A method and an apparatus for estimation of a Doppler frequency in a wireless telecommunication system

Technical Background

In a wireless telecommunication system a data transmission channel can be time varying. This can be due to the movement of the transmitter, the movement of the receiver and/or changes in the telecommunication's environment. For cellular systems and broadcast systems the major cause for a large Doppler frequency spread is relatively high speed movement of the communication's terminal. In wireless telecommunication systems the transmitted signal typically undergoes refractions, shadowing and reflections due to the presence of various objects such as buildings, trees etc. in the data transmission channel. In consequence, the signal waves emitted by the transmitter arrive at the receiver antenna over multiple paths, a phe- nomen also known as multipath propagation. A complete set of propagation paths between a transmitter and a receiver forms a multipath data transmission channel. The data transmission path can be characterized by parameters, in particular delay, attenuation and phase shift. The path delay depends on the path length and on the speed at which the wave is propagating in different media. Along the paths signal attenuation and phase shift is caused by fading. To cope with severe channel conditions without using complex equalization filters signals can be transmitted as Orthogonal Frequency Division Multiplexing OFDM signals by means of mul- ticarriers. In an OFDM system a channel comprises a number of subcarriers which can be independently modulated, each by its own data. The modulation can be in accordance with a number of well-known modulation techniques such as Quadrature Amplitude Modulation QAM or Phase-Shift Keying PSK. A basement signal in an OFDM system is formed by the sum of these modulated subcarriers. A large number of closely spaced orthogonal subcarriers can be used to carry the data. The data is divided into several parallel data streams or channels, one for each subcarrier. Each subcarrier can be modulated with a conventional modula- tion scheme such as QAM or PSK at a low symbol rate obtaining total data rates similar to conventional single carrier modulations schemes. The employed low symbol rate makes the use of a guard interval between symbols affordable making it possible to handle time- spreading and eliminate Inter-Symbol Interference ISI. Intersymbol Interference ISI is caused in a signal when it passes through a frequency selective channel. In OFDM systems this can cause a loss of orthogonality of a subcarrier resulting in Inter-Carrier Interference ICI. To handle large delay spreads for a system based on OFDM one can make use of Guard Intervals GI. Guard Intervals GI are also referred to a cyclic prefix CP. A cyclic prefix CP is a copy of the last part of the OFDM symbol that is pre-appended to the transmitted symbol and re- moved before demodulation. The Long Term Evolution LTE wireless system is upgrading a conventional UMTS system towards OFDM based wireless broadband systems. A LTE eUTRAN system is designed to be compatible and to co-exist with existing wireless standards such as GSM, Edge, UMTS or HSPA. Orthogonal Frequency Division Multiplexing OFDM can be used for a downlink connection between a base station BS and a mobile station MS as well as for an uplink connection between a mobile station MS and a corresponding base station BS. The structure of a conventional LTE system consists of a 10 ms long radio frame containing twenty slots each of 0,5 ms duration. This frame is valid for both TDD and FDD transmissions.

To improve performance OFDM receivers typically include channel estimators provided to dynamically determine a channel response. This information can then be used to enable the receiver to process the received signal in a way that compensates for time dispersion effects of the channel. A conventional way of determining a channel response in an OFDM receiver is to dedicate certain ones of the carriers for use in conveying pilot signals or pilot symbols PS. Pilot signals PS contain known information that permits the channel estimator to determine a channel response on that carrier frequency by comparing the actual received signal with a signal known to have been transmitted. Carriers conveying the pilot symbols PS are spaced apart in frequency by an amount that permits the channel response of carriers laying in be- tween the pilot carriers to be accurately estimated by interpolating the channel responses determined for the pilot carriers.

For a wireless link Multi Input and Multi Output (MIMO) uses multiple antennas at both the transmitter and receiver to improve the communication performance. MIMO is a part of wire- less telecommunication standards such as 3 GPP Long Term Evolution, WiMAX and HSPA.

In BO ZHOU "Doppler estimation for uplink LTE", Master of Science Thesis, Stockholm, Sweden, 2008, a method for Doppler frequency estimation in MIMO-OFDM systems has been described. In this approach as Zero Crossing Rate (ZCR) or a Level Crossing Rate (LCR) algorithm which counts the number of crossings of an axis or level by fading sample is used. It has been proposed by Leif Wilhelmsson "Doppler spread estimation for OFDM system" US patent US 7599 453, 2009, that hysteresis can be used to improve performance for a low signal to noise ratio SNR and a low velocity of the user. Another conventional approach has been presented by Schober H., Jondral F., "The velocity estimation for OFDM based communication systems", Eng. Lab, Karlsruhe University, Germany, VTC 2002, IEEE 56th publication date 2002, vol. 2, pages 715-718, Vol. 2, describing the calculation of a number of samples of an autocorrelation function R (0), R (2 F _D x), R (2πΡο2τ), ... where τ is the time interval between two adjacent pilot symbols PS, searching for a first negative sample R (2πΡπ1οτ). In this conventional method the Doppler frequency fD is calculated as:

wherein this formula is calculated with the first zero crossing point of an autocorrelation function J ₀ (X) happening in a region of X ~ 2,405. Further, "A propagation channel estimation in mobile telecommunication systems" at the International Science Technical Conference, Voronezh, 2001 , vol. 2, pages 958 - 965, considers that an autocorrelation function is known at some parts of delays, and to use linear or square piece approximation of the autocorrelation function. After having found the first zero crossing point the Doppler frequency can be approximated by the above mentioned formula.

A main disadvantage of conventional methods for Doppler estimation in particular in uplink Long Term Evolution LTE wireless systems is a low performance in case of a low signal to noise ratio SNR and a low speed or velocity v of the user, e.g. a mobile station MS. Under these conditions the number of crossings greatly increase not due to signal fading but due to phase breathing at a low signal to noise ratio SNR. The method employing hysteresis, especially designed for low signal to noise ratio SNR can slightly improve the performance, but the achieved performance is still very low. The approach presented by Schober H., Jondral F., "The velocity estimation for OFDM based communication systems", Karlsruhe University, Germany, VTC 2002 has the disadvantage that in some modern telecommunication systems such as Long Term Evolution LTE a dynamic scheduling is used. Consequently, it is impossible to find enough samples to calculate first zero crossing points of the autocorrelation func- tion. The approach presented by Kajukov I., Manelis, V. "Propagation Channel Estimation in Mobile Telecommunication Systems", International Science, Technical Conference Voronezh, 2001, vol. 2, pages 958 - 965, uses piece approximation of the autocorrelation function which are known at some parts of delay only. The disadvantage of this approach is that it is difficult to cover a huge dispersion of Doppler frequencies. For example from 0 - 300 Hz, autocorrelation function is to be known at the special positions to avoid multiple meanings or ambiguity of the autocorrelation function. For example, for a Long Term Evolution LTE FDD uplink system pilot signals would have to be placed with a delay equal to 2 ms, however, his condition is not acceptable for commercial Long Term Evolution LTE wireless telecommunication systems.

SUMMARY OF THE INVENTION A goal of the present invention is to provide a method and an apparatus for estimation of a Doppler frequency in a wireless telecommunication system with an improved performance of Doppler frequency estimation. According to a first aspect this can be achieved by a method for estimation of a Doppler frequency in a wireless telecommunication system,

wherein data blocks each having at least one pilot symbol are retransmitted with a predetermined retransmission delay in response to a hybrid automatic repeat request, and

wherein the Doppler frequency is calculated in a first operation mode depending on a time interval between two pilot symbols and in a second operation mode depending on the retransmission delay.

In a first possible implementation of the method according to the first aspect each data block comprises a sub-frame having at least two slots each comprising a predetermined number of data symbols and a pilot symbol.

In a second possible implementation of the first possible implementation of the method according to the first aspect the symbols of said data block are formed by Orthogonal Frequency Division Multiplexing, OFDM, symbols.

In a third possible implementation of the method according to the first aspect as such or any of the aforementioned first and second implementations each OFDM symbol comprises a cyclic prefix. In a fourth possible implementation of the method according to the first aspect or any of its aforementioned implementations the retransmission delay is longer by a predetermined factor than the time interval between two pilot symbols.

