Академический Документы
Профессиональный Документы
Культура Документы
The Maximum Processing Gain of The Spreading Signal in Multipath Fading Channels
Wanzhi Ma, Jun Wang, Ying Shen, Youxi Tang
National Key Lab of Science and Technology on Communications, University of Electronic Science and Technology of China, Chengdu, China Email: tangyx@uestc.edu.cn
AbstractConsidering the joint algorithm of coherent detection and non-coherent detection, we focus on the maximum processing gain of spreading signal, which is 2-dimensional spread spectrum signal, relative to binary phase shift keying (BPSK) and transmitted in rich scattering multipath fading channels. In this paper, we first investigate the processing gain of spreading signal and analyze its expression. Then, by increasing signal bandwidth or symbol period infinitely, we obtain that the processing gain is bounded, and the maximum processing gain is determined by the channel parameters and the energy-per-bit to noise density ratio (Eb/N0). By analysis and simulation, it is shown that the processing gain is not more than 30dB, when the spreading signal relative to BPSK, with Eb/N0 20dB, the maximum Doppler shift 100Hz, and the negative exponential factor 5E-6, is transmitted in rich scattering multipath fading channels. Keywords-Rich scattering; spreading signal; processing gain; signal bandwidth; symbol period
gain of spreading signal in a multipath fading channel. Our contribution in this paper can be described as follows. Considering the joint algorithm of coherent detection and non-coherent detection, we derive the expression of the processing gain with the signal bandwidth and the symbol period, and analyze the maximum of it. The rest of the paper is organized as follows. In section II, the system model is introduced. In section III, the expression of processing gain is derived with the signal bandwidth and the symbol period. Afterward, in section IV, some numerical and simulation results are shown. Finally, section V concludes this paper.
I. INTRODUCTION
With the development of wireless communications technology, the spreading signal is widely used in the field of wireless communications. On one hand, spreading signal is the most common basic signal which helps finishing time and frequency synchronization of wireless communications, on the other hand, wireless communication systems with spreading signal have anti-interference and low probability of intercept properties. The processing gain of spreading signal is defined as the ratio between the input and the output signal to noise ratio (SNR) as in [1]. Furthermore, in an additive white Gaussian noise (AWGN) channel, the processing gain of spreading signal is proportional to the ratio between the signal bandwidth and the symbol rate as in [1] and [2]. By considering direct sequence spread spectrum system under the AWGN channel, given signal bandwidth and bit rate, the author analyzes the processing gain and the bit error probability (BER), of binary phase shift keyed (BPSK), quadrature phase shift keyed (QPSK), and 16 quadrature amplitude shift keyed (16QASK) spreading signal in [3], and concludes that the processing gain of 16QASK spreading signal is maximum while BER of QPSK spreading signal is minimum. Besides, by applying the parallel data transmission with n orthogonal spreading sequences other than the serial data transmission with one spreading sequence, the processing gain grows by n times as in [4]. However, there has been little research on the processing
978-0-7695-4357-4/11 $26.00 2011 IEEE DOI 10.1109/CMC.2011.77 362
bD = 1 is the input data symbol, which is spread by 2-dimensional spread spectrum matrix DD , where is the pseudorandom spreading sequence in DS-SS unit, and is the pseudorandom spreading sequence in DMC-SS unit. The output signal X P and XD is represented respectively as XP = bPP P (1)
XD = bD DD
bP
X
P
(2)
X
s (t )
bD
b D
r (t )
Figure 1. System model. Where bP and bD denote the input pilot symD denotes the bol and the data symbol of transmitter respectively, and b output data symbol of receiver.
in
D
= ( , ) , and = ( , ) . Based on block type pilot arrangement, XP and XD are combined as in [6]. In detail, one column of pilot matrix XP is inserted between two columns of data matrix XD . After adding pilot symbol, an output matrix X with M rows and 2 N columns is produced. As in Fig. 2, is a chip of pilot, is a chip of data. After X passing through the IDFT unit, being processed by adding guard time and up-sampling, the output signal s ( t ) is fed into antenna.
D 1 D 2 D N
D 1
D 2
D T M
P which P = ( 1P , 2P N ) ,
P p = (1P , 2P M )
t
Figure 2. Block type pilot arrangement. is a chip of pilot, is a chip of data.
