
The invention relates to a method and a device for determining the beginning of a data frame in an OFDM data stream.

For synchronization between transmitter and receiver in an Orthogonal Frequency Division Multiplexing (OFDM) transmission system often the starting position of data frame within the OFDM data stream is used.

From the
US 2009/0103667 A1 For example, a method for determining the start position of OFDM symbols and OFDM data frames is known. The detection of the start position of an OFDM data symbol is performed on the basis of a cyclic guard interval correlation (Cyclic Prefix Correlation (CPC)), based on which the detection of the start position of an OFDM data frame by means of power measurement at the beginning of a OFDM data frame zero symbol takes place.

With the Worldwide Interoperability for Microwave Access (WiMAX) standard IEEE 802.16, no null symbols are transmitted at the beginning of an OFDM data frame, so that detection of the beginning of OFDM data frames on the basis of a power measurement is eliminated.

From the
US 7,660,371 B2 the features of the preamble of claim 1 are known. Furthermore, it is known from this document to determine a start position of an OFDM data frame as the sampling time associated with the maximum value of the metric.

From the
US 2009/0190510 A1 are also the features of the preamble of claim 1 known.

The object of the invention is therefore to provide a method and a device for determining the beginning of OFDM data frames in an OFDM data stream of a data transmission according to the WiMAX standard.

The object is achieved by a method for determining the first sample value of a data frame in an OFDM data stream having the features of patent claim 1 and by a device for determining the first sample value of a data frame in an OFDM data stream having the features of patent claim 12. Advantageous technical extensions are listed in the dependent claims.

According to the invention, for each sample of the OFDM data stream in a first step, a first metric by means of "cyclic guard interval correlation" and in a second step second metric values of a second metric by adding one of the number of OFDM symbols transmitted within a data frame interval of timeoffset first metric values whose intervals correspond in each case to the distances between the first samples of the OFDM symbols transmitted within the data frame interval. Finally, in a third step, the maximum value of the second metric is detected within a data frame interval as the first sample of the OFDM data frame.

When determining the first metric by means of "guardinterval cyclic correlation", use is made of the fact that the data transmitted in a guard interval are identical to the samples spaced by the number of samples provided for useful data transmission in an OFDM symbol , are positioned at the end of an OFDM symbol.

For this purpose, in a "cyclic guardinterval correlation" cyclically a certain number of successive samples of the OFDM data stream, which is smaller than or equal to the samples contained in the guard interval, with an identical number of consecutive samples of the OFDM data stream , which are each offset by the number of samples provided for user data transmission in an OFDM symbol, superimposed and checked for constructive overlay  in the presence of samples of the guard interval  or destructive overlay  in the presence of samples for user data transmission.

The cyclic overlay is performed by weighting a certain number of consecutive samples of the OFDM data stream with an identical number of consecutive samples of the OFDM data stream, each by the number of times for payload data transmission provided samples in an OFDM symbol and subsequent summation of the weighted samples.

In the case of constructive superimposition, a maximum valueideally a triangular maximum peak valueof the first metric forms at a sample positioned at the beginning of the guard interval within an OFDM symbol.

When determining the second metric, use is made of the fact that the individual OFDM data frames according to the WiMAX standard directly adjoin each other and each have an identical data frame interval, preferably of 5 milliseconds, but which is not a multiple of an OFDM symbol length. In the OFDM data stream, therefore, null values are inserted in the OFDM data stream at the end of each OFDM data frame instead of OFDM symbols.

For each individual sample of the OFDM data stream, in each case a second metric value is determined which results from the summation of first metric values at sampling times whose time intervals correspond to the time intervals between the first samples of the individual OFDM symbols. The second metric values thus determined in each case at the individual sampling times result in a second metric which likewise has maximum values at intervals of an OFDM symbol length, but which, in contrast to the course of the first metric, have a descending course.

The decreasing profile of the maximum values in the second metric results from the fact that, on the one hand, due to the temporal offset of the individual first metric values during the addition of the individual first metric values to form the second metric value, an increasingly smaller number of maximum values, the OFDM Symbols in the currently considered OFDM data frame of the OFDM data stream include, and on the other hand, the maximum values of the individual first metric values associated with the first samples of the OFDM symbols in the subsequent OFDM data frames due to the occupancy of the ends of the individual OFDM data frames Zero values are no longer in the time frame of multiples of an OFDM symbol length and thus make no or a lesser contribution in the currently considered OFDM data frame in the addition of the individual first metric values to form the second metric value.

In this way, the position of the largest maximum value in the course of the second metric gives the starting position of the currently considered OFDM data frame in the OFDM data stream.

Become for the determination of the initial position of the currently considered OFDM data frame not only second metric values, which are determined by addition of temporally offset first metric values of OFDM symbols transmitted in the currently considered OFDM data frame, but also takes into account second metric values, which by means of addition of temporal offset first metric values of OFDM symbols transmitted in subsequent OFDM data frames, the largest maximum values belonging to the start positions of the individual OFDM data frames are clearly different from the valuereduced maximum values and can therefore be identified more easily.

If the receiver does not know at which positions in the individual OFDM data frames OFDM symbols are transmitted and at which positions in the individual OFDM data frames no OFDM symbols are transmitted, then all possible hypotheses for the assignment of the individual positions of the individual ones must be given OFDM data frames are set up and the individual hypotheses are compared with regard to the selection of the hypothesis which corresponds to the actual occupancy of the individual positions in the individual OFDM data frames.

