CN101232303B - Low complex degree equalization method based on iteration jam deleting in spread spectrum communication system - Google Patents

Abstract

The invention relates to a low-complexity balancing method which is used in a spread spectrum communication system based on iterative interference cancellation, and belongs to the technical field of wireless communication. The method includes the following steps: in the pilot stage when receiving a signal, a channel is estimated using such methods as glide correlation and least square to get an estimation value of compound channel responses; for each compound channel response, L paths with maximum amplitude are selected from KNc paths, and the values of the other paths are assigned as zero; the channel responses are fragmented, Nc paths in each section form a channel response vector, and M vectors belonging to the k section form a channel response matrix; a correlation value register and a provisional decision variable register are initialized; in a numeric data code element stage when receiving the signal, the iterative interference cancellation is acted according the sequence of thenumeric data code element to get a final decision value of an emitting code element. The invention greatly reduces the calculation amount of the tap selection and the coefficient training, but the performance thereof surpasses a linear balancing method which arranges the tap according to the maximum path position.

Description

In the spread spectrum communication system based on the low complex degree equalization method of iteration jam deleting

Technical field

The invention belongs to technical field of wireless communication, particularly the low complex degree equalization method in the spread spectrum communication system.

Background technology

Wireless communication system uses electromagnetic wave to finish the transmission of information on the diverse location of space, and it generally comprises and transmits and receives two parts.At radiating portion, information bit waiting for transmission is transformed into radio frequency transmissions at last through Base-Band Processing such as error correction coding, spread spectrum, modulation.Transmitting to be subjected to the influence of various monkey wrench in the wireless channel before arriving receiving terminal, become during as multipath interference, multi-user interference, channel etc.Therefore at receiving unit, earlier the baseband signal of sampling to be handled, suppress multipath interference etc., just can demodulate the information bit that transmitting terminal is sent here then, finish whole message transmitting procedure as using equalization methods.

Spread spectrum is a kind of information transmission technology commonly used in the wireless communication system, and the spectrum width that it transmits by expansion improves the performance of system.Spread spectrum transmits the bandwidth that takies generally much larger than the needed minimum bandwidth of transmission information, and the spreading code that uses is independent of information to be transmitted.

Because video stretching, under the constant situation of transmitting power, the power spectral density of spread-spectrum signal reduces greatly, even can be lower than the power spectral density of white noise, and promptly signal is submerged among the noise, and this greatly reduces its probability detected or that intercept and capture.Along with the increase of signal bandwidth, originally the multipath that can not differentiate in the propagation channel becomes distinguishable gradually, thereby receiving terminal can carry out frequency diversity, reduces the influence of multipath fading to systematic function.Originally may influence the narrow band interference of whole signal frequency range, and can only influence the band segment of signal now, therefore the influence to systematic function also reduces greatly.Spread spectrum can be allowed a plurality of users shared frequency range simultaneously, and the signal of stack can be separated by various signal processing technologies effectively at receiving terminal mutually.Because the increase of bandwidth, spread-spectrum signal also has higher temporal resolution and stationkeeping ability.

In the middle of radio communication, the application of spread spectrum is very extensive.A lot of international standards, three big standard WCDMA, CDMA2000 and TD-SCDMA as the third generation (3G) mobile communication, the standard IEEE 802.11b of WLAN (wireless local area network), the standard IEEE 802.15.4 of Wireless Personal Network etc. has used the information transmission technology of spread spectrum as core.At present just in some flourish emerging technologies,, also mostly adopted spread spectrum as ultra-wideband communications etc.

(TH), frequency hopping (FH) and linear frequency modulation (Chirp) etc. were four kinds when basic spectrum spreading method comprised straight preface (DS), jumping.Wherein straight preface and time-hopping systems reach the purpose of spread spectrum by shorter chip of introducing cycle in code element, and frequency hopping and linear frequency modulation system are then by introducing the purpose that more carrier frequency point reaches spread spectrum.In the direct sequence spread spectrum system, a user can have only a spreading code, also a plurality of spreading codes can be arranged.When using a plurality of spreading code, each code element is selected to use one of them spreading code according to bit waiting for transmission, and spreading code itself is promptly carrying information like this, and the availability of frequency spectrum is higher.If the waveform of whole code element behind the spread spectrum is regarded as an impulse waveform, data bit directly is modulated on this waveform, then only use the transmission means of a spreading code to can be regarded as the monophasic waveform modulation, and use the transmission means of a plurality of spreading codes to can be regarded as many waveform modulated.Obviously, the monophasic waveform modulation is a special case of many waveform modulated.Many waveform modulated method commonly used has polynary quadrature modulation (MOK) and polynary biorthogonal modulation (MBOK) etc., and MOK only selects spreading code according to the emission bit, and MBOK also launches the phase place of the whole spreading code of bit modulation according to other after selecting spreading code.

The THSS system is based on pulse modulated system, transmit to have low duty ratio, thereby the position of pulse can change flexibly.In such system, except pulse amplitude modulation, pulse position modulation is also usually used, thereby improves the availability of frequency spectrum of system.Jump time-code and defined the launch time of an interior pulse of code element, and can think a frequency spreading wave, be sometimes referred to as signature waveform according to this impulse waveform of jumping the whole code element of time-code generation.After the pulse position modulation, also can think to have used another frequency spreading wave from the angle of whole code element, promptly the pulse position modulation in the THSS system also can be regarded a kind of many waveform modulated as.Therefore, from the angle of many waveform modulated, various the transmitting of direct sequence spread spectrum and THSS can be handled with unified method.

