EP1023784A2

EP1023784A2 - Method and receiving device for estimating channels in communications systems

Info

Publication number: EP1023784A2
Application number: EP98954234A
Authority: EP
Inventors: Leo Rademacher
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 1997-10-14
Filing date: 1998-09-29
Publication date: 2000-08-02
Also published as: WO1999020061A2; JP2001520492A; WO1999020061A3; AU1143599A; CN1281602A

Abstract

According to the invention, a receive signal consisting of data symbols is received by a receiving station. Said receive signal is split up into individual sampling values at the receive end and compared with known data symbols in order to determine channel coefficients, individual known data symbols of the receive signal being stored in the receive station. Since less channel coefficients are determined for fast-moving mobile stations, the estimation accuracy for these channel coefficients improves.

Description

description

Method and receiving device for channel estimation in communication systems

The invention relates to a method and a receiving device for channel estimation in communication systems with transmission channels between radio stations, at least one of which is movable.

In communication systems, messages (for example voice, image information or other data) are transmitted via transmission channels, in radio communication systems this is done with the aid of electromagnetic waves via a radio interface. The electromagnetic waves are emitted at carrier frequencies that lie in the frequency band provided for the respective system. With GSM (Global System for Mobile Communication), the carrier frequencies are in the range of 900 MHz. For future radio communication systems, for example the UMTS (Universal Mobile Telecommunication System) or other 3rd generation systems, frequencies in the frequency band of approx. 2000 MHz are provided.

The emitted electromagnetic waves are attenuated due to losses due to reflection, diffraction and radiation due to the curvature of the earth and the like. As a result, the reception power that is available at the receiving radio station decreases. This damping is location-dependent and also time-dependent for moving radio stations. If the mobile station moves very quickly, the channel conditions of a transmission channel change considerably even over a short period of time. In the case of multipath propagation, several signal components arrive at the receiving radio station with different delays. The influences described describe the connection-specific transmission channel. A radio communication system is known from DE 195 49 148, which uses CDMA subscriber separation (CDMA Code Division Multiple Access), the radio interface additionally lending a time-division multiplex subscriber separation (TDMA Time

Multiple Access Division). A JD method (joint detection) is used at the receiving end in order to carry out improved detection of the transmitted data with knowledge of CDMA codes of several participants. It is known that at least two data channels can be assigned to a connection via the radio interface, each data channel being distinguishable by an individual spreading code.

It is known from the GSM mobile radio network that transmitted data are transmitted as radio blocks (bursts) within time slots, mid-ambiences with symbols known in the receiving radio station being transmitted within a radio block. These midambles can be used in the sense of training sequences for tuning the radio station on the reception side. The receiving radio station uses the midambles to perform a channel estimate, i.e. an estimate of the channel impulse responses for different transmission channels. Alternatively, channels designated by different CDMA codes can be set up in parallel for training sequences and user data.

In the case of rapidly changing channel conditions, the channel impulse response of a transmission channel also changes within a transmitted radio block, so that only parts of a radio block are used for channel estimation. Regardless of the channel conditions, a fixed number of equidistant channel coefficients is determined. However, this channel estimate is inaccurate.

The invention has for its object to provide a method and a receiving device for channel estimation, which is reliable despite the fast movement of the mobile station Allow determination of channel coefficients. The object is achieved by the method with the features of claim 1 and the receiving device with the features of claim 11. Advantageous developments of the invention can be found in the subclaims.

According to the invention, in the method for channel estimation in communication systems with transmission channels between radio stations, at least one of which is movable, a receiving station consists of a data symbol

Receive signal received. The data symbols are modified by a transmission method, e.g. through spreading, modulation and channel distortion.

The reception signal is broken down into individual samples on the reception side and compared with known data symbols to determine channel coefficients, with individual known data symbols of the reception signal being stored in the reception station. The fact that a reduced number of channel coefficients is determined for fast-moving mobile stations improves the estimation accuracy for these channel coefficients. The disproportion of short sub-blocks of the received signal to the long desired channel impulse responses is eliminated and the accuracy of the estimated channel impulse response is improved. With mobile stations moving more slowly, a larger number of channel coefficients is determined.

This type of channel estimation can also be used to shift limits for the speed of a radio station and the permissible channel conditions without simultaneously having to lengthen the training sequences or reduce the data rate. The invention is applicable both to a transmission between a base station and a mobile station, and to a transmission between mobile stations. According to advantageous refinements of the invention, the known data symbols are compared with samples of data symbols in the transmitted training sequence or the known data symbols are stored after data detection and then compared with the originally received samples. In this way, it is possible either to use undistorted known data symbols which allow a more precise channel estimation at the expense of a lower user data rate. In the second variant, an accurate data estimate is a prerequisite for the subsequent comparison of the channel coefficients. However, this can be used to improve the channel estimation during the evaluation of parts of a radio block that do not contain any training sequences.

A further embodiment of the invention provides that when determining the channel coefficients a reduced number of values of the known data symbols, i.e. the training sequence or the estimated data symbols. In this way, accurate and current estimates for the channel coefficients are obtained even with a shortened period of time for sample values to be evaluated.