For this purpose, for each hypothesis, an associated second metric is calculated from second metric values which are each formed from the sum of timeoffset first metric values, wherein the time offset of the respective temporally offset first metric value is the time offset between the respective transmitted in the OFDM data frame according to the respective hypothesis OFDM symbol and the first transmitted in the same OFDM data frame OFDM symbol is determined.

For each hypothesis, the sum of the two differences between the largest maximum value of the determined second metric and in each case the two maximum values of the determined second metric offset by one OFDM symbol length on the right and left is determined within the individual OFDM data frame intervals. The hypothesis for which the largest sum value is determined, represents the correct hypothesis, the actual occupancy of the individual positions within the respective OFDM data frame with Represents OFDM symbols. The position of the largest maximum value within the second metric of the correct hypothesis gives the starting position of the respective OFDM data frame within the OFDM data stream.

For the correctness of this hypothesisbased procedure, the following conditions are required:
 An OFDM subdata frame is either completely occupied with OFDM symbols or contains no OFDM symbols at all,
 Each first OFDM subframe at the beginning of the data transmission area provided for the data transmission from a base station to a mobile station or the data transmission area of an OFDM data frame provided for the data transmission from the mobile station to the base station is occupied with OFDM symbols,
 • in each case only contiguous OFDM subdata frames within the data transmission area provided for the data transmission from a base station to a mobile radio device or within the data transmission range of an OFDM data frame provided for the data transmission from the mobile radio to the base station are occupied with OFDM symbols.

Before or after examination of the individual hypotheses, it is determined in a preliminary step whether OFDM symbols are ever transmitted within a data frame interval. For this purpose, it is checked whether a first metric calculated for this purpose within the data frame interval has at least one number of maximum values corresponding to the number of OFDM symbols per subdata frame, which are above a preselected threshold value and are each spaced apart by one OFDM symbol interval. If no OFDM symbols are identified in the examined data frame interval, the subsequent data frame interval is entered.

The method according to the invention and the device according to the invention for determining the beginning of a data frame in an OFDM data stream are explained in detail below with reference to the drawing. The figures of the drawing show:

1A a representation of a data structure of an OFDM data stream consisting of individual OFDM symbols,

1B a representation of a data structure of an OFDM data frame,

2A a timing diagram of an OFDM data stream,

2 B a time diagram of an associated first metric,

2C a time diagram of an associated second metric in consideration of OFDM symbols of a single OFDM data frame,

2D a time diagram of an associated second metric considering OFDM symbols of several OFDM data frames,

2E a representation of a data structure occupied by OFDM symbols and unoccupied positions in an OFDM data stream,

2F a time diagram of an associated first metric under real transmission conditions,

2G a timing diagram of an associated second metric for a first hypothesis,

2H a timing diagram of an associated second metric for a second hypothesis,

3 a flow chart of the inventive method for determining the beginning of a data frame in an OFDM data stream and

4 a block diagram of an inventive device for determining the beginning of a data frame in an OFDM data stream.

The mathematical foundations required for understanding the invention are derived below:
In the first step, first metric values x _{CPC} (n) of a first metric are determined by means of "cyclic guardinterval correlation" according to a previously used method. For this purpose, for each sampling instant n according to equation (1), a sequence of received samples r _{iARx} (n + i) of the OFDM data stream with a sequence of received samples r _{iARx} (n + i _{N} _{FFT} ), each by the number N _{FFT} of samples provided for user data transmission in an OFDM symbol are transmitted, weighted and then added together. The number N _{Corr} of samples r _{iARx} (n + i) of the OFDM data stream weighted with the identical number N _{Corr} of staggered samples r _{iARx} (n + i _{N} _{FFT} ) of the OFDM data stream is less than or equal to the number N _{CP} of samples in a guard interval. Since the OFDM transmission system is a multipleinputmultipleoutput (MIMO) transmission system, the computation of the first metric values x _{CPC} (n), as shown in equation (1), takes place at one Number N _{R} of receive antennas total N _{R} sequences of samples r _{iARx} (n + i) of the OFDM data stream which are weighted with a total of N _{R} sequences of staggered samples r _{iARx} (n + i) of the OFDM data stream.


N _{FFT}   In a number N _{EmpfangAbtastwerte} of received samples of the OFDM data stream n = 1, ..., N, for each sampling _{empfangAbtastwerte} _{Corr} N + 1, a first metric value x calculated _{CPC} (s).

In order to implement the _{method according to} the invention on a computer unit with limited computing accuracy, for example a signal processor, a normalization is to be carried out on the basis of very different signal levels of the received samples r _{iARx} (n + i) of the OFDM data stream. For this purpose, a normalized first metric value x _{CPCNorm} (n) is determined according to equation (2), which results from an unnormalized first metric value x _{CPC} (n) by magnitude formation and subsequent division by the squared power P (n) ^{2} in the considered interval.


N _{FFT}   In a number N _{EmpfangAbtastwerte} of received samples of the OFDM data stream n = 1, ..., N _{empfangAbtastwerte,} for each sampling time N + 1 _{Corr} a normalized first metric value computed x _{CPCNorm} (s).