In a fifth implementation being a possible implementation of the latter implementation of the method according to the first aspect the transmitted data blocks are received by a predetermined number N of reception antennas of a MIMO receiver.

In a sixth implementation being a possible implementation of the latter implementation of the method according to the first aspect each reception antenna provides a stream of data blocks having a predetermined number of complex correlation responses in the time domain.

In a seventh implementation being a possible implementation of one of the fifth and sixth implementations of the method according to the first aspect, from each data block comprising complex correlation responses in the time domain the cyclic prefix of each response is removed and the remaining responses are converted by discrete Fourier transformation into the frequency domain to provide a stream of data blocks having a predetermined number K of complex correlation responses for each reception antenna of the receiver in the frequency domain.

In a eighth implementation being a possible implementation of the seventh implementation of the method according to the first aspect, from the provided stream of complex correlation responses in the frequency domain of each reception antenna data blocks of K complex correlation responses of pilot symbols are selected to form a stream of data blocks of K complex correlation responses of pilot symbols for each reception antenna.

In a ninth implementation being a possible implementation of the eighth implementation of the method according to the first aspect, each data block comprising K complex correlation responses of pilot symbols an averaging over the number Nsc of subcarriers used by the OFDM symbols is performed to form a stream of data blocks of averaged complex correlation responses of K/Nsc pilot symbols for each reception antenna of the receiver.

In a tenth implementation being a possible implementation of the ninth implementation of the method according to the first aspect, each data block of K/Nsc averaged complex correlation responses of pilot symbols provided by each antenna of the predetermined number of reception antennas is put into K/Nsc " N delay lines each providing a time delay of a predetermined time length to form at the output of each delay line a stream of data blocks of K/Nsc ' N delayed averaged complex correlation responses of pilot symbols for each reception antenna of the receiver.

In a eleventh implementation being a possible implementation of the tenth implementation of the method according to the first aspect a stream of autocorrelation function estimations is calculated on the basis of the stream of data blocks of averaged complex correlation responses and the stream of data blocks of delayed averaged complex correlation responses of pilot symbols for each reception antenna of the receiver.

In a twelfth implementation being a possible implementation of the eleventh implementation of the method according to the first aspect, in the first operation mode the stream of first autocorrelation function estimations is calculated by:

multiplying of K/Nsc averaged complex correlation responses of pilot symbols of a first slot of a sub-frame for each reception antenna with corresponding conjugated averaged complex correlation responses of pilot symbols of a second slot of said sub-frame, averaging the real parts of the K/Nsc obtained multiplication results for each reception antenna of the N reception antennas to provide an average value for each reception antenna of the receiver, and

averaging the obtained average values of all reception antennas of the receiver to provide the respective first autocorrelation function estimation.

In a thirteenth implementation being a possible implementation of the twelfth implementation of the method according to the first aspect, in the first operation mode the first autocorrelation function estimation i?, is calculated according to the formula:

wherein

τ is the time interval between two pilot symbols in the same data block,

K/Nsc is the number of averaged complex correlation responses of pilot symbols of each data block,

N is the number of reception antennas of the receiver,

C is the averaged complex correlation response of a pilot symbol in the frequency domain, and

C* is the conjugated averaged complex correlation response of a pilot symbol in the frequency domain.

In a fourteenth implementation being a possible implementation of one of the eleventh to thirteenth implementations of the method according to the first aspect, in the second operation mode a set of second autocorrelation function estimations R ₂ is calculated according to the following formula: for several values of an integer number n between a minimum retransmission delay (n _m j _n - τ) and a maximum retransmission delay (n _max - τ),

wherein

η τ is the retransmission delay of a data block in response to a hybrid automatic repeat request,

K/Nsc is the number of averaged complex correlation responses of pilot symbols PS of each data block, N _n is the number of available pairs of averaged complex correlation responses of pilot symbols PS of a first and second data block of each reception antenna and corresponding delayed averaged complex correlation responses C* of pilot symbols PS of the first and second data block of the same reception antenna with a delay corresponding to the retransmission delay η·τ,

C is the averaged complex correlation response of a pilot symbol PS in the frequency domain, and

C* is the conjugated averaged complex correlation response of a pilot symbol PS in the frequency domain, and

T _j is a template coefficient being equal to 1 if transmission in data block j is present in the channel, otherwise T _j being equal to 0.

In a fifteenth implementation being a possible implementation of one of the twelfth to thirteenth implementations of the method according to the first aspect in the first operation mode the first autocorrelation function estimations Rj are filtered in time by means of an alpha filter of a first type comprising a first alpha parameter, al, to form a stream of first filtered autocorrelation function estimations .

In a sixteenth implementation being a possible implementation of the method according to the first aspect or any of its aforementioned implementations, a first stream of first power estimations PI is calculated in the first operation mode by taking a square module of each of the K/Nsc averaged complex correlation responses of pilot symbols PS of a first and second data block for each reception antenna and

by averaging all obtained square modules of all data blocks of all reception antennas of the receiver.

In a seventeenth implementation being a possible implementation of the method according to the first aspect or any of its aforementioned implementations, a stream of second power estimations P ₂ is calculated in the second operation mode by taking a square module of each of the K/Nsc averaged complex correlation responses C of pilot symbols PS of a first and second data block for each reception antenna,

by taking a square module of each of K/Nsc delayed averaged complex correlation responses C of pilot symbols PS of a first and second block for each reception antenna, and

by averaging of all obtained square modules of all data blocks of all reception antennas of the receiver.

In a eighteenth implementation a possible implementation of the sixteenth implementation of the method according to the first aspect, in the first operation mode from the calculated stream of first power estimations Pi a noise power value is subtracted and the result is filtered in time by means of an alpha filter of a first type comprising a first alpha parameter al to form a stream of first filtered power estimations Ρ _λ .

In a nineteenth implementation being a possible implementation one of the seventeenth and eighteenth implementations of the method according to the first aspect, in the first operation mode the Doppler frequency estimations foi are calculated by means of a Bessel function in response to a stream of first ratio values 7?, / P _i calculated by dividing the first filtered autocorrelation function estimations by the first filtered power estimations .

In a twentieth implementation being a possible implementation of the nineteenth implementation of the method according to the first aspect, in the first operation mode the Doppler frequency estimations foi are calculated by defining on an interval from zero to a first extremum of the Bessel function the closest corresponding argument of a zero order Bessel function wherein the Bessel function of this found argument is close to the calculated respective first ratio value R _x i ' P _l .

In a twenty-first implementation being a possible implementation of the twentieth implementation of the method according to the first aspect, the defined argument is multiplied by a constant value V.

In a twenty-second implementation being a possible implementation of the twenty-first implementation of the method according to the first aspect, the multiplied constant value V is given by:

1

wherein τ being the time interval between two pilot symbols (PS) of a data block (DB), to provide the Doppler frequency estimations foi of the first operation mode. In a twenty-third implementation being a possible implementation of one of the twentieth to twenty-second implementation of the method according to the first aspect, in the first operation mode the provided Doppler frequency estimations foi are filtered by an alpha filter of a second type comprising a second alpha parameter a2 to form a stream of filtered Doppler frequency estimations of the first operation mode.

In a twenty-fourth implementation being a possible implementation of one of the fourteenth to twenty-third implementations of the method according to the first aspect, in the second opera- tion mode each autocorrelation function estimation from the set of second autocorrelation function estimations R ₂ is filtered in time by means of an alpha filter of a third type comprising a third alpha parameter a3 to form a stream of sets of second filtered autocorrelation function estimations R ₂ .

In a twenty-fifth implementation being a possible implementation of one of the sixteenth to twenty-fourth implementations of the method according to the first aspect, in the second operation mode from the second power estimations P2 a noise power value is subtracted and the result is filtered in time by means of an alpha filter of a third type comprising a third alpha parameter a3 to form a stream of second filtered power estimations P ₂ .

