CN101286958B - Terrestrial broadcasting channel estimating method for digital television and device thereof - Google Patents

Abstract

The invention relates to a method for estimating channel of digital TV terrestrial broadcasting and a device thereof, which pertains to the technical filed of digital information transmission and is applicable to national standard digital TV terrestrial broadcasting (GB 20600-2006) systems, digital communication systems, telemetry and telecontrol systems and other relevant systems and technical fields. Under the frame head modes of PN420 and PN945 of digital TV terrestrial broadcasting, the multipath channel impulse response which falls into the prefix length range of the frame head PN sequence can be estimated by the following processing of sampling sequence of symbol rate after synchronization: (a) synchronization and pre-processing; (b) cyclic correlation; (c) channel impulse molding. The method and the device of the invention utilize diversity gain brought by the PN sequence correlation, have good anti-noise performance, can effectively reduce interference resulting from non-idealautocorrelation of the PN sequence , accurately and fast estimate the multipath channel impulse response which falls into the prefix length range of the frame head PN sequence and have the advantagesof high estimation accuracy, small calculation amount and low implementation complexity compared with other methods.

Description

A kind of terrestrial broadcasting channel estimating method for digital television and device thereof

Technical field

The invention belongs to digital information transmission technical field, be applicable to related system and technical fields such as Digital Television Terrestrial Broadcasting (GB 20600-2006) system, digital communication system, telemetering and remote control system, be specifically related to the channel estimation technique of a kind of Digital Television Terrestrial Broadcasting (GB20600-2006) PN420 and PN945 frame head mode.

Background technology

Digital TV broadcast signal is propagated in the multipath channel of ground, because the signal that reflected by objects etc. makes receiving terminal receive in the surrounding environment is many stacks that transmit that different propagation delays and channel fading are arranged, it is the time delay expansion, especially the time delay expansion can cause serious intersymbol interference (ISI) concerning the wideband digital television terrestrial broadcasting, and then cause serious reception decoding error, therefore need the influence of multipath channel is compensated at receiving terminal; Compensation process generally comprises channel estimating and balanced two steps, wherein channel estimating is that receiving terminal is in order to know channel condition information (CSI, Channel Status Information) processing, balanced promptly the reduction according to the channel condition information that obtains transmits.Channel estimating and equilibrium can also can be carried out from time domain from frequency domain; If can accurately estimate channel impulse response theoretically, and do not consider implementation complexity, time domain equalization and frequency domain equalization are equivalent so.In addition, the system that becomes when the terrestrial wireless channel is, its impulse response changes over time, thus the channel estimating of receiving terminal and balanced want can adaptive channel time-varying characteristics, i.e. adaptive equalization; Usually by in transmitting, periodically inserting the channel fading that known frequency pilot sign or training sequence are estimated the frequency domain response characteristic of channel impulse response or equivalence for receiving terminal and then become when compensating adaptively.

As previously mentioned, if can know the time domain impulse response or the frequency domain response characteristic of equal value of channel, then can carry out equilibrium to received signal in time domain or frequency domain easily.One is carried out channel estimating at frequency domain is the digital television standard DVB-T in Europe with balanced exemplary, the COFDM multicarrier transmission systems that it adopts inserts continuous pilot and scattered pilot signal at frequency domain especially, the position of continuous pilot in each COFDM symbol all fixed, for example in the 8k pattern, insert 177 continuous pilot, in the 2k pattern, inserted 45 continuous pilot.The position of scattered pilot is different in different COFDM symbols, and with four COFDM symbols is loop cycle, and as shown in Figure 1, wherein gray circles is represented the scattered pilot subcarrier, black circles is represented the TPS pilot sub-carrier, and the oblique line circle is represented the continuous pilot subcarrier.Can estimate to obtain the channel frequency characteristic of discrete time point and subcarrier frequency during system works by known pilot symbols, can obtain whole T/F two dimensional channel frequency response characteristic by interpolation again, and then can be in the single order isostatic compensation channel multi-path decline of frequency domain by each subcarrier.

Digital tv ground broadcasting is a scheme that the single and multi-carrier modulation is merged, single and multi-carrier modulating mode frame structure is identical, be the Hierarchichal frame structure of digital television ground broadcast transmission system as shown in Figure 2, its elementary cell-signal frame is made up of frame head and data block, wherein frame head is generated by the PN sequence, and data block adopts single-carrier modulated or OFDM multi-carrier modulation.

Difference according to frame head can be divided into PN420, three kinds of frame head modes of PN595 and PN945, the main distinction of different frame head mode frame structures are different frame head PN sequences, but they all are to be got by PN sequence cyclic-extension (as PN420 and PN945) or intercepting (as PN595).Wherein the frame head PN sequence under PN420 and the PN945 frame head mode is respectively by 255 and 511 length of PN sequence (be called and generate the PN sequence), and adds respectively that in front and back the Cyclic Prefix of certain-length and cyclic suffix obtain, shown in Fig. 3 (a); The phase deviation PN sequence that the frame head PN sequence of this structure of while also can be regarded 255 or 511 length as adds Cyclic Prefix or suffix composition, respectively as Fig. 3 (b), shown in Fig. 3 (c).Why being referred to as phase deviation PN sequence, is because for the generation PN among Fig. 3 (a), this PN sequence has fixing initial phase offset with it.