2 1
where Tchip
( )
(7)
where
pi = Pi
P
j =0
Q 1
, and
2 1
F ( a, b; c; z ) is the Gaussian
X mn is the mth row and nth column of X . In rich scattering multipath fading channels, the received signal is combined with many multipath components as in [7]. When the maximum multipath delay is larger than time resolution of receiver, multipath components can be divided into many scattering clusters, each of which is one distinguishable path, and the interval between two adjacent clusters is the time resolution of receiver, which is reciprocal of signal bandwidth Bs as in [7]. In one distinguishable path, assuming the multipath components are continuous, there are infinite multipath components, so the amplitude distribution of the signal can be approximated very closely by the Rayleigh distribution as in [8]. The received signal over a rich scattering multipath fading channel can be expressed as r ( t ) = hs i (t i ) + n (t )
i =0 Q 1
hyper- geometric function. When Bs is infinite, the time resolution of receiver is infinitesimal, so (6) can be simplified as
=
then
1e 2 tm
( t + ) tm
cos [ 2 ( t ) Bc ] dtd
(8)
Bs >
0.0496 . lim Bc = t
m
(9)
(4)
where i = i Bs is the delay of ith path, hi denotes the fading channel coefficient of ith path, and n ( t ) is the additive white complex Gaussian noise at time t . In this paper, the channel holding time follows a negative exponential distribution, for it is commonly used in wireless channels as in [7] and [8]. The negative exponential * i tm , in distribution can be expressed as P i = E ( hi hi ) = e which tm is the negative exponential factor. Channel time-selectivity is described by coherence time Tc , which is affected by the maximum Doppler shift as in [9].
on the assumption that the receiver has acquired time and frequency synchronization, having been processed by down-sampling and de-adding guard time, the received signal r ( t ) pass through the DFT unit, and produce the 2-dimensional received matrix R with M rows and 2N columns in the time-frequency plane. Based on block type pilot arrangement, we extract the chips of pilot from R to generate R P , and extract the chips of data from R to generate R D . Both R P and R D are M D D P P n = 1 , m n = 1 , rows and N columns. Assuming m , and the the dot product of R P and pp is defined as H
. The mth dot product of R D and DD is defined as S are expressed and S row and nth column of H respectively as mn = hmn P P h + nmn (10) D D mn = hmn bD + nmn s (11)
D P where hmn , hmn denote respectively the fading channel
P D , nmn are the respective additive white coefficient, and nmn complex Gaussian noise, of the mth row and nth column of R P and R D . Besides, two nmn with different m or n are statistically independent. Given the channel parameters, Bc and Tc are obtained by (5), (6) and (7). The number of coherence rows B and the number of coherence columns T of R are readily obtained, and expressed respectively as B= (12) min(Bs, Bc)Tofdm
= 0.423 . (5) 16 fd2 fd Channel frequency-selectivity is described by coherence bandwidth Bc , which is affected by the channel holding time distribution in [9]. Assuming the time-domain correlation coefficient = 0.9 , Bc is determined by (6) and (7) in [10] Tc
= pi p j cos [ 2 ( i j ) Bc ]
i =0 j =0 Q 1 Q 1
(6)
(13)
363
where Bs is the signal bandwidth, Ts is one symbol period, in which one bit is transmitted, is the ratio of the symbol period minus the total guard time and symbol period, Tofdm is the length of one chip in time-domain after adding guard time, min ( ) denotes the operation of extracting the minor one, and denotes the operation of rounding down to the closest integer. The number of coherence rows B P and the number of are obtained respectively as coherence columns T P of H BP = B (14) T P = T 2. (15) D The number of coherence rows B and the number of are obtained respectively as coherence columns T D of S BD = B (16) T D = T 2. (17) are and S From top to bottom and from left to right, H divided into N B NT sub-matrixes, respectively, each of which is called one coherence area, where NB = B (18) BT s ofdm NT = (19) Ts TTofdm . Without loss of generality, it is assumed that the fading channel coefficients are identical over one coherence area, but statistically independent when located in different coherence areas. Then the channel estimation for each coherence area is expressed as
2 * D = 1 B2T 2bD Psi b H xy + 1 BT Psi H xy nxyk + 4 2 x =1 y =1 x =1 y =1 k =1 BT 2 BT 2 NB NT N B NT BT 2 P * P * D + 1 BTbD Psi H xy (nxyk + (nxyk nxyk ) ) 2 x =1 y =1 k =1 x=1 y =1 k =1 k =1
N B NT
N B NT
BT 2
(24)
denotes the noise of the kth pilot element in the coherence D area of the xth row and yth column, and nxyk denotes the noise of the kth data element in the coherence area of the xth row and yth column. It is readily obtained that
H
m =1 n =1
N B NT
2 mn