It is generally acknowledged that spread spectrum system has anti-multipath interference capability preferably, but that is a multidiameter delay conclusion relatively more in short-term.When maximum multipath time delay during less than a code-element period, the autocorrelation performance good according to frequency expansion sequence, multipath disturbs and can be suppressed effectively after the despreading.If use high specific to merge (MRC) Rake receiver, can also collect the multipath energy that merging receives, further improve the performance of system.But after multidiameter delay surpassed a code-element period, despreading and MRC-Rake can not solve the problem that multipath disturbs between code element, and systematic function can rapid deterioration.

The situation that this long multipath disturbs all can occur in underwater sound communication, digital television broadcasting, the super 3G cdma communication system of 3G/ and ultra-wideband communication system.But so-called long multipath disturbs and is not meant the absolute growth of multidiameter delay, and is meant the relative length of the code-element period that transmits relatively, if multipath disturbs and covered a plurality of signal elements, promptly belongs to long multipath and disturbs.For example, in 3G (Third Generation) Moblie TD-SCDMA system, the spreading rate of employing 3.84Mcps and length are 8 spreading code, and code-element period is about 2us, and when radius of society was 3 kilometers, multidiameter delay can reach 10us, so multipath disturbs five code elements of covering.And in the indoor ultra-broadband communication system, the multidiameter delay maximum 100ns that only has an appointment, if code-element period is 20ns, then multipath disturbs and also can cover five code elements.

Long multipath disturbs and can only suppress by equalization methods.Equalization methods can be divided into linear and non-linear two big classes, and linear equalization uses a transversal filter to be handled to received signal, and the coefficient of filter can according to ZF (ZF) or least mean-square error criterions such as (MMSE) be calculated or training obtains.Nonlinear equalization then has various way of realization, and the maximum likelihood Sequence Detection is a kind of equalization methods of optimum, and it uses the whole emission bit sequence of viterbi algorithm direct estimation according to channel estimation value.Decision feedback equalization uses a feedforward filter and a feedback filter, and wherein feedforward filter carries out the albefaction matched filtering, and multipath disturbs between feedback filter deletion code element, and its performance is better than linear equalizing method.

The implementation complexity of equalization methods comprises two parts, and a part is at its coefficient training process, and another part is in data detection process, and the difference in distribution of its complexity of different equalization methods between these two parts is very big.Linear equalization needs very big amount of calculation to train the coefficient of filter, but when Data Detection, its amount of calculation is identical with MRC-Rake.And the maximum likelihood Sequence Detection only needs channel estimation value, does not need to train filter coefficient, but its complexity is exponential increase with the length of intersymbol interference when Data Detection, so generally do not use the maximum likelihood Sequence Detection in real system.Decision feedback equalization is because comprise the feedforward filter of a linearity, so also will very big amount of calculation train the coefficient of filter.In the situation of long multipath interference, no matter be that filter coefficient is trained or Data Detection all becomes quite complicated.In addition, for the system of many waveform modulated, receiving terminal needs a plurality of filters to carry out Data Detection, therefore needs the many groups of training filter coefficient, and this can further increase the complexity of balancing procedure.

In spread spectrum communication system, the method for existing reduction complexity of equalization mainly contains two classes.The first kind is that chip-level equalisation is converted into the symbol level equilibrium, and promptly received signal is carried out the matched filtering of multipath channel earlier, again with the chip rate down-sampling, then the signal behind the down-sampling is carried out equilibrium.This method has reduced the complexity of equalizing coefficient training, but the complexity of Data Detection part but raises to some extent, because in fact the process of multipath channel matched filtering and down-sampling is equivalent to a MRC-Rake receiver, just this a part of complexity has been equivalent to chip-level equalisation, in addition the back also has the equilibrium of one-level symbol level.

The method that second class reduces complexity of equalization is to reduce the number of taps of filter.In general outdoor transmission environment, because the density of reflector is lower, multipath channel often shows quite sparse characteristic.Utilize this characteristic, filter can only be placed tap in position, can significantly reduce the number of tap like this.And in the indoor ultra-broadband channel, multipath generally is that sub-clustering arrives, though bunch in be intensive, often show certain sparse characteristic between bunch, therefore also can reduce the number of filter tap.For linear equalization, best tap position selection problem is a multivariate joint probability search problem, and its computation complexity is difficult to bear often.Though place tap in the maximum diameter position simple, performance but is under some influence.

For traditional decision feedback equalization, received signal is through after the feedforward filter, and the sparse characteristic of multipath channel can be destroyed, causes feedback filter can not adopt sparse tap.Adopt the decision feedback equalization of inverted sequence to address this problem, promptly allow signal disturb through the feedback filter deletion earlier, pass through feedforward filter again, feedback filter can adopt sparse tap like this, and the length of intersymbol interference significantly shortens after the feedback filtering, thereby can reduce the complexity of feedforward filter.But this way has only solved the problem of complexity, does not but consider receiving Effect on Performance.Because feedback filter is deleted greatly multipath energy in advance, so the signal to noise ratio during bit decision has bigger decline, when the signal to noise ratio of received signal was low, the bit error rate of system can be than higher.

Summary of the invention

The objective of the invention is to propose a kind of equalization methods, disturb with the long multipath that suppresses in straight preface and the THSS system based on iteration jam deleting for overcoming the weak point of prior art.It has inherited the advantage of decision feedback equalization in interference delete and tap system of selection, but has avoided the shortcoming of its loss multipath energy.Compare with linear equalization, method of the present invention does not need to calculate equalizing coefficient fully, only needs channel estimation value just can suppress multipath interference between code element, and when tap is selected, the maximum diameter position promptly is the optimum position, greatly reduces the amount of calculation of tap selection algorithm.In the Data Detection part, amount of calculation of the inventive method and selectivity tap MRC-Rake receive quite, but its performance has but surpassed the linear equalizing method of placing tap by the maximum diameter position.