Frame head is periodically inserted in the frame emission sequence as known training sequence, and a kind of channel multi-path fading compensation scheme that can be used for the Digital Television Terrestrial Broadcasting reception is: based on the time domain impulse response of known frame head PN sequencal estimation channel and equilibrium subsequently.Special construction in particular for PN420 and PN945 frame head, and because the circulation autocorrelation performance of the near ideal of PN sequence, the multipath that PN420 and PN945 pattern are fallen into frame head PN sequence prefix length scope can utilize the relevant correlated series that obtains of conjugation of the frame head symbol sebolic addressing of local circular correlation PN sequence and reception, is calculated the estimation of channel time domain impulse response sequence again by this correlated series.For from correlated series calculating channel impulse response sequence, existent method or ignore the autocorrelative non-ideal characteristic of PN sequence cycles is directly dwindled with equal proportion (peak value of PN sequence cycles auto-correlation function) by correlated series and is obtained the estimation of channel impulse response; Or the way of employing iteration, big based on the path that signal power is big to the influence in other paths, other path influences insignificant hypothesis to it simultaneously, from correlated series, find the path of amplitude maximum successively, and eliminate its influence to other paths, look for time the strongest path to do identical processing again, and the like.As can be known, though first method is simple, error has been introduced in an approximate processing artificially; Second method adopts the processing of iteration can reduce the imperfect relevant influence that brings of PN sequence to a certain extent, but complicated difficult realization of this alternative manner particularly searched for the maximum path process and be unsuitable for real-time processing.

Summary of the invention

The objective of the invention is to propose under a kind of Digital Television Terrestrial Broadcasting PN420 and the PN945 frame head mode, by to processing, the multipath channel impulse response that falls into frame head PN sequence prefix length scope is carried out fast and the low-complexity method and the device of accurately estimation through synchronous character rate sample sequence later.

Terrestrial broadcasting channel estimating method for digital television of the present invention has identical processing method and step for two kinds of frame head modes of PN420 and PN945, and just system parameters is different, as shown in table 1.

Table 1

A kind of terrestrial broadcasting channel estimating method for digital television of the present invention, under Digital Television Terrestrial Broadcasting PN420 or PN945 frame head mode, can be for the multipath channel impulse response that falls into frame head PN sequence prefix length scope by the following processing of character rate sample sequence is synchronously later estimated that it contains following steps successively:

(1) reach preliminary treatment synchronously,

(a) according to frame synchronization information, the synchronous character rate sample sequence F of i signal frame later _iBe decomposed into frame head P _iWith data block D _i, and then obtain the circular correlation sequence P of this frame _i'={ P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI).F wherein _i={ F _i(1), F _i(2) ..., F _i(L _F), P _i={ P _i(1), P _i(2) ..., P _i(L _PI), D _i={ D _i(1), D _i(2) ..., D _i(3780) } be the synchronous i signal frame character rate sample sequence of warp that receives respectively, the frame head part and the data block portions of i signal frame sequence have relation between them:

P _i(j)＝F _i(j)，j＝1，2，...，L _PI

D _i(j)＝F _i(j+L _PI)，j＝1，2，...，3780

Wherein 3780 correspondences is the IDFT length of digital television ground broadcast signal frame data section.

(b), generate the relevant PN sequence PN of local initial cycle of i frame according to the synchronization parameter of present frame _{I, 1}={ PN _{I, 1}(1), PN _{I, 1}(2) ..., PN _{I, 1}(L _PN), relevant PN sequence of the initial cycle that this this locality produces and transmitting terminal constitute the back L of corresponding i frame frame head _PNThe PN sequence that individual symbol constituted is identical, promptly is a phase deviation PN sequence;

(2) circular correlation calculates correlated series c _i={ c _i(1), c _i(2) ..., c _i(L _C), contain following steps successively:

(a) count initialized device j=1;

(b) with circular correlation sequence P _i' and PN _{I, j}It is relevant to do conjugation, obtains c _i(j), promptly

$c_i\left(j\right)=\underset{k=1}{\overset{L_{\mathrm{PN}}}{\mathrm{\Σ}}}{P}_i(k+L_C)\×{\mathrm{PN}}_{i,j}^{*}\left(k\right)$

(c) ring shift right upgrades local circular correlation PN sequence, obtains PN _{I, j+1}:

PN _i，j+1＝{PN _i，j(L _PN)，PN _i，j(1)，PN _i，j(2)，...，PN _i，j(L _PN-1)}

(d) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish;

(3) channel impulse response moulding is from correlated series c _i={ c _i(1), c _i(2) ..., c _i(L _C) calculate the current frame channel impulse response and estimate h _i={ h _i(1), h _i(2) ..., h _i(L _C), contain successively and have the following steps:

(a) count initialized device j=1;

(b) calculating channel impulse response pre-estimation h _i' (j)

$h_i^{\&prime;}\left(j\right)\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1+c_i\left(j\right)/{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00042.png"\; state="\{\{state\}\}">C</figure-callout>}}_2$

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * and P, P is the power normalization factor, is the symbol power P of the transmitting terminal PN sequence that receives _STThe symbol power P of the circular correlation PN sequence that produces with this locality _SRLong-pending.

(c) abandon h _i' (j) in insecure little level value to reduce The noise, obtain channel impulse response estimation h _i(j), that is:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array}\right.$

H wherein _TBe one and abandon decision threshold that desired different antinoise of visual application and multi-path resolved sensitivity decide.

D) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish.

