SEQUENTIAL ENERGY DETECTION FOR TOUCH INPUT DETECTION. Youngchun Kim, Ahmed H. Tewfik. Nikhil Kundargi. Electrical and Computer ...
SEQUENTIAL ENERGY DETECTION FOR TOUCH INPUT DETECTION Youngchun Kim, Ahmed H. Tewfik
Nikhil Kundargi
Electrical and Computer Engineering
National Instruments
The University of Texas at Austin, Texas, USA
Austin, Texas, USA
by a single ADC. The main objective of
ABSTRACT In this paper we propose a novel detection algorithm that de livers impressive savings in the sensing time of the capacitive touchscreen systems using sequential energy detection meth ods. We also show that these savings can translate into a sig nificant boost to the operational battery life of today's mobile devices. We provide numerical analysis of sequential energy detection scheme, and the average sample number of mea surements required to decide the presence of a touch input is derived. The analysis and simulation results confirm that the proposed scheme can save at least
70%
or more computa
tional resources for performing touch detection under realistic noise condition as compared to the conventional fixed sample size detection scheme.
Index Terms-
[3 , 5]
is to recover
exact signals by modulating the sparse input signals in fre quency domain using pseudo sequences. In both cases, the energy savings are achieved at sensing side, but the savings are degraded because of the recovery methods which gener ally require complex or iterative computations. The sequential energy detector was proposed in
sequential detection, capacitive touch screen
tension of the work, and we derive the average sample num bers (ASNs) of the proposed sequential scheme.
With the
extended result, relative efficiency of the proposed method shows high efficiency in touch detection problem under real istic SNR conditions.
2 we briefly
introduce the standard touch detection using the energy de tector. Section
3
presents the proposed sensing and detection
schemes and provides performance analysis. In Section
1.
for
systems. We consider the touch detection problem as an ex
This paper is organized as follows: In Section
binary hypothesis test, energy detector,
[6]
an efficient detection of spectrum sensing in cognitive radio
4
we
present the simulation results of the proposed schemes and an
INTRODUCTION
example of applications.
The capacitive touchscreen display is one of the biggest drain on the battery of a mobile handset. In this paper we present
2.
a highly efficient touch detection scheme that can reliably de
TOUCH DETECTION BY ENERGY DETECTOR
tect finger touches at a fraction of the energy of conventional
Signals with low SNR require averaging multiple measure
touch detection techniques. In touch screen, as the number
ments for accurate detection.
of sensing points is increased, high power consumption or in
capacitive touch screens,
20
creased sensing latency are expected in processing. To over
quired to obtain
dB SNRs to detect finger touches
20
rv
30
For example, in the case of rv
100
measurements are re
come these limitations, a fast sensing and detection scheme
using today's analog circuit technology
is required. A fast sensing circuit can be designed with in
we monitor
creased power consumption. In addition, the circuit increases
the existence of touch signal, the detection problem can be
the hardware complexity. Thus, engineers need to consider a
modeled as a binary hypothesis testing problem: Under the
tradeoff between sensing latency and power consumption to
null hypothesis Ho, the noisy measurement of a sensor signal
meet system requirements. In mobile devices such as smart
N
[7,8].
Suppose that
measurements on each touch nodes to decide
without touch input, and under the alternative hypothesis HI,
phones and tablet PCs, major portion of energy is consumed
the noisy measurement of a sensor signal with touch input.
in user interfaces (LCD display and touch input processing)
This hypothesis problem can be written as
[1]. For accurate detection and better user interface, energy
Ho
efficient sensing and detection schemes are necessary with where
manage multiple sensor inputs. Compressed Sensing (CS) based systems use fewer num ber measurements than the standard sampling strategy
[2-4].
However, the CS based method requires a specially designed modulation circuit for measurement. In
978-1-4673-6997-8/15/$31.00 ©2015 IEEE
;r(t)
=
v(t),
HI
:
;r(t)
=
s(t) + v(t),
denotes a unknown touch signal,
v(t)
(1)
is a noise
process that is modeled as a discrete time zero-mean white Gaussian noise with covariance The event signal
Ev
E[v(t)v(s)]
=
EvOk,l
with
Ok,I' s(t) could be either a deterministic or a ran
noise power density
[5] , the authors show
that multi-channel sparse signal can be effectively sampled
s(t)
:
and Kronecker delta function
dom variable. For generalization, we assume that the event
3941
ICASSP2015
signal is a real random variable, and the sign of the signal is unknown. Under this formulation, it is well known that the energy detector is the optimal detection method energy of
s(t) can be computed by Es
[9-11].
The
� ft:o s2(t)dt [12].
=
During a sampling interval (O,T), the energy is approximated by
·T
l
_
t-O
where;[i
=
;[(i/2B).
2B
=
=
u.
[s[l], s[2]' . . , s[N]]. ·
=
Thus,
The energy of each signal can
If
N