For the calculation of a second metric value x _{FSM} (n) of the second metric at the sampling time n, in each case a timeoffset normalized first metric value x _{CPCNorm} (n + v (m)) is determined for each OFDM symbol transmitted in an OFDM data frame, where the time offset v (m) of the mth OFDM symbol transmitted in the sequence of OFDM symbols within the OFDM data frame is derived from the distance of the first sample of the mth transmitted OFDM symbol to the first sample of the first transmitted OFDM symbol results. In principle, any value can be used for the time offset v (1) of the first OFDM symbol transmitted in an OFDM data frame, since in the sequence of received OFDM symbols the beginning of a data frame must first be identified, while only one assignment sequence of several successive OFDM symbols in the sequence of received samples. Preferably zero is used as the value for the time offset v (1) of the first OFDM symbol transmitted in an OFDM data frame. Given a N _{number} of symbols of OFDM symbols transmitted in an OFDM data frame, the second metric value x _{FSM} (n) of the second metric at sampling time n results from knowing the relative positions of the first samples of the individual OFDM symbols transmitted in the OFDM data frame the addition of the total of N _{Nested symbols} to a respective transmitted OFDM symbol and timeshifted first metric values x _{CPC Norm} (n + v (m)) according to equation (3).


In a number N _{EmpfangAbtastwerte} of received samples of the OFDM data stream and a number of N _{samples / data frame} of samples per OFDM data frame is used for each sampling time n = 1, ..., N _{empfangAbtastwerte}  N _{samples / data frames,} a second metric value x _{FSM} ( n) of the second metric.

By the time offset of the individual normalized first metric values x _{CPCNorm} (v (n + m)), which with increasing position m of the transmitted in the OFDM data frame OFDM symbol in each case is lower by a factor of one number of maxima in the distance of an OFDM symbol length N _{S} from the starting position of the OFDM data frame, in each case an OFDM symbol length N _{S} and subsequent addition results in a sequence of maxima with decreasing maximum values in each OFDM data frame in the course of the second metric values composed of second metric values x _{FSM} (n) , The largest maximum value of this sequence of maxima is positioned at the start position of the respective OFDM data frame.

To even better dampen the maxima with smaller maximum values than the maximum with the largest maximum value, the second metric composed of second metric values x _{FSM} (n) is still optimized. For this purpose, temporally offset first metric values which belong to OFDM symbols in subsequent OFDM data frames are taken into account in the calculation of an optimized second metric composed of optimized second metric values x _{FSMOpt} (n). Since the OFDM symbols in subsequent OFDM data frames do not occur in the time frame of multiples of an OFDM symbol length of the OFDM symbols of the first OFDM data frame, the first metric values associated with the OFDM symbols transmitted in subsequent OFDM data frames carry x _{CPCNorm} (n + v (m)) does not add a constructive contribution to the second metric values x _{FSM} (n) of the second metric in the case of a summation over all temporally offset first metric values. Rather, those timeoffset first metric values x _{CPCNorm} (n + v (m)), which belong to one of the subsequent OFDM data frames, belong to one of the respective OFDM data frames and to second metric values x _{FSM} (n + i * N _{samples / data frame)} to add the composite second metric and then those from the optimized second metric values x _{FSMOpt} (s) composite optimized second metric from the addition of a total of N for all _{data frames} transmitted in an OFDM data stream OFDM data frames each determined second metric values x _{FSM} (n + i · N _{samples / data frame} ) according to equation (4).


The starting position n ^ _{data frame} of the first OFDM data frame to be identified, according to equation (5), the sampling instant of the maximum value of the optimized second metric values x _{FSMOpt} (n) is obtained.


In the event that the assignment of the OFDM data frame with OFDM symbols is not known in advance, all hypotheses for the assignment of the OFDM data frame must be established in advance. According to equation (6), an associated second metric composed of second metric values x.sub.j _{, FSM} (n) or according to equation (7) is an optimized second metric composed of optimized second metric _{values x.sub.j, FSMOpt} (n) for each individual hypothesis j the time offsets v _{j} (2), v _{j} (3), ..., v _{j} (m), ... v _{j} (N occupied _{symbols} ) of the 2 _{nd} to N occupied _{symbols} th in an OFDM Data frame transmitted OFDM symbols relative to the position of the first transmitted in an OFDM data frame OFDM symbol.


Since each second metric composed of optimized second metric values x _{j, FSMOpt} (n) has a maximum value for each hypothesis j, an associated start position n _{j, data frame} for an OFDM data frame according to equation (8) can be used for each hypothesis j Sampling time of the maximum value of the optimized second metric values x _{j, FSMOpt} (n) determined for the respective hypothesis j.


The hypothesis j for the assignment of an OFDM data frame with OFDM symbols corresponds to the actual occupancy of the OFDM data frame with OFDM symbols, whose optimized second metric value x _{j, FSMOpt} (n) has a maximum maximum value x _{j, FSMOpt} (n ^ _{j, data frame} ) which stands out most clearly compared to the adjacent maximum values. Thus, for the determination of the correct hypothesis j, the difference between the respective maximum maximum value is _{correct} for each hypothesis j x _{j, FSMOpt} (n ^ _{j, data frame} ) and the respective maximum value offset on the left by an OFDM symbol length of N _{S} probing values x _{j, FSMOpt} (n ^ _{j, data frame}  N _{S} ) to the difference between the respective maximum maximum value x _{j, FSMOpt} (n ^ _{j, data frame} ) and the respective maximum value offset by one OFDM symbol length of N _{S} samples on the right x _{j, FSMOpt} (n ^ _{j, data frame} + N _{S} ) and the correct hypothesis j _{correctly identified} according to equation (9) as the largest sum hypothesis j.


