turn on only at the end of the
second load pulse, instead of turning on after the first
pulse. This allows greater voltage to build up across the 100
storage capacitor. For the particular load acting on the
sensor (consisting of four extremums), VCsmax is
approximately four times the VCsmax for first extremum 50

(6.04V). When compared to the modified “Max voltage

switch” controller, it is seen from figure 11 and figure 18
that the available voltage has increased by a factor of over 0
0 1 2 3 4 5
1.5. Subsequently from figure 20, the peak power is Time(s)
obtained to be 36.3 mW (an increase of 136%) and the Figure 20: Cumulative energy transferred to a 1K Load with
total energy transferred to the load equals 188 µJ (an modified “Switched Inductor”
increase of 169%). D. Comparison of the three algorithms for first
extremum

Algorithm Approximate V Csmax

Fixed Threshold Switching Vhigh

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

 4  C piezo  used in the switching circuit constrains the off-threshold

modified Max Voltage
Switching

C C 

  Vstrain
max
 2Vd  to 1.75V. As a result, SWL turns on when the capacitor
 s piezo  voltage VCs reaches 2.75V and turns off when VCs falls
 4  C piezo 
modified Switched
Inductor

C C 

  Vstrain
max
 2Vd  below 1.75V.
The electronic circuits were modified so that SWL
 s piezo 
turns on when VCs reaches a maximum. In order to detect
Table 1: Comparison of maximum storage capacitor voltage for the a global maximum, and to collect energy from both axles,
3 algorithms SWL was modified to turn on when VCs does not increase
It must be noted that for the “Fixed Threshold for a period of 100ms. The off-threshold was once again
Switching” algorithm to work reliably, Vhigh cannot be chosen to be 1.75V. Figure 23 show the results from
arbitrarily large. Vhigh must necessarily be chosen a volt or “Max Switching algorithm”. It is seen that the two axles
would generate a combined voltage of 4V. This agrees
two lower than the lowest VCsmax that can be expected well with the simulations in section IV.B. Figure 24 show
corresponding to the set of all possible Vstrain
max
i . Hence the results from the “Switched Inductor Algorithm”
for “Fixed Threshold Switching” algorithm is experiments and the capacitor voltage is found to be 5.9V.
Vhigh
This agrees well with the simulations in section IV.C.
necessarily smaller than VCsmax for the modified Max 10

Voltage Switching. From equation (32a) it is clear that 9

  1 . Thus, VCsmax for “Fixed Threshold Switching” <
8
VCsmax for modified “Max Voltage Switching” < VCsmax for

Storage capacitor voltage (V)

7
modified “Switched Inductor”.
6

V. EXPERIMENTAL RESULTS 5

The three algorithms presented in section IV were 4

implemented with a self-contained electronic circuit. The 3
switches SWL and SWP were implemented using
2
MOSFETs. The extremely small amount of power
required by the control system was derived entirely from 1
the charge energy stored in the capacitor. Sets of 0
experiments were carried out with each of the three 0 2 4 6
Time(s)
8 10 12 14

circuits connected to the piezo. Each experiment consisted Figure 22: “Fixed Threshold algorithm” with a high threshold of
of driving a compact car over the sensor at 20 kmph. This 2.75V
resulted in separate loading from the two axles, first by
the front tires and then by the rear tires. Figure 21 shows a 10

schematic of the test setup. 9

The speed of 20 kmph (12 mph) was chosen for the
8
tests since it was the lowest speed with which the car
Storage capacitor voltage (V)

could be made to move consistently over the sensor in 7

multiple tests. Higher speeds will result in higher loads
6
on the sensor and are likely to increase the energy
harvested. 5
Lane centerline
4

0
0 2 4 6 8 10 12
Asphalt below Time(s)
the Road Surface Figure 23: "Max Switching Algorithm"
Sensor
Road Surface From both the experimental data and the simulation
Figure 21: "Schematic of test setup" (figure not drawn to scale) results, it can be seen that the operating range for the
Figure 22 show the results from the testing of the capacitor voltage is expected to be between 0V and 10V
“Fixed threshold algorithm”. The on-threshold for the and the load current is expected to be between 0 and 10
algorithm was chosen at 2.75 V so that the sensor would mA.
detect light vehicle such as motorcycles. The MOSFET