The Digital Television Terrestrial Broadcasting channel estimating apparatus structure that the present invention proposes includes:

(a) state controller, the input synchronous control signal, according to synchronous control signal and system clock produce, the update system status signal is in order to indicate synchronously and the preliminary treatment state, sub-state under circular correlation state or channel impulse response completed state and each treatment state carries out respective handling to control other module operating state to input data sequence;

(b) the character rate sampled signal frame sequence after frame head separator, its input connect synchronously has a circular correlation sequence output port; Reach the preliminary treatment state synchronously, the frame head separator is with circular correlation sequence P _i' from continuous character rate sampled signal frame F _iIn isolate, and export to the correlated series buffer unit;

(c) correlated series buffer unit, it is a L _PNThe register series of level is in order to storage circular correlation sequence, and its input connects the circular correlation sequence output of frame head separator, and with the L that stores in the register _PNIndividual data symbol is exported to the circular correlation device; Reach the preliminary treatment state synchronously, receive successively by the frame head separator and separate the circular correlation sequence P that obtains _i', and with the L of this sequence _PNIndividual symbol { P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI) send into register series storage successively; At the circular correlation state, the circular correlation sequence symbol data of storage are exported to the circular correlation device in order to carry out the circular correlation computing;

(d) circular correlation PN generator produces circular correlation PN sequence according to state control signal, exports to the circular correlation device in order to carry out the circular correlation computing;

(e) circular correlation device, it has two inputs: by the circular correlation sequence of correlated series buffer unit input and the circular correlation PN sequence of being imported by circular correlation PN sequence generator; At the circular correlation state, finish common L _CThe related operation of inferior circular correlation sequence and circular correlation PN sequence conjugation obtains correlated series c _i(j), j=1,2 ..., L _C

(f) channel impulse response former, input is from the correlated series c of circular correlation device _i(j), j=1,2 ..., L _C, carry out the channel impulse response forming processes, calculate output estimated channel impulse response sequences h _i(j), j=1,2 ..., L _C

Beneficial effect of the present invention: the channel estimation methods of the present invention's definition and device thereof have utilized the relevant diversity gain that brings of PN sequence, noise robustness is good, and can effectively reduce the interference that the imperfect auto-correlation of PN sequence is brought, accurately with the multipath channel impulse response of estimating to obtain falling into frame head PN sequence head prefix length scope apace, and compare additive method and have the estimated accuracy height, amount of calculation is little, the advantage that implementation complexity is low.

Description of drawings:

Fig. 1 be among the European DVB-T pilot tone the time-the frequency locus;

Fig. 2 is the digital television ground broadcast transmission system Hierarchichal frame structure;

Fig. 3 is the frame head structure under Digital Television Terrestrial Broadcasting PN420 and the PN945 frame head mode;

Fig. 4 is a channel impulse response estimation method flow of the present invention;

Fig. 5 is that the present invention reaches pretreatment process synchronously;

Fig. 6 is a circular correlation handling process of the present invention;

Fig. 7 is a channel impulse response forming processes flow process of the present invention;

Fig. 8 is a channel impulse response estimation structure drawing of device of the present invention;

Fig. 9 is that the time domain of character rate sample frame header sequence under the multipath channel constitutes schematic diagram;

Figure 10 is that length is the circular correlation function curve of the PN sequence of L;

Figure 11 (a) is a multipath channel impulse response that the strong footpath of three 0dB is arranged;

Figure 11 (b) is the channel estimating simulation result of channel estimation methods of the present invention;

Figure 12 is that channel estimation methods of the present invention and other channel estimation methods performances compare;

Figure 13 is that channel estimation methods of the present invention compares with other channel estimation methods performances under the Complex Channel.

Embodiment

As previously mentioned, for the broadcasting of the broadband wireless in the terrestrial wireless channel, receiving terminal is not done compensation as the ISI (intersymbol interference) that multipath channel is caused, will cause serious decoding error; Estimate and then the frequency selective fading that multipath causes compensated that Digital Television Terrestrial Broadcasting has periodically been inserted frame head PN sequence as known training sequence in its frame structure definition for ease of CSI to time varying channel; Therefore concerning the reception of Digital Television Terrestrial Broadcasting, a kind of suitable channel multi-path fading compensation scheme can be divided into two steps: based on the channel estimating of each signal frame frame head PN sequence and equilibrium subsequently.Wherein for channel estimating, can be similar in the duration at frame head and to think that channel remains unchanged, thereby obtain each frame head channel time domain impulse response constantly based on frame head PN sequencal estimation.Three kinds of frame head mode PN420 in Digital Television Terrestrial Broadcasting, among PN595 and the PN945, the frame head of PN420 and PN945 pattern has its special structure: the phase deviation PN sequence that they can regard 255 or 511 length respectively as adds the Cyclic Prefix composition, sees Fig. 3 (b).The present invention has utilized this special construction of frame head PN sequence under PN420 and two kinds of frame head modes of PN945, for the multipath channel time domain impulse response that falls into frame head PN sequence cycles prefix length scope, adopt processing method of the present invention, can accurately and apace obtain channel estimation results with low implementation complexity.