In a twenty-sixth implementation being a possible implementation of the twenty-fourth or twenty-fifth implementation of the method according to the first aspect, in the second operation mode the set of Doppler frequency estimations for different retransmission delays is calculated by means of a Bessel function in response to a stream of sets of second ratio values R ₂ 1 P ₂ calculated by dividing each second filtered autocorrelation function estimations R ₂ from the set of the filtered autocorrelation function estimations by the second filtered power estimations P ₂ . In a twenty-seventh implementation being a possible implementation of the twenty-sixth implementation of the method according to the first aspect, in the second operation mode the set of Doppler frequency estimations for different retransmission delays is calculated by defining on an interval from zero to a first extremum of the Bessel function a set of the closest corresponding arguments of a zero order Bessel functions wherein the Bessel functions of these found arguments are close to the set of calculated respective second ratio values R ₂ 1 P ₂ . In a possible implementation of the method according to the first aspect each found argument is multiplied by a constant value V.

In a twenty-eighth implementation being a possible implementation of the twenty-seventh implementation of the method according to the first aspect the multiplied constant value V is given by:

2π · η · τ wherein η·τ is a retransmission delay of a data block in response to a hybrid automatic repeat request, to provide the Doppler frequency estimations fo ₂ of the second operation mode. In a twenty-ninth implementation being a possible implementation of one of the twenty-sixth to twenty-eighth implementations of the method according to the first aspect, in the second operation mode the Doppler frequency estimations are calculated by averaging of all Doppler frequency estimations from the set of Doppler frequency estimations for different retransmis- sion delays to provide the Doppler frequency estimations fo ₂ of the second operation mode.

In a thirtieth implementation being a possible implementation of the twenty-ninth implementation of the method according to the first aspect in the second operation mode the provided Doppler frequency estimations fp ₂ are filtered by an alpha filter of a fourth type comprising a fourth alpha parameter <x4 to form a stream of second filtered Doppler frequency estimations f _D2 of the second operation mode.

In a thirty-first implementation being a possible implementation of one of the fifth to thirtieth implementations of the method according to the first aspect, switching from the first operation mode to the second operation mode is performed according to switching conditions SC.

In a thirty-second implementation being a possible implementation of the thirty-first implementation of the method according to the first aspect, the switching conditions SC comprise as a first switching condition SQ an availability of a hybrid automatic repeat request HARQ, as a second switching condition SC ₂ that a current filtered Doppler frequency estimation calculated by the receiver in the current operation mode is less than a predetermined frequency threshold value fiH,

as a third switching condition SC ₃ that the value of the second filtered autocorrelation function estimation R ₂ calculated in the second operation mode for the minimum retransmission delay (n _m j _n -T) is higher than the value of the second filtered autocorrelation function estimation R ₂ calculated in the second operation mode for the maximum retransmission delay

(nmax-τ).

In a thirty-third implementation being a possible implementation of one of the fifth to thirty- second implementations of the method according to the first aspect, switching from the second operation mode to the first operation mode is performed also according to switching conditions SC.

In a thirty-fourth implementation being a possible implementation of the thirty-third imple- mentation of the method according to the first aspect, the switching conditions SC comprise as a fourth switching condition SC4 that a time interval between reception of the last hybrid automatic repeat request HARQ and a current time is less than a predetermined maximum time period M _max , as a fifth switching condition SC5 that a current filtered Doppler frequency f _m ; f _D2 j calculated by the receiver in the current operation mode is more than a predetermined frequency threshold value f _t h and

as a sixth switching condition SC6 that the value of the second filtered autocorrelation func- tion R ₂ calculated in a second operation mode for the maximum retransmission delay (n _m i _{n T} ) is lower than the value of the second filtered autocorrelation function R ₂ calculated in the second operation mode for the maximum retransmission delay (n _{max T} ).

In a thirty-fifth implementation being a possible implementation of one of the fifteenth or eighteenth to thirty-fourth implementations of the method according to the first aspect, the alpha parameters of the employed alpha filters comprise

the first alpha parameter ai being set to 0.01 ,

the second alpha parameter a ₂ being set to 0.03,

the third alpha parameter a ₃ being set to 0.007 and

the fourth alpha parameter a ₄ being set to 0.3.

The above mentioned goal is also according to a second aspect achieved by a wireless telecommunication system comprising at least one receiver performing the method according to the first aspect of the present invention or any conceivable implementation of that method including those described above.

In a first possible implementation of the wireless telecommunication system according to the second aspect the wireless telecommunication system is formed by a Long Term Evolution LTE system.

In a second possible implementation being an implementation of the first implementation of the wireless telecommunication system according to the second aspect, the wireless telecommunication system is formed by a Long Term Evolution LTE FDD system.

In a third possible implementation being an implementation of the first implementation of the wireless telecommunication system according to the second aspect, the wireless telecommunication system is formed by a Long Term Evolution LTE TDD system.

In a forth possible implementation being an implementation of the wireless telecommunication system according to the second aspect or any of its aforementioned implementations, the wireless telecommunication system is a Long Term Evolution LTE system performing the method according to the first aspect or any conceivable implementation of that method including those described above for an uplink UL between a mobile station and a base station. In a fifth possible implementation being an implementation of the wireless telecommunication system according to the second aspect or any of its aforementioned implementations, the wireless telecommunication system is a Long Term Evolution LTE system performing the method according to the first aspect or any conceivable implementation of that method including those described above for a downlink DL between a base station and a mobile station of the wireless telecommunication system.

In a sixth possible implementation being an implementation of the second implementation of the wireless telecommunication system according to the second aspect, possibly also in combination with the additional features of one of the fourth and fifth implementations, the wireless telecommunication system is formed by a Long Term Evolution FDD uplink system wherein the time length L of the delay lines is set to 8 ms. In a seventh possible implementation being an implementation of the second implementation of the wireless telecommunication system according to the second aspect, possibly also in combination with the additional features of one of the fourth, fifth and sixth implementations, the wireless telecommunication system is formed by a Long Term Evolution LTE FDD uplink system wherein the frequency threshold value f™ is set to 70 Hz.

In a eighth possible implementation being an implementation of the third implementation of the wireless telecommunication system according to the second aspect, possibly also in combination with the additional features of one of the fourth and fifth implementations, the wireless telecommunication system is formed by a Long Term Evolution LTE TDD uplink system wherein the time length L of the delay lines is set to 10 ms.

In a ninth possible implementation being an implementation of the third implementation of the wireless telecommunication system according to the second aspect, possibly also in combination with the additional features of one of the fourth, fifth and eighth implementations, the wireless telecommunication system is formed by a Long Term Evolution TDD uplink system wherein the frequency threshold value frH is set to 60 Hz.

In a tenth possible implementation being an implementation of the wireless telecommunication system according to the second aspect or any of its aforementioned fist to ninth imple- mentations, the wireless telecommunication system comprises a maximum time period M _max which is set to 250 ms. According to a third aspect the invention provides a receiver for a wireless telecommunication system performing the method according to the first aspect or any of the aforementioned first to thirty-fifth implementations thereof.

According to a fourth aspect the invention provides a base station of a wireless telecommunication system comprising a receiver according to the third aspect.

BRIEF DESCRIPTION OF FIGURES

In the following possible implementations of embodiments of different aspects of the present invention are described with reference to the enclosed figures.