The starting position n
_{data frame of} the respective OFDM data frame results according to equation (10) as the sampling instant of the maximum value of the optimized second metric values
_{correctly} belonging to the correct hypothesis j
,


In undisturbed operation of the OFDM transmission system as well as time invariance of the OFDM transmission channel is only once the starting position n ^ _{data frame} of a single OFDM data frame according to equation (10). The starting positions n ^ _{data frame} (i) The following total of N _{data frames} in an OFDM data stream transmitted OFDM data frame i arise due to the fixed OFDM data frame length, preferably of 5 milliseconds, and the directly without gap contiguous OFDM data frame according to equation (11). Of course, the invention also covers other data frame lengths other than 5 milliseconds that will be introduced in future standards. n ^ _{data frame} (i) = n ^ _{data frame} + (i1) × N _{samples / data frame} (11)

If the transmission channel has little or no contamination, a hypothesis test is not required. In this case, it is sufficient to collect the start positions of the individual OFDM symbols in the vector v (m) and to perform the determination of the second metric composed of second metric values x _{FSM} (n) as stated above.

On the other hand, if the OFDM transmission is frequently disturbed, for example interrupted, or if there is a timevariant OFDM transmission channel, then those are for the calculation of the start positions n ^ _{data frame} (i) the subsequent OFDM data frame i required conditions of a fixed data frame length, preferably of 5 milliseconds, and a seamless transmission of successive OFDM data frames may not be given. In this case, for each subsequent OFDM data frame within the OFDM data stream, an individual identification of the associated start position n ^ _{data frame} (i) based on equation (10) be advantageous.

If the receiver does not know the availability of each OFDM data frame with OFDM symbols, all possible hypotheses j _{i} for the assignment of each OFDM data frame i must be examined using OFDM symbols also in this case and the correct hypothesis j _{correctly, i} for the assignment of the respective OFDM data frame i with OFDM symbols on the basis of equation (9) to identify. Typically, the assignment of all OFDM data frames is identical to OFDM symbols, so that for the subsequent OFDM data frames the assignment structure of the OFDM data frame determined by hypothesis comparison in the first OFDM data frame can be adopted with OFDM symbols and thus the determination of the start positions n ^ _{data frame} (i) for the following OFDM data frames i clearly simplified. But even the case of a different assignment of the respective subsequent OFDM data frame i with OFDM symbols is covered by the invention. In this case, a hypothesis comparison is to be carried out for each successive OFDM data frame i.

Is the starting position n ^ _{data frame} determined for each OFDM data frame, according to equation (12A), the start position n ^ _{symbol} (i, l) the Lth OFDM symbol in the ith OFDM data frame for the operation of a frequency division duplex (FDD) in which in the respective data frame either a data range intended for transmission from the mobile to the base station (uplink data area ) or a data area (downlink data area) intended for the transmission from the base station to the mobile station is determined.

For the operation case of a time division duplex (TDD), in which in the respective data frame at the data frame beginning a certain for the transmission of the base station to the mobile data area and followed by a gap provided for the transmission from the mobile to the base station data area is, the starting position results n ^ _{symbol} (i, l) the ith OFDM symbol in the data area of the ith OFDM data frame destined for transmission from the mobile station to the base station also according to equation (12A) and the start position n ^ _{symbol} (i, l) the lth OFDM symbol in the data area of the ith OFDM data frame intended for the transmission from the mobile station to the base station according to equation (12B). n ^ _{symbol} (i, l) = n ^ _{data frame} (i) + (l1) * N _{S} (12A) n ^ _{symbol} (i, l) = n ^ _{data frame} (i) + N _{symbols / data area} 1 × N _{S} + N _{gap} + (1  1) × N _{S} (12B)

The starting positions n ^ _{symbol} (i, l) or n ^ _{symbol} (i, l) of the individual OFDM symbols in an OFDM data frame according to Equations (12A) and (12B) can be used for the determination of a coarse estimate Δf ^ be used for the residual carrier frequency offset .DELTA.f. For this purpose, it is assumed that the sampling time value r (n) of the received OFDM signal at the sampling time n according to equation (13) is corrupted by the residual carrier frequency offset Δf compared to the sampling time value s (n) of the associated transmitted OFDM signal at the sampling time n. r (n) = s (n) * e ^{j2π * Δf * n} (13)

Starting from equation (1), the nonnormalized first metric value x _{CPC} (n) of the nonnormalized first metric results at the sampling instant n for a received OFDM signal falsified by the residual carrier frequency offset Δf compared to the transmitted OFDM signal according to equation (14). ,


Since the sample value s (n _{symbol} + i) of the transmitted OFDM signal within a guard interval corresponds to the sampling value s (n _{symbol} + i _{N FFT} ) of the transmitted OFDM signal offset in time by the number N _{FFT} of useful data transmission samples , the nonnormalized first metric composed of nonnormalized first metric values x _{CPC} (n _{symbol} ) results at the start time n _{symbol of} an OFDM symbol according to equation (15).


The rough estimate Δf ^ for the residual carrier frequency offset .DELTA.f thus results from equation (15) according to equation (16).