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

10 
Veffective t   sign Vstrain t   VCpiezo t    (34)
9

max V strain t   VCpiezo t   VCs t   2Vd ,0 
8 In modeling overall dynamics, the first order
Storage capacitor voltage (V)

7 nonlinear electrical dynamic equations (11-16) are

dominated by the much slower dynamics of the
6
mechanical system. The system exhibits a two time scale
5 property and the faster electrical dynamics needs to be
modeled by its quasi-steady state value ([11], [14]) which
4
corresponds to Veffective t   0 . Equivalently
3
V strain t   VCpiezo t   VCs t   2Vd (35)
2
C s and C piezo are two capacitors connected in series.
1
If Vstrain t  does not change signs
0
0 2 4 6 8 10 12 14 VCpiezo t  Cs
Time(s)
 (36)
Figure 24: " Switched Inductor Algorithm" VCs t  C piezo
It is apparent that if SWP is controlled as prescribed, Thus
the “Switched Inductor” offers the best energy harvesting  C piezo 
performance.    V strain  2Vd  if V strain  2Vd
 C  C piezo 
VCs   s 
VI. CONCLUSIONS 
 0 otherwise
This paper developed a battery-less wireless traffic
sensor that operates by harvesting vibration energy from and
the passing of a vehicle over the sensor. The sensor is  Cs 
   V strain  2V d  if V strain  2V d
significantly smaller than an inductive loop detector and  C  C piezo 
has other advantages of easier installation, low cost, no VCpiezo   s 
wiring and ability to do vehicle classification. Three 
 0 otherwise
different control algorithms were developed to improve
the efficiency of energy harvesting from the vibrations. (37)
Of the three algorithms, the switched inductor algorithm
was shown to be the best at maximizing the harvested B. CALCULATION OF VCsmax FOR SWITCHED INDUCTOR
energy. Experimental results showed that the developed ALGORITHM
traffic sensor was able to harvest adequate energy from
the passing of every vehicle over the sensor to enable the The V max
Cs
for Switched Inductor algorithm (equation
powering of electronics for wireless transmission. (31)) can be derived as follows. When SWP is closed in
the LCR circuit shown figure 16, the effective voltage
Appendix (Calculation of VCsmax ) driving the resistive and inductive components of the
circuit is given by
A. CALCULATION OF VCsmax FOR MAX VOLTAGE Veffective t   signVstrain t   VCpiezo t 
(38)
SWITCHING ALGORITHM 
max V strain t   VCpiezo t   VCs t   2Vd ,0 
The V for Max Voltage Switching algorithm
max
Cs
In the absence of SWP, the overall dynamics is
(equation (28)) can be derived as follows. It is clear that dominated by the mechanical system and VCsmax would be
Vstrain  V piezo  VCpiezo (refer figure 3). Thus if given by equation (28). There would be no gain in VCs . If
Vstrain  VCpiezo  VCs  2Vd , the bridge circuit rectifies the SWP is closed at some V strain  V strain
max
, it would be result in
piezo current and charges the storage capacitor. When a step input to the electrical circuit. If V strain t 0   2Vd , the
V strain  VCpiezo  VCs  2Vd , the diodes block the flow of diode bridge will begin to conduct when SWP is closed.
di p
current thus preventing storage capacitor from L  Rtotal i p  Veffective t 0  (39)
discharging. If i p does not change signs and dt
When i p is unidirectional, the electrical dynamics
V strain  VCpiezo  VCs  2V d , the effective voltage driving
the current through the resistive element in the circuit is  
can be written in terms of Q  i p dt and Q  i p dt .
given by Hence
1 Q (40)
VCpiezo 
C piezo i p dt 
C piezo