Fig. 1 shows a diagram illustrating an uplink in a wireless telecommunication system between a mobile station and a base station for illustrating the method for estimation of a Doppler frequency in a wireless telecommunication system according to an aspect of the present invention;

Fig. 2 shows a frame structure for a Long Term Evolution wireless telecommunication system as employed in a possible implementation of the method for estimation of a Doppler frequency in a wireless telecommunication system according to an aspect of the present invention;

Fig. 3 shows a signal diagram for illustration of a dynamic scheduling in a Long

Term Evolution wireless telecommunication system and for illustration of the method for estimation of a Doppler frequency according to the first aspect of the present invention;

Fig. 4 shows a block diagram of a receiver which can be employed in a wireless telecommunication system according to the second aspect of the present invention;

Fig. 5 shows a simple state diagram for illustrating the functionality of a method and an apparatus for estimation of a Doppler frequency within a wireless telecommunication system according to an aspect of the present invention;

Fig. 6 shows a flow chart of a possible implementation of a method for estimation of a Doppler frequency in a wireless telecommunication system according to the second aspect of the present invention; shows a circuit diagram for illustrating a possible implementation of an alpha filter as employed within a method and an apparatus for estimation of a Doppler frequency in a wireless telecommunication system according to an aspect of the present invention; shows a diagram for illustrating a dependence of an normalized autocorrelation function from the number of subcarriers for illustrating the functionality of a method for estimation of a Doppler frequency in a wireless telecommunication system according to the first aspect of the present invention;

Fig. 9 shows a signal diagram indicating the dependence of an average output value calculated in a first operation mode from a Doppler frequency in the propagation channel illustrating the functionality of a method for estimation of a Doppler frequency in a wireless telecommunication system according to the first aspect of the present invention;

Fig. 10 shows autocorrelation functions for different Doppler frequencies illustrat- ing the functionality of a method for estimation of a Doppler frequency in a wireless telecommunication system according to a first aspect of the present invention; shows a diagram for illustrating a Bessel function which can be used in a possible implementation of a method for estimation of a Doppler frequency in a wireless telecommunication system according to an aspect of the present invention;

Fig. 12a - 12h show diagrams for illustrating the results provided by the method for esti- mation of a Doppler frequency in a wireless telecommunication system according to the first aspect of the present invention in comparison with conventional methods according to the state of the art;

Fig. 13a - 13f show snap shots of Doppler frequency estimations for different parameters provided by the method for estimation of a Doppler frequency in a wireless telecommunication system according to the first aspect of the present invention DETAILED DESCRIPTION OF EMBODIMENTS

In the following possible implementations and embodiments of different aspects of the method and apparatus for estimation of a Doppler frequency in a wireless telecommunication system according to the present invention are described in more detail.

As can be seen in fig. 1 a wireless telecommunication system employing a method for estimation of a Doppler frequency according to the first aspect of the invention the system can comprise at least one base station BS to which one or several mobile stations MS can be con- nected via a wireless link. In Fig. 1 an uplink UL between a mobile station MS and a base station BS is illustrated. The mobile station MS can move with a velocity v to the base station BS. For example, a typical user can move with a speed ranging from 3 km/h to 250 km/h provided that the mobile station MS is arranged within a vehicle. The movement of the mobile station MS relative to the base station BS can cause a frequency shift so that it is necessary to perform a Doppler frequency estimation. For example, in an uplink of a Long Term Evolution LTE wireless telecommunication system due to the lack of pilot signals on the same frequency in conventional systems there is a problem of having an uncompleted estimation of the autocorrelation function. The autocorrelation function is known only on some limited set of delay values. To overcome this disadvantage the method for estimation of Doppler fre- quency in a wireless telecommunication system according to the first aspect of the present invention uses hybrid automatic repeat request, HARQ, transmissions which extends the set of possible delay values, on which an autocorrelation function is known. The use of such delay values increases significantly the performance of the estimation of the Doppler frequency. In a first aspect of the present invention a method is provided which estimates a Doppler fre- quency in the wireless telecommunication system in two different operation modes OM _\ ,

OM ₂ . The Doppler frequency estimation is performed by a receiver receiving data blocks DB from a transmitter. The receiver can be for example provided within a base station BS as shown in fig. 1. A hybrid automatic repeat request HARQ is used in wireless telecommunication systems with downlink DL or uplink UL for high speed data transmission. The hybrid automatic repeat request HARQ can be used for facilitating fast error detection and correction. HARQ is a stop and wait protocol wherein subsequent transmissions can take place after receiving an acknowledgement signal from the receiving entity. After an acknowledgement is received a further transmission of the next data block is done otherwise a retransmission of the same data block is performed. In Long Term Evolution LTE systems HARQ can be im- plemented on a MAC level module called HARQ entity. HARQ is synchronous and the mobile station of the user entity UE receives the uplink grant for transmission on a control channel. In the method for estimating a Doppler frequency in a wireless telecommunication system according to the first aspect of the present invention data blocks DB each having at least one pilot symbol PS are retransmitted with a predetermined transmission delay in response to a hybrid automatic repeat request HARQ. The Doppler frequency is calculated in the first op- eration mode OMj depending on a time interval τ between two pilot symbols PS. In the second operation mode OM ₂ the Doppler frequency fo is calculated depending on the retransmission delay. The switching between the two operation modes OMj, OM ₂ is performed according to switching conditions or switching criteria SC. Fig. 2 shows a frame structure for illustrating a possible implementation of the method for estimation of a Doppler frequency fo in a wireless telecommunication system according to the first aspect of the present invention. In the shown exemplary data structure a frame comprises sub-frames (or data blocks) wherein each sub-frame has two time slots. The frame as employed by a LTE wireless telecommunication system comprises ten sub-frames (or data blocks) each having two slots. The frame can comprise a length of 10 ms. Each slot comprises a predetermined number of data symbols and a pilot symbol PS. It can be seen in fig. 2 that each subframe comprises in the shown example seven symbols wherein one of the symbols is formed by a pilot symbol PS. In the shown example the fourth symbol of the seven symbols is formed by a pilot symbol PS. The symbols can be formed by Orthogonal Frequency Division Multiplexing OFDM symbols each having a cyclic prefix CP. The use of OFDM symbols increases the immunity to multipath propagation and increases the ability of usage of adaptive modulation on different subcarriers. All OFDM symbols comprise a cyclic prefix CP which is added to avoid multipath interference. Fig. 3 illustrates a dynamic frequency scheduling which is for example employed in a Long Term Evolution LTE wireless telecommunication system. Fig. 3 shows sub-frames each having two slots having a pilot symbol PS wherein the time τ is the time interval between two adjacent pilot symbols PS. The time interval τ can be for example 0,5 ms. As shown in the example of fig. 3 four data blocks or sub-frames each having two time slots DBi, DB ₂ , DB ₃ , DB ₄ are transmitted. It can be seen that the first two data blocks DBi, DB ₂ are transmitted without that a transmission error occurs. It can be seen that the different data blocks DBj are transmitted on different frequencies. Since in the shown example the third data block DB ₃ is transmitted with a transmission error it is retransmitted with a retransmission delay as data block DB ₄ . As can be seen in fig. 3 the retransmission delay η·τ is longer by a predetermined factor n than the time interval τ between two pilot symbols PS for the same data block DB. With the method according to the first aspect of the present invention the hybrid automatic repeat request HARQ transmissions as shown in fig. 3 are used for extending a set of possible delay values on which an autocorrelation function is known. The transmitted data blocks as shown in fig. 3 are received by a receiver which can be located within a node of a wireless telecommunication system, for example a receiver within a base station. In a possible implementation the transmitted data blocks DB are received by a predetermined number of reception antennas of a MIMO receiver. In a possible implementation each reception antenna provides a stream of data blocks having a predetermined number of complex correlation responses in the time domain.

Fig. 4 shows a block diagram of a possible specific implementation for a receiver 1 according to a further aspect of the present invention which performs a method for estimation of a Dop- pler frequency fo. In the shown implementation of fig. 4 the receiver 1 is a MIMO receiver in having several reception antennas 2-1 to 2-N. Each reception antenna 2-i provides a stream of data blocks having a predetermined number of complex correlation responses in the time domain. The data blocks received in the time domain by the reception antennas are pre- processed to increase the signal to noise ratio SNR. From each data block comprising complex correlation responses in the time domain the respective cyclic prefix CP of each response is removed by a CP removal unit 3-i being connected to the respective reception antenna 2-i as shown in fig. 4. The remaining responses are converted by discrete Fourier transformation DFT in a corresponding DFT unit 4-i into the frequency domain to provide a stream of data blocks having a predetermined number K of complex correlation responses for each reception antenna of the MIMO receiver 1 in the frequency domain.