The rough estimate Δf ^ for the residual carrier frequency offset Δf is a normalized quantity which, according to equation (17), is obtained by normalizing the associated nonnormalized quantity Δf ^ [Hz] is calculated with the sampling frequency f _{S.}


The nonnormalized rough estimate
Δf ^ [Hz] for the residual carrier frequency offset .DELTA.f can according to equation (18) as a multiple
Δf ^ ' the bandwidth
a frequency carrier will be described.


Thus results between the normalized coarse estimate
Δf ^ for the residual carrier frequency offset Δf and the multiple
Δf ^ ' the normalized rough estimate
Δf ^ for the residual carrier frequency offset Δf from the bandwidth
a frequency carrier based on equation (17) and (18) a mathematical relationship according to equation (19).


The normalized rough estimate
Δf ^ for the residual carrier frequency offset .DELTA.f as a multiple
Δf ^ ' the bandwidth
of a frequency carrier thus results according to equation (20).
Δf ^ '= 1 / 2π * arg {x _{CPC} (n _{symbol} )} (20)

Since equation (20) is equivalent to the normalized first metric composed of nonnormalized first metric values x
_{CPC} (n
_{symbol} + v (m)) associated with the start position of all OFDM symbols transmitted in an OFDM data stream, an improvement of the normalized one may be coarse estimate
Δf ^ for the residual carrier frequency offset Δf as Vilefaches
Δf ^ ' the bandwidth
of a frequency carrier according to equation (21), by averaging over the nonnormalized first metric values x
_{CPC} (n
_{symbol} + v (m)) of all OFDM symbols transmitted in the OFDM
_{data stream} in total of N
_{symbols / data stream} . In this case, the time offset v
_{i} (m) of the mth OFDM symbol transmitted in the ith OFDM data frame is used in relation to the first transmitted OFDM symbol.


The method according to the invention for determining the beginning of a data frame in an OFDM data stream will be described below with reference to the flowchart in FIG 3 and the associated apparatus according to the invention for determining the beginning of a data frame in an OFDM data stream on the basis of the block diagram in FIG 4 explained in detail:
In the first method step S10 is in a unit 1 for determining a first metric a normalized first metric value x _{CPCNorm} (n) for each sampling time n of the OFDM data stream by means of "cyclic guard interval correlation" according to equation (2) determined from the individual OFDM symbols of the received OFDM signal. The individual OFDM symbols can thereby be received by a single receiving antenna of a singleantenna system or by a plurality of receiving antennas of a multiantenna system. For this purpose, for each sampling time n of the OFDM data stream, a number N _{Corr} of successive samples of the received OFDM signal, each weighted by a sample of the OFDM data stream offset by the number N _{FFT} of samples for user data in an OFDM symbol , added together. The number N _{Corr} of successive samples of the received OFDM signal for weighting and subsequent addition is less than or equal to the number N _{CP} of samples in the guard interval of an OFDM symbol.

In this way, results in accordance
2 B a time course for the amount  x
_{CPCNorm} (n)  the normalized first metric composed of normalized first metric values x
_{CPCNorm} (n), which is at the start positions
(With i = 1, 2, ...) of the individual OFDM symbols each have a maximum value  neglecting interference components in the received OFDM signal ideally a triangular or pointed maximum value  has. These maxima result from the fact that for the calculation of the normalized first metric composed of normalized first metric values x
_{CPC Norm} (n) at the individual start positions
the individual OFDM symbols are weighted a sequence of samples within the guard interval of a received OFDM symbol with a sequence of identical samples at the end of the same received OFDM symbol (see crossed hatched areas in the OFDM data stream in FIG
1A ) and thus a maximum possible constructive superposition of the two sample sequences is present.

In the next method step S20, if the assignment of the individual OFDM data frames with OFDM symbols is unknown to the receiver, it is determined in a preliminary step whether OFDM symbols are transmitted at all in the data frame interval currently to be considered or an OFDM data frame without OFDM symbols  ie a nonactivated OFDM data frame  is present. For this purpose, the normalized first metric composed of normalized first metric values x _{CPCNorm} (n) is used and, based on a suitably selected threshold value, the number of maximum values of the normalized first metric values x _{CPCNorm} (n) which is greater than the threshold value is determined. If the number of identified maximum values of the normalized first metric composed of normalized first metric values x _{CPC Norm} (n) which are greater than the threshold value is greater than a fixed number N _{min} , then an OFDM data frame occupied by OFDM symbols lies an activated OFDM data frame  before and with the determination of the start position of this OFDM data frame can be continued. Otherwise, the OFDM data stream must be examined for existing OFDM symbols in the next data frame interval. As a typical value for the comparison number N _{Min} , according to the WiMAX standard, the minimum number of OFDM symbols in a single OFDM subframe, namely five, can be selected.

The check as to whether OFDM symbols are ever transmitted in the currently considered data frame interval can also be carried out before the hypothesis test is carried out.

If there is an activated OFDM data frame, it is then in a unit 2 to form hypotheses for the first OFDM data frame to be identified in the case of an identical allocation structure of all OFDM data frames with OFDM symbols or for all OFDM data frames to be identified in the case of a different assignment structure of all OFDM data frames with OFDM symbols, each possible hypothesis an assignment of the respective OFDM data frame determined with OFDM symbols.

