Background technology
In the transmission of broadband communications system, especially in the wireless broadband communication transmission system, narrow-band interference signals such as the same frequency of analog or digital or adjacent interference frequently can have a strong impact on transmission performances.Therefore, filtering just becomes a kind of the most basic and extremely important technology of electronic information process field, utilizes filtering technique to extract needed signal from the signal of complexity, suppresses noise or interference simultaneously.Filter has in fact also just become a kind of frequency-selective system, and it gives very little decay to some frequency signal, and other unwanted signals are given very big decay, stops unwanted signal to pass through as far as possible.At the electronic system median filter is a kind of basic element circuit, uses a lot, and technology is also comparatively complicated, and the performance of the good and bad directly decision product of while filter is so much country pays much attention to filter theory research and product development.
If interference signal a little less than, do not influence system synchronously, can binding data, adopt several different methods, modes such as for example traditional feed forward equalization or feedback equalization, soft mapping, iteration equalizing even turbo equilibrium are come anti-interference, can both obtain good performance.There are many argumentations relevant for these anti-disturbance methods in recent some for CDMA and ofdm system.But when interference signal is very strong, will have influence on the synchronous of system.
In three more than ten years in the past, people have proposed many circuit or method strengthens useful signal, or separation/elimination noise and interference signal.In these methods, different circuit structures suppresses or eliminates the ability difference of interference signal.Although those utilize the method for training sequence or known pilot signal that certain effect is arranged, blind processing mode seems more attractive.Wherein the method for the blind processing of a class is exactly to utilize trapper to go to eliminate this interference signal after obtaining the centre frequency of interference signal.This estimation utilizes power spectrum method (after this signal is carried out the FFT conversion, asking its energy spectrum), even but carry out FFT continually and estimate, follow the tracks of the still non-constant of variation of interference signal.
The blind processing method of other types mainly adopts Adaptive Signal Processing.Sef-adapting filter belongs to modern filter category, sef-adapting filter is for fixed filters, fixed filters belongs to classical filter, its frequency filtering is fixed, the frequency of sef-adapting filter filtering is a self adaptation input signal and changing, so its scope of application is wider.Under without any condition about the priori of signal and noise, sef-adapting filter utilizes previous moment to obtain the filter parameter that filter parameter is regulated now automatically, with the statistical property of adaptation signal and noise the unknown or change at random, thereby realize optimal filter.Self adaptive filter is exactly in fact that a kind of himself transmission characteristic of regulating has reached optimized Weiner filter.
Generally speaking, sef-adapting filter is made up of two parts, the one, filter construction, the 2nd, the adaptive algorithm of adjustment filter coefficient, the structure of sef-adapting filter adopts FIR or IIR structure all can, because there is stability problem in iir filter, the therefore general structure that adopts the FIR filter as sef-adapting filter.
Fig. 1 is the structural representation of the existing sef-adapting filter of using always, wherein x (n) is an input signal, after input signal x (n) is postponed, carries out Error Calculation with output signal y (n) again, carry out the filtering parameter adjusting thereby send into sef-adapting filter behind the acquisition control information e (n), that is:
h
n(i)=h
n-1(i)+μ·conj(x(i))·e(i)
E (i)=x (n-D)-y (n), h
n(i) be transfer function, μ is a constant, and conj asks conjugation, elongatedness when D is, and the purpose of adjustment is to make error signal e (i) minimum.
As shown in Figure 2, it is the lateral type Filter Structures schematic diagram of using always, and sef-adapting filter also has a cover adaptive algorithm except comprising a filter according to certain structural design, and adaptive algorithm is judged according to certain and designed.The algorithm of sef-adapting filter is mainly basic as calculating with various criterion conditions.Two kinds of criterion conditions are arranged usually: least mean-square error criterion and least square method criterion.(it is that the mean-square value of output error sequence e (i)=x (n-D)-y (n) is minimized, and comes power of amendment coefficient according to this criterion for Least Mean Square, LMS) algorithm for least mean-square error.