Based on the low complex degree equalization method of iteration jam deleting, may further comprise the steps in the spread spectrum communication system that the present invention proposes:

1) in the pilot tone stage of received signal, use the methods such as relevant or least square of sliding to carry out channel estimating, obtain the estimated value h of compound channel response _m[i], m=0,1 ..., M-1, i=0,1 ..., KN _c-1, wherein M represents the number of the frequency spreading wave that communication system is used, and K represents the code element number that the compound channel response covers, N _cRepresent the sampling number in the code-element period;

2) to each compound channel response h _m[i], i=0,1 ..., KN _c-1, from KN _cChoose the L bar footpath of amplitude maximum in the bar footpath, it is zero that the value in other footpath is composed; Here, 0＜L≤KN _c

3) with the Baud Length be the cycle to the channel response segmentation, the N in each section _cBar directly constitutes a channel response vector, wherein k channel response vector h _{K, m}=[h _m[kN _c], h _m[kN _c+ 1] ..., h _m[(k+1) N _c-1]] ^T, k=0,1 ..., K-1, m=0,1 ..., M-1, [] ^TThe transposition of representing matrix or vector; By M vector h that belongs to the k section together _{K, m}Constitute channel response matrix H _k=[h _{K, 0}, h _{K, 1}..., h _{K, M-1}], k=0,1 ..., K-1;

4) initialization correlation register c _Reg(n, m)=0, n=0,1 ..., N+K-2, m=0,1 ..., M-1 and interim judgement amount register ${\tilde{a}}_n=0,$ n＝-K+1，…，0，…，N+K-2。Here N represents a code element number in the packet,

Be the vector of a M dimension, wherein

Be

M element;

5) in the data symbols stage of received signal, by receiving the data symbols order, from y ₀, y ₁To y _N+K-2, carry out iteration jam deleting, obtain transmit symbol a ₀, a ₁To a _N-1The conclusive judgement value, demodulate the information bit that transmitting terminal is sent here, finish whole message transmitting procedure.The concrete steps of carrying out iteration jam deleting are as follows:

(1) when receiving data symbols y ₀The time, carry out the operation of following three steps:

A. the known symbols a in the packet front end pilot frequency code element sequence _-1, a _-2..., a _-K+1Write interim judgement amount register

These code elements of deletion are to a from received signal ₀The interference of signal obtains signal behind the interference delete

${\stackrel{\‾}{y}}_{\mathrm{0,0}}=y_0-\underset{k-1}{\overset{K-1}{\mathrm{\Σ}}}{H}_k{\tilde{a}}_{-k}$

B. calculate signal behind the interference delete respectively

With the correlation of M channel response vector, write correlation register c _Reg(0, m):

$c_{\mathrm{reg}}(0,m)={\stackrel{\‾}{y}}_{\mathrm{0,0}}^T{h}_{0,m}$

Wherein, m=0,1 ..., M-1;

C. in M correlation, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

$m^{\′}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(0,m)\left|\right\}$

Value and c according to m ' _RegThe symbol of (0, m ') writes interim judgement amount register

${\tilde{a}}_0={[{\tilde{a}}_{\mathrm{0,0}},{\tilde{a}}_{\mathrm{0,1}},\·\·\·,{\tilde{a}}_{0,M-1}]}^T$

${\tilde{a}}_{0,m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(0,m^{\′})\right\},& m=m^{\′}\\ 0,& \mathrm{otherwise}\end{array}\right.$

(2) when receiving data symbols y _n, during 1≤n≤N+K-2, to n '=n, n-1 ... n-K+1 (0≤n '≤N-1) carry out the operation of following three steps respectively:

A. according to the interim judgement amount

The signal of other code element of deletion is to a from received signal _{N '}The interference of signal obtains signal behind the interference delete

${\stackrel{\‾}{y}}_{n^{\′},n}=y_n-\underset{k\≠n-n^{\′}}{\overset{K-1}{\underset{k=0}{\mathrm{\Σ}}}}{H}_k{\tilde{a}}_{n-k}$

B. calculate signal behind the interference delete respectively

With the correlation of M channel response vector, write correlation register c _Reg(n ', m):

$c_{\mathrm{reg}}(n^{\′},m)=\{\begin{array}{cc}{\stackrel{\‾}{y}}_{n^{\′},n}^T{h}_{n-n^{\′},m},& n^{\′}=n\\ c_{\mathrm{reg}}(n^{\′},m)+{\stackrel{\‾}{y}}_{n^{\′},n}^T{h}_{n-n^{\′},m},& \mathrm{otherwise}\end{array}$

Wherein, m=0,1 ..., M-1;

C. in M correlation, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

$m^{\′}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(n^{\′},m)\left|\right\}$

Value and c according to m ' _RegThe symbol of (n ', m ') writes interim judgement amount register

${\tilde{a}}_{n^{\′}}={[{\tilde{a}}_{n^{\′},0}{,\tilde{a}}_{n^{\′},1},\·\·\·,{\tilde{a}}_{n^{\′},M-1}]}^T$

${\tilde{a}}_{n^{\′},m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(n^{\′},m^{\′})\right\},& m=m^{\′}\\ 0,& \mathrm{otherwise}\end{array}\right.$

If n '=n-K+1 obtains the conclusive judgement value of transmit symbol:

$a_{n-K+1}={\tilde{a}}_{n-K+1}$

The principle of this method of the present invention is based on iteraction multiple users and detects and iteration equalizing.In spread spectrum system, the interior received signal of code-element period is the stack of several transmit symbol multipath signals, so multipath disturbs the multi-user interference that can regard as between several Virtual User.The interference delete algorithm can be used for suppressing these virtual multi-users and disturb.But, because same transmit symbol can all cause multipath to disturb in adjacent several code elements, the signal that is corresponding Virtual User in these code elements derives from same emission information, prior information when therefore the interim judgement value that obtains of code element place, front can be used as the back symbol detection, and the code element new signal component that merges in back can be improved these interim judgement values.By continuous iteration, the prior information of relevant transmit symbol can be more and more accurate.The conclusive judgement statistic has merged all multipath energies that receive, and has avoided the influence of intersymbol interference again, therefore has good detection performance.

