WO2006106474A2

WO2006106474A2 - Method and apparatus for estimating channel in mobile communication system

Info

Publication number: WO2006106474A2
Application number: PCT/IB2006/051010
Authority: WO
Inventors: Yanmeng Sun
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2005-04-08
Filing date: 2006-04-04
Publication date: 2006-10-12
Also published as: WO2006106474A3

Abstract

The present invention provides a channel estimation method and apparatus for use in a mobile communication system receiver. The method and apparatus uses a transmission signal with special structure as provided in the present invention, to estimate channel parameters. The transmission signal according to the present invention comprises a burst. The burst comprises multiple training sequence segments and multiple data segments arranged alternately. Compared with conventional burst structure that only includes a single training sequence segment, this novel burst structure can fully reflect the characteristics of channel change experienced during the communication procedure of the whole burst. According to the method and apparatus for estimating channel parameters in the present invention, flexible combination processing can be performed on the multiple training sequence segments in the received burst according to the channel change mode caused by the receiver's mobility, to obtain the channel parameters for satisfying different CCT requirements and restoring data correctly. Therefore, it overcomes the limitation on the length of the burst caused by the CCT of the channel fading change caused by the mobility of the receiver and thus improves the overall performance of the communication system effectively.

Description

METHOD AND APPARATUS FOR ESTIMATING CHANNEL IN MOBILE

COMMUNICATION SYSTEM

FIELD OF THE INVENTION The present invention relates generally to mobile communication systems, and more particularly, to a burst for use in mobile communication systems and a channel estimation method and apparatus using the training sequences in the burst.

BACKGROUND OF THE INVENTION In mobile communication systems, multiple users usually share common radio resources by employing various multiple access techniques, such as FDMA (Frequency Division Multiple Access), TDMA (Time Division Multiple Access) or CDMA (Code Division Multiple Access) or their combinations, to communicate with each other. To transmit user data effectively and facilitate data reception at the receiving side, frame structure is used typically in the system to divide the time domain and further divide the frame or sub-frame into multiple timeslots according to the requirements for services and transmission performance. For example, TD-SCDMA (Time Duplex-Synchronous Code Division Access) in 3G mobile communication system, adopts frame structure, wherein each sub-frame comprises 7 timeslots and different users can transmit data simultaneously in a certain timeslot with different channel codes in bursts mode.

On the other hand, when propagating along wireless channels, signal fading and distortion usually occur due to the propagation distance, barrier obstruction and the relative motion between the receiver and the transmitter, which cause the data received at the receiving side is different from that transmitted from the transmitting side. To restore data correctly at the receiving side, known training sequences are usually inserted in the data segments to be transmitted, such that the channel parameters can be estimated for data restoration by processing the desired and distorted training sequences. Fig.l illustrates a burst structure in TD-SCDMA system, which includes data segment 1, data segment 2 and training sequence 3. Data segments 1 and 2 carry user data and training sequence 3 is a known pseudo random code sequence mainly for estimating channel parameters. For ease of illustration, the control segments or GP (Guard Period) possibly inserted in two ends or at other positions in the burst are omitted herein. The premise for estimating channel parameters by using the training sequences in the burst so as to restore data therein is that the channel fading undergone by the training sequences for channel estimation and that by the user data segment for data restoration by using the channel estimation results in the same burst should have a close correlation, i.e. the span between them in time domain should be confined strictly in the CCT (Channel

Coherent Time). Here, the CCT refers to the maximum time interval in which channel parameters can maintain correlation. When the receiver is in low mobility, the channel parameters change slowly and the CCT is long; conversely, when the receiver is in high mobility, the channel parameters change fast and the CCT is short. When the moving speed of the receiver exceeds a certain value, it's very likely that the channel parameters estimated by using the training sequences can't represent the channel characteristics experienced by the whole burst during transmission procedure, which thus will affect the accuracy and effectiveness of data restoration unavoidably.

With the development of mobile communication technologies, higher requirements for the speed supported by mobile communication systems and the quality of the received signals are desired. The system design and the equipment specification of the 3G terrestrial cellular mobile communication systems is required to support data transmission when the mobile terminal moves at the speeds of 120 km/h (UTRAN TDD mode) and 500 km/h (UTRAN FDD mode). Fig.2 illustrates the channel fading characteristics when the receiver moves at different speeds. Fig. 2(a) illustrates the channel fading scenario when the receiver moves at a speed of 120 km/h and Fig. 2(b) illustrates the channel fading scenario when the receiver moves at a speed of 500 km/h. It can be observed from the figures that the fluctuation of channel fading caused by the high mobility of the receiver is much more rapid and severe than that by the low mobility of the receiver; and the CCT when the receiver moves at a high speed is far less than that when the receiver moves at a low speed.