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

1 Q mechanical system (38Hz)

(41)
VCs 
C piezo i p dt 
Cs
 damping ratio associated with the vibrating
mechanical system (0.7)
Now signV max
  signQ  . When C s and C piezo are width of the support beam (25mm)
strain bs
initially not charged, also the width of the piezo
Veffective t 0   Vstrain
max
 2Vd  signVstrain
max
 (42) d 31 strain produced per unit applied Electric Field
( - 190  10 -12 m V )
The piece-wise linear dynamics of the LCR system
can be written as g 31 open circuit electric field produced for an unit
  R Q  1 C
piezo  1 C s Q
LQ applied stress ( g 31  -11.6  10 -3 Vm N )
total (43) 1
 Vstrain  2Vd  signVstrain 
max max g piezo constant defined as the open circuit
voltage developed per unit applied strain
Now i p  Q and i p is unidirectional up to the first ( g = g 31 EVpiezo = 1.4623 10 5 V )
maximum of Q at the end of the first half-oscillation. hs thickness of the support beam ( 6.25mm)
Since equation (43) is valid when i p iL load current through RL
is unidirectional, it can be used to determine this first ip current through the piezo
maximum Qmax . For this second order system, l length scale associated with the mechanical
Qmax  C    Vstrain
0
(44) system
Hence m mass associated with the mechanical system
 C piezo  u displacement of the mechanical system
   Vstrain
 C s  C piezo 
max

 2Vd if Vstrain
max

 2Vd m
u mid the magnitude of displacement of the
VCs  
max mechanical system at the point where the force
 is applied
 0 otherwise C piezo capacitance of the piezo
(45) Cs storage capacitor
 C s 

C  C
  Vstrain

max

 2Vd  signV strain
max
  F is the force applied on the mechanical system
by the passing automobile
 s piezo 
 EVpiezo modulus of elasticity of the piezo at constant
1
VCpiezo  max
if Vstrain  2Vd
 voltage (constant electric field)
 ( EVpiezo  6.6  1010 N m 2 )
 0 otherwise elastic modulus of the main beam (200GPa)
 Em
(46) Es elastic modulus of the support beam (200GPa)
Note: Im area moment of inertia main beam
If the piezo is charged sufficiently in the reverse ( 1.2957  10 -7 mm )
direction such that VCpiezo1
 V strain
max
 VCs1  2Vd , the system area moment of inertia support beam
Is
would continue to oscillate. When the piezo current flows ( 5.0863 10 -10 mm )
in the reverse direction, the change in VCs and VCpiezo at L value of inductance used in “Switched
the end of the second half oscillation ( dVCs2 and dVCpiezo
2
) Inductor” algorithm (section C) (10mH)
Lm effective length of the main beam (1.7125m)
can be obtained by replacing Vmax
strain
with Ls effective length of the support beam (0.2m)
V 1
V max
V . 1 Rtotal total resistance in the circuit due to switches
Cpiezo strain Cs
and other components (not including RL ) (327
ohms)
C. NOMENCLATURE
RL load resistance (1000 ohms)
 ratio of peak response to steady state response Rd diode resistance
of the electrical RC circuit
SWL load switch
 thickness of piezo (0.191mm)
SWP piezo switch
 strain in the piezo, from eq. (4)
VCs voltage across the storage capacitor C s
 constant associated with the mechanical system
equaling 1  n2 ml  VCsmax maximum voltage across the storage capacitor
 elec constant associated with the mechanical system VCpiezo voltage across the piezo capacitor C piezo
model arising due to electro-mechanical Vd forward voltage drop across each diode (1.1 V
coupling , for the diode used)
6EVpiezo d 31  E s h s  , elec  6.0192  10-8
n open-circuit natural frequency of the 1
: The constant “e” is not available for the piezo material, hence an
equivalent constant is used