The present invention has the following advantages:

(1) implementation complexity is low, does not need to calculate or the training equalizer coefficients, and whole implementation complexity and high specific merge Rake and receive quite;

(2) performance is good, and the performance of the inventive method is much better than the decision feedback equalization method of inverted sequence; When all multi-path location is placed equalizer tap, the performance of performance of the inventive method and linear minimum mean-squared error equalizer is suitable, when only placing equalizer tap in part maximum diameter position, the performance of the inventive method is better than the performance of linear minimum mean-squared error equalizer;

(3) wide adaptability can be used for the wide-band spread spectrum communication system of monophasic waveform or many waveform modulated, also can be used for the ultra-wideband communication system of pulse amplitude or position modulation.

Description of drawings

Fig. 1 is the schematic diagram of transmitted waveform in the direct sequence spread spectrum system.

Fig. 2 is the schematic diagram of transmitted waveform in the THSS system.

The performance that Fig. 3 is the inventive method and other method when full tap relatively.

Fig. 4 is that the inventive method and the performance of other method when the part tap compare.

Embodiment

Reaching embodiment in conjunction with the accompanying drawings based on the low complex degree equalization method of iteration jam deleting in the spread spectrum communication system that the present invention proposes is described in detail as follows.

The inventive method may further comprise the steps:

At first, use the methods such as relevant or least square of sliding to carry out channel estimating, obtain the estimated value h of compound channel response in the pilot tone stage of received signal _m[i], m=0,1 ..., M-1, i=0,1 ..., KN _c-1, wherein M represents the number of the frequency spreading wave that communication system is used, and K represents the code element number that the compound channel response covers, N _cRepresent the sampling number in the code-element period;

To each compound channel response h _m[i], i=0,1 ..., KN _c-1, from KN _cChoose the L bar footpath of amplitude maximum in the bar footpath, it is zero that the value in other footpath is composed, here, and 0＜L≤KN _c(complexity of the big more system of L is high more, so the size of its value complexity that system can bear can be according to practical application the time decides, also can account for the percentage of channel response gross energy and decide) by the maximum diameter energy;

With the Baud Length be the cycle to the channel response segmentation, the N in each section _cBar directly constitutes a channel response vector, the wherein N in the k section _cBar directly constitutes channel response vector h _{K, m}=[h _m[kN _c], h _m[kN _c+ 1] ..., h _m[(k+1) N _c-1]] ^T, k=0 here, 1 ..., K-1, m=0,1 ..., M-1, [] ^TThe transposition of representing matrix or vector; By M vector h that belongs to the k section together _{K, m}Constitute channel response matrix Hk=[h _{K, 0}, h _{K, 1}..., h _{K, M-1}], k=0 here, 1 ..., K-1;

Initialization correlation register c then _Reg(n, m)=0, n=0,1 ..., N+K-2, m=0,1 ..., M-1 and interim judgement amount register ${\tilde{a}}_n=0,$ N=-K+1 ..., 0 ..., N+K-2, N represents a code element number in the packet here,

Be the vector of a M dimension, wherein

Be

M element;

Last in the data symbols stage of received signal, by the data symbols order of received signal, from y ₀, y ₁To y _N+K-2, carry out iteration jam deleting, obtain transmit symbol a ₀, a ₁To a _N-1The conclusive judgement value, demodulate the information bit that transmitting terminal is sent here, finish whole message transmitting procedure.

The above-mentioned concrete grammar that carries out iteration jam deleting is:

When receiving data symbols y ₀The time, the known symbols a in the packet front end pilot frequency code element sequence _-1, a _-2..., a _-K+1Write interim judgement amount register

These code elements of deletion are to a from received signal ₀The interference of signal obtains signal behind the interference delete ${\stackrel{\‾}{y}}_{\mathrm{0,0}}=y_0-\underset{k-1}{\overset{K-1}{\mathrm{\Σ}}}{H}_k{\tilde{a}}_{-k};$ Calculate signal behind the interference delete respectively

With the correlation of M channel response vector, write the correlation register $c_{\mathrm{reg}}(0,m)={\stackrel{\‾}{y}}_{\mathrm{0,0}}^T{h}_{0,m},$ Here m=0,1 ..., M-1; In M correlation, search for maximum, thereby obtain the numbering of the pairing frequency spreading wave of maximum related value $m^{\′}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(0,m)\left|\right\}$ Value and c according to m ' _RegThe symbol of (0, m ') writes interim judgement amount register

${\tilde{a}}_0={[{\tilde{a}}_{\mathrm{0,0}},{\tilde{a}}_{\mathrm{0,1}},\·\·\·,{\tilde{a}}_{0,M-1}]}^T$

${\tilde{a}}_{0,m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(0,m^{\′})\right\}& m=m^{\′}\\ 0,& \mathrm{otherwise}\end{array};\right.$

When receiving data symbols y _n, during 1≤n≤N+K-2, to n '=n, n-1 ... n-K+1,0≤n '≤N-1 carries out the operation of three steps respectively:

According to the interim judgement amount

The signal of other code element of deletion is to a from received signal _{N '}The interference of signal obtains signal behind the interference delete ${\stackrel{\‾}{y}}_{n^{\′},n}=y_n-\underset{k\≠n-n^{\′}}{\overset{K-1}{\underset{k=0}{\mathrm{\Σ}}}}{H}_k{\tilde{a}}_{n-k};$