That is, the moving speed of the mobile terminal and its dynamic change put higher requirements on the structure design of the burst.

Theoretically, the drawbacks of rapid change in signal fading and short CCT when the receiver is in high mobility can be overcome by shortening the length of the burst, so as to obtain the channel parameters for restoring the data in the same burst effectively by performing channel estimation using the training sequences in the burst. However, a too short burst may bring many negative effects to the whole communication system. For example, a too short burst will decrease the effective load on the physical channels and increase the complexity of modulation and demodulation and the difficulty of system implementation. Furthermore, when the receiver is in relatively low mobility, a too short burst will cause waste of radio resources and restrict the highest transmission rate of the user's data communication. It has become a bottleneck for further optimization and development of mobile communication systems as how to design the burst structure so that the receiver can transmit data effectively and restore data correctly within a larger range of dynamic change in mobility.

It is, therefore, necessary to provide a more effective burst and a method and apparatus for estimating the channel parameters by using the training sequences in the bursts, so as to overcome the above problems.

SUMMARY OF THE INVENTION

An object of the present invention is to provide a method and apparatus for estimating channel parameters for use in a wireless communication system receiver, to adapt to the change in channel coherent characteristics caused by change in the mobility of the receiver and improve the efficiency for estimating channel parameters.

A channel estimation method for use in a mobile communication system receiver according to the present invention, comprises the steps of: processing a burst extracted from received signals, to separate a plurality of data segments and a plurality of training sequence segments; processing the plurality of separated training sequence segments, to determine a channel fading mode caused by the receiver's mobility; and processing the plurality of separated training sequence segments according to the determined channel fading mode, to obtain channel parameters required for restoring the plurality of separated data segments.

A channel estimation apparatus for use in a mobile communication system according to the present invention, comprises: a signal separating means, for processing a burst extracted from the received signals, to separate a plurality of data segments and a plurality of training sequence segments; a mode determining means, for processing the plurality of separated training sequence segments, to determine a channel fading mode caused by the receiver's mobility; and a parameter processing means, for processing the plurality of separated training sequence segments according to the determined channel fading mode, to obtain channel parameters required for restoring the plurality of separated data segments.

A transmission signal according to the present invention, is generated by modulating a burst onto a radio carrier, the burst comprising multiple data segments and multiple training sequence segments, the multiple training sequence segments being generated by segmenting a compound training sequence and inserted at different positions in the transmission data so that the multiple training sequence segments and multiple data segments are arranged in order, thus to optimally reflect the channel fading change caused by the receiver's mobility. A transmitting apparatus transmits the transmission signal.

As described above, the transmission signal for use in mobile communication systems as provided in the present invention includes a burst. The burst includes multiple training sequence segments and multiple data segments, and the multiple training sequence segments and the multiple data segments are arranged alternately. Compared with conventional burst structure only including a single training sequence segment, this novel burst structure can fully reflect the characteristics of channel change experienced during the communication procedure of the whole burst, so that the receiving side can perform flexible combination processing on the multiple training sequence segments in the received burst according to the channel change mode caused by the receiver's mobility, to obtain the channel parameters for satisfying different CCT requirements and restoring data correctly.

It overcomes the limitation on the length of the burst caused by the CCT of the channel fading change caused by the mobility of the receiver and thus improves the overall performance of the communication system effectively.

Other objects and attainments together with a fuller understanding of the invention will become apparent and appreciated by referring to the following descriptions and claims taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 illustrates a burst in TD-SCDMA mobile communication system;

Fig. 2 illustrates the channel fading characteristics when the receiver is in different mobility; wherein Fig. 2(a) illustrates the channel fading scenario when the receiver moves at a speed of 120 km/h and Fig. 2(b) illustrates the channel fading scenario when the receiver moves at a speed of 500 km/h;

Fig.3 shows a burst according to an embodiment of the present invention;

Fig.4 shows the construction of multiple training sequence segments in a burst according to an embodiment of the present invention;

Fig.5 is the flowchart showing the method for estimating channel parameters according to an embodiment of the present invention;

Fig.6 shows the combination of channel parameters according to an embodiment of the present invention; Fig.7 shows the apparatus for estimating channel parameters according to an embodiment of the present invention;

Fig.8 illustrates the channel parameter processing means in the scenario of fast channel fading in the apparatus for estimating channel parameters according to an embodiment of the present invention; Fig.9 illustrates the channel parameter processing means in the scenario of slow channel fading in the apparatus for estimating channel parameters according to an embodiment of the present invention;

Throughout all the above drawings, like reference numerals will be understood to refer to like, similar or corresponding features or functions.

