978-1-4673-0963-9/10/$26.00 ©2012 IEEE 1750
2012 5th International Congress on Image and Signal Processing (CISP 2012)
A Method for Ultra-Wideband Signal Representation
Based on Adaptive Redundancy Dictionary
Fuhua Fan
State Key Laboratory of Pulse Power Laser Technology
Electronic Engineering Institute of Hefei
Hefei, China
Xuezhong Yin
State Key Laboratory of Pulse Power Laser Technology
Electronic Engineering Institute of Hefei
Hefei, China
Abstract—Redundancy dictionary is often used for signal sparse
representation. Impulse radio ultra-wideband (IR-UWB)
communication signals formed in narrow pulse are naturally
sparse in time domain. However, the received signals become
approximately compressible because of UWB communication
channel multipath effect. Therefore, a method for constructing
redundancy dictionary, with atoms matched UWB signals, is
firstly presented using statistic learning for UWB signal sparse
representation in this paper. Simulation experiment results show
that the signal to noise ratio (SNR) of IR-UWB signals after
sparse representation is apparently improved using the adaptive
redundancy dictionary constructed by the proposed method. For
example, the SNR of IR-UWB sparse representation signal is
about 5dB when the received signal is -10dB.
Keywords- redundancy dictionary; IR-UWB signal; sparse
representation
I.
I
NTRODUCTION
UWB communication is one of the key technologies of
next-generation wireless communication. UWB signal is
characteristic of absolute bandwidth above 500MHz, and even
up to several GHz, which puts high demands on ADC device
for IR-UWB digital receiver and thus difficult to realize under
the conditions of current technology. Therefore, it is urgent
and necessary to develop compressive sampling methods for
UWB signal based on compressive sensing, and relevant
research findings are given in [1 2]. IR-UWB signal
compressive sampling is depended on the signal sparsity, and
thus the signal sparse representation determines the
compressive sampling efficiency. As redundancy dictionary is
suitable for signal sparse representation, redundancy
dictionary appropriate for UWB signals should be constructed
according to signal structure, in order to represent the signals
in simple form.
In general, sparse representation dictionary is obtained by
scale transformation and translation of orthogonal basis
functions. For specific signals, dictionary atoms are designed
as especial functions characteristic of the signals, and these
dictionaries are called as parameterized dictionary, often
applied for UWB signal representation in many publication
papers. For example, UWB signal pulse and its delayed time
waveform are used as redundancy dictionary atoms, which
request the delayed atom waveform matching the UWB
communication multipath delay time parameters. When there is
a little mismatch, the signal sparse representation is bad, and
even invalid. Therefore, dictionary learning algorithms are
introduce in [3 4 5]. In this paper, a method combing sparse
representation error weighting and recursive least square
criterion, i.e., EWRLSC method is present to construct the
redundancy dictionary for IR-UWB signal sparse
representation. A larger number of simulation experiments are
conducted on PPM-TH-UWB signals, and the results show that
much improvement on signal sparse representation SNR is
obtained especially for received UWB signals with low SNR.
II. IR-UWB
S
IGNAL
S
TRUCTURE
A
NALYSIS
A. IR-UWB signal
The signal waveform is especially designed in IR-UWB
communication system for high-speed data transmission, and
extreme narrowing Gaussian pulse with low power is often
used as basic waveform. includes There are two ways, i.e.,
time hopping and direct sequence spread spectrum applied for
IR-UWB communication, and correspondingly typical UWB
signal modulations are PPM-TH-UWB and PAM-DS-UWB.
The above PPM-TH-UWB and PAM-DS-UWB signals are
expressed as respectively
1
0
() ( )
ε
−
∞
−−
=−∞ =
=−−−−
∑∑
p
N
PPM TH UWB s p j c i
ij