The Digital Television Terrestrial Broadcasting PN420 that the present invention proposes and the channel estimating apparatus structure of PN945 frame head mode as shown in Figure 8, it includes:

(a) state controller is a finite state machine, the input synchronous control signal, according to synchronous control signal and system clock produce, the update system status signal; With a signal frame is the work period, when the character rate sampled data arrival of a new frame through Synchronous Processing, channel estimating apparatus reaches preliminary treatment, circular correlation and channel impulse response forming processes synchronously to it successively and estimates to obtain current demand signal frame frame head channel time domain impulse response constantly; Correspondingly, state controller generation and update system state control signal carry out respective handling to control other module operating state to input data sequence in order to indicate synchronously and the sub-state under preliminary treatment state, circular correlation state or channel impulse response completed state and each treatment state;

(b) frame head separator, the character rate sampled signal frame sequence after its input connection synchronously, input is through synchronous character rate sampled signal frame sequence later ..., F _I-1, F _i, F _I+1..., a circular correlation sequence is arranged ..., P _I-1', P _i', P _I+1' ... output port; Reach the preliminary treatment state synchronously, the frame head separator is with the frame header divided data sequence P of present frame _iBack L _PNIndividual symbol, i.e. circular correlation sequence P _i' from character rate sampled signal frame F _iIn isolate, and export to the correlated series buffer unit;

(c) correlated series buffer unit, it is a L _PNThe register series of level is in order to the circular correlation sequence of storage present frame, and its input connects the circular correlation sequence output of frame head separator, and with the L that stores in the register _PNIndividual data symbol is exported to the circular correlation device; Reach the preliminary treatment state synchronously, receive successively by the frame head separator and separate the circular correlation sequence P that obtains _i', and with the L of this sequence _PNIndividual symbol { P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI) send into register series storage successively; At the circular correlation state, the circular correlation sequence symbol data of storage are exported to the circular correlation device in order to carry out the circular correlation computing;

(d) circular correlation PN generator produces circular correlation PN sequence according to state control signal, exports to the circular correlation device in order to carry out the circular correlation computing; Synchronous preliminary treatment state generates the relevant PN sequence PN of local initial cycle of current i frame correspondence _{I, 1}={ PN _{I, 1}(1), PN _{I, 1}(2), PN _{I, 1}(3) ..., PN _{I, 1}(L _PN); During the circular correlation state, the ring shift right by sequence after the circular correlation device is finished related operation produces corresponding local circular correlation PN sequence for related operation next time, promptly

PN _{I, j}={ PN _{I, j-1}(L _PN), PN _{I, j-1}(1), PN _{I, j-1}(2) ..., PN _{I, j-1}(L _PN-1) } the circular correlation PN sequence of Chan Shenging is exported to the circular correlation device;

(e) circular correlation device, it has two inputs: by the circular correlation sequence of correlated series buffer unit input and the circular correlation PN sequence of being imported by circular correlation PN sequence generator; At the circular correlation state, finish common L _CThe related operation of inferior circular correlation sequence and circular correlation PN sequence conjugation obtains correlated series c _i(j), j=1,2 ..., L _C, that is:

$c_i\left(j\right)=\underset{k=1}{\overset{L_{\mathrm{PN}}}{\mathrm{\&Sigma;}}}{P}_i(k+L_C)\&times;P{N}_{i,j}^{*}\left(k\right),j=\mathrm{1,2},...,L_C$

Correlated series c _iExport to the channel impulse response former;

(f) channel impulse response former, input is from the correlated series c of circular correlation device _i(j), j=1,2 ..., L _C, and finish following processing and calculate channel impulse response pre-estimation sequence:

$h_i^{\&prime;}\left(j\right)\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1+c_i\left(j\right)/{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00042.png"\; state="\{\{state\}\}">C</figure-callout>}}_2,j=\mathrm{1,2},...,L_C$

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * P.Wherein P is the power normalization factor, is the symbol power P of the transmitting terminal PN sequence that receives _STThe symbol power P of the circular correlation PN sequence that produces with this locality _SRLong-pending.

Because there is noise jamming, h _i' little level value in (j) is unreliable, and these little level values will be dropped to reduce The noise, promptly finish following operation:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\left(j\right)\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array}\right.,j=\mathrm{1,2},...,L_C$

H wherein _TBe one and abandon decision threshold that desired different antinoise of visual application and multi-path resolved sensitivity decide.

The present invention proposes the channel estimation methods under a kind of Digital Television Terrestrial Broadcasting PN420 and the PN945 frame head mode simultaneously, by the processing to the synchronous character rate sample sequence later of warp, implementation complexity that can be low carries out accurately and estimation fast the multipath channel impulse response that falls into frame head PN sequence prefix length scope.As shown in Figure 4, this estimation procedure mainly is divided into three steps: (a) reach preliminary treatment synchronously; (b) circular correlation (Cyclic Correlation) obtains correlated series; (c) channel impulse response moulding (Channel impulse molding) calculates channel impulse response by correlated series.

Be described in detail below with reference to the concrete enforcement of accompanying drawing this method.

Impulse response continuous time of wireless channel can be modeled as:

$h\left(t\right)=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}h\left({\mathrm{\&tau;}}_i\right)\mathrm{\&delta;}(t-{\mathrm{\&tau;}}_i)+n\left(t\right)---\left(1\right)$

H (τ wherein _i) be the channel fading of respective path, N represents the multidiameter path number, n (t) is the channel additive noise.It may be noted that represented channel is the combination of all systems from transmitting terminal to the receiving terminal process here, for example this aggregate channel may be the cascade of the filtering of transmitting terminal raised cosine roll off, waveform channel, receiving terminal band pass filter and matched filter etc.