Calculate signal behind the interference delete respectively

With the correlation of M channel response vector, write the correlation register $c_{\mathrm{reg}}(n^{\′},m):c_{\mathrm{reg}}(n^{\′},m)=\left\{\begin{array}{cc}{\stackrel{\‾}{y}}_{n^{\′},n}^T{h}_{n-n^{\′},m}{,}^{}& n^{\′}=n\\ c_{\mathrm{reg}}(n^{\′},m)+{\stackrel{\‾}{y}}_{n^{\′},n}^T{h}_{n-n^{\′},m},& \mathrm{otherwise}\end{array},\right.$ Here m=0,1 ..., M-1;

In M correlation, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value: $m^{\′}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(n^{\′},m)\left|\right\},$ Value and c according to m ' _RegThe symbol of (n ', m ') writes interim judgement amount register ${\tilde{a}}_{n^{\′}}={[{\tilde{a}}_{n^{\′},0}{,\tilde{a}}_{n^{\′},1},\·\·\·,{\tilde{a}}_{n^{\′},M-1}]}^T,{\tilde{a}}_{n^{\′},m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(n^{\′},m^{\′})\right\},& m=m^{\′}\\ 0,& \mathrm{otherwise}\end{array}\right.$

When n '=n-K+1, obtain the conclusive judgement value of transmit symbol $a_{n-\mathrm{<figure-callout\; id="1"\; label="K\; +"\; filenames="S2008100579875E00011.png"\; state="\{\{state\}\}">K</figure-callout>}+1}={\tilde{a}}_{n-K+1}.$

Based on the low complex degree equalization method of iteration jam deleting, be applicable to the communication system of various straight prefaces or THSS in the spread spectrum communication system that the present invention proposes.It can be used for the wide-band spread spectrum communication system of monophasic waveform or many waveform modulated, also can be used for the ultra-wideband communication system of pulse amplitude or position modulation.

Consider general spread spectrum communication system, it transmits and can be expressed as

$s\left(t\right)=\sqrt{E_s}\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}\underset{m=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{a}_{n,m}{\mathrm{\&phi;}}_m(t-n{T}_s)$

E wherein _sThe energy of expression transmit symbol, N represents the code element number of launching, the waveform number that M expresses possibility, a _{N, m}Represent in n the code element amplitude of modulating in m the transmitted waveform.In a code element, only can select a waveform to launch, suppose that selecteed waveform is d in n code element _nIndividual, then to m=0 ..., M-1 has only the m=d of working as _nThe time a _{N, m}Value just be not 0.φ _m(t) be m possible transmitted waveform, its energy is normalized, promptly ${\&Integral;}_0^{\mathrm{Ts}}{\mathrm{\&phi;}}_m^2\left(t\right)=1,$ T _sThe expression code-element period.Generally speaking, a _{N, m}Be ambipolar, promptly its value is+1 or-1, can modulate 1 bit, and waveform is selected to modulate log ₂M bit, so each code element can be modulated (1+log ₂M) individual bit, the data transfer rate of system are R _b=(1+log ₂M)/T _s

Such expression formula has reflected various possible modulation and spread spectrum mode, and wherein the information of spreading code, jumping time-code and shaped pulse is included in transmitted waveform φ _m(t) lining.Fig. 1 has provided an example of transmitted waveform in the direct sequence spread spectrum system, waveform number M=4 wherein, and the length of direct sequence spread spectrum sign indicating number is 8, shaped pulse is a rectangular pulse, here waveform 1 φ in the formula above waveform 4 is promptly corresponding ₀(t) to φ ₃(t).Fig. 2 has provided an example of transmitted waveform in the THSS system, waveform number M=4 wherein, and the length of THSS sign indicating number is 4, shaped pulse is the second order Gauss pulse, with the waveform in the master drawing 1 φ in the formula above waveform 4 is promptly corresponding ₀(t) to φ ₃(t).As can be seen from Figure 2, in fact the relative position between the pulse does not change, for no other reason than that pulse position modulation makes the code element waveform that integral translation take place, so pulse position modulation can be regarded as a special case of many waveform modulated.Certainly as long as, different modulation waveforms can have different jumping time-codes, at this moment jumping the time-code message reflection to φ _m(t) just passable in.When M=1, this expression formula has just reflected direct sequence spread spectrum system that has only a spreading code or the THSS system that has only pulse amplitude modulation.

If definition emission amplitude vector

a _n＝[a _n，0，a _n，1，…，a _n，M-1] ^T

Wherein [] ^TThe transposition of representing matrix or vector, definition transmitted waveform vector

(t)＝[φ ₀(t)，φ ₁(t)，…，φ _M-1(t)] ^T

Then transmit and to be expressed as another kind of compacter form

Be the embodiment that example illustrates the inventive method with a short pulse ultra-wideband communication system below.According to the expression formula that transmits, the ENERGY E of transmit symbol is set _s=1, the transmit symbol number N=1000 of each packet, frequency spreading wave number M=4.Adopt the mode of pulse position modulation in conjunction with pulse amplitude modulation, at first select four possible pulse transmitting sites with two bits to be launched, from the angle of code element waveform, different pulse transmitting sites have promptly been represented different frequency spreading waves; And then with the polarity of a bit modulation pulse to be launched, so each code element can be transmitted three bits.Code-element period T is set _s=20ns, pulse width T _p=1ns, can obtain spreading gain is G=T _s/ T _p=20, the system data rate is R _b=(1+log ₂M)/T _s=150Mbps.

The impulse response of supposing multipath channel is expressed as h (t), and the impulse response of receiver matched filter is p (t), then can define the compound channel response

h _m(t)＝φ _m(t)*h(t)*p(t)

Wherein symbol " * " is represented convolution algorithm, and the compound channel response has reflected the common influence of transmitted waveform, multipath channel and receiver matched filtering.Because the base band signal process of modern receiver generally all is digitized, so the later received signal of matched filtering at first can become digital signal through over-sampling and quantification.From the angle of baseband signal, the channel response of equivalence also disperses, promptly like this

h _m[i]＝h _m(iT _c)，i＝0，1，…，KN _c-1

T wherein _cIn the expression sampling interval, K represents that the length of compound channel response is K code-element period, N _c=T _s/ T _cRepresent the sampling number in the code-element period.For the short pulse ultra-wideband communication system, T _cGenerally get pulse duration, at this moment N _cAlso can think all possible pulse position number.