DETAILED DESCRIPTION OF THE INVENTION

One of the main features of the method and apparatus for estimating channel parameters for use in mobile communication system receivers as provided in the present invention is that the method and apparatus can choose different combination of multiple training sequence segments in a burst according to the channel change mode related to the receiver's mobility and process the combination to obtain channel parameters satisfying the requirements on the CCT from different channel change modes and required for restoring data correctly, thus to improve the overall performance of the receiver effectively. Another main feature of the method and apparatus for estimating channel parameters for use in mobile communication system receivers as provided in the present invention is that the method and apparatus uses a transmission signal with a special structure as provided in the present invention to estimate channel parameters. The transmission signal includes a burst that includes multiple training sequence segments inserted at different positions in the transmission data. Compared with conventional burst structure only including a single training sequence segment, this burst structure can fully reflect the characteristics of channel change experienced during the communication procedure of the whole burst, and the estimation performed based on the segmented short training sequences and their combinations may satisfy the CCT requirements in different mobility of the receiver, thus to overcome the limitation on the burst length from the CCT in conventional burst structure.

Detailed descriptions will be given below to the burst provided in the present invention and the method and apparatus for estimating channel parameters by using multiple training sequences in the burst.

Fig.3 shows a burst according to an embodiment of the present invention. For ease of illustration, GP (Guard Period) segments, which are likely to exist in the burst, are omitted here. It can be seen from Fig.3 that the burst includes multiple data segments and multiple training sequence segments, wherein the multiple training sequence are generated by segmenting a compound training sequence and are inserted at different positions in the transmission data in order, so as to be arranged alternately with the multiple data segments. Fig.4 shows the construction for multiple training sequence segments in a burst according to an embodiment of the present invention. As shown in the figure, the compound training sequence is obtained by performing bitwise operation on training sequence A (with length M) that is repeated for K times and training sequence B (with length N), wherein training sequence A and training sequence B are not incoherent with each other and their lengths satisfy the requirement N - K - M . Specifically, when the values of the training sequences are {+1, -1 }, the above bitwise operation is scalar multiplication; when the values of the training sequences are logical value of {0, 1 }, the above bitwise operation is logical XOR operation. Multiple training sequence segments can be obtained by segmenting the compound training sequence, and the multiple segmented training sequence segments may have same length or different lengths, but the lengths should be integer times of the length M of training sequence A. The burst shown in Fig.3 can be obtained by inserting the multiple segmented training sequences at different positions in the transmission data. Descriptions will be given to the construction of the burst shown in Fig.4, in conjunction with mathematical equations.

First, choose and generate two incoherent training sequences A and B with lengths M and N respectively, and M and N satisfy:

N = KM (1)

Typically, it can be assumed that the value of each chip is {-1, +1} and each sequence is defined as follows:

c^m = [C^, c{«,-, C^₁] (3)

An extended training sequence with length equal to the long pseudo random sequence B can be obtained by repeating the short pseudo random sequence A for K times.

Then said compound training sequence can be obtained by performing scalar multiplication operation on the extended training sequence and the long pseudo random sequence B:

m = [m_o,m_ι,---,m_κ__ι] (4) where m₍ is: _m -V₀W iB) (A) (B) (A) (B) (A) (B) ₁ ,^

m =\c^(A) -c^(B) c^(A)-c^(B) c^(A)-c^{(B) ■■■} c^(A) -c^(B) 1 (6)

_m - r_c(A) (B) _C(A) (B) _CW . ΛB) ... (A) (B) -, _π)

¹¹¹K-I L^L0 ¹^(K-I)M '^Ll ^L(A:-1)M+1'¹'2 ^L(A:-1)M+2' '^LM-1 ^KM-Ii V'/

Multiple training sequence segments (for example, L training sequence segments) can be obtained by segmenting the compound training sequence m, and the L training sequence segments all have the length integer times of M, that is, each of the segmented training sequences is the combination of a sub-sequence m₍ or its adjacent sub-sequences, for example, M₀=m₀ and M₁=[Tn_1^m₂) are among the multiple segmented training sequences respectively.