In addition, be american documentation literatures such as 4052559,4238746,005325204,4673982,4524424,4420815,6976044 in the patent No.; IEEE journal the 63rd volume of publishing in December, 1975, the 12nd phase, in the papers of people about " adaptive noise elimination " such as 1692-1716 page or leaf and Widrow, all related to the algorithm approximate to the real-time lowest mean square of wiener-Hoff filter, all these documents all are to adopt IIR or FIR filter to eliminate interference signal, and upgrade the coefficient of filter according to the result of the time delay cross-correlation of the signal of the output of filter and input.But the convergence rate of these methods is very slow, and can only resist interference signal with specific frequency spectrum shape, this means that these methods can not be widely used in scene more generally, if any the fast-changing environment of the centre frequency of frequency hopping interference or interference signal.
Therefore, press for a kind of more excellent sef-adapting filter.
Embodiment
See also Fig. 3, the blind adaptive filter that can eliminate narrow band interference of the present invention mainly can comprise: sef-adapting filter, first delay cell and error calculation unit etc.
Described sef-adapting filter is used for and will be exported y (n) after input signal x (n) filtering, and it can be the FIR filter, also can be iir filter.The internal structure of FIR filter and iir filter is known to those skilled in the art to be known, so be not described in detail in this; And those skilled in the art can select different sef-adapting filters for use voluntarily according to different application scenarios.
Described first delay cell is coupled in described sef-adapting filter, is used for the output signal according to the default described sef-adapting filter of the first time delay length delay.Usually, elongatedness can be determined according to the application scenarios of reality in first o'clock, as when being applied to many subbands situation in the communication system and single frequency band scenario, can adopt different time delays.And the output signal that postpones can be current output signal y (n), also can comprise the signal of exporting before the sef-adapting filter.
Described error calculation unit is coupled in the described sef-adapting filter and first delay cell, be used for the current output signal of described sef-adapting filter and the output signal of described first delay cell are carried out extracting first control information behind the related operation, so that described first control information is fed back to described sef-adapting filter, adjust its parameter for described sef-adapting filter, and then eliminate the interference in the back input signal that described sef-adapting filter inserts.When calculating target function, can adopt the above norm of second order or second order.Need to prove that the norm exponent number of employing is high more, anti-interference is good more, can be complicated more but implement circuit, and therefore, during practical application, can adopt the norm of suitable exponent number to carry out correlation computations as required.In the present embodiment, for example:
Then the output y (n) of sef-adapting filter can be expressed as:
Wherein N is a sef-adapting filter length, and T is a transposed operator, and * is an adjoint operator, minimizes for making control information (being target function), just needs to make C (n
1, n
2..., n
k)=E{y (n) y (n+n
1) ... y (n+n
k) norm minimize, in other words
f=E{‖C(n
1,n
2,…,n
k)‖
p}→minf
Wherein, ‖ ‖
pBe p rank norms, especially, ‖ ‖ is the second order norm.
Calculate for simplifying, select the second-order statistic definition:
f=E{‖y(n)y
*(n-S/2)‖}→minf,
Therefore:
If adopt least mean square algorithm LMS, under the FIR framework, can obtain following result:
e(i)=y(n-1)
*‖y(n-S/2-1)‖x(n-i)+y(n-1-S/2)
*‖y(n-1)‖x(n-i-S/2),i=0,…,N-1,h(i)=h(i)-μe(i)。
If output vector is:
In order to weigh treatment effect, then
The correlated results of output vector just is:
In addition, need to prove, above-mentioned is to adopt least mean square algorithm (being LMS) to calculate minimal error in time domain, also can adopt recurrent least square method (being RLS), or other forms of algorithm, and be not limited only to calculate in time domain, also can calculate in the conversion territory, as in frequency-domain calculations etc., describe in detail no longer one by one at this.
Have again, at some application scenarios, also can in device, set up one second delay cell again, be used for that input signal x (n) with the described sef-adapting filter of input sends into described error calculation unit after according to the second default time delay length delay and adjust its parameter for described sef-adapting filter so that carry out obtaining corresponding second control information behind the related operation with the output signal y (n) of described sef-adapting filter, this is similar with prior art, so this also no longer stating more.
Also have, if the correlation of sef-adapting filter output signal y (n) is very weak, for example be lower than a certain preset threshold, in the case can be the bypass of adaptive narrow-band interference cancellation circuit to save power consumption, thus, one judging unit can be set in device, be used to judge whether first control information that described error calculation unit calculates is lower than preset threshold value so that determine that whether needing that input signal is sent into described sef-adapting filter carries out interference filtering, if first control information that calculated be lower than preset threshold value, then input signal x (n) is directly sent into unit such as subsequent decoding and directly decode, and do not need through sef-adapting filter filtering.