In a further stage from the provided stream of K complex correlation responses in the frequency domain each reception antenna 2-i data blocks of K complex correlation responses of pilot symbols PS are selected by a corresponding PS selection unit 5-i to form a stream of data blocks of K complex correlation responses C of pilot symbols PS for each reception antenna 2-i. Each data block comprising K complex correlation responses C of pilot symbols PS undergoes an averaging over a number Nsc of subcarriers used by the OFDM symbols in an averaging unit 6-i as shown in fig. 4. The averaging is performed to form a stream of data blocks of K/Nsc averaged complex correlation responses C of pilot symbols PS for each reception antenna 2-i of the MIMO receiver 1 as shown in fig. 4.

Each data block of K/Nsc averaged complex correlation responses C of pilot symbols PS provided by each reception antenna 2-i of the predetermined number N of reception antennas is put into K/Nsc ' delay lines 7-i for providing a time delay of a predetermined time length L to form at the output of each delay line 7-i a stream of K/Nsc " N delayed averaged complex correlation responses C of pilot symbols PS for each reception antenna of the MIMO receiver 1 as shown in fig. 4. After this pre-processing a stream of autocorrelation function estimations R is calculated in the shown implementation on the basis of the stream of averaged complex correlation responses C and on the basis of the stream of data blocks of delayed averaged complex correlation responses C of pilot symbols PC for each reception antenna of the MIMO receiver 1. The MIMO receiver 1 as shown in fig. 4 is switchable between two operation modes OM.

In the first operation mode OMi the stream of autocorrelation function estimations R _\ is calculated within the calculation unit 8 as shown in fig. 4. In the autocorrelation calculation unit 8 in a possible implementation a stream of autocorrelation function estimations Ri is calcu- lated by multiplying K/Nsc averaged complex correlation responses C of pilot symbols PS of a first slot of a sub-frame for each reception antenna with corresponding K/Nsc conjugated averaged complex correlation responses C* of pilot symbols PS of a second slot of the same sub-frame. Further in a possible implementation the K/Nsc real parts of the obtained K/N _S c multiplication results for each reception antenna of the N reception antennas are averaged to provide an average value for each reception antenna 2-i of the receiver 1. Further, the obtained average values of all reception antennas of the receiver 1 are averaged again to provide the respective autocorrelation function estimation R _\ output by the autocorrelation calculation unit 8 as shown in fig. 4. In a possible specific implementation of the MIMO receiver 1 as shown in fig. 4 in the first operation mode OMj the autocorrelation function estimation Ri is calculated within the autocorrelation calculation unit 8 according to the formula: wherein

τ is the time interval between two pilot symbols PS in the same data block DB,

K/Nsc is the number of averaged complex correlation responses C of pilot symbols PS of each data block DB,

N is the number of reception antennas 2-i of the receiver 1,

C is the averaged complex correlation response of a pilot symbol PS in the frequency domain, and

C* is the conjugated averaged complex correlation response of a pilot symbol PS in the frequency domain.

In the second operation mode OM2 of the MIMO receiver 1 as shown in fig. 4 a set of auto- correlation function estimations R ₂ is calculated by the autocorrelation calculation unit 9 as shown in fig. 4. In a possible embodiment the autocorrelation function estimations R ₂ are calculated by the autocorrelation calculation unit 9 according to the following formula: N K/H _sc L

^R > ^{N ' R) =} ^ N -N∑ ,∑ , ∑ , ^RE P- j ^{■ c} -- ^» . ^< · ^τ > ^{■ TJ→} } for several values of an integer number n between a minimum retransmission delay (n _m j _n x) and a maximum retransmission delay (n _max T),

wherein

η·τ is the retransmission delay of a data block DB in response to a hybrid automatic repeat request HARQ,

KTNsc is the number of averaged complex correlation responses C of pilot symbols PS of each data block DB,

N _n is the number of available pairs of averaged complex correlation responses C of pilot symbols PS of a first and second data block DB of each reception antenna 2-i and corresponding delayed averaged complex correlation responses C of pilot symbols PS of the first and second data block of the same reception antenna with a delay corresponding to the retransmission delay η τ,

C is the averaged complex correlation response of a pilot symbol PS in the frequency domain, and

C* is the conjugated averaged complex correlation response of a pilot symbol PS in the frequency domain, and

T _j is a template coefficient being equal to 1 if transmission in data block j is present in the channel, otherwise T _j being equal to 0.

In a possible implementation of the MIMO receiver 1 as shown in fig. 4 a stream of power estimations Pi is calculated in the first operation mode OMi by a power calculation unit 10 wherein the power calculation unit 10 calculates in the first operation mode OMi a stream of power estimations Pi by taking a square module of each of the K/Nsc averaged complex correlation responses of pilot symbols PS of a first and second data block for each reception antenna 2-i and by averaging all obtained square modules of all data blocks DB of all reception antennas 2-i within the MIMO receiver 1 as shown in fig. 4. According to a possible implementation of the MIMO receiver 1 as shown in fig. 4 a stream of power estimations P ₂ is calculated in the second operation mode OM ₂ by a power calculation unit 1 1 wherein the power calculation unit 1 1 calculates in the second operation mode OM ₂ a stream of power estimations by taking a square module of each of the K/Nsc averaged complex correlation responses C of pilot symbols PS of a first and second data block for each reception antenna. Further, power estimations P ₂ are calculated by taking a square module of each of the K/Nsc delayed averaged complex correlation responses C of pilot symbols of a first an second data block of each reception antenna and by averaging of all obtained square modules over all reception antennas of the MIMO receiver 1.

In a possible implementation of the receiver 1 as shown in fig. 1 in the first operation mode OMi the autocorrelation function estimations Rj calculated by the autocorrelation calculation unit 8 can be filtered in time by means of an alpha filter 12 of a first type comprising a first alpha parameter al to form a stream of filtered autocorrelation function estimations ( ?, ).

Further, in the first operation mode OMi from the calculated stream of power estimations Pi a noise power value σ ² can be subtracted by means of subtractor 13 wherein the difference re- suit can be filtered in time by means of an alpha filter 14 of a first type comprising a first alpha parameter al to form a stream of filtered power estimations P _x as shown in fig. 4. In the shown exemplary implementation of the MIMO receiver 1 in the first operation mode OMi of the receiver 1 the Doppler frequency estimations foi are calculated by a Doppler calculation unit 15 by means of a Bessel function in response to a stream of ratio values R / P _x calculated by dividing the filtered autocorrelation function estimations R _x received from the alpha filter 12 by the filtered power estimations P received from the other alpha filter 14. In the first operation mode OMi the Doppler frequency estimations foi are calculated by the Doppler calculation unit 15 by defining on an interval from zero to a first extremum of the Bessel function the closest corresponding argument of a zero order Bessel function wherein the Bes- sel function of this found argument is close to the calculated respective ratio value R _x l P _x .

In a possible implementation the found argument can be multiplied by a constant value V. This constant value V is given in a possible implementation by r = . ¹

2π · τ wherein τ being time interval between two pilot symbols (PS) of a data block (DB), to provide the Doppler frequency estimations foi of the first operation mode OMi output by the Doppler frequency calculation unit 15 as shown in fig. 4. For example, in a FDD LTE uplink the calculated estimations of Doppler frequency fo in the first operation mode OMi can be equal to:

wherein

wherein

alpha filterl (Ri(l ·τ)) is the result of averaging of estimations of the autocorrelation function within the first operation mode OMi provided by an alpha filter of the first type and wherein y=f(x,l τ) the Bessel inverse function which calculates an argument of zero order Bessel function on an interval xe (0,3.85) corresponds to: y = arg {J ₀ (2π\ · τ · z) = x)

ze(0,3.85)

In a possible implementation the provided Doppler frequency estimations foi output by the Doppler frequency calculation unit 15 are filtered by an alpha filter 16 of a second type comprising a second alpha filter a2 to form a stream of filtered Doppler frequency estimations f _m in the first operation mode OMi of the receiver 1.