In order to assess the performance of the inventive method, produce the impulse response of one group of ultra broadband multipath channel by Computer Simulation, statistical data packet is through the average error bit rate of receiving terminal after the multipath channel then.IEEE 802.15.4a channel model is used in emulation, and it is based on Saleh-Valenzuela sub-clustering channel model and actual measurement data is set up.Use " CM2 " environment wherein, just do not have inhabitation place, direct projection path, produce 1000 secondary channels altogether and realize with statistical average bit error rate (BER).Under this environment, the root mean square time delay of multipath channel expands to 19ns, and maximum delay expands to 100ns, so intersymbol interference covers 5 code elements.

H _m[i] is periodic segment, the wherein N of k section with the Baud Length _cBar is directly formed the channel response vector

h _k，m＝[h _m[kN _c]，h _m[kN _c+1]，…，h _m[(k+1)N _c-1]] ^T

M channel response vector that belongs to the k section together formed channel response matrix

H _k＝

h _k，0，h _k，1，…，h _k，M-1」

Therefore received signal can be write as vector form

$y_n=\underset{k=0}{\overset{K-1}{\mathrm{\&Sigma;}}}{H}_k{a}_{n-k}+z_n,n=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;,N+K-2$

Y wherein _nThe received signal vector of representing n code element, z _nThe noise vector of representing n code element, they all comprise N _cIndividual element.z _nObey the Gaussian Profile of zero-mean, its covariance matrix is IN _c, wherein

Be the bilateral power spectral density of receiver front end thermal noise, the noise of its value during according to emulation is recently given, IN _cBe N _c* N _cThe unit matrix of dimension.Because the reason of multidiameter delay, N transmit symbol can obtain the received signal of N+K-2 code element at receiving terminal.In this expression formula, for a _N-kValue, when n-k＜0, a _N-kValue that should data fetch packet front end pilot frequency code element, and behind n-k＞N-1, because no longer include transmit symbol, so a _N-kShould get full zero vector, what at this moment receive all is multipath signal.

Use the low complex degree equalization method based on iteration jam deleting of the present invention's proposition and handle to received signal, may further comprise the steps:

1) in the pilot tone stage of received signal, use the methods such as relevant or least square of sliding to carry out channel estimating, obtain the estimated value h of compound channel response _m[i], m=0,1,2,3, i=0,1 ..., 99;

2) to each compound channel response h _m[i], i=0,1 ..., 99, from 100 footpaths, choose the L bar footpath of amplitude maximum, it is zero that the value in other footpath is composed, and in this embodiment, gets L=100 respectively, and 32,16 have carried out performance simulation, and when L=100, in fact all multipaths have all participated in data demodulates;

3) with the Baud Length be the cycle to the channel response segmentation, 20 footpaths in each section constitute a channel response vector, wherein k channel response vector h _{K, m}=[h _m[20k], h _m[20k+1] ..., h _m[20 (k+1)-1]] ^T, k=0,1 ..., 4, m=0,1,2,3, [] ^TThe transposition of representing matrix or vector.By 4 vector h that belong to the k section together _{K, m}Constitute channel response matrix H _k=[h _{K, 0}, h _{K, 1}, h _{K, 2}, h _{K, 3}], k=0,1 ..., 4;

4) initialization correlation register c _Reg(n, m)=0, n=0,1 ..., 1003, m=0,1,2,3 and interim judgement amount register ${\tilde{a}}_n=0,$ n＝-4，…，0，…，1003。Here, ${\tilde{a}}_n={[{\tilde{a}}_{n,0},{\tilde{a}}_{n,1},{\tilde{a}}_{n,2},{\tilde{a}}_{n,3}]}^T$ Be the vector of one 4 dimension, wherein

Be

M element;

5) in the data symbols stage of received signal, by receiving the data symbols order, from y ₀, y ₁To y ₁₀₀₃, carry out iteration jam deleting, obtain transmit symbol a ₀, a ₁To a ₉₉₉The conclusive judgement value, demodulate the information bit that transmitting terminal is sent here, finish whole message transmitting procedure.The concrete steps of carrying out iteration jam deleting are as follows:

(1) when receiving data symbols y ₀The time, carry out the operation of following three steps:

A. the known symbols a in the packet front end pilot frequency code element sequence _-1, a _-2, a _-3, a _-4Write interim judgement amount register These code elements of deletion are to a from received signal ₀The interference of signal obtains signal behind the interference delete

${\stackrel{\&OverBar;}{y}}_{\mathrm{0,0}}=y_0-\underset{k=1}{\overset{4}{\mathrm{\&Sigma;}}}{H}_k{\tilde{a}}_{-k}$

B. calculate signal behind the interference delete respectively

With the correlation of 4 channel response vectors, write correlation register c _Reg(0, m):

$c_{\mathrm{reg}}(0,m)={\stackrel{\&OverBar;}{y}}_{\mathrm{0,0}}^T{h}_{0,m}$

Wherein, m=0,1,2,3;

C. in 4 correlations, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

$m^{\&prime;}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(0,m)\left|\right\}$

Value and c according to m ' _RegThe symbol of (0, m ') writes interim judgement amount register

${\tilde{a}}_0={[{\tilde{a}}_{\mathrm{0,0}},{\tilde{a}}_{\mathrm{0,1}},{\tilde{a}}_{\mathrm{0,2}},{\tilde{a}}_{\mathrm{0,3}}]}^T$

