WO2009069083A2

WO2009069083A2 - Pn phase recovery in a dmb-t system

Info

Publication number: WO2009069083A2
Application number: PCT/IB2008/054961
Authority: WO
Inventors: Ming GONG
Original assignee: Nxp B.V.
Priority date: 2007-11-30
Filing date: 2008-11-26
Publication date: 2009-06-04
Also published as: CN101874392A; CN101453554A; WO2009069083A3; EP2225864A2

Abstract

A PN Phase Recovery (PPR) method is used to acquire PN sequence phase synchronization in a system such as DMB-T. The time offset of the positions of the basic PN sequence in successive signal frames is estimated robustly. An accurate decision of the signal frame index is made based on multiple time offsets measured in the successive signal frames through a voting mechanism with modest calculation complexity.

Description

PN PHASE RECOVERY IN A DMB-T SYSTEM

BACKGROUND

Time domain synchronous OFDM (TDS-OFDM) is one of the fundamental physical schemes of the DMB-T specification for terrestrial broadcasting in China, in which a multiplex frame structure is used as shown in FIG. 2. The frame structure is important for time synchronization in the DMB-T receiver.

A signal frame is the basic element of the multiplex frame structure. As shown in FIG. 3, a signal frame consists of two time-domain signal parts, namely frame header and frame body. The frame header and frame body have the same baseband symbol data rate (7.56Msym/sec). In the signal frame header, a PN sequence is sent for purposes of synchronization and channel estimation. At the same time, the PN header also serves as a guard time interval for the following OFDM frame body in the place of the conventional cyclic prefix (CP).

The PN header with length of LPN includes three parts: a complete period of the PN sequence with length of N_PN symbols, a length L_pre PN preamble and a length L_post PN post-amble. Three types of signal frame options are defined in the specification with different parameter combinations.

In type 2, the same segment of the PN sequence is used in every frame, which makes PN phase synchronization easier than in types 1 and 3, where different PN sequences are sent in different signal frames. The present invention concerns PN phase recovery for types 1 and 3.

The PN sequences used in signal frames will change from frame to frame. Knowledge of PN sequences used in each frame is necessary for the synchronization of sampling frequency and sampling time phase, and for channel estimation. Also, it is necessary to maintain PN sequence phase synchronization throughout the reception procedure. As described herein, the object of PN phase recovery (PPR) is to recognize the index/of the^th signal frame given the received data sampled from several signal frame headers.

In an AWGN channel, if in the current signal frame there is a peak in the correlation signal at time t], and in the next signal frame another peak in the correlation result can be found at time ^, the distance between these two peaks is a function of the index of the frame, enabling the index of the signal frame to be determined.

However, the receiver may not necessarily always operate in an AWGN channel. The radio channel may experience severe fading as well as strong interference. In TDS-OFDM, especially when the length of the multipath channel is long, the OFDM frame body part may unfortunately interfere with the frame header part. The peak of the correlation is not always easy to determine correctly due to possible strong noise and interference. Because of the PN phase change from frame to frame, traditional averaging or accumulation of the correlation result within multiple successive signal frames to suppress the noise and interference becomes impossible.

What makes the measurement of the time difference between the two peaks more difficult is sampling frequency error. There may be large sampling frequency error if a low-cost crystal is adopted. In digital processing receivers, the above correlation will be done using samples of the received signal. Large sampling frequency error will add more difficulties to PN phase recovery, because large sampling frequency error will cause uncertainty in the determination of the frame header from the analysis based on the sampled data.

To get correct information about the signal frame index as early as possible, the PN phase needs to be captured as fast as possible. However, at the stage when the receiver begins to work, the receiver has little knowledge about the radio channel. It is therefore a challenge to capture the PN phase quickly and robustly, even in a severe wireless environment.

SUMMARY

PN Phase Recovery (PPR) methods and apparatus are described to acquire PN sequence phase synchronization in a system such as DMB-T, where the time offset of the positions of the basic PN sequence in successive signal frames is estimated robustly. An accurate decision of the signal frame index is made based on multiple time offsets measured in successive signal frames through a voting mechanism with little calculation complexity. In this manner, a DMB-T receiver can be made more robust and can be rapidly synchronized with the transmitter in PN sequence phase, even in an environment with very low signal to noise ratio or in the presence of large sample frequency errors.