The above-mentioned blind adaptive filter that can eliminate narrow band interference is widely used, and for example, can disturb (ACI) filter combination with the neighbour frequently, to eliminate arrowband and adjacent interference frequently, also can be applicable in the multi-sub-band system, is used to eliminate the narrow band interference in each subband.
As shown in Figure 4, it is a kind of transmission of broadband communications system schematic, in described system, comprise the above-mentioned blind adaptive filter that can eliminate narrow band interference, this the blind adaptive filter that can eliminate narrow band interference is arranged on before the adjacent interference filter frequently, input signal x (n) is admitted to demodulator, so that demodulate the communication information after handling through narrow band interference and adjacent frequency interference successively.
Again as shown in Figure 5, it is another kind of transmission of broadband communications system schematic, with different being of transmission of broadband communications system shown in Figure 4, the blind adaptive filter that can eliminate narrow band interference is arranged on after the adjacent interference filter frequently, after input signal x (n) handles through adjacent frequency interference earlier, again by the blind adaptive filter filtering narrow band interference that can eliminate narrow band interference, then by the demodulator demodulation.
See also Fig. 6 again, it is many subbands transmission of broadband communications system schematic, this many subbands transmission of broadband communications system comprises: a plurality of subband access units, it is subband access unit 1, subband access unit 2, subband access unit 3 ... subband access unit N, be respectively applied for and insert different subband signals, each subband access unit all is coupled with the blind adaptive filter that can eliminate narrow band interference, the blind adaptive filter 1 that can eliminate narrow band interference is coupled in subband access unit 1, the blind adaptive filter 2 that can eliminate narrow band interference is coupled in subband access unit 2, the blind adaptive filter 3 that can eliminate narrow band interference is coupled in subband access unit 3 ... the blind adaptive filter N that can eliminate narrow band interference is coupled in subband access unit N, each subband signal of having eliminated narrow band interference forms full communication information after being admitted to the assembled unit combination.
Need to prove, above-mentioned Fig. 4-communication system shown in Figure 6, each is self-contained can eliminate narrow band interference the blind adaptive filter structure all as previously mentioned, promptly all comprise sef-adapting filter, first delay cell, reach error calculation unit etc., certainly, also can set up second delay cell and judging unit etc. again according to reality.
See also Fig. 7 again, it is the workflow schematic diagram that blind adaptive is eliminated the method for narrow band interference in the transmission of broadband communications system.At first sef-adapting filter (for example being FIR filter or iir filter etc.) will be exported y (n) after current input signal x (n) filtering, then, signal (comprising current output signal and the signal of exporting before) with the described sef-adapting filter output of first o'clock elongatedness (it can be determined by the application scenarios of reality) delay, again with the signal y (n) of the current output of described sef-adapting filter and the signal after postponing carry out related operation (for example can be in time domain, also can change the territory at other adopts LMS or RLS scheduling algorithm to calculate second order or the norm more than the second order) to obtain control information e (n) (can adopt LMS or RLS to extract), then, described control information e (n) is fed back to described sef-adapting filter, adjust filter parameter by described sef-adapting filter filtering according to the control information e (n) that feeds back to, so that eliminate the interference among the follow-up input signal x (n+i).
In sum, the blind adaptive filter of narrow band interference of eliminating of the present invention is by carrying out related calculation to sef-adapting filter output signal and output delay of output signal signal, extract control information thus, adjust the filtering parameter of sef-adapting filter, it is applicable to China Mobile multimedia broadcasting (CMMB) receiver and other DTV receiver; Also can be used for the broadband reception device, comprise Digital Television and telecommunication receiver etc., the situation for a plurality of strong narrow band interference are arranged also can be suitable for; For sweep rate is tens of milliseconds a plurality of strong narrow band interference, also can resist.The present invention compares with existing adaptive approach, and it is less to the channel effect of superposition, and is less to the receiver performance damage; Secondly, it has response speed faster, and certain follow-up control is arranged.And compare with traditional notch filter, it does not need to predict the spectral shape of interference signal, the flase drop problem in also can avoiding simultaneously detecting.
The foregoing description just lists expressivity principle of the present invention and effect is described, but not is used to limit the present invention.Any personnel that are familiar with this technology all can make amendment to the foregoing description under spirit of the present invention and scope.Therefore, the scope of the present invention should be listed as claims.