${\tilde{a}}_{0,m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(0,m^{\&prime;})\right\},& m=m^{\&prime;}\\ 0,& \mathrm{otherwise}\end{array}\right.$

(2) when receiving data symbols y _n, 1≤n≤1003 o'clock, to n '=n, n-1 ... n-4 (0≤n '≤999) carries out the operation of following three steps respectively:

A. according to the interim judgement amount The signal of other code element of deletion is to a from received signal _{N '}The interference of signal obtains signal behind the interference delete

${\stackrel{\&OverBar;}{y}}_{n^{\&prime;},n}=y_n-\underset{k\&NotEqual;n-n^{\&prime;}}{\overset{4}{\underset{k=0}{\mathrm{\&Sigma;}}}}{H}_k{\tilde{a}}_{n-k}$

B. calculate signal behind the interference delete respectively

With the correlation of 4 channel response vectors, write correlation register c _Reg(n ', m):

$c_{\mathrm{reg}}(n^{\&prime;},m)=\left\{\begin{array}{cc}{\stackrel{\&OverBar;}{y}}_{n^{\&prime;},n}^T{h}_{n-n^{\&prime;},m},& n^{\&prime;}=n\\ c_{\mathrm{reg}}(n^{\&prime;},m)+{\stackrel{\&OverBar;}{y}}_{n^{\&prime;},n}^T{h}_{n-n^{\&prime;},m},& \mathrm{otherwise}\end{array}\right.$

Wherein, m=0,1,2,3;

C. in 4 correlations, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

$m^{\&prime;}=\underset{m}{\arg \max }\left\{\right|c_{\mathrm{reg}}(n^{\&prime;},m)\left|\right\}$

Value and c according to m ' _RegThe symbol of (n ', m ') writes interim judgement amount register

${\tilde{a}}_{n^{\&prime;}}={[{\tilde{a}}_{n^{\&prime;},0},{\tilde{a}}_{n^{\&prime;},1},{\tilde{a}}_{n^{\&prime;},1},{\tilde{a}}_{n^{\&prime;},3}]}^T$

${\tilde{a}}_{n^{\&prime;},m}=\left\{\begin{array}{cc}\mathrm{sign}\left\{c_{\mathrm{reg}}(n^{\&prime;},m^{\&prime;})\right\},& m=m^{\&prime;}\\ 0,& \mathrm{otherwise}\end{array}\right.$

If n '=n-4 obtains the conclusive judgement value of transmit symbol:

$a_{n-4}={\tilde{a}}_{n-4}$

That is to say a _nThe conclusive judgement value to wait until receiving symbol y _N+4Demodulation just can obtain after finishing.According to vector a _nValue, can know position and polarity that transmitting terminal is modulated in each code element, thereby can demodulate two bits of modulation positions and a bit of polarity of modulation.When obtaining a ₉₉₉The conclusive judgement value and demodulate after all bits, the message transmitting procedure of whole packet finishes.

As can be seen, in the low complex degree equalization method that the present invention proposes, main amount of calculation is the related operation in the iteration jam deleting process, promptly calculate the correlation of signal and M channel response vector behind the interference delete, need K interference delete operation because detect each transmit symbol, so the total related operation in each code element place is MK.In asking the process of correlation, main amount of calculation is multiplying, in the above example, if L=KN _c=100, promptly do not use sparse tap, then each code element place needs 400 multiplication; If adopt sparse tap, as L=16, then each code element place only needs 64 multiplication.When adopting identical tap number, it is identical that the multiplication computation amount of iteration jam deleting equalization methods receives with MRC-Rake.Linear MMSE equilibrium amount of calculation when Data Detection also is so many, but it will consume a large amount of amounts of calculation in tap selection and coefficient training stage.For example, use coupling to pursue and attack algorithm and select 16 taps, its amount of calculation is about 16 ²The magnitude of * 100=25600 multiplication; Use least mean square algorithm (LMS) to train equalizer coefficients adaptively, if finish the coefficient training after the iteration 100 times, then its amount of calculation is about the magnitude of 1600 multiplication.The amount of calculation of inverted sequence decision feedback equalization method depends on the length that the feedback fraction deletion is disturbed, when feedback fraction is deleted all inter symbol interference, the feedforward part only need be carried out 4 related operations at each code element place, do not carry out tap and only need to select 80 multiplication yet, but it will be appreciated that below, because wasted too much multipath energy, the performance of this equalization methods is very bad.In the middle of ultra-wideband communication system, the amount of calculation of channel estimating part can be accomplished very little, if design pilot frequency code element place does not have inter-pulse interference, then only with simply on average just obtaining channel estimation value.

Fig. 3 has provided the iteration jam deleting equalization methods of the present invention's proposition and the performance of other method of reseptance compares, wherein the solid line of " rhombus " mark is represented the performance curve of inverted sequence decision feedback equalization method (Rev-DFE), the dotted line of mark is represented the performance curve of high specific merging Rake method of reseptance (MRC-Rake) " to go up triangle ", the solid line of " cross " mark is represented the performance curve of iteration jam deleting equalization methods (IIC), the dotted line of " circle " mark is represented the performance curve of linear MMSE equalization methods (MMSE-EQU), the solid line of " rectangle " mark is represented matched filtering circle (MFB), it is meant the getable best receptivity of channel matched filtering when not having intersymbol interference, this is the optimal performance under the ideal conditions, is the performance upper bound (error rate lower bound) of various equalization methods.Abscissa is with E among the figure _b/ N ₀The signal to noise ratio of formal definition, E _bThe emitted energy of representing every bit, N ₀The one-sided power spectrum density of expression noise, ordinate is bit error rate (BER).In the emulation of this figure, the whole bag of tricks has all used whole multi-path locations to place tap, and number of taps is 100.As can be seen, the bit error rate of inverted sequence decision feedback equalization is the highest, secondly is that high specific merging Rake receives, and the balanced performance with linear MMSE equilibrium of iteration jam deleting is suitable, is much better than the performance that high specific merges Rake reception and inverted sequence decision feedback equalization.