Other features and advantages will be understood upon reading and understanding the detailed description of exemplary embodiments, found herein below, in conjunction with reference to the drawings, a brief description of which is provided below.

BRIEF DESCRIPTION OF THE DRAWING

FIG. 1 is a diagram of a DMB-T receiver in which the present invention can be used.

FIG. 2 is a diagram showing the multiplex frame structure for DMB-T.

FIG. 3 is a diagram showing the structure of a signal frame.

FIG. 4 is a diagram showing the PN phase offset for the signal frames in a super frame.

FIG. 5 is a block diagram of a time offset estimator.

FIG. 6 is a diagram showing a constellation for obtaining information about the signal frame index.

FIG. 7 is a block diagram of a PN phase recovery module.

DESCRIPTION

There follows a more detailed description of the present invention. Those skilled in the art will realize that the following detailed description is illustrative only and is not intended to be in any way limiting. Other embodiments of the present invention will readily suggest themselves to such skilled persons having the benefit of this disclosure. Reference will now be made in detail to embodiments of the present invention as illustrated in the accompanying drawings. The same reference indicators will be used throughout the drawings and the following detailed description to refer to the same or like parts.

The general structure of a DTV receiver is shown in FIG. 1. A signal is received by an RF module (not shown) and sampled in an ADC 105. The sampled signal 107 is applied to a digital front end 109 that performs synchronization in response to information 110 from a synchronization block 111. The synchronization block performs calculations to enable, for example, PN phase recovery I l ia, carrier offset recovery 111b, symbol offset recovery 111c, and sampling frequency recovery 11 Id. An output signal 113 of the digital front end 109 is applied to the synchronization block 111 and to a channel estimation and equalization block 115, which receives information from and provides information to the synchronization block 111 via lines 117 and 119, respectively. An output signal 121 of the channel estimation and equalization block is applied to a decoder 123, which decodes and outputs received information 125.

The present invention is concerned particularly with PN phase recovery (11 Ia).

Referring to FIG. 2 and FIG. 3, in the DMB-T specification for terrestrial broadcasting in China, there are F signal frames in a super frame. PN(f,i) ,i = 0,\,...,L_PN -\ is the PN sequence used in/-th signal frame in a super frame.

Three types of signal frame options are defined in the specification with different parameter combinations. Table 1 lists the parameters of the three types.

Table 1: The parameters of the three types of the signal frame header

In type 2, the same segment of the PN sequence is used in every frame, which makes PN phase synchronization easier than in types 1 and 3, where different PN sequences are sent in different signal frames. The present invention concerns PN phase recovery for types 1 and 3. Types 1 and 3 are similar; type 1 (PN420) will be taken as an example for illustration.

The PN sequence used in the/-th signal frame may be defined as PN(f,i) , i = 0,l,...,L_PN -l . In the PN420 sequence, L_PN = 420. According to the DMB-T specification, for the PN420 sequence, the PN sequence used is derived from an m-sequence and has a special initial phase in an LFSR PN generator for each signal frame of a super frame. The PN sequence in 0-th signal frame, PN(OJ) , can be generated in an LFSR with the initial phase of "10110000" (binary number). The different PN sequences used in F signal frames have some internal relations.

A basic PN sequence P₀(i) may be defined which satisfies

P₀ (i) = PN(o, /+ 82) , Ϊ = O, I, ..., N - I ( i )

where N is the period of the m-sequence. For PN420, N = 255. All the PN sequences used in different signal frames can be regarded as a derivative of the basic PN sequence P₀(i) . From analysis of the initial phases of the LFSR for the signal frames in a super frame as tabulated in the DMB-T specification, the PN sequence with length of 420 used in the/-th signal frame can be generated as

PN(f,i) = P₀([i -O(f) -82]modN) , i = 0,l,...,L_PN -1 ( 2 )

where O(f) is a variable PN phase offset for the/^th signal frame. O(f) can be calculated using the following formula:

O(f) is shown in FIG. 4.

The PN sequences used in signal frames will change from frame to frame. Knowledge of PN sequences used in each frame is necessary for the synchronization of sampling frequency and sampling time phase, and for channel estimation. Also, it is necessary to maintain PN sequence phase synchronization throughout the reception procedure. As described herein, the object of PN phase recovery (PPR) is to recognize the index/of the/-th signal frame given the received data sampled from several signal frame headers.

