Blind Chip Rate Estimation in Multirate Blind Chip Rate Estimation in Multirate CDMA Transmissions Using Multirate Sampling at Slow Flat Fading Channels Sampling at Slow Flat Fading Channels A U T H O R S S i A U T H O R S : S i a v a s h G h a v a m i h G h i B a h m a n A b o l h a s s a n i D e p a r t m e n t o f E l e c t r i c a l E n g i n e e r i n g D e p a r t m e n t o f E l e c t r i c a l E n g i n e e r i n g I r a n U n i v e r s i t y o f S c i e n c e a n d T e c h n o l o g y T e h r a n , I r a n N o v e m b e r 2 0 0 8
Outlines Motivations System Model Estimation of Channel Impulse Response p p Blind Chip Rate Estimation Blind Bit Rate Estimation Simulation Results Conclusions
Motivations 3G cellular systems capability of supporting transmission data as diverse as voice, packet bili f i i i d di i k data, low-resolution video, and compressed audio. heterogeneous services produce digital information streams with different data rates digital information streams with different data rates Multi Rate Transmission variable spreading length (VSL) technique where all users employ sequences with the same chip period sequences with the same chip period the data rate is tied to the length of the spreading code of each user. Another way to view a multirate CDMA transmission is to consider a constant spreading length where users employ sequences with diff different chip periods. hi i d In general, chip time and spreading sequence length of users can be selected variable.
Motivations In the literature, different methods have been presented for chip time estimation. many number of those methods are based on cyclic cumulant method, which have been presented in literatures [3-7]. , p [3 7] Cyclic cumulant method in multi-rate and multi-user system doesn’t exhibit good performance because cyclic frequencies of different users overlap. li f i f diff l chip time estimation of different users is not possible. Since, In this paper blind chip time estimation Since, In this paper blind chip time estimation based on channel impulse response estimation using singular value decomposition of estimated received signal covariance matrix is proposed. matrix is proposed.
Motivations For spreading sequence estimation in spectrum surveillance systems over slow flat Rayleigh fading channel plus AWGN, length of code must be determined. It needs to determined. It needs to Bit time Fluctuations of received signal covariance matrix chip time hi ti Multi-rate sampling of channel impulse response. This needs no prior knowledge about transmitter in p g the receiver side; it is typically the case in blind signals interception in the military field or in spectrum surveillance spectrum surveillance.
System Model Down link scenario of a multi-rate DS-CDMA system. Channel model is slow flat Rayleigh fading plus Additive white Gaussian Noise (AWGN) Signal power is lower than noise power in receiver (SNR<0 in dB) due to no prior knowledge about (SNR i dB) d t i k l d b t spreading sequences A symbol time is equal to the spreading sequence A symbol time is equal to the spreading sequence (Short Spreading Code) Symbols have zero mean and are uncorrelated Symbols have zero mean and are uncorrelated
Estim ation of Channel Im pulse Response Estimation of Received Signal Covariance Matrix g Covariance matrix C i t i H H H of Noise R E rr E xx E nn R R r x n Received Signal d S g Covariance matrix Covariance matrix Received Signal R i d Si l N i Noise without noise of Signal Expansion of Received Signal Covariance Matrix p g K 1 2 0 0 * 1 1 * R (1 ) v ( v ) v ( v ) I n k k k k k k k k 0 Eigen vectors of f SNR of k th user Delay of k th user Noise Variance k th user In Downlink scenario, users are synchronous, hence , y , K 1 * 2 R { v v I } 0 n k k k k k 1
Blind Chip Rate Estim ation Channel impulse response p p Sampling Frequency mL m i T F 1/ 2 c s h ( ) t N c p ( t j m ) . m k i , k i , R Chip Rate j 0 c convolution of the transmitter filter , channel filter and receiver filter l i f h i fil h l fil d i fil Normalized Channel impulse response mL m i T T 1/ 2 1/ 2 c c c c sgn( sgn( h h ( )) ( )) t t N N c u ( ( t t ( ( j j 1) 1) ) ) u ( ( t t j j ) ) , k i , k i , m m j 0 Step Function Differential of Normalized Channel impulse response d d sgn( sgn( h h ( ( t t )) )) T T mL m i i k k i , 1/ 2 c N c [ ] ( m t j ) Impulse Function k i , dt m j 0
Over Sampling Ratio = 4 Blind Chip Rate Estim ation Over Sampling Ratio = 1
Blind Chip Rate Estim ation It is obvious that in sampling frequencies above the Nyquist rate, by increasing the sampling frequency number of zero crossings on differential increasing the sampling frequency number of zero crossings on differential of spreading sequence remains constant. Since Number of zero crossings ZCR L Length of diffrentiated Sequence th f diff ti t d S can be used for determining oversampling ratio and then chip rate estimation.