The multiplicity of possible hypotheses is reduced if the following prerequisites for the assignment of an OFDM data frame with OFDM symbols are met:
 An OFDM subframe of an OFDM data frame is either completely occupied with OFDM symbols or contains no OFDM symbols at all,
 The transmission of an OFDM subframe fully occupied with OFDM symbols in an OFDM data frame commences according to 2 at the start position of the data area provided for the transmission from the base station to the mobile radio device (downlink data area, starting at the start position of the respective OFDM data frame) or at the start position of the data area provided for the transmission from the mobile radio device to the base station (uplink data area, starting at a position within the OFDM data area offset from the start position of the respective OFDM data frame by the downlink data area and a zerovalued gap) and
 The individual OFDM subframes, which are completely occupied by OFDM symbols, are transmitted coherently in the downlink and / or uplink data area of the respective OFDM data frame.

Thus, for the operating case of a frequency division duplex (FDD), in which all eight subframes of an OFDM data frame are assigned to either the downlink data area or the uplink data area, there are a total of eight hypotheses for occupying one OFDM data frames with OFDM symbols (hypothesis 1: only first OFDM subdata frame of the downlink or uplink data area occupied; 2nd hypothesis: occupying only first and second OFDM subdata frames of the downlink or uplink data area; ....; 8. hypothesis: all eight subdata frames of the downlink or uplink data area occupied).

For the operation case of a time division duplex (TDD), in a first variant, four OFDM subframes of an OFDM data frame are assigned to the downlink data area and four OFDM subframes of an OFDM data frame to the uplink. Data area assigned. In the event that only the downlink data area or only the uplink data area is occupied, there are four hypotheses each. In the event that the downlink data area and at the same time the uplink data area are occupied, this results in a total of 16 hypotheses. In a second variant of the operating time of a time duplex, six OFDM subdata frames are assigned to the downlink data area and two OFDM subdata areas are assigned to the uplink data area. In the event that only the downlink data area is occupied, there are a total of six hypotheses. In the event that only the uplink data area is occupied, there are a total of two hypotheses. In the event that the downlink data area and at the same time the uplink data area are occupied, there are a total of twelve hypotheses.

In the same method step S20 is by a unit 3 for determining a second metric in the knowledge of the assignment structure of each OFDM data frame with OFDM symbols by the receiver for each sampling time n each have a second metric value x _{FSM} (n) a second metric according to equation (3) and ignorance of the occupancy structure of each OFDM data frame with OFDM symbols by the receiver for each determined hypothesis j and for each sampling time n respectively a second metric value x _{j, FSM} (n) of a second metric according to equation (6) determined. For this purpose, a time offset v (m) between the mth OFDM symbol transmitted within an OFDM data frame and the first OFDM symbol transmitted within an OFDM data frame is determined for each position occupied by an OFDM symbol within an OFDM data frame a normalized first metric value x _{CPCNorm} (n + v (m)) offset in time from the respectively determined time offset v (m) is determined. In the case of knowledge of the assignment structure of an OFDM data frame by the receiver, the second metric value x _{FSM} (n) of the second metric results from the addition of all timeshifted normalized first metric values x _{CPCNorm} (n + v (m)), which are in each case in For each hypothesis j, the corresponding second metric value x _{j, FSM} (n) of the second metric at the sampling time n is obtained from the OFDM data frame transmitted by an OFDM data frame Addition of all timeshifted normalized first metric values x _{CPCNorm} (n + v _{j} (m)), which were determined in each case for an OFDM symbol transmitted in an OFDM data frame in accordance with the hypothesis j.

The course of the amount  x
_{FSM} (n)  or 
_{xj, FSM} (n)  the second metric composed of second metric values x
_{FSM} (n) and x
_{j, FSM} (n), respectively
2C also has maximum values at the individual start positions
(with i = 1, 2, ...) of the individual OFDM symbols. However, the individual maximum values of the second metric composed of second metric values x
_{FSM} (n) and x
_{j, FSM} (n) have a decreasing value curve starting with the largest maximum value at the start position
of the first OFDM symbol within an OFDM data frame, which is also the start position
of the respective OFDM data frame. This decreasing value curve of the individual maximum values of the second metric composed of second metric values x
_{FSM} (n) or x
_{j, FSM} (n) results from two effects:
 The fixed OFDM data frame length of preferably 5 milliseconds is not an integer multiple of an OFDM symbol length of N _{S} samples and thus zero values are transmitted in the OFDM data frame at the end of an OFDM data frame (see simply hatched areas in FIG 1A ). Due to the time offset caused by the zero values, the proportions of the staggered and normalized first metric values x _{CPCNorm} (n + v (m)) and x _{CPCNorm} (n + v _{j} (m)) are those of the following OFDM data frames transmitted are no longer in the cyclic time frame of an OFDM symbol length of N _{S} samples and therefore carry no or only a small contribution to the additive, ie constructive superposition of all timeshifted and normalized first metric values x _{CPC Norm} (n + v (m)) or x _{CPCNorm} (n + v _{j} (m)) in the determination of the second metric values x _{FSM} (n) and x _{j, FSM} (n) of the second metric at the sampling time n.
 Due to the fact that the OFDM symbols transmitted in an OFDM data frame are positioned contiguously at the beginning of the downlink data area and / or the uplink data area, the individual staggered and normalized first metric values x _{CPCNorm} (n + v (m )) or x _{CPCNorm} (n + v _{j} (m)) with increasing time offset a smaller contribution to the additive ie constructive overlay to determine the second metric values x _{FSM} (n) and x _{j, FSM} (n) at start positions of the individual OFDM symbols at an increasing time distance to the starting position of the respective OFDM data frame.