In a possible implementation of the receiver 1 as shown in fig. 4 in the second operation mode OM ₂ of the receiver 1 each autocorrelation function estimation from the set of autocorrelation function estimations R ₂ provided by the autocorrelation calculation unit 9 is filtered in time by means of an alpha filter 17 of a third type comprising a third alpha parameter a3 to form a stream of filtered autocorrelation function estimations R ₂ as shown in fig. 4. Further, in the second operation mode OM ₂ from the calculated power estimations P ₂ provided by the power calculation unit 1 1 a noise power value σ is subtracted by subtracting means 18 and the difference result is filtered in time by means of an alpha filter 19 of a third type comprising a third alpha parameter a3 to form a stream of filtered power estimations P ₂ . In a possible im- plementation of the receiver 1 as shown in fig. 4 in the second operation mode OM ₂ the set of Doppler frequency estimations for different retransmission delays is calculated by a Doppler calculation unit 20 as shown in fig. 4. The Doppler frequency calculation unit 20 calculates in the second operation mode the Doppler frequency fo ₂ by means of a Bessel function in response to a stream of sets of ratio values R ₂ 1 P ₂ calculated by dividing each filtered autocor- relation function estimation R ₂ filtered by the alpha filter 17 from the set of filtered autocorrelation function estimations by the filtered power estimations P ₂ output by the other alpha filter 19.

In the second operation mode OM ₂ the Doppler frequency estimations fo ₂ can be calculated by defining on an interval from zero to a first extremum of the Bessel function a set of the closest corresponding arguments of a zero order Bessel functions wherein the Bessel functions of these found arguments are close to the set of calculated respective ratio values R ₂ I P ₂ , wherein each found argument can be multiplied by a constant value V being: ιπ · η · τ wherein n x is the retransmission delay of a data block DB in response to a hybrid automatic repeat request HARQ to provide the Doppler frequency estimations fo ₂ of the second opera- tion mode OM ₂ output by the Doppler frequency calculation unit 20 as shown in fig. 4. In the second operation mode OM ₂ the Doppler frequency estimations are calculated by averaging of all Doppler frequency estimations from the set of Doppler frequency estimations for different retransmission delays to provide the all Doppler frequency estimations fo ₂ of the second operation mode OM ₂ The Doppler frequency estimations calculated in the second operation mode OM ₂ by the Doppler frequency calculation unit 20 can be filtered in a possible implementation by an alpha filter 21 of a fourth type comprising a fourth alpha parameter a4 to form a stream of filtered Doppler frequency estimations f _D2 of the second operation mode

OM ₂ as shown in fig. 4.

For example, for a FDD Long Term Evolution LTE uplink UL the Doppler frequency f _D calculated by the Doppler frequency calculation unit can be calculated as follows: wherein

Alphafilter3(R ₂ (n - T))

R n - r) =

Alphafilter3(P ₂ - σ ² ) ^' wherein n = 15, 16, 17 and

wherein alpha filter (R ₂ (nx)) is the result of averaging the sets of estimations of autocorrelation functions in operation mode OM ₂ in an alpha filter of the third type and J=f(x,n t) is the Bessel inverse function which calculates a set of arguments of a zero order Bessel function on an interval xe(0,3.85) which corresponds to y = arg {j ₀ (2m ^■ τ ^■ z) = x)

re(0,3.85)

Value x~3,85 corresponds to a position of a first extremum of the autocorrelation function. As can be seen in fig. 4 the current filtered Doppler frequency estimations f _m , f _m can be applied to a switching control unit 22 which can control the switching between the operation modes OMi, OM ₂ according to switching conditions SC. In a possible exemplary embodiment the switching from the first operation mode OMi to the second operation mode OM ₂ is performed according to switching conditions SC comprising as a first switching condition SCi the availability of a hybrid automatic repeat request HARQ. As a further second switching condition SC ₂ a switching from the first operation mode OMi to the second operation mode OM ₂ the switching condition can comprise that a current filtered Doppler frequency estimation f _m , f _in calculated by the MIMO receiver 1 in the current operation mode is less than a predetermined frequency threshold value fiH- wherein the predetermined frequency threshold value fxH being set in a specific exemplary embodiment for LTE FDD UL system to 70 Hz and for a LTE TDD UL system to e.g. 60 Hz. Further, the switching from a first operation mode OMi to the second operation mode OM ₂ is performed according to a third switching condition SC ₃ comprising that the value of the filtered autocorrelation function estimation R ₂ calculated in the second operation mode OM ₂ for the minimum retransmission delay of n _m j _n -x is higher than the value of the filtered autocorrelation function estimation R ₂ calculated in the second operation mode OM ₂ for the maximum retransmission delay n _max T. Further, the switching from the second operation mode OM ₂ to the first operation mode OMi can be performed according to switching conditions SC as well as under the control of the switching control unit 22. For switching from the second operation mode OM ₂ to the first operation mode OMi the switching control unit 22 uses as a fourth switching condition SC ₄ that a time interval between reception of the last hybrid automatic repeat request HARQ and a current time is less than a predetermined maximum time period M _max . In a specific implementation this maximum time period M _max can be set to 250 ms for a LTE FDD UL system and for a LTE TDD UL system. Switching from the second operation mode OM ₂ to the first operation mode can be performed according also to a fifth switching condition SC ₅ , i.e. if a current filtered Doppler frequency estimation f _m , f _D2 calculated by the MIMO receiver 1 in the current operation mode is more than a predetermined frequency threshold value fiH- Switching from the second operation mode OM ₂ to the first operation mode OMi can also be performed according to a sixth switching criterioumSC ₆ comprising that the value of the filtered auto correlation function R ₂ calculated in the second operation mode OM ₂ for the minimum retransmission delay (n _m i _{n τ} ) is lower than the value of the filtered autocorrelation function R ₂ calculated in the second operation mode OM ₂ for the maximum retransmission delay (n _max x).

Fig. 5 shows a state diagram for illustrating the functionality of a method and an apparatus for performing an estimation of a Doppler frequency fo in a wireless telecommunication system. For example the MIMO receiver 1 as shown in fig. 4 can be switched between two operation modes OMi, OM ₂ according to switching conditions SC as illustrated in fig. 5. Switching between the two operation modes OMi, OM ₂ can be performed under the control of the switching control unit 22 within the MIMO receiver 1 as shown in fig. 4. In a possible embodiment the switching conditions SC can be configured for example by means of a configuration interface. It is also possible that the switching conditions SC are preconfigured and stored in a configuration memory. The modules shown in fig. 4 comprise pre-processing modules 3-i to 7-i which can be implemented by hardware modules. In a possible embodiment the pre-processing modules 3-i to 7-i can be also implemented by software modules. The calculation modules for the two operation modes OMi, OM ₂ , i.e. the modules 8 to 21 as well as the switching control module 22 can be implemented either by hardware modules or by software modules. The estimated noise supplied to the subtracting means 13,18 can be provided by a noise power estimation unit not shown in fig. 4. The number N of reception antennas 2-i employed by the MIMO receiver 1 can vary. In a possible embodiment the MIMO receiver can comprise N = 4 reception antennas 2-i. In an alternative embodiment the MIMO receiver 1 employs for example two reception antennas 2-i. In a possible specific implementation the time length L of the delay lines 7-i are configurable and can be adapted to the respective telecommunication system. For example, for a LTE FDD uplink system the time length L of the delay lines 7-i can be set to 8 ms. Further, for example for a LTE TDD uplink system the time length L of the delay lines 7-i can be set to 10 ms. The implementation as shown in fig. 4 the MIMO receiver 1 comprises several filters 12, 14, 16 which are employed in the first operation mode OMi and filters 17, 19, 21 which are employed in the second operation mode OM ₂ . In a possible embodiment alpha filters are used. In a specific implementation the alpha parameter values of alpha filters of the first type, i.e. filters 12, 14 are set to 0.01. Further, in a specific implementation the second alpha parameter a2 the alpha filter 16 is set to 0.03. In a possible implementation the third alpha parameter a3 the alpha filters 17, 19 of the third type are set to 0.007. In a possible implementation the fourth alpha parameter a4 of the alpha filter 21 of the fourth type is set to 0.3. In a possible implementation the alpha parameter values are adaptable as well. In a possible implementation the alpha parame- ter values are stored in a configuration memory connected to the alpha filters. Further, it is possible that the alpha parameter values are adapted via the configuration interface by an external control unit.