From random signal theory it can be seen that the compound training sequence m is still a pseudo training sequence if the above training sequences A and B are pseudo random sequences and has good self-correlation characteristics and can satisfy the basic requirements of training sequences for estimating channel parameters.

It's to be noted that selection of parameters M, N and K in equation (1) should consider the performance requirement for estimating channel parameters in different channel change modes. First, selection of the length M of the short pseudo random sequence A should consider the requirement for channel correlation performance in high mobility of the receiver, that is, the interval between two adjacent training sequence segments in time domain (including the time span occupied by the two training sequences themselves) in the burst should be no large than the CCT when the receiver is in high mobility, such that the channels experienced by adjacent training sequence segments are correlated closely. Second, selection of the repetition times K of the short pseudo random sequence A should be restricted within a reasonable range such that the time span of the multiple training sequences in said burst should be no large than the CCT in low mobility scenario of the receiver. For example, the CCT is T₁ when the receiver is in high mobility and the CCT is AT₁ when the receiver is in low mobility, then K may be 4. More specifically, selection of K should be suitable for the depth of the receiver buffer and the design complexity of the correlator.

Fig.5 is the flowchart showing the method for estimating channel parameters by using multiple training sequences in the above burst according to an embodiment of the present invention. Fig.6 illustrates the combination of channel parameters according to an embodiment of the present invention. Detailed descriptions will be given below to the method for estimating channel parameters and its specific steps in conjunction with Fig.5 and Fig.6.

First, extract a burst from the received signals and process the burst to separate multiple data segments and multiple training sequence segments (step S20). Then, process the multiple separated training sequence segments, to determine the channel fading mode caused by the mobility of the receiver (step S30). Finally, process the multiple separated training sequence segments according to the determined channel fading mode, to obtain the channel parameters required for restoring the separated data segments (step S40). Detailed descriptions will be given below to the steps in the method shown in Fig.5 in conjunction with mathematical expressions.

Generally, multiple bursts can be multiplexed properly to transmit multiple users data in a same timeslot. Since the frame structure, burst structure in the system and the multiplex method of the system are known at the receiving side, the receiving side can easily extract a burst from the received signals.

The acquired burst is processed according to known burst structure, and the multiple training sequence segments and multiple data segments can be separated and cached respectively. Suppose that the multiple separated training sequences can be represented as M₁ (?),/ = 0,"-,L-I , wherein M _t {t) corresponds to multiple desired training sequences

M₁ and the symbol '-' represents the distortion caused by passing the propagation channel.

It can be seen from random signal theory that the channel fading change mode is closely related to the mobility of the receiver. Generally, the higher the mobility of the receiver is, the shorter the CCT will be and the faster the channel fading change will be.

Thus, the channel fading change mode can be determined by estimating the mobility of the receiver or computing the CCT, or using information about the mobility of the receiver and the CCT. The determination procedure for the channel fade mode will be explained below, taking the calculation of the CCT as an example. First, calculate the channel coherent coefficients of a pair or multiple pairs of predefined training sequence segments in the multiple separated training sequence segments in the burst. Then, compare the calculated channel coherent coefficients with the predefined channel coherent threshold, to determine the channel fading change mode, wherein the predefined channel coherent threshold is one or more predefined values for the statistical characteristics of the CCT change caused by the mobility of the signal receiver.

Generally, the calculation can be made by selecting any two training sequence segments from the acquired multiple training sequence segments, and the calculated coherent coefficients indicate the correlation between the channel fading experienced by the two selected training sequence segments. The channel coherent coefficients can be calculated from the following equation, taking the calculation of the first training sequence segment and the last one in said multiple training sequence segments: p = -^f_o ^Λ ^τM_o (t) - M_L__ι(t)dt (8)

where p represents channel coherent coefficients, M₀(t) represents the first training sequence segment in the acquired multiple training sequence segments, M_L__L(/) represents the L^th ,i.e. the last training sequence segment, in the acquired multiple training sequence segments, T_A represents the width of the acquired training sequence segment in time domain, i.e. the time interval in the burst and E_s represents the energy of a sequence symbol in the embodiment. The channel coherent coefficients of any two training sequences can be computed by selecting other training sequences with different sub- script numbers as in equation (8). The channel fading change mode can be determined by comparing the obtained channel coherent coefficients and the predefined channel coherent threshold.