In the continuous time domain, the received signal around the header of the/-th signal frame header is given as:

[∑P₀(iβ(t -iT_s)]®δ(t-[82 + O(f)])®g(t) + n(t)

where g(t) is the impulse response of the combined equivalent channel, which includes the effect of the SRRC pulse shaping filters at the transmitter and receiver and the effect of the radio channel. The impulse response g(t) is given by:

g(t) = SRRC(t) ® h(t) ® SRRC(t) = RC(t) ® A(O ( 5 )

where RC(t) is the impulse response function of a raised cosine filter, SRRC(t) is the impulse

L-I response function of a squared root raised cosine filter, h(t) = V o^δ^ -τJ is the channel

impulse response function, and n(t) is AWGN noise.

The receiver calculates the correlation between received data samples and the local basic PN sequence P₀ (i) . In the continuous time domain, the correlation is

R(t) = R₀(t) ® g(t) ®8 (t -[Z2 + O(f)]T_s -fl'_FO + M.t) ( 6 )

where R₀ (t) = V R₀ (k)δ (t - kT_s ) . R₀ (k) is the auto -correlation function of P₀ (k) and has a conspicuously large value only when k=0.

In an AWGN channel, there is a peak in the correlation signal at time ti,

t_l = {%2 + O(J) + fl._F ) T_s + \ ( 7 ) where Δ_o is a constant time reference point. In the next signal frame, another peak in the correlation result can be found at t₂ = (82 + O(f + 1) + (/ + \)L_F ) T_s + A₀. The distance between these two peaks is a function of the index of the frame, enabling the index of the signal frame to be determined.

However, as described above, the receiver may not necessarily always operate in an AWGN channel. The radio channel may experience severe fading as well as strong interference. In TDS- OFDM, especially when the length of the multipath channel is long, the OFDM frame body part may unfortunately interfere with the frame header part. The peak of the correlation is not always easy to determine correctly due to possible strong noise and interference. Because of the PN phase change from frame to frame, traditional averaging or accumulation of the correlation result within multiple successive signal frames to suppress the noise and interference becomes impossible.

Furthermore, what makes the measurement of the time difference between the two peaks more difficult is sampling frequency error. There may be large sampling frequency error if a low-cost crystal is adopted. In digital processing receivers, the above correlation R(t) will be done using samples of the received signal.

The result of such digital correlation, R{k) , is the same as sampling the continuous time signal R(t) if the effect of limited PN length is ignored:

R(k) = R(kT_s +z) ( 8 )

where ε is a data sampling time shift and ε = ε₀ + / • Δ • T_s includes two parts: an initial sampling reference point ε₀ and an accumulated sampling time error due to the sampling frequency error.

If the sampling frequency error is df = — f —'- —f (f_s ' is the sampling frequency and^ is the stated

J s sampling frequency aligned with the transmitter), the accumulated sampling time error in one signal frame is Lp*df*T_s . Lp is the length of a signal frame in units of T_s; in a PN420-type signal, Lp=4200. Large sampling frequency error will add more difficulty to PN phase recovery, because large sampling frequency error will cause uncertainty in the determination of the frame header from the analysis based on the sampled data.

To get correct information about the signal frame index as early as possible, the PN phase needs to be captured as fast as possible. However, at the stage when receiver begins to work, the receiver has little knowledge about the radio channel. It is therefore a challenge to capture the PN phase quickly and robustly, even in a severe wireless environment.

In an exemplary embodiment, PN phase recovery is achieved using two main modules. One module performs time offset estimation (TOE) of the basic PN sequence in two successive signal frames. The other module is a decision module that decides the signal frame index.

1. Measurement of the time offset of basic PN sequences

Measurement of the time offset of the basic PN sequence is done by finding the correlation of the header part of two successive signal frames, represented as follows:

RR_f (t) = z(f,t) ® z(f + \,t) ( 9 ) where θ stands for the correlation operator. After some mathematical manipulation,

RR₁ (t) = R₀ (t) ® R₀ (t) ® g(t) ®g(t) ® δ(t -[0(f + l) - 0(f)]T -L_FT ) + u(t) ^{( 1 O )}

where R₀ (t) ® R₀ (t) is almost a wave of impulse shape and u(t) is a noise term. The quantity git) ® g(t) = h{t) ® hit) ® RC it) ® RC it) has the strongest peak at t=0, no matter what kind of wireless channel, single path or multipath.