Fig. 6 shows a flow chart of an exemplary implementation of a method for estimation of a Doppler frequency fo in a wireless telecommunication system according to the first aspect of the present invention.

It can be seen in fig. 6 that in a first step S 1 a data block DB is received, for example by the MIMO receiver 1 as shown in fig. 4. In the embodiment as shown in fig. 6 the receiver 1 ini- tially is set to the first operation mode OMi. In a further step S2 correlation values are calculated in the first operation mode OM _\ depending on the time interval τ between two pilot symbols PS, for example for τ = 0.5 ms. Alpha parameter values are set in step SI to e.g. 0.01. In a further step S3 the first switching criterion SCi is checked, i.e. the switching control unit 22 checks as in a first switching condition SCi whether a hybrid automatic repeat request HARQ has been received and is available. When a hybrid automatic repeat request HARQ has been received the switching condition SCI is fulfilled and the process continues with step S4. In step S4 the correlations are calculated for a retransmission delay time η·τ which can be for example 7 ms. The alpha parameter value of an alpha filter of the third type can be set for example to 0.007. If it is decided in step S3 that the first switching condition SCi is not met, i.e. that no HARQ has been received, it is checked in step S5 whether the receiver is in the first operation mode OMi or in the second operation mode OM ₂ . If the receiver is in the second operation mode OM ₂ a further switching criterium SC ₄ is checked in step S6. The fourth switching condition SC ₄ can be that a time interval between reception of the last hybrid automatic repeat request HARQ and a current time is less than a predetermined maximum time period M _max - In a possible specific implementation this maximum time period M _max can be set for example to 250 ms. If the fourth switching condition SC4 is not met the first operation mode OMi is set in step S7. Otherwise, the process continues with step S8 as shown in fig. 6. Having performed the calculation in step S4 the process checks in step S9 whether a second and a third switching condition SC ₂ , SC ₃ is met. The second switching condition SC2 can be in a possible implementation that a current filtered Doppler frequency estimation f _Dl , f _D2 calculated by the receiver 1 in the current operation mode OM is less than a predetermined frequency threshold value fiH- The frequency threshold value fju can be for example in a LTE FDD uplink UL system set to 70 Hz. Further, in a LTE TDD uplink UL system the frequency threshold value can be for example be set to 60 Hz. Further, it is checked in step S9 whether a third switching condition SC3 is met. In a possible implementation this third switching condition SC3 is whether the value of the filtered autocorrelation function R ₂ calculated in a second operation mode OM ₂ with a minimum retransmission delay (n _m j _n -x) is higher than the value of the filtered autocorrelation function R ₂ calculated in the second operation mode

OM ₂ for the maximum retransmission delay (n _max x). For example, for a LTE uplink n _m j _n can be 15 and n _max can be 17. If the second switching condition SC ₂ and the third switching condition SC ₃ are met the second operation mode OM ₂ is set in step S10 as shown in fig. 6. Otherwise, the operation mode OM is set to the first operation mode OMi in step SI 1. If the re- ceiver is in the first operation mode OMi the calculation of the Doppler frequency is performed in step S 12 as shown in fig. 6. Accordingly, in step S8 the calculation of the Doppler frequency fo is performed in the second operation mode OM ₂ whereas in step S 12 the calculation of the Doppler frequency fois performed in the first operation mode OMi . The calculation of the Doppler frequencies in step S8, i.e. in the second operation mode OM _2; can for example be performed by the modules 9, 1 1 , 17, 18, 19, 20, 21 in the MIMO receiver 1 as shown in fig. 4. The calculation of the Doppler frequencies fo in the first operation mode OM) in step S 12 can for example be performed by the modules 8, 10, 12, 13, 14, 15, 16 within the MIMO receiver 1 as shown in fig. 4. Accordingly, the modules 8, 10, 12, 13, 14, 15, 16 form a first calculation unit 23 for the first operation mode OMi and the modules 9, 1 1 , 17, 18, 19, 20, 21 form a calculation unit 24 for the second operation mode OM ₂ . The calculation units 23, 24 for the two different operation modes OMi, OM ₂ can be implemented by hardware modules or by software modules. In a possible embodiment step S8 is performed by the calculation unit 24 for the second operation mode OM ₂ whereas step S12 is performed by the calculation unit 23 for the first operation mode OMj. In a possible implementation the two calculation units 23, 24 for the two different operation modes OMi, OM ₂ can be formed by separate entities provided for the respective operation mode OMi or OM ₂ . In an alternative embodiment the two operation mode calculation units 23, 24 can be formed by the same entity being switched and adapted to the current operation mode OMi, OM ₂ . In this specific embodiment the first alpha parameter values al for the alpha filters 12, 14 can be switched to the third alpha parameter values a3 of the corresponding alpha filters 17, 19. Further the second alpha parameter a2 for the alpha filter 16 can be switched to the fourth alpha parameter value a4 of the alpha filter 21. Accordingly, the mode 1 autocorrelation calculation unit 8 is switched in this embodiment to form the mode 2 autocorrelation calculation 9. In the same manner the power calculation unit 10 can be switched to form the power calculation unit 1 1 of the second operation mode OM ₂ . Accordingly, in one embodiment two different calculation units 23, 24 for the different operation modes OM are provided whereas in an alternative embodiment one single calculation unit is provided for both operation modes OMj, OM ₂ being adapted accordingly.

Fig. 7 shows a circuit diagram of a possible implementation of an alpha filter which can be used with a MIMO receiver 1 as shown in fig. 4. The alpha filter as shown in fig. 7 is a digital filter receiving an input signal x(n) and as an output a filtered signal y(n). It comprises an internal output register REG. The received signal x(n) is multiplied with the respective alpha parameter value a by multiplication means and added to a feedback signal comprising the multiplication result the previous output signal y(n-l) with a multiplication factor (1-a). Accordingly, the output signal y(n) output by the alpha filter is given by: y(n)=ax(n) + (l-a)y(n-l)

The alpha filter as shown in fig. 7 can be used for example for implementing alpha filters 12, 14, 16, 17, 19, 21 as shown in the exemplary embodiment of the MIMO receiver 1 as shown in fig. 4.

Fig. 8 shows a diagram for illustrating the estimation of a Doppler frequency f _D in a wireless telecommunication system according to a first aspect of the present invention. Fig. 8 shows a dependence of a normalized autocorrelation function from a number of subcarriers Nsc- It can be seen from fig. 8 that for example for a Long Term Evolution LTE wireless telecommunication system where the number of subcarriers Nsc is for example chosen to be 6 no considerable change of the propagation channel takes place, i.e. the performance losses are less than 0.3 decibel. Fig. 9 shows a dependence of a average output 1 from a Doppler frequency in the propagation channel. Curve I shows the output signal to noise ration SNR = 3 decibel. Curve II shows the output value for a signal to noise ratio SNR = 10 decibel and curve III shows the output for a signal to noise ratio SNR = 30 decibel. Curve IV shows the ideal output. In the first operation mode OM] the estimation shows low performance in case of a low Doppler frequency and a low signal to noise ratio SNR. In this situation the method an apparatus according to the present invention switches to the second operation mode OM ₂ to increase its performance.