The following description is given to the method for estimating the corresponding parameters with different channel fading modes in conjunction with Fig.5 and Fig.6, taking fast channel fading and slow channel fading as examples.

In fast channel fading mode, the receiver is usually in high mobility and the CCT is shorter accordingly. And at this time, the receiving side has to estimate channel parameters by using multiple training sequences in the burst respectively, to obtain the channel parameters for restoring the user data of their adjacent data segments. It can be seen from Fig.5 that the step S40A of estimating channel parameters further comprises the sub-steps of: performing correlation processing on a known short pseudo random sequence (training sequence A) and multiple separated training sequence segments { M _t {t),i = O,---,L-1 } respectively, to obtain multiple initial channel parameters (step S402); and determining the combination range of the initial channel parameters according to the obtained channel coherent coefficients (step S406), wherein the basic concept of the combination range of channel parameters is shown in Fig.6 and will be described later in detail. In a simple word, every combination range includes at least one training sequence segment and at least one data segment separated from the burst, and the training sequence segment and the data segment in the combination range should have good channel coherence. Numerical processing is performed respectively on the initial channel parameters in each combination range, to obtain the channel parameters for restoring data in each data segment within the corresponding combination range (step S408).

The method for estimating initial channel parameters at the above step S402 is similar to the conventional match filter method, and is available from correlation processing, which is well known to those skilled in the art and thus is omitted herein. The characteristics of CCT change caused by the mobility of the receiver must be taken into consideration in determining the combination range of the channel parameters at the above step S404.

It can be seen from the previous discussion to the selection of parameters M, N and K in equation (1) that the span in time domain of two adjacent training sequence segments in the burst (including the time span occupied by the two training sequence segments themselves) should be less than the CCT in high mobility scenario of the receiver, that is, the adjacent training sequence segments have better channel coherence. Since the determination of the above parameters usually refers to the desired highest moving speed of the receiver, when the actual moving speed of the receiver is lower than the designed reference value of the desired speed, it's very likely that several training sequence segments with consecutive numbers experience highly correlated channels in multiple training sequence segments in the received burst. The channel parameters that reflect the actual channel changes more accurately can be obtained, by performing further numerical processing on different combinations of the initial channel parameters, obtained at step S402, corresponding to several training sequence segments with consecutive numbers. In the description to the embodiment, combination of the several determined training sequence segments with consecutive numbers and their adjacent data segments is termed as combination range of channel parameters. As shown in Fig.6, the training sequence 0

( M₀ ) and the training sequence 1 ( M ₁ ) have good coherence and the channel parameters obtained from M ₀ and M₁ can be used to restore the user data in data segment 0 and data segment 1, thus the corresponding channel parameter combination range P-I includes training sequences M₀ and M₁ and data segments 0 and 1.

Numerical processing of channel parameters (step S408) is to perform combination computation on the initial parameters in each combination range of channel parameters. The numerical processing may be simple averaging computation on the initial channel parameters in the combination range, or complicated numerical approximation processing.

The several results obtained from the numerical processing can be used as channel parameters for restoring the data segments in each combination ranges of channel parameters.

It can be seen from Fig.5 that the receiver is usually in low mobility under slow channel fading mode and the CCT is usually longer. At this time, the receiving side can estimate the channel parameters by using the compound sequence generated by concatenating multiple training sequences in the burst, to obtain the channel parameters for restoring all user data in the burst. It can be seen from Fig.5 that the step S40B of estimating channel parameters further comprises the sub-steps of: concatenating multiple training sequence segments separated from the burst into a combined sequence

M = JM₀₅M₁,-- -,M_1-1 J corresponding to the desired compound training sequence (step

S412); and performing correlation processing on the combined sequence with a known compound training sequence (such as the compound training sequence m =

described in equation 4), to obtain estimation results of channel parameters, for restoring the data in all data segments separated from the burst. Wherein, the method for estimating channel parameters is similar to the conventional match filter method, and is available from correlation processing, which is well known to those skilled in the art and thus is omitted herein.

The above method for estimating channel parameters for use in a mobile communication system by using the training sequences in the burst taken in conjunction with Fig.5 can be implemented in software or hardware, or in combination of both. The corresponding apparatus for implementing the above method is shown in Fig.7. In the following, a detailed description will be given to the apparatus for estimating channel parameters in the present invention, in conjunction with Fig.7.