The performance that Fig. 4 has provided several method when selecting the maximum diameter position as tap position compares, wherein the solid line of " star " mark represents that tap number is the performance curve that 16 high specific merges Rake method of reseptance (MRC-Rake-16), the dotted line of " following triangle " mark represents that tap number is the performance curve that 32 high specific merges Rake method of reseptance (MRC-Rake-32), the solid line of " cross " mark represents that tap number is the performance curve of 16 linear MMSE equalization methods (MMSE-EQU-16), the dotted line of " circle " mark represents that tap number is the performance curve of 16 iteration jam deleting equalization methods (IIC-16), the solid line of " rhombus " mark represents that tap number is the performance curve of 32 linear MMSE equalization methods (MMSE-EQU-32), the dotted line of " go up triangle " mark represents that tap number is the performance curve of 32 iteration jam deleting equalization methods (IIC-32), the solid line of " rectangle " mark is still represented matched filtering circle (MFB), and matched filter circle also is to use whole 100 taps, represents the performance upper bound of the whole bag of tricks.Because the performance of inverted sequence decision feedback equalization method is bad, so there is not its performance when tap is selected of emulation again.When doing the tap selection,, in real system, can not realize, so we have used with the method for maximum diameter position as tap position because optimum multivariate joint probability searching method amount of calculation is too big.Abscissa is with E among the figure _b/ N ₀The signal to noise ratio of formal definition, ordinate are bit error rates.As can be seen from the figure, use after the tap selection, under identical number of taps condition, the performance of iteration jam deleting equalization methods is better than linear MMSE equalization methods, more is better than high specific and merges the Rake method of reseptance.From with the contrast of accompanying drawing 3 as can be seen, when using 32 taps, several method can be near they performances when using whole tap.Comprehensive, the iteration jam deleting equalization methods that the present invention proposes has low complex degree and high performance characteristics concurrently, is fit to various forms of spread spectrum communication systems, has excellent application value and potentiality.

Claims (1)

1. In the spread spectrum communication system based on the low complex degree equalization method of iteration jam deleting, it is characterized in that, may further comprise the steps:

   1) in the pilot tone stage of received signal, use is slided relevant or least square method is carried out channel estimating, obtains the estimated value h of compound channel response _m[i], m=0,1 ..., M-1, i=0,1 ..., KN _c-1, wherein M represents the number of the frequency spreading wave that communication system is used, and K represents the code element number that the compound channel response covers, N _cRepresent the sampling number in the code-element period;

   2) to each compound channel response h _m[i], i=0,1 ..., KN _c-1, from KN _cChoose the L bar footpath of amplitude maximum in the bar footpath, it is zero that the value in other footpath is composed; Wherein, 0＜L≤KN _c

   3) with the Baud Length be the cycle to the channel response segmentation, the N in each section _cBar directly constitutes a channel response vector, wherein k channel response vector h _{K, m}=[h _m[kN _c], h _m[kN _c+ 1] ..., h _m[(k+1) N _c-1]] ^T, k=0,1 ..., K-1, m=0,1 ..., M-1, [] ^TThe transposition of representing matrix or vector; By M vector h that belongs to the k section together _{K, m}Constitute channel response matrix H _k=[h _{K, 0}, h _{K, 1}..., h _{K, M-1}], k=0,1. ..., K-1;

   4) initialization correlation register c _Reg(n, m)=0, n=0,1 ..., N+K-2, m=0,1 ..., M-1 and interim judgement amount register N=-K+1 ..., 0 ..., N+K-2, N represents a code element number in the packet here, Be the vector of a M dimension, wherein Be

   

   M element;

   5) in the data symbols stage of received signal, by receiving the data symbols order, from y ₀, y ₁To y _N+K-2, carry out iteration jam deleting, obtain transmit symbol a ₀, a ₁To a _N-1The conclusive judgement value, demodulate the information bit that transmitting terminal is sent here, finish whole message transmitting procedure;

   The concrete steps of carrying out iteration jam deleting in the described step 5) are as follows:

   (1) when receiving data symbols y ₀The time, carry out the operation of following three steps:

   A. the known symbols a in the packet front end pilot frequency code element sequence _-1, a _-2..., a _-K+1Write interim judgement amount register

   

   These code elements of deletion are to a from received signal ₀The interference of signal obtains signal behind the interference delete

   

   

   B. calculate signal behind the interference delete respectively With the correlation of M channel response vector, write correlation register c _Reg(0, m):

   

   Wherein, m=0,1 ..., M-1;

   C. in M correlation, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

   

   Value and c according to m ' _RegThe symbol of (0, m ') writes interim judgement amount register

   

   

   

   (2) when receiving data symbols y _n, during 1≤n≤N+K-2, to n '=n, n-1 ... n-K+1, and 0≤n '≤N-1, carry out the operation of following three steps respectively:

   A. according to the interim judgement amount

   

   The signal of other code element of deletion is to a from received signal _{N '}The interference of signal obtains signal behind the interference delete

   

   

   B. calculate signal behind the interference delete respectively

   

   With the correlation of M channel response vector, write correlation register c _Reg(n ', m):

   Wherein, m=0,1 ..., M-1;

   C. in M correlation, search for maximum, thereby obtain the numbering m ' of the pairing frequency spreading wave of maximum related value:

   

   Value and c according to m ' _RegThe symbol of (n ', m ') writes interim judgement amount register

   

   

   

   If n '=n-K+1 obtains the conclusive judgement value of transmit symbol:

   

Images