As mentioned, when dealing with digital samples, the accumulated sampling time error in one signal frame due to the sampling frequency error, which is unknown to the receiver at this stage, must be considered. It can be represented as x_f = A ■ T_s . A strong peak can be found in the correlation of the sampled signal around the position t_f = [0(f + l) -0(f)]T_s +L_FT_s +τ_f + e ( 11 )

where e is the measurement error. This value t_/ indicates the distance between the same PN segment in the two successive signal frames, which may indicate the index of the signal frame.

A block diagram of a time offset estimator (TOE) 500 for performing time offset estimation is shown in FIG. 5. Received signal samples 501 are applied to a first correlation block 503, a buffer 505, and a second correlation block 507. A generator 502 for generating a basic PN sequence Po is coupled to the correlation block 503 such that a correlation is performed between the received signal samples 501 and the basic PN sequence Po. The results of this correlation operation determine a peak correlaton time ti and determine which portion of the received signal samples stored in the buffer 505 will be used in a subsequent correlation operation. The correlation block 503 is coupled to the buffer 505 in order to perform this selection. The selected portion of the received signal samples is delayed by one signal frame by a delay element 509. The correlation block 507 then performs a further correlation, this time between received signal samples of the current frame and the selected portion of the received signal samples of the preceding frame. The results of this correlation operation determine a peak correlaton time t₂. A time offset calculation block 511 then calculates the time offset t_f as the difference t₂ - ti.

The time offset estimator (TOE) implements a sequence of steps that is repeated in every frame as summarized below.

Calculate the correlation between the signals in the/-th signal frame header and the basic PN sequence P₀(i)

where N is the period of the basic PN sequence. (For PN420, N=255)

Search for significant correlation peaks.

Save R₁(O in S(O , S(O = Ri(O Set a empty path set φ .

Find the path with the maximum energy if any, which is greater than R_1n^ - Ih_x , (e.g. the threshold i = argmax {\S(i)\}

If no path is greater than

■ , go to (g)

Let S(i" ) = 0, S(i" ± l) = 0,S(i^* ±2N) = 0, S(i" ± 2N ± 1) = 0

In this step, if a certain path is searched in the middle part of the result, its correlated parts are forced to zeros to avoid unnecessary redundancy. The result with index i* is correlated with the following results, which are forced to zero: i*+/-l , i*+/-l+/-2N, i*+/- 2N.

Add jz^*} to the paths set φ , goes to step (d) Sort the elements in set φ , get Ji₁, i₂,...., Ϊ_L}

Save a data segment covering length of L_c= [2N+(Ϊ_L-ii)]T_s, where the start time point of the recording of the segment is namely D(k) = z(f,i_l — + k—) , k=0,l,2, ..., L_c -1

T In the n+l-th signal frame, calculate the correlation between D(k) and z(f + l,k—) , i.e.:

Find the integer position h for the peak of the correlation /₂ = argmax{|i?₂(0|}

and record the peak time point of the segment

where Co is a constant for calibration for the time detector. (co= 0.387). Calculate the time offset t_/= trfi

2. Determine the index of the signal frame in a super frame

Define a sequence Q(f) , letting Q(f) = O(f + 1) - 0{f) . According to (3), after some manipulation,

Based on (7), theoretically the time offset of PN sequences in two successive signal frames is t_/ = [β(/)] T_s + L_FT_s +x_f + e . Because Q(f) is a one-to-one mapping function, with the measured t_j- , the index of signal frame/can be determined if % , , e can be ignored. However, the receiver may experience strong noise and interference. At the same time, sampling frequency error may exist to introduce a large value x , in the time measurement. Therefore accurate and robust determination can only be done according to multiple measurements in multiple successive signal frames.

Suppose Mtime offsets, t_f+m , m=0,l,2, ..., M-I , are measured in successive M+l signal frames.

There are ^hypotheses in this index detection problem. If the traditional minimum distance method is used,

/ = _aτgmml∑[t_f+m -Q(f + m)-L_FT ]²\ ( 12 )

/ [ m J

Because the error e from noise and interference and the error τ_f due to sampling frequency error is not noise similar to AWGN noise, minimum distance detection is not necessarily the optimum detection method. Moreover, there is considerable calculation burden entailed by minimum distance detection due to too many hypotheses.