Continuous wave is propagated through multipath channel, forms the stack of a plurality of different copies that transmit that postpone and decline, and receiving terminal receives this signal and behind over-sampling, its baseband equivalence is expressed as follows:

$y\left(n\right)=s\left(n\right)*h\left(n\right)+n\left(n\right)=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}h\left(i\right)s(n-i)+n\left(n\right)---\left(2\right)$

Wherein " * " represents linear convolution, and s (n) is a base band transmit, and y (n) is a baseband receiving signals, and n (n) represents additive noise.Therefore equivalence is to base band, and channel can be modeled as one the FIR filter of additive noise at random, has tap coefficient h (n), n=1,2 ..., K, the K correspondence be the longest multidiameter.

Suppose at a signal frame frame head duration section channel impulse response constantly, then i signal frame is through after the multipath channel under Digital Television Terrestrial Broadcasting PN420 or the PN945 pattern, and multipath number of path K is smaller or equal to L _CThe time, the frame header divided data of the i signal frame that receiving terminal receives is:

$P_i\left(k\right)\underset{j=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{h}_i\left(j\right)\&CenterDot;f(k-j)+n\left(k\right),k=\mathrm{1,2},...,L_{\mathrm{PI}}---\left(3\right)$

{ h wherein _i(j)=0|j＞k}, $f(k-j)\left\{\begin{array}{cc}{p}_i(k-j+1),& k-j>=0\\ d_{i-1}(3781+k-j),& k-j<0\end{array}\right.$ Sequence p _i(k) (i=1,2 ...; K=1,2 ..., L _P) the expression transmitting terminal forms the frame head sequence of i frame; d _i(k) (i=1,2 ...; K=1,2 ..., 3780) the expression transmitting terminal forms the sequence of data segments of i frame.

(3) meaning represented of formula as shown in Figure 9.As can be seen from Figure, since the design feature of frame head, the frame header sub-sequence P of reception data _i={ P _i(1), P _i(2) ..., P _i(L _PI) back L _PNThe sequence P that individual symbol is formed _i'={ P _i' (1), P _i' (2) ..., P _i' (L _PN)={ P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI) except interchannel noise, do not sneak into the multipath interference of data segment, but obtain by the phase deviation PN sequence stack of a plurality of different weights, the phase deviation PN sequence here is meant that also the generation PN with respect to present frame has the PN sequence of initial phase offset.Thereby utilize the circulation autocorrelation performance of PN sequence near ideal, pass through P _iThe relevant channel fading coefficient h that can each path is corresponding of conjugation of the phase deviation PN sequence that ' section character rate sample sequence and each path are corresponding _i(j) extract, and obtain, therefore claim P owing to each phase deviation PN sequence can be moved to right by initial phase offset PN sequence cycles among Fig. 3 (b) _i' be the circular correlation sequence.

From the above, channel impulse response information can obtain by the conjugation relevant treatment to circular correlation sequence and phase deviation PN sequence, therefore as shown in Figure 5, reaching preliminary treatment synchronously is when each signal frame arrives, carry out before circular correlation and the channel impulse response forming processes, according to synchronizing signal with frame header divided data sequence P _iFrom continuous character rate sampled signal frame F _iIn isolate, and then obtain the circular correlation sequence P of corresponding frame _i'; Produce simultaneously and constitute the relevant PN sequence PN of the pairing initial cycle of this signal frame frame head _{I, 1}(the phase deviation PN sequence shown in Fig. 3 (b)) is so that obtain when circular correlation in order to each relevant phase deviation PN sequence PN _{I, j}, j=1,2 ..., L _C

Circular correlation is in order to calculate correlated series.The correlated series of i frame is defined as a length L _CSequence, i.e. c _i={ c _i(1), c _i(2) ..., c _i(L _C), be that the correlation computations by circular correlation sequence and each phase deviation PN sequence conjugation obtains, promptly

$c_i\left(j\right)=\underset{k=0}{\overset{L_C}{\mathrm{\&Sigma;}}}{P}_i^{\&prime;}\left(k\right)\&CenterDot;P{N}_{i,j}^{*}\left(k\right),k=\mathrm{1,2},...,L_C---\left(4\right)$

By formula (3) or Fig. 9 as can be known, wherein be used between each phase deviation PN sequence of related operation following relation being arranged:

PN _i，j＝{PN _i，j-1(L _PN)，PN _i，j-1(1)，PN _i，j-1(2)，...，PN _i，j-1(L _PN-1)} (5)

Therefore as shown in Figure 6, circular correlation should comprise following steps:

(a) count initialized device j=1;

(b) with circular correlation sequence P _i' and PN _{I, j}It is relevant to do conjugation, obtains c _i(j), promptly

$c_i\left(j\right)=\underset{k=1}{\overset{L_{\mathrm{PN}}}{\mathrm{\&Sigma;}}}{P}_i(k+L_C)\&times;P{N}_{i,j}^{*}\left(k\right)$

(c) ring shift right upgrades local circular correlation PN sequence, obtains PN _{I, j+1}:

PN _i，j+1＝{PN _i，j(L _PN)，PN _i，j(1)，PN _i，j(2)，...，PN _i，j(L _PN-1)}

(d) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish;

Because the almost desirable autocorrelation performance of PN sequence, therefore can be similar to and think that correlated series that (4) calculate just can be similar to through the equal proportion convergent-divergent and obtain channel impulse response, promptly

h _i(j)＝c _i(j)/(L _PN·P) (6)