In the next method step S30 is also in a unit 3 for determining a second metric in the knowledge of the allocation structure of each OFDM data frame with OFDM symbols by the receiver for each sampling time n each optimized second metric value x _{FSMOpt} (n) of the optimized second metric according to equation (4) and ignorance of the allocation structure of each OFDM Data frame with OFDM symbols by the receiver for each determined hypothesis j and for each sampling time n respectively an optimized second metric value x _{j, FSMOpt} (n) of the optimized second metric according to equation (7) determined. For this purpose, the second metric values x _{FSM} (n) or x _{j, FSM} (n) of the second metric at the sampling time n respectively determined for each of the consecutive OFDM data frames are offset by the time offset i × N _{sample values / data frames} between the respective OFDM data frames i and the first OFDM data frame identified in the OFDM data stream offset in time and the respective time offset second metric values x _{FSM} (n + i * N _{samples / data frame} ) or x _{j, FSM} (n + i * N _{samples / data frame} ) added.

The time course of the amount  x
_{FSMOpt} (n)  or 
_{xj, FSMOpt} (n)  the optimized second metric composed of optimized second metric values x
_{FSMOpt} (n) and x
_{j, FSMOpt} (n)
_{, respectively} 2D ideally points to the starting position
of the respective OFDM data frame a maximum value, which is opposite to the at the individual starting positions
(with i = 2, 3,...) of the remaining OFDM symbols positioned maximum values of the optimized second metric values composed of optimized second metric values x
_{FSMOpt} (n) and x
_{j, FSMOpt} (n), respectively.

In the subsequent method step S40 is in a unit
4 for determining a maximum value when the allocation structure of each OFDM data frame with OFDM symbols is known by the receiver, the largest maximum value of the optimized second metric values x
_{FSMOpt} (n) according to equation (5) in the respective OFDM data frame and thus the associated starting position
of the respective OFDM data frame in the OFDM data stream.

Ignoring the assignment structure of each OFDM data frame with OFDM symbols by the receiver is in one unit 4 in order to determine a maximum value for each determined hypothesis j, respectively the largest maximum value of the optimized second metric values x _{j, FSMOpt} (n) are determined according to equation (8) in the respective OFDM data frame.

In the next step S50, in the case of ignorance of the allocation structure of each OFDM data frame with OFDM symbols by the receiver by one unit
5 to detect the correct hypothesis, the occupancy structure of each OFDM data frame with OFDM symbols corresponding to the actual occupancy of each OFDM data frame with OFDM symbols. For each hypothesis j, a difference between the maximum value determined in the respective OFDM data frame in the previous method step S40 is provided for this purpose
the second metric composed of optimized second metric values x
_{j, FSMOpt} (n) and the maximum value of the optimized second metric value x
_{j, FSMOpt} (n) offset on the left by an OFDM symbol length of N
_{S} samples to a difference between that in the respective OFDM data frame determined maximum value
the optimized second metric values x
_{j, FSMOpt} (n) and the maximum value of the optimized second metric values x
_{j, FSMOpt} (n) added to the right by an OFDM symbol length of N
_{S} samples and the correct hypothesis j
_{correctly} determined according to equation (9), where the sum of both differences has the largest value.

While in the case of a correct hypothesis j _{correct} only those staggered in time and normalized first metric values (v _{j} (m) n +) in the determination of the optimized second metric values x _{j, FSMOpt} (s) are taken into account x _{CPCNorm,} on the one hand, a symmetry of the individual Maxima of the optimized second metric values x _{j, FSMOpt} (n) to the largest maximum of the optimized second metric values x _{j, FSMOpt} (n) and on the other hand a marked difference between the largest maximum of the optimized second metric values x _{j, FSMOpt} (n) and the two right and left side next offset positioned maxima of the optimized second metric values x _{j, FSMOpt} (n) causes the other hypotheses have lead to an asymmetry of the single maxima of the optimized second metric values x _{j, FSMOpt} (n) to the largest maximum of the optimized second metric values x _{j, FSMOpt} (n) and on the other hand to a much weaker difference between the largest Ma Maximum of the optimized second metric values x _{j, FSMOpt} (n) and the two maxima of the optimized second metric values x _{j, FSMOpt} (n) positioned next offset to the right and left.

For actual occupancy of consecutive OFDM data frames with OFDM symbols according to 2E are the amount  x _{CPCNorm} (n)  the associated normalized first metric composed of normalized first metric values x _{CPCNorm} (n) 2F , the amount  x _{1, FSMOpt} (n)  the second metric of a first assignment _{hypothesis,} composed of second metric values x _{1, FSMOpt} (n), in which the first three positions of an OFDM data frame are occupied by OFDM symbols, in 2G and the amount  x _{2, FSMOpt} (n)  the second metric of a second assignment _{hypothesis} composed of second metric values x _{2, FSMOpt} (n), in which all four positions of an OFDM data frame are occupied by OFDM symbols, in 2H shown.