As shown in Fig.7, the apparatus for estimating channel parameters for use in a mobile communication system according to an embodiment of the present invention mainly comprises a signal separating means 200, a mode determining means 300, a parameter processing means 400 and two switching means 302 and 306. First, the signal separating means 200 extracts a burst from the received signals and processes the burst, to separate multiple data segments and multiple training sequence segments. Then, the mode determining means 300 performs a calculation similar to equation (8) on the multiple separated training sequence segments, and compares the calculated channel coherence coefficients and the predefined channel coherence threshold, to judge whether the current channel fading characteristic is fast channel fading mode or slow channel fading mode, last, the parameter processing means 400 processes the multiple separated training sequence segments according to the determined channel fading mode, to obtain the channel parameters required for restoring the multiple data segments separated from the burst. Wherein, the channel parameter processing means 400 comprises two relatively independent means: a fast channel fading processing means 420 and a slow channel fading processing means 460, respectively for estimating parameters in fast channel fading mode and for estimating parameters in slow channel fading mode.

There are two independent switching means between the channel mode determining means 300 and the channel parameter processing means 400, wherein the first switching means 302 is operable to switch the channel parameter processing means to fast channel fading processing means 420 or slow channel fading processing means 460 according to the determined channel fading mode, and the second switching means 306 is operable to output the channel parameters for restoring data and the corresponding combination ranges of the channel parameters selectively, according to the determined channel fading mode and channel parameter processing results.

Fig.8 shows a fast channel fading processing means 420 and Fig.9 shows a slow channel fading processing means 460. Detailed descriptions will be given below to the two means in conjunction with Fig.8 and Fig.9.

It can be seen from Fig.8 that the fast channel fading processing means 420 further comprises a correlation processing means 424, for performing correlation processing on a known short pseudo random sequence and the multiple separated training sequence segments respectively, to obtain multiple initial channel parameters; a range determining means 426, for determining the combination ranges of the initial channel parameters according to the channel coherence coefficients; and a numerical processing means 428, for performing numerical processing on the initial channel parameters in each combination range, to obtain the channel parameters for restoring the data in each data segment in the corresponding combination range. The numerical processing may be a simple averaging calculation, or a complicated numerical approximation processing and is well known to those skilled in the art and thus the description is omitted herein.

It can be seen from Fig.9 that the slow channel fading processing means 460 comprises a combining means 464, for concatenating the multiple separated training sequence segments into a combined sequence M = JM₀, M₁,-- -,M_1-1 J corresponding to the desired compound training sequence, and a correlation processing means, for performing correlation processing on the combined sequence with a known compound training sequence m =

to obtain the channel parameters required for restoring the user data in all data segments in the burst.

It is to be understood by those skilled in that art that the burst for use in mobile communication systems and the method and apparatus for estimating channel parameters by using the training sequences in the burst as disclosed in the present invention may be applied not only to mobile communication systems, but also to wireless LAN communication systems and various communication systems communicating with the burst where there is relative motion between the transmitter and the receiver.

It is to be understood by those skilled in the art that various improvements and modifications can be made to the burst for use in mobile communication systems and the method and apparatus for estimating channel parameters by using the training sequences in the burst as disclosed in the present invention without departing from the basis of the present invention, the scope of which is to be defined by the attached claims herein.

Claims

CLAIMS:

1. A channel estimation method for using in a mobile communication system receiver, comprising the steps of: (a) processing a burst extracted from received signals, to separate a plurality of data segments and a plurality of training sequence segments;

(b) processing said plurality of separated training sequence segments, to determine a channel fading mode caused by said receiver's mobility; and

(c) processing said plurality of separated training sequence segments according to the determined channel fading mode, to obtain channel parameters required for restoring said plurality of separated data segments.

2. The method according to claim 1, wherein said extracted burst comprises a desired burst undergoing wireless channel fading, said desired burst comprising a plurality of data segments and a plurality of training sequence segments, said plurality of training sequence segments being generated by segmenting a compound training sequence.

3. The method according to claim 2, wherein said compound training sequence is a pseudo random sequence computed from two incoherent pseudo random sequences, where in said two incoherent pseudo random sequences, the length N of a long pseudo random sequence is K times of the length M of the other short pseudo random sequence and K is a natural number more than 1.

4. The method according to claim 3, wherein said compound training sequence is derived by repeating said short pseudo random sequence for K times and then performing bitwise scalar multiplication algebraic operation or XOR logical operation on the repeated pseudo random sequence and said long pseudo random sequence.