L wherein _PNBe the length of PN sequence, the power normalization factor $P=(P_{\mathrm{TS}}\&CenterDot;{\mathrm{<figure-callout\; id="4"\; label="S"\; filenames="DEST\_PATH\_H200810104247201E00011.png"\; state="\{\{state\}\}">S</figure-callout>}}_{4\mathrm{QAM}})\&CenterDot;(P_{\mathrm{RS}}\&CenterDot;{\mathrm{<figure-callout\; id="4"\; label="S"\; filenames="DEST\_PATH\_H200810104247201E00011.png"\; state="\{\{state\}\}">S</figure-callout>}}_{4\mathrm{QAM}}^{*})=P_{\mathrm{TS}}\&CenterDot;P_{\mathrm{RS}},{\mathrm{<figure-callout\; id="4"\; label="S"\; filenames="DEST\_PATH\_H200810104247201E00011.png"\; state="\{\{state\}\}">S</figure-callout>}}_{4\mathrm{QAM}}$ It is the 4QAM constellation symbol of power normalization $S_{}$ ${}_{4\mathrm{QAM}}\&CenterDot;{\mathrm{<figure-callout\; id="4"\; label="S"\; filenames="DEST\_PATH\_H200810104247201E00011.png"\; state="\{\{state\}\}">S</figure-callout>}}_{4\mathrm{QAM}}^{*}=1,$ P _TSAnd P _RSBe respectively the symbol power of the transmitting terminal PN sequence that receives and the symbol power of the local circular correlation PN sequence that produces.

But as shown in figure 10, symbol power is that the circulation auto-correlation function R (n) of 1 PN sequence is not to be desirable δ function, but

$R\left(n\right)=\left\{\begin{array}{cc}{L}_{\mathrm{PN}}& n=0\\ -1,& n\&NotEqual;0\end{array},n=\mathrm{0,1},...,L_{\mathrm{PN}}-1---\left(7\right)\right.$

Therefore the approximate processing of the channel impulse response of (6) formula definition has been introduced error artificially, but by analyzing as can be known correlated series c _iAnd channel impulse response h _iBetween certain corresponding relation is arranged, can obtain channel impulse response estimation more accurately by channel impulse response forming processes as described below.

Consider the influence of interchannel noise, so the correlated series c of i frame _i={ c _i(1), c _i(2) ..., c _i(L _C) and i frame channel impulse response h _i={ h _i(1), h _i(2) ..., h _i(L _C) between relation is arranged

$c_i\left(j\right)/P=L_{\mathrm{PN}}\&CenterDot;h_i\left(j\right)-\underset{k=0,k\&NotEqual;j}{\overset{L_C}{\mathrm{\&Sigma;}}}{h}_i\left(k\right)+n_i\left(j\right)---\left(8\right)$

N wherein _i(j) error that causes for interchannel noise.At first can obtain the channel impulse response pre-estimation:

$h_i^{\&prime;}\left(j\right)=\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1+c_i\left(j\right)/C_2---\left(9\right)$

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * P.

In addition, because have noise jamming, h _i' little level value in (j) is unreliable, and these little level values are dropped to reduce The noise, promptly finish following operation:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\left(j\right)\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array},j=\mathrm{1,2},...,L_C---\left(10\right)\right.$

H wherein _TBe one and abandon decision threshold that desired different antinoise of visual application and multi-path resolved sensitivity decide.

The channel impulse response forming processes promptly is to correlated series c _i={ c _i(1), c _i(2) ..., c _i(L _C) processing of each coefficient correlation calculates the current frame channel impulse response and estimate h _i={ h _i(1), h _i(2) ..., h _i(L _C), therefore by aforementioned analysis as can be known channel impulse response forming processes process can comprise as shown in Figure 7 step:

(a) count initialized device j=1;

(b) calculating channel impulse response pre-estimation h _i' (j)

$h_i^{\&prime;}\left(j\right)=\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00032.png"\; state="\{\{state\}\}">C</figure-callout>}}_1+c_i\left(j\right)/{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="DEST\_PATH\_H200810104247201E00011.png,S2008101042472E00042.png"\; state="\{\{state\}\}">C</figure-callout>}}_2$

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * and P, P is the power normalization factor, is the symbol power P of the transmitting terminal PN sequence that receives _STThe symbol power P of the circular correlation PN sequence that produces with this locality _SRLong-pending.

(c) abandon h _i' (j) in insecure little level value to reduce The noise, obtain channel impulse response estimation h _i(j), that is:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array}\right.$

(d) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish.

Because PN sequence auto-correlation can obtain extra diversity gain, therefore can effectively suppress interchannel noise and disturb correlated series c _i={ c _i(1), c _i(2) ..., c _i(L _C) very high precision arranged; And (9) calculating of formula definition can not introduced extra noise in the evaluated error that effectively reduces the imperfect auto-correlation introducing of PN sequence, therefore can obtain better channel estimating performance; This method amount of calculation is little from the above in addition, and implementation complexity is low.

For the channel estimation methods that proposes is estimated, and verify above-mentioned derivation and computational process, we have carried out Computer Simulation.