Fig. 10 shows autocorrelation functions for different Doppler frequencies fo and different time delays ranging from 0 - 9 ms. It can be seen that several circles of autocorrelation functions can fit in an interval from 0 - 7 ms delay in case of high Doppler frequencies, i.e. a Doppler frequency of more than 70 Hz. The average autocorrelation function of fading samples can theoretically be described by a Bessel function of zero order, however, practically this often results in that the autocorrelation function considerably differs from the theoretical autocorrelation function. This depends on the environment around the base station and the type of channel as well as on the concrete geographical locations of buildings around the base stations or receiver. Accordingly, only a main part of autocorrelation function of a Doppler frequency estimation, namely J ₀ (x), xe (0,3.85) can be taken into account, wherein x=3.85 corresponds to the position of the first extremum of the function as shown in fig. 1 1. Determination of an exact value of a border frequency, i.e. a predetermined threshold frequency value frH at which the method and apparatus of the invention switches its operation mode OM can be derived from fig. 10. For example, for an LTE FDD uplink mode up to a frequency of 70 Hz the autocorrelation function has only one meaning, i.e. if the Doppler frequency fo is further increased then the autocorrelation function has several meanings, i.e. is ambiguous. For example, for a frequency of 150 Hz the autocorrelation function crosses the autocorrelation value level 0.25 three times, i.e. at about 2 ms, 7 ms and 8 ms. To avoid multiple meanings of the autocorrelation function additional delays for autocorrelation function can be used for frequencies below the predetermined threshold frequency values fjH of e.g. 70 Hz in case of a FDD LTE uplink and below e.g. 60 or 50 Hz for a TDD LTE uplink.

Fig. 12a - 12h show Doppler frequency estimations and noise power estimations for different environments and in particular different signal to noise ratios SNRs for the method according to operation mode OMj of the present invention (A) in comparison to a conventional Zero Crossing Rate ZCR algorithm (Z) or a conventional Level Crossing Rate algorithm with hysteresis (H). As can be seen from fig. 12a - 12h the method according to the first aspect even of the operation mode OMi of present invention can outperform the conventional Zero Crossing Rate method (Z) or Level Crossing Rate method with hysteresis (H). Fig. 13a - 13f show different snap shots of a Doppler frequency estimation for different parameters.

Fig. 13a shows a Doppler estimation for a Doppler frequency 10 Hz, a modulation 64 QAM, 10 resource blocks and a signal to noise ratio SNR= - 5 decibel and a Repetition Code Rate = 1/33.75. Fig. 13a shows a Doppler estimation for the first operation mode OMi as well as the second operation mode OM ₂ wherein for this operation mode the Doppler frequency estimation is shown both for dynamic scheduling DS and not dynamic (static) scheduling SS. Fig. 13b shows Doppler estimations of a time for a Doppler frequency fo = 30 Hz with a signal to noise ratio of SNR=-15 decibel for the same modulation resource blocks and Repetition Code Rate as in fig. 13a.

Fig. 13c shows a Doppler estimation for a Doppler frequency Hz with a signal to noise ratio of SNR=-5 decibel for the same modulation resource blocks and Repetition Code Rate as in fig. 13a, 13b.

Further, Fig. 13d shows a Doppler estimation of a time for a Doppler frequency ίπ=100 Hz for the same modulation resource blocks and Repetition Code Rate as in fig. 13a - 13c.

Fig. 13e shows a Doppler estimation of a time for a Doppler frequency fo=l 50 Hz for the same modulation resource blocks and Repetition Code Rate as in fig. 13a - 13d.

Fig. 13f shows a Doppler estimation for a Doppler frequency fo=10 Hz for a signal to noise ratio of SNR=-4 decibel or 10 resource blocks and a Repetition Code Rate of 1/33.75.

Fig. 13 a-f show also effectiveness of the switching conditions. As one can see in Fig. 13d, (Doppler frequency 100 Hz), there is no difference between the two operation modes of algorithms - OMI and OM2; even sometimes the output of the algorithm goes to values below f _T H (e.g. 70 Hz in case of a FDD LTE uplink), switching condition SC3 prohibits a wrong switching from operation mode OMi to operation mode OM ₂ .

Fig. 13 a-f show effectiveness of the algorithm in case of different values of SNR. When SNR is high, then operation mode OMi of algorithm provides good performance, and when SNR is low, that HARQ's are rising, the algorithm does switch to the second operation mode OM ₂ which can improve performance of Doppler estimation. The method according to the first aspect of the present invention for estimation of a Doppler frequency fb can be used for different wireless telecommunication systems, in particular for a Long Term Evolution LTE wireless telecommunication system in particular for a Long Term Evolution LTE FDD or TDD system. The method and apparatus for estimation of a Doppler frequency fb in a wireless telecommunication system can be used for other wireless telecommunication systems using a hybrid automatic repeat request HARQ as well.

The Doppler frequency calculations in a propagation channel as shown in fig. 9 indicate that in the first operation mode OMi the method provides a low performance in case of a low speed of the user and low signal to noise ratio SNR. This happens because for a low speed of user changes of 0.5 ms the autocorrelation function values are very small as shown in table 1. T =lms τ = 2ms .· T = 3ms τ = 4ms

3 7.2 0.9999 0.9995 0.9979 0.9954 0.9918

15 36.1 0.9968 0.9872 0.9492 0.8875 0.8044

30 72.2 0.9872 0.9492 0.8044 0.5877 0.3312

60 144.4 0.9492 0.8044 0.3312 -0.1524 -0.3945

120 288.9 0.8044 0.3312 -0.3945 -0.0256 0.2913

350 842.6 -0.1 187 -0.0778 -0.2288 -0.1631 0.0061

For example, when the speed of user changes is by 10 times, e.g. from 3 km/h - 30 km/h the autocorrelation value for 0.5 ms changes only by 1.3 %. In case of a noised autocorrelation value such a small change of the autocorrelation function value prohibits to have a high quality estimation value of the Doppler function. To overcome this drawback the estimation results of operation mode OMi the method and apparatus switches to the second operation mode OM ₂ which uses the HARQ samples for estimation of the autocorrelation function as addi- tional delays. The method and apparatus for estimation of a Doppler frequency for a wireless telecommunication system can be implemented either by hardware or software modules. The method can be performed by executing a software program executing the software modules. The method according to the present invention comprises in the shown embodiments two operation modes OMi, OM ₂ and in particular a first operation mode OMi, for example for a high Doppler frequency, and another operation mode OM ₂ for a low Doppler frequency. Switching between operation modes OM is based on the availability of the hybrid automatic repeat request HARQ of the current Doppler frequency estimation as well as a lower boundary of possible Doppler frequency estimation. In alternative embodiments more than two operation modes OM can be provided and the switching between the operation modes is performed ac- cording to switching conditions SC. In a possible embodiment the method is employed by a wireless telecommunication system using OFDM symbols. The method and apparatus for estimation of a Doppler frequency in a wireless telecommunication system is not restricted to a system using OFDM symbols but can be used for any multicarrier system employing pilot symbols PS. Accordingly, the time interval τ between pilot symbols PS can vary in different embodiments. Further, the retransmission delay being higher by a predetermined factor n than the time interval τ between the two pilot symbols PS can also vary depending on the factor n and can be adapted to the respective wireless telecommunication system. The method for estimation of a Doppler frequency in a wireless telecommunication system can be implemented in any receiver or transceiver provided for example in a base station BS but also within an- other node of a wireless telecommunication system, in particular also in a transceiver of a mobile station MS. According to a further aspect of the present invention a receiver for a wireless telecommunication system is provided for performing the method for estimation of a Doppler frequency according to the first aspect of the present invention. This receiver can be a MIMO receiver as shown in fig. 4. Further, it is also possible that the receiver does com- prise another number of reception antennas 2-i. A switching control between the operation modes OMi, OM ₂ can be performed under the control of an integrated switching control unit 22 as shown in fig. 4. Further, it is also possible that the switching between the two operation modes OMi and OM ₂ is performed in response to a control signal applied to the receiver from an external control unit. The different parameters for the alpha filters, for the calculation modules, for the autocorrelation functions and the Doppler frequencies can be preconfigured, but it is also possible that they are adapted during reception of the data blocks. Moreover, it is possible that the modules in the receiver can be adapted or configured for different wireless telecommunication systems with different requirements for data transmission. The adaption and configuration can be performed in response to a configuration set up and applied to the receiver from an external control unit. The calculation of the Doppler frequency estimations can also be calculated by other functions than a Bessel function. This function can also be adapted and configured according to a configuration signal or can be preconfigured in a confirmation memory of the receiver. Instead of alpha filters Kalman filters or some other type of filters can be used in a possible implementation.