5. The method according to claim 4, wherein the lengths of said plurality of training sequence segments are integer times of the length M of said short pseudo random sequence.

6. The method according to claim 1, wherein step (b) comprises: performing coherence operation on at least two predefined training sequence segments in said plurality of separated training sequence segments, to obtain a corresponding channel coherence coefficient; and comparing said channel coherence coefficient with at least one predefined channel coherence threshold, to determine a corresponding channel fading mode.

7. The method according to claim 6, wherein, if said channel fading mode is fast fading mode, step (c) comprises the steps of: performing correlation processing on a known short pseudo random sequence and said plurality of separated training sequence segments respectively, to obtain a plurality of initial channel parameters; determining combination ranges of said plurality of initial channel parameters based on channel coherence characteristics according to said channel coherence coefficient, each of said combination range comprising at least one of said separated training sequence segments and at least one of said separated data segments; and processing the initial channel parameters in each of said combination ranges respectively, to obtain the channel parameters for restoring data in each data segment within the corresponding combination range.

8. The method according to claim 6, wherein if said channel fading mode is slow channel fading mode, step (c) comprises: concatenating said plurality of separated training sequence segments into a combined sequence corresponding to a known compound training sequence; and performing correlation processing on said known compound training sequence and said combined sequence, to obtain the channel parameters for restoring the data in said separated data segments.

9. A channel estimation apparatus for using in a mobile communication system, comprising: a separating means, for processing a burst extracted from received signals, to separate a plurality of data segments and a plurality of training sequence segments; a determining means, for processing said plurality of separated training sequence segments, to determine a channel fading mode caused by said receiver's mobility; and a processing means, for processing said plurality of separated training sequence segments according to the determined channel fading mode, to obtain channel parameters required for restoring said plurality of separated data segments.

10. The apparatus according to claim 9, wherein said extracted burst comprises a desired burst undergoing wireless channel fading, said desired burst comprising a plurality of data segments and a plurality of training sequence segments, said plurality of training sequence segments being generated by segmenting a compound training sequence.

11. The apparatus according to claim 10, wherein said compound training sequence is a pseudo random sequence computed from two incoherent pseudo random sequences, where in said two incoherent pseudo random sequences, the length N of a long pseudo random sequence is K times of the length M of the other short pseudo random sequence and K is a natural number more than 1.

12. The apparatus according to claim 11, said compound training sequence is derived by repeating said short pseudo random sequence for K times and then performing bitwise scalar multiplication algebraic operation or XOR logical operation on the repeated pseudo random sequence and said long pseudo random sequence.

13. The apparatus according to claim 12, wherein the lengths of said plurality of training sequence segments are integer times of the length M of said short pseudo random sequence.

14. The apparatus according to claim 9, wherein said mode determining means comprises: a computing means, for performing coherence operation on at least two predefined training sequence segments in said plurality of separated training sequence segments, to obtain a corresponding channel coherence coefficient; and a comparing means, for comparing said channel coherence coefficient with at least one predefined channel coherence threshold, to determine a corresponding channel fading mode.

15. The apparatus according to claim 14, wherein if said channel fading mode is fast channel fading mode, said parameter processing means comprises: a first correlating means, for performing correlation processing on a known short pseudo random sequence and said plurality of separated training sequence segments respectively, to obtain a plurality of initial channel parameters; a range determining means, for determining combination ranges of said plurality of initial channel parameters based on channel coherence characteristics according to said channel coherence coefficient, each of said combination range comprising at least one of said separated training sequence segments and at least one of said separated data segments; and a numerical processing means, for processing the initial channel parameters in each of said combination ranges respectively, to obtain the channel parameters for restoring data in each data segment within the corresponding combination range.

16. The apparatus according to claim 14, wherein if said channel fading mode is slow channel fading mode, said parameter processing means comprises: a combining means, for concatenating said plurality of separated training sequence segments into a combined sequence corresponding to a known compound training sequence; and a second correlating means, for performing correlation processing on said known compound training sequence and said combined sequence, to obtain the channel parameters for restoring the data in said separated data segments.

17. A mobile terminal, comprising: a receiving means, for receiving signals undergoing wireless channel fading; an estimating means, for estimating the received signals, said estimating means for channel parameters comprising: a separating means, for processing a burst extracted from received signals, to separate a plurality of data segments and a plurality of training sequence segments; a determining means, for processing said plurality of separated training sequence segments, to determine a channel fading mode caused by said receiver's mobility; and a processing means, for processing said plurality of separated training sequence segments according to the determined channel fading mode, to obtain channel parameters required for restoring said plurality of separated data segments; and a post-processing means, for performing post-processing on the received signals according to the obtained channel parameters.