An alternative approach to detection is therefore described. To illustrate the approach clearly, consider the case of four successive signal frames. Within the first three successive signal frames, two time offsets can be measured, i.e.: T_f = t_f -L_FT_s = [Q(f)]T_s +τ₀ +e₀

T_f+ι = t_f+ι -L_FT_s = [Q(J + ϊ)]T_s +τ_ι +e_ι ^{( U )}

Let

u Df = ¹T f+i - ¹Tf ( 14 )

If there is no noise or error in TV₊₁ and T_f , then for every signal frame/, a unique pair ( TV₊₁ , T_f ) exists. Two time offsets of successive signal frames are therefore put into a pair, which is then used for the determination of the signal frame index/ For illustration, the pair is drawn in a 2D plane in FIG. 6. The pair ( 7V₁ , T_f ) is like a constellation point for information '/". The pair

( T_f+1 ,T_f ) is illustrated as an asterisk, and the number beside it is the index of the signal frame.

For example, beginning at the signal frame with index of "0", the pair is (-1 ,2). That is to say, the time offset between signal frame 0 and signal frame 1 is -1 and the time offset between signal frame 1 and signal frame 2 is 2. When it comes to signal frame with index of "1", the pair is (2,- 3). This indicates that the time offset between signal frame 1 and signal frame 2 is 2 and the time offset between signal frame 2 and signal frame 3 is -3. Only a small portion of the constellation points are drawn in FIG. 6. Other points are not shown. The constellation points are found to be symmetrically distributed on the two sides of line "y=-x".

The variable D\ derived from the result of measurements helps to determine which zone (parallel to the line "y=x ") the target constellation point lies in. Due to the short-term stability property of the sampling frequency, it can be supposed thatτ_{0 ^}T₁. So for D_f ^l much of the influence of the sampling frequency error can be avoided. The determination of the zone is therefore more robust. Based on determination of the zone, two candidate constellation points can be determined, i.e.: S(I) = {/(1),/₂(1)} .

With the next three signal frames, similar results can be acquired. T_f+l = t_f+l -L_FT_s = [Q(f + I)]^ ₊T_{1 +} e_x T_f+2 = t_f+2 -L_FT_s = [Q(f + 2)]T_s +τ₂ + e₂

Another two candidates can be decided, S (2) = {/(2) - 1,/₂(2) - 1} .

There will be an overlap between S(I) and S(2). So the final decision is

f ={S(l)}f]{S(2)}

Of course, more measured time offsets may be involved in the detection of the PN phase procedure for robust detection. The more measured time offsets are involved, the stronger the confidence level of the final detection is.

The block diagram of a PN phase recovery (PPR) module is shown in FIG. 7. Successive time offset estimates from the time offset estimator 500 are applied to successive delay elements 701, 703, 705, etc., to 7Ox. A corresponding number of slicers 711, 713, 715, etc., to 71x are provided. Each sheer receives a different pair of time offset estimates delayed by one signal frame and maps the pair to the constellation of FIG. 6. Outputs of the slicers are applied to a voting machine 721 to produce a final time offset estimate/.

The steps performed in the PPR module are summarized below.

Within the successive M signal frames, measure the M-I time offsets T f+0 ' T f+l ' '"' T f+M-2

Set S as an empty set.

In slicers : Loop a),b),c) from m=0 to m=M-2, Calculate T = T_f+m+l - T_f+m

_{Z = int[}£zi_]

V = 4K + l If(V>0) {m= (V-l)/4, n₂=2*m+l} Else {m= -l *(V+3)/4, n₂=2*m }

cii = (ri2-l-m+F) mod F and a₂ =(F-l-n₂-l-m+F) mod F

add {ai,a₂} to the set S

Find in the set S the element/* with the most occurrences. If the number of occurrences p is beyond a pre-defined threshold PTH, the completion of PN phase recovery can be assumed. Otherwise, the detection procedure will continue, involving more signal frames and more measured time offsets.

Besides robust performance, the foregoing approach shows low calculation complexity.

Although embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions and alternations can be made without departing from the spirit and scope of the inventions as defined by the appended claims.

Claims

C L A I M SWhat is claimed is:

1. A method of estimating timing offset of a common PN segment between a signal portion of a frame of a transmission signal coded using a first PN sequence of a family of PN sequences and a signal portion of a subsequent frame of the transmission signal coded using a second related PN sequence of the family of PN sequences, the first PN sequence and the second PN sequence both containing the common PN segment, the timing offset being used by the transmission signal to convey information about the transmission signal, the method comprising: performing a first correlation between a PN sequence characteristic of the family of PN sequences and a received signal corresponding to the transmission signal; analyzing results of the first correlation; storing a portion of the received signal selected based on results of the first correlation; performing a second correlation between the received signal during the subsequent frame and the stored portion of the received signal; estimating the timing offset based on a correlation peak of the second correlation; and using the timing offset to obtain an indication of the information about the transmission signal.