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
Vhigh on (high) threshold
Vlow off (low) threshold
V piezo voltage measured across the piezo
Vstrain voltage open circuit voltage generated due to
the strain
max
V strain maximum value of Vstrain corresponding to the
maximum value of the strain
V  max
strain i
extremum value of Vstrain corresponding to the
ith extremum of displacement ‘u’
VII. REFERENCES
[1] C. Chen, J. Kwon, J. Rice, A. Skabardonis and P. Varaiya. ,
“Detecting errors and imputing missing data for single-loop
surveillance systems.”, Transp.Res.Rec., pp.160-167, 2003.
[2] S. Chen, Z. Sun and B. Bridge. , “Traffic Monitoring Using Digital
Sound Field Mapping”, IEEE Transactions on Vehicular
Technology, vol. 50, no. 6, pp.1582-1589, 2001.
[3] S. Chen, Z. P. Sun and B. Bridge. , “Automatic traffic monitoring
by intelligent sound detection”, Intelligent Transportation System,
1997.ITSC 97.IEEE Conference on, pp.171-176, 1997.
[4] B. Coifman. , “Improved Velocity Estimation Using Single Loop
Detectors”, Transportation Research Part A: Policy and Practice,
vol. 35, no. 10, pp.863-880, 2001.
[5] C. W. De Silva, “Mechatronics: An Integrated Approach”, CRC
Press, 2004.
[6] N. Ferrier, S. Rowe and A. Blake. , “Real-time traffic monitoring”,
Applications of Computer Vision, 1994., Proceedings of the
Second IEEE Workshop on, pp.81-88, 1994.
[7] W. E. L. Grimson, C. Stauffer, R. Romano and L. Lee. , “Using
adaptive tracking to classify and monitor activities in a site”,
Computer Vision and Pattern Recognition, 1998.Proceedings.1998
IEEE Computer Society Conference on, pp.22-29, 1998.
[8] P. Horowitz, W. Hill, “The art of electronics”, Cambridge
University Press, 1989.
[9] S. Kamijo, Y. Matsushita, K. Ikeuchi and M. Sakauchi. , “Traffic
monitoring and accident detection at intersections”, Intelligent
Transportation Systems, IEEE Transactions on, vol. 1, no. 2,
pp.108-118, 2000.
[10] M. S. Ketchum, “Structural Engineers Handbook” 2nd edition,
McGraw-Hill, 1981
[11] H. K. Khalil. , “Nonlinear Systems”, Upper Saddle River, New
Jersey, Prentice Hall, 2002.
[12] S. W. Kim, Y. Eun, H. Kim, J. Ko, W. J. Jung, Y. Choi, Y. G. Cho
and D. Cho. , “Performance comparison of loop/piezo and
ultrasonic sensor-based traffic detection systems for collecting
individual vehicle information”, Proc.of 6th World Congress on
ITS (CD-ROM), 1998.
[13] S. W. Kim, K. Kim, J. Lee and D. D. Cho. , “Application Of Fuzzy
Logic to Vehicle Classification Algorithm In Loop/Piezo-Sensor
Fusion Systems”, Asian Journal of Control, vol. 3, no. 1, pp.64-68,
2001.
[14] P. V. Kokotovic, J. O'Reilly and H. K. Khalil. , “Singular
Perturbation Methods in Control: Analysis and Design”, Society
for Industrial and Applied Mathematics Philadelphia, 1986.
[15] C. Li, K. Ikeuchi and M. Sakauchi. , “Acquisition of Traffic
Information Using Video Camera with 2D Spatio-Temporal Image
Transformation Technique”, IEEE Intelligent Transportation,
pp.634-638, 1999.
[16] E. Lefeuvre, A. Badel, C. Richard, D. Guyomar, “Piezoelectric
Energy Harvesting Device Optimization by Synchronous Electric
Charge Extraction”, J Intell Mater Syst Struct, vol. 16, no. 10, pp.
865-876, 2005.
[17] L.J. Mountain, J. B. Garner. , “Application of photography to
traffic surveys”, The Jour of the Inst. of Highway Engineers,
pp.12-19, 1980.
[18] P. T. Martin, Y. Feng and X. Wang. , “Detector Technology
Evaluation”, Mountain-Plains Consortium, 2003.
Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.
[19] P. G. Michalopoulos and C. A. Anderson , “Costs and benefits of
vision-based, wide-area detection in freeway applications”,
Transp.Res.Rec., pp.40-47, 1995.
[20] O. D. Nwokah, “Mechanical Systems Design Handbook Modeling
Measurement and Control”, CRC Press, 2001.
[21] S. Oh, S. G. Ritchie and C. Oh. , “Real Time Traffic Measurement
from Single Loop Inductive Signatures”, 81 stTRB Annual
Meeting, January, 2002.
[22] G. K. Ottman, H. F. Hofmann, A. C. Bhatt and G. A. Lesieutre. ,