The symmetry of the individual maxima of the second metric values x _{1, FSM} (n) to the largest maximum of the second metric values x _{2, FSM} (n) and the marked difference between the largest maximum of the second metric values x _{2, FSM} (n ) and the two maxima of the second metric values x _{1, FSM} (n) of the correct first assignment hypothesis, positioned next offset on the right and left side 2G ,

On the other hand, in 2H the asymmetry of the individual maxima of the second metric values x _{2, FSM} (n) to the largest maximum of the second metric values x _{2, FSM} (n) and on the other hand the much weaker difference between the largest maximum of the second metric values x _{2, FSM} (n) and the maxima of the second metric values x _{2, FSM} (n) of the incorrect second assignment hypothesis due to the erroneous contribution of the timeoffset first metric values x _{CPCNorm} (n + v) belonging to the fourth OFDM symbol of the OFDM data frame are positioned next offset on the right and left side (4)) to the composite to the largest maximum of the (second metric values x _{2, FSM} n) second metric left side offset of the second maxima metric values x _{2, FSM} (s) to be recognized.

Contains the analyzed OFDM receive signal superimposed noise components due to a low signaltonoise ratio of the OFDM transmission channel or due to inaccuracies in the signal path of the receiver  for example, not mutually compensated inphase and quadrature channel in the quadrature modulator of the receiver so erroneous maximum values in the normalized first Metric values x _{CPCNorm} (n) at a start position of an OFDM symbol, as in 2F in the dotted Störsignalverlauf indicated by arrow occur.

These erroneous maximum values at a start position of an OFDM symbol in the normalized first metric values x _{CPC standard} (n) falsify the asymmetry of the individual maxima of the second metric values x _{FSM} (n) to the largest maximum of the second metric values x _{FSM} (n) and the difference between the largest maximum of the second metric values x _{FSM} (n) and the two maxima of the second metric values x _{FSM} (n) positioned next offset to the right and left in the case of a correct assignment hypothesis, in the case of a correct assignment hypothesis, the distortions with respect to the symmetry of the individual maxima of the second metric x _{FSM} (n) to the largest maximum of the second metric values x _{FSM} (n) and with respect to the difference between the largest maximum of the second Metric values x _{FSM} (n) and the two maxima of the second metric values x _{FSM} (n) positioned next offset on the right and left side are smaller.

In the same method step S50, after the detection of the correct assignment hypothesis j, the starting position is
_{correctly} determined
n ^ _{data frame} of the respective OFDM data frame according to equation (10) as a sampling time of the largest maximum value of the correct assignment hypothesis j
_{correctly} associated optimized second metric values
determined.

In the next method step S60, the start positions n ^ _{data frame} (i) determined in each case on the first identified OFDM data frame following OFDM data frames. In an undisturbed  ie uninterrupted  operation of the OFDM transmission and a timeinvariant OFDM transmission channel are the starting positions n ^ _{data frame} (i) of the individual OFDM data frames are arranged in a cyclic time frame and are determined according to equation (11). In a disturbed operation of the OFDM transmission and a timevariant OFDM transmission channel, the starting positions n ^ _{data frame} (i) the individual OFDM data frames are no longer arranged in a cyclic time grid. If there is no longer a cyclic time grid, then the start positions n ^ _{data frame} (i) the individual OFDM data frames therefore continuously as in the case of the starting position n ^ _{data frame} of the first OFDM data frame detected in the OFDM data stream are respectively determined according to method steps S10 to S50.

In the same method step S60 is in one unit 6 for determining the starting position of OFDM symbols starting from the starting position n ^ _{data frame} or n ^ _{data frame} (i) the individual OFDM data frames the starting positions n ^ _{symbol} (i, l) determined in the individual OFDM data frame i at the lth position respectively transmitted OFDM symbols. For this purpose, for the operating case of a frequency duplex (FDD) for all transmitted in an OFDM data frame OFDM symbols and for the operating case of a time division duplex (TDD) for all transferred in the downlink data area OFDM symbols start position n ^ _{symbol} (i, l) of the ith OFDM symbol in the ith OFDM data frame according to equation (12A). For the operation case of a time division duplex (TDD), the start position for all OFDM symbols transmitted in the uplink data area n ^ _{symbol} (i, l) of the ith OFDM symbol in the ith OFDM data frame according to equation (12B).

In an optional method step S70 is in an optional unit
7 to estimate the carrier frequency offset a rough estimate
Δf ^ for the carrier frequency offset .DELTA.f starting from the phase of the first metric x
_{CPC} (n
_{symbol} ) at a start position n
_{symbol of} an OFDM data symbol transmitted in the OFDM data stream according to equation (15). A normalized rough estimate
Δf ^ for the residual carrier frequency offset .DELTA.f as a multiple
Δf ^ ' the bandwidth
of a frequency carrier is calculated according to equation (20). An improvement of the standardized rough estimate
Δf ^ for the residual carrier frequency offset .DELTA.f as a multiple
Δf ^ ' the bandwidth
of a frequency carrier is achieved according to equation (21) as a phase over a plurality of average first values of values x
_{CPC} (n
_{symbol} ) belonging to a respective one start position n
_{symbol} of a plurality of OFDM symbols transmitted in the OFDM data stream.

The invention is not limited to the illustrated embodiment with all variants. Other types of standardization and other, in particular future, allocation structures of OFDM data frames with OFDM symbols are also covered by the invention.