18. The mobile terminal according to claim 17, wherein said extracted burst comprises a desired burst undergoing wireless channel fading, said desired burst comprising a plurality of data segments and a plurality of training sequence segments, said plurality of training sequence segments being generated by segmenting a compound training sequence.

19. The mobile terminal according to claim 18, wherein said compound training sequence is a pseudo random sequence computed from two incoherent pseudo random sequences, where in said two incoherent pseudo random sequences, the length N of a long pseudo random sequence is K times of the length M of the other short pseudo random sequence and K is a natural number more than 1.

20. The mobile terminal according to claim 19, wherein said compound training sequence is derived by repeating said short pseudo random sequence for K times and then performing bitwise scalar multiplication algebraic operation or XOR logical operation on the repeated pseudo random sequence and said long pseudo random sequence.

21. The mobile terminal according to claim 20, wherein the lengths of said plurality of training sequence segments are integer times of the length M of said short pseudo random sequence.

22. The mobile terminal according to claim 17, wherein said mode determining means comprises: a computing means, for performing coherence operation on at least two predefined training sequence segments in said plurality of separated training sequence segments, to obtain a corresponding channel coherence coefficient; and a comparing means, for comparing said channel coherence coefficient with at least one predefined channel coherence threshold, to determine a corresponding channel fading mode.

23. The mobile terminal according to claim 22, wherein if said channel fading mode is fast fading mode, said parameter processing means comprises: a first correlating means, for performing correlation processing on a known short pseudo random sequence and said plurality of separated training sequence segments respectively, to obtain a plurality of initial channel parameters; a range determining means, for determining combination ranges of said plurality of initial channel parameters based on channel coherence characteristics according to said channel coherence coefficient, each of said combination range comprising at least one of said separated training sequence segments and at least one of said separated data segments; and a numerical processing means, for processing the initial channel parameters in each of said combination ranges respectively, to obtain the channel parameters for restoring data in each data segment within the corresponding combination range.

24. The mobile terminal according to claim 22, wherein if said channel fading mode is slow channel fading mode, said parameter processing means comprises: a combining means, for concatenating said plurality of separated training sequence segments into a combined sequence corresponding to a known compound training sequence; and a second correlating means, for performing correlation processing on said known compound training sequence and said combined sequence, to obtain the channel parameters for restoring the data in said separated data segments.

25. A transmission signal, said transmission signal being generated by modulating a burst on a wireless carrier, said bust comprising a plurality of data segments and a plurality of training sequence segments, said plurality of training sequence segments being generated by segmenting a compound training sequence.

26. The transmission signal according to claim 25, wherein said compound training sequence is a pseudo random sequence computed from two incoherent pseudo random sequences, where in said two incoherent pseudo random sequences, the length N of a long pseudo random sequence is K times of the length M of the other short pseudo random sequence and K is a natural number more than 1.

27. The transmission signal according claim 26, wherein said compound training sequence is derived by repeating said short pseudo random sequence for K times and then performing bitwise scalar multiplication algebraic operation or XOR logical operation on the repeated pseudo random sequence and said long pseudo random sequence.

28. The transmission signal according to claim 27, wherein the lengths of said plurality of training sequence segments are integer times of the length M of said short pseudo random sequence.

29. An apparatus for transmitting a transmission signal, said transmission signal being generated by modulating a burst on a wireless carrier, said bust comprising a plurality of data segments and a plurality of training sequence segments, said plurality of training sequence segments being generated by segmenting a compound training sequence.

30. The transmitting apparatus according to claim 29, wherein said compound training sequence is a pseudo random sequence computed from two incoherent pseudo random sequences, where in said two incoherent pseudo random sequences, the length N of a long pseudo random sequence is K times of the length M of the other short pseudo random sequence and K is a natural number more than 1.

31. The transmitting apparatus according to claim 30, wherein said compound training sequence is derived by repeating said short pseudo random sequence for K times and then performing bitwise scalar multiplication algebraic operation or XOR logical operation on the repeated pseudo random sequence and said long pseudo random sequence.

32. The transmitting apparatus according to claim 31, wherein the lengths of said plurality of training sequence segments are integer times of the length M of said short pseudo random sequence.