Blind Chip Rate Estim ation Sampling frequencies less than Nyquist rate p g q yq loss some of data samples and decrease length of differential of sequence, it causes ZCR increases for sampling frequency lower than the i i f li f l h h Nyquist rate.
Blind Chip Rate Estim ation Since, by weighting ZCR, it is possible to reduce ZCR , y g g , p for sampling frequencies less than the Nyquist rate . we propose weighted zero crossing ratios, which is defined as WZCR The Number of zero crossings g ZCR
Effect of code length increament Increasing the code length g g increases the computational complexity of subspace decomposition of the received signal covariance matrix due to increasing the dimension of covariance matrix increasing the dimension of covariance matrix, instead, output SNR in output of covariance matrix estimator increases by the factor of SNR N out 2 2 N SNR 1 out out 1
Blind Bit Rate Estim ation For blind bit rate estimation, we use fl fluctuation i of f correlation estimator [9]. To compute the fluctuations, we divide the received signal di id h i d i l into M temporal windows with duration of T F for each window window. T F 1 ˆ m * R ( ) r ( t ) r ( t ) dt r m m T F Fluctuations of correlation estimator 0 Taking expectation of above equation over M window 1 M 1 2 2 ˆ ˆ ˆ ˆ ˆ ( ˆ ( ) E R ( ) R ( ) R ( ( ) ) R ( ) ) R ( ) ( ) r r M M r r s s n n m 0 K S 1 i ˆ R ( ) R ( ) s s k i , i 0 k 0
Spreading Sequence Length Code length of each user is determined using their g g corresponding chip times bit times. bi i Estimated parameter using this method is useful and applicable and applicable spectrum surveillance in very low SNR (negative SNR in dB). Modern wireless systems such as cognitive radios. y g
Estimated Channel Impulse Response Estimated eigenvector corresponding to the maximum eigenvalue of the received signal covariance matrix g Exact and estimated spreading sequence of user 1 The differential of the estimated eigenvector. The differential of the estimated eigenvector.
Zero crossing Number of Zero crossings in terms of sampling frequency Zero crossing ratio in terms of sampling frequency weighted zero crossing ratio in terms of sampling frequency all weighted zero crossing ratio in terms of sampling frequency, all of them are related to user1
Chip Rate Detection Performance 97.5% Probability of chip rate detection related to the first user for 10000 times Monte Carlo Test. 97% 97% Probability of chip rate detection related to the second user for 10000 times Monte Carlo Test. d f ti M t C l T t
Conclusions This paper considered the problem of blind chip rate estimation of direct sequence spread spectrum (DS-SS) signals in multi-rate and multiuser direct-sequence code division multiple accesses (DS-CDMA) division multiple accesses (DS-CDMA). The estimation is based on the multi-rate sampling of the differential of estimated channel impulse response. p p Simulation results showed that the chip rate and bit rate can be determined exactly in very low SNR (-5 dB) and in multi-rate and multiuser field. lti t d lti fi ld Therefore it is possible to blindly estimate the code length of each user in such systems length of each user in such systems.
Recommend
More recommend