2. The method of Claim 1, wherein the information is a signal frame index.

3. The method of Claim 1, wherein the stored portion of the received signal is selected based on timing of significant correlation peaks obtained from the first correlation.

4. The method of Claim 1, wherein, in determining significant correlation peaks obtained from the first correlation, cyclic repetitions are suppressed.

5. The method of Claim 1, comprising using a plurality of timing offset values to obtain an indication of the information about the transmission signal.

6. The method of Claim 5, comprising: forming pairs of timing offset values estimated from closely-neighboring signal frames; for each pair of timing offset values, determining a plurality of possible information values; and selecting a possible information value having a greatest frequency of occurrence.

7. In a system in which a signal portion of a frame of a transmission signal is coded using a first PN sequence of a family of PN sequences and a signal portion of a subsequent frame of the transmission signal is coded using a second related PN sequence of the family of PN sequences, the first PN sequence and the second PN sequence both containing a common PN segment, a timing offset of the common PN segment between the signal portion of the frame and the signal portion of the subsequent frame being used by the transmission signal to convey information about the transmission signal, a method of using a plurality of such timing offset values to obtain an indication of the information about the transmission signal, the method comprising: forming pairs of timing offset values estimated from closely-neighboring signal frames; for each pair of timing offset values, determining a plurality of possible information values; and selecting a possible information value having a greatest frequency of occurrence.

8. The method of Claim 7, comprising: mapping a pair of timing offset values to a zone within a scatter diagram; and determining the plurality of possible information values according to said zone.

9. Apparatus for estimating timing offset of a common PN segment between a signal portion of a frame of a transmission signal coded using a first PN sequence of a family of PN sequences and a signal portion of a subsequent frame of the transmission signal coded using a second related PN sequence of the family of PN sequences, the first PN sequence and the second PN sequence both containing the common PN segment, the timing offset being used by the transmission signal to convey information about the transmission signal, the apparatus comprising: means for performing a first correlation between a PN sequence characteristic of the family of PN sequences and a received signal corresponding to the transmission signal; means for analyzing results of the first correlation; storage for storing a portion of the received signal selected based on results of the first correlation; means for performing a second correlation between the received signal during the subsequent frame and the stored portion of the received signal; means for estimating the timing offset based on a correlation peak of the second correlation; and means for using the timing offset to obtain an indication of the information about the transmission signal.

10. The apparatus of Claim 9, wherein the information is a signal frame index.

11. The apparatus of Claim 9, wherein the stored portion of the received signal is selected based on timing of significant correlation peaks obtained from the first correlation.

12. The apparatus of Claim 9, wherein, in determining significant correlation peaks obtained from the first correlation, cyclic repetitions are suppressed.

13. The apparatus of Claim 9, wherein a plurality of timing offset values are used to obtain an indication of the information about the transmission signal.

14. The apparatus of Claim 13, comprising: means for forming pairs of timing offset values estimated from closely-neighboring signal frames; means for, for each pair of timing offset values, determining a plurality of possible information values; and means for selecting a possible information value having a greatest frequency of occurrence.

15. In a system in which a signal portion of a frame of a transmission signal is coded using a first PN sequence of a family of PN sequences and a signal portion of a subsequent frame of the transmission signal is coded using a second related PN sequence of the family of PN sequences, the first PN sequence and the second PN sequence both containing a common PN segment, a timing offset of the common PN segment between the signal portion of the frame and the signal portion of the subsequent frame being used by the transmission signal to convey information about the transmission signal, an apparatus using a plurality of such timing offset values to obtain an indication of the information about the transmission signal, the method comprising: means for forming pairs of timing offset values estimated from closely-neighboring signal frames; means for, for each pair of timing offset values, determining a plurality of possible information values; and means for selecting a possible information value having a greatest frequency of occurrence.

16. The apparatus of Claim 16, wherein said means for determining a plurality of possible information values comprises: means for mapping a pair of timing offset values to a zone within a scatter diagram; and means for determining the plurality of possible information values according to said zone.