A given multipath channel, its impulse response amplitude are shown in Figure 11 (a), and this is the multipath channel that the strong footpath of three 0dB is arranged that a simulation single frequency network (SFN) is used.Be the channel impulse response amplitude that estimation obtained when the receive channel signal to noise ratio (snr) was for 5dB under the PN945 pattern shown in Figure 11 (b), wherein abandon decision threshold h _TBe set at channel impulse response pre-estimation sequences h _i' (j), and j=1,2 ..., L _C/ 10th of a sequence amplitude peak.As can be seen from the figure, under the very noisy disturbed condition, this channel estimation methods can estimate the multipath channel impulse response that falls in the frame head PN sequence prefix ranges exactly, eliminates multipath channel for equilibrium the condition that provides is provided.For the autocorrelative non-ideal characteristic of PN sequence, the way of the scheme that has is to ignore its non-ideal characteristic as previously mentioned, promptly is similar to by correlated series according to (6) formula and obtains channel impulse response; The scheme that also has is to adopt the way of iteration, and is big to the influence in other paths based on the path that signal power is big, and other path influences insignificant hypothesis to it simultaneously, successively from correlated series c _iIn find the footpath of amplitude maximum, and eliminate its influence to other paths, look for time the strongest path to do identical processing again, and the like.As can be known, though first method is simple, error has been introduced in an approximate processing artificially; Second method adopts the processing of iteration can reduce the imperfect relevant influence that brings of PN sequence to a certain extent, but complicated difficult realization of this alternative manner particularly searched for the maximum path process and be unsuitable for real-time processing.Correlated series that the present invention obtains according to derivation and the corresponding relation between the channel impulse response (9) formula are carried out simple channel impulse response forming processes to correlated series and can be reduced the interference that the imperfect auto-correlation of PN sequence is brought, and can obtain better channel impulse response estimation.Be respectively as shown in figure 12 in test channel (Brazilian B channel), different received signal to noise ratio (SNR) adopt the channel estimation methods of the inventive method down, based on the channel estimation methods of Ideal Cycle auto-correlation hypothesis and based on the channel estimating mean square deviation (MSE) of the channel estimation methods of iteration, wherein for ease of contrast h _TBe set at 0.Channel estimation methods of the present invention and suitable under each channel signal to noise ratio condition as can be known based on the channel estimation methods performance of iteration, and all be better than channel estimating based on the relevant hypothesis of Ideal Cycle, particularly preferably under the situation, can obviously improve the estimated accuracy of channel impulse response at channel condition.In addition, suppose that channel condition is more complicated, promptly the path is abundant and strong path is more, and the interference that the imperfect auto-correlation of PN sequence this moment causes is big and can not ignore, thereby will be apparent in view based on the channel estimation errors of Ideal Cycle auto-correlation hypothesis; No longer set up to a certain extent simultaneously based on the aforementioned hypothesis of the channel estimation methods of iteration, thus its to the improvement of precision of channel estimation also with limited.Be in the complicated channel as shown in figure 13, different received signal to noise ratio (SNR) adopt the channel estimation methods of the inventive method down and based on the channel estimating mean square deviation of the channel estimation methods of iteration, in the relative mean square deviation (MSE relatively) based on the channel estimating mean square deviation of Ideal Cycle auto-correlation hypothesis channel estimation methods.As we know from the figure, in this Complex Channel condition, compare with channel estimation methods of the present invention, limited based on the channel estimation methods of iteration to the raising of precision of channel estimation, and the raising performance along with received signal to noise ratio no longer includes further improvement when being better than certain received signal to noise ratio; And channel estimation methods performance of the present invention all is better than the channel estimation methods based on iteration under each channel signal to noise ratio condition, and along with the raising precision of channel estimation of channel receive channel ratio is also constantly improved.

In sum, channel estimation methods of the present invention has utilized the nonideal autocorrelation performance of PN sequence cycles, therefore computation complexity that can be low effectively reduces the evaluated error that the imperfect auto-correlation of PN sequence is brought, can be more accurate and estimation obtains falling into frame head PN sequence prefix length scope apace multipath channel impulse response; And than channel estimation methods based on iteration, channel estimation methods of the present invention all shows the estimated accuracy that is not inferior to it under each channel condition, particularly show more excellent performance under the Complex Channel condition, and its implementation complexity is lower simultaneously, computing is simpler, is more suitable for real-time processing.

Top combination is derived and accompanying drawing has been described in detail concrete enforcement of the present invention, but the present invention is not restricted to the foregoing description, for obtaining better channel estimation ability and noise robustness, under the spirit and scope situation of the claim that does not break away from the application, those skilled in the art can make various modifications or remodeling.

Claims (3)

1. terrestrial broadcasting channel estimating method for digital television, it is characterized in that, under Digital Television Terrestrial Broadcasting PN420 or PN945 frame head mode, can be for the multipath channel impulse response that falls into frame head PN sequence prefix length scope by the following processing of character rate sample sequence is synchronously later estimated that it contains following steps successively:

(1) reaches preliminary treatment synchronously; According to frame synchronization information, the synchronous i signal frame character rate sample sequence F later of warp _iBe decomposed into frame head P _iWith data block D _i, and then obtain the circular correlation sequence P ' of this frame _i={ P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI), wherein, L _PIBe frame head length, L _CBe the circulating prefix-length of frame head PN sequence, and, generate the initial local circular correlation PN sequence PN of i frame according to the synchronization parameter of present frame _{I, 1}, PN _{I, 1}={ PN _{I, 1}(1), PN _{I, 1}(2) ..., PN _{I, 1}(L _PN);

(2) circular correlation calculates correlated series c _i={ c _i(1), c _i(2) ..., c _i(L _C);

(a) count initialized device j=1;