“Adaptive piezoelectric energy harvesting circuit for wireless
remote power supply”, Power Electronics, IEEE Transactions on,
vol. 17, no. 5, pp.669-676, 2002.
[23] J. C. Rojas and J. D. Crisman. , “Vehicle detection in color
images”, Intelligent Transportation System, 1997.ITSC 97.IEEE
Conference on, pp.403-408, 1997.
[24] Santiram Kal, “Basic Electronics: Devices, Circuits and IT
Fundamentals” Prentice Hall of India, 2004.
[25] H. A. Sodano, D. J. Inman and G. Park. , “Comparison of
Piezoelectric Energy Harvesting Devices for Recharging
Batteries”, J Intell Mater Syst Struct, vol. 16, no. 10, pp.799-807,
2005.
[26] H. A. Sodano, D. J. Inman and G. Park. , “A Review of Power
Harvesting from Vibration Using Piezoelectric Materials”, The
Shock and Vibration Digest, vol. 36, no. 3, pp.197-205, 2004.
[27] C. Sun, S. G. Ritchie and S. Oh. , “Inductive Classifying Artificial
Network for Vehicle Type Categorization”, Computer-Aided Civil
and Infrastructure Engineering, vol. 18, no. 3, pp.161-172, 2003.
[28] Urban ITS Center at Polytechnic University Under sub-contract
with Calspan University at Buffalo Research Center. , “User’s
Manual For Cost Estimates of Intelligent Transportation
Technologies (CEIT)”,
[29] N. Ushio and T. Shimizu. , “Loop vs ultrasonic in Chicago:
Ultrasonic vehicle detector field test isolating diffused reflection
and enduring harsh environment”, Proc.5th World
Congr.Intelligent Transport Systems, 1998.
[30] M. A. Vicencio, B. Qiu and M. G. Hartley. , “Algorithms and
architectures for the analysis of road-trafficmovements”, Image
Processing and its Applications, 1989., Third International
Conference on, pp.182-186, 1989.
[31] K. Vijayaraghavan and R. Rajamani, “Active Control Based
Energy Harvesting for Battery-Less Wireless Traffic Sensors,”
Proceedings of the 2007 American Control Conference.
[32] M.S. Weinberg, “Working equations for piezoelectric actuators
and sensors”, Journal of Microelectromechanical Systems, vol. 8,
no4, pp. 529-533, 1999.
Krishna Vijayaraghavan obtained his M.S degree from University
Of Minnesota at Minneapolis in 2005 and his B.Tech degree from the
Indian Institute of Technology at Madras in 2003. He is currently
working towards his PhD. at the University of Minnesota. His research
interests include fault-tolerant control, battery-less wireless sensors,
signal processing and real-time software. Krishna’s awards include the
ITS MN Graduate Student Award (2008) in recognition of outstanding
research contributions to Intelligent Transportation Systems (ITS)
technology and the Sivasailam Merit Prize for Best Individual B.Tech
Project in Mechanical Engineering in IIT Madras, (2003).
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

Rajesh Rajamani obtained his M.S. and Ph.D. degrees from the
University of California at Berkeley in 1991 and 1993 respectively and
his B.Tech degree from the Indian Institute of Technology at Madras in
1989. Dr. Rajamani is currently Professor of Mechanical Engineering
at the University of Minnesota. His active research interests include
sensors and control systems for automotive and biomedical applications.
Dr. Rajamani has authored over 75 journal papers and is a co-
inventor on 7 patent applications. He is the author of “Vehicle
Dynamics and Control” published by Springer Verlag in 2005. Dr.
Rajamani has served as Chair of the IEEE Technical Committee on
Automotive Control and on the editorial boards of the IEEE
Transactions on Control Systems Technology and the IEEE/ASME
Transactions on Mechatronics. Dr. Rajamani has been a recipient of the
CAREER award from the National Science Foundation, the 2001
Outstanding Paper award from the journal IEEE Transactions on
Control Systems Technology, the Ralph Teetor Award from SAE, and
the 2007 O. Hugo Schuck Award from the American Automatic Control
Council.

Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Park College of Engineering and Technology. Downloaded on August 16,2010 at 05:10:50 UTC from IEEE Xplore. Restrictions apply.