(b) with circular correlation sequence P ' _iAnd PN _{I, j}It is relevant to do conjugation, obtains c _i(j), wherein

PN _{I, j}={ PN _{I, j-1}(L _PN), PN _{I, j-1}(1), PN _{I, j-1}(2) ..., PN _{I, j-1}(L _PN-1) }, then

$c_i\left(j\right)=\underset{k=1}{\overset{L_{\mathrm{PN}}}{\mathrm{\&Sigma;}}}{P}_i(k+L_C)\&times;{\mathrm{PN}}_{i,j}^{*}\left(k\right)$

(c) ring shift right upgrades local circular correlation PN sequence, obtains PN _{I, j+1}

(d) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish;

(3) channel impulse response moulding calculates the channel impulse response sequencal estimation,

(a) count initialized device j=1;

(b) calculating channel impulse response pre-estimation h ' _i(j)

$h_i^{\&prime;}\left(j\right)=\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/C_1+c_i\left(j\right)/C_2$

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * and P, L _PNBe the PN sequence length in order to the generation frame head, P is the power normalization factor;

(c) abandon h ' _i(j) insecure little level value obtains channel impulse response estimation h to reduce The noise in _i(j), that is:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array}\right.$

H wherein _TBe one and abandon decision threshold that desired different antinoise of visual application and multi-path resolved sensitivity decide;

(d) if j＜L _C, j=j+1 then, and get back to step (b);

Otherwise finish.

2. a kind of terrestrial broadcasting channel estimating method for digital television according to claim 1 is characterized in that, has identical processing method for two kinds of frame head modes of Digital Television Terrestrial Broadcasting PN420 and PN945, and just system parameters is different.

3. a Digital Television Terrestrial Broadcasting channel estimating apparatus is characterized in that it includes: state controller, frame head separator, correlated series buffer unit, circular correlation PN generator, circular correlation device, channel impulse response former; Its relation is as follows:

(a) state controller, the input synchronous control signal, according to synchronous control signal and system clock produce, the update system status signal is in order to indicate synchronously and the preliminary treatment state, sub-state under circular correlation state or channel impulse response completed state and each treatment state carries out respective handling to control other module operating state to input data sequence;

The signal frame sequence of the character rate sampling after (b) frame head separator, its input connect synchronously has a circular correlation sequence output port; Reach the preliminary treatment state synchronously, the frame head separator is with circular correlation sequence P ' _iFrom continuous character rate sampled signal frame F _iIn isolate, and export to the correlated series buffer unit;

(c) correlated series buffer unit is a L _PNThe register series of level is in order to storage circular correlation sequence P ' _i, its input connects the circular correlation sequence output of frame head separator, and with the L that stores in the register _PNIndividual data symbol is exported to the circular correlation device; Reach the preliminary treatment state synchronously, receive successively by the frame head separator and separate the circular correlation sequence P ' that obtains _i, and with the L of this sequence _PNIndividual symbol { P _i(L _C+ 1), P _i(L _C+ 2) ..., P _i(L _PI) send into register series storage, L successively _PIBe frame head length, L _CCirculating prefix-length for frame head PN sequence; At the circular correlation state, the circular correlation sequence symbol data of storage are exported to the circular correlation device in order to carry out the circular correlation computing;

(d) circular correlation PN generator produces circular correlation PN sequence according to state control signal, exports to the circular correlation device in order to carry out the circular correlation computing;

(e) circular correlation device, it has two inputs: by the circular correlation sequence of correlated series buffer unit input and the circular correlation PN sequence of being imported by circular correlation PN sequence generator; At the circular correlation state, finish common L _CThe related operation of inferior circular correlation sequence and circular correlation PN sequence conjugation obtains correlated series c _i(j), j=1,2 ..., L _C, that is:

$c_i\left(j\right)=\underset{k=1}{\overset{L_{\mathrm{PN}}}{\mathrm{\&Sigma;}}}{P}_i(k+L_C)\&times;{\mathrm{PN}}_{i,j}^{*}\left(k\right)$

In the formula: L _PNBe PN sequence length in order to the generation frame head; P ' _iBe the circular correlation sequence; PN _{I, j}Be circular correlation PN sequence;

(f) channel impulse response former, input is from the correlated series c of circular correlation device _i={ c _i(1), c _i(2) ..., c _i(L _C), carry out the channel impulse response forming processes, calculate output estimated channel impulse response sequences h _i={ h _i(1), h _i(2) ..., h _i(L _C), be specially:

Calculating channel impulse response pre-estimation sequence:

$h_i^{\&prime;}\left(j\right)=\left(\underset{k=1}{\overset{L_C}{\mathrm{\&Sigma;}}}{c}_i\left(k\right)\right)/C_1+c_i\left(j\right)/C_2,$ j＝1，2，...，L _C

C wherein ₁=(L _PN-L _C+ 1) * (L _PN+ 1) * and P, C ₂=(L _PN+ 1) * and P, P is the power normalization factor;

With h ' _i(j) insecure little level value abandons in, obtains:

$h_i\left(j\right)=\left\{\begin{array}{cc}{h}_i^{\&prime;}\left(j\right)& \left|h_i^{\&prime;}\left(j\right)\right|\&GreaterEqual;h_T\\ 0& h_i^{\&prime;}<h_T\end{array}\right.,$ j＝1，2，...，L _C

H wherein _TBe one and abandon decision threshold that desired different antinoise of visual application and multi-path resolved sensitivity decide.

Images