It is also advantageous to select the channel coefficients to be determined in such a way that on average they are more powerful than unselected values. This ensures that even a reduced number of channel coefficients describe the transmission channel with sufficient accuracy and forms a good basis for a subsequent data estimation using the channel coefficients. For example, the selected values can be taken from a previously determined larger number of channel coefficients. Consequently, according to the invention, only the significant channel coefficients are repeatedly calculated from a preceding channel estimate. The selection of the channel coefficients to be determined is repeated at larger intervals in order to take into account the development of the meaning of individual channel coefficients which are not constantly calculated. > ⁾ IV ⁾ 'y-> no Cn O Cn o Cn

ö ^y QZW tX σi cn 53 p cn 0 ω PQ P.? o Λ σ P- yQ J t 0 P. cn

PJ φ 0 φ N PJ P ⁾ PJ P- P- φ P- PJ Ω Φ Φ P- J * PJ 0 v 3 Q tr 0 J P- P- Φ tr 0 cn ti • 0 0 0 <Q Ω P- rt 0 tr 0 φ 0 CΛ 0 Hj 0 Ό Φ Ω 0 Ω φ 3

Φ P. Hh Od 0 PJ PJ 0 tr rt 0 ΣV σ pι: • J ω rt PJ Ω P- 0 • tr J tr p:

P- Φ 0: ^y Q • J rt Φ Φ rt φ rt 0 cn Ω Φ tr P. P- rt g &>

0 tr φ α ΪV P- H N M P- Φ tr 3 P- cn 0 Φ P-

N Hj 0 p- P- OO <£} tr P- P- 0 ω iQ 0 PJ o Q. CΛ P ⁾ PJ ^ P P- 3 pj: cn Φ

Φ INI 0 Φ φ φ Φ 0 P- Φ P- 0 0? PJ 0 Φ Φ rt 0 0 rt 0 Ό Λ rt P-

P- Φ 0 Hi Hl HS 0 Hf φ vP p- Φ 0 rt II Hl Φ PJ rt P> Φ 0 0 P>: 0 Q P- yQ TI Hi Hi N yQ • 0 cn P- P- Φ n Hi 0 ≤ P "α P- 0 Φ

Φ Ω cn rt l_l. PJ P- P- O: P- a 0 tr rt rt yQ P- tr P- V N O P 0 yQ cn P α Hj

0 tr tr PJ o 0 NN ^y Q <CΛ tr} Φ + N Φ HO ι-i HS PJ P- Φ

0 φ cn HS PJ P- Φ O rt Φ φ Hj Φ φ ti P- CL 3 rt φ O O 0 cn <

0 p- rt P- φ φ Hj 0 Hj 0: tr Φ P- _* Φ 0 0 Hi tr tP rt rt PL O cn Z φ 0 0 0 Φ tr Ω Φ Hj r J 0 N & ⁾ Hi φ Z PJ Φ Hj

Ό φ 0 3 rt rt 0 P ¹ cn Φ X "0 0 3 rt P- P- P- P. o 0 HS rt φ P- HS • ö Φ Φ IJ cn Φ CΛ ω 03 Hi P- Φ Φ cn N P- NH rt Φ φ rt o 0 0 0 cn <rt H P- O Φ Φ ιQ rt 0 Hj PJ P- Φ Φ rt • PJ P-

P "Φ • N vQ O P- Ω Cn Hj Φ 0 tr cn Φ P- 0

0 φ 0 PJ: Λ P- <Φ Φ H 3 et tr <rt cn Ώ Z 0 Φ P- 0 £ rt <ö cn tr pj: • PJ 3 Φ P- 0 rt 3 tr rt O P- «rt yQ P- yQ rt Φ O PJ yQ PJ tr N - - 0 Hj rt Φ rt Φ H 3 P> Φ PJ ti ti Φ tr 0 3 φ Hi

Φ Φ 0 rt yQ yQ Φ CΛ P- tr H yQ rt 3 0 h ^ P. P.? 0 Hj Ω P- Ξ rt

Hj HS HS tr Z Hj 0 P- Φ rt rt Φ 0 J Λ φ Φ Hi z tr 0 rt P>: Φ

Φ o φ O: yQ tr P- H Φ P- 0 P 'rt Φ H P- tsi φ Φ 0 tr 0 φ w tX Hj Hj P tr 0 P> rt PJ ti ^• - • yQ cn • O p- 0 iQ rt Z

Hj Hj PJ 0: rt Ω φ Φ P ⁾ Hh »» yQ tr Ω tr Φ P> Φ t> Φ rt S

P- Hi 0 Ω t HN r 0 S PJ 3 tr α P- & 0 03 PJ 0 HS Hj Φ Φ pj: P- PJ? Ϊ PJ Φ rt O _^ 0 Hl P- pi: 0 Ό cn NF> rt Φ 0 cn Cn 0 P-

0 0 cn 0 0 yQ φ O and others rt rt rt rt Hj CΛ rt PJ 0: yQ cn Ω rt rt P- cn P- Ω Z φ Hj P- Ω ω PJ - * N Ω Φ PJ tr Ω PJ tr φ tr φ S Φ

Φ 0 Ω Ω tr £ P- ti N PJ tr? cn 0 tr X) Hj 1— '? 0 ti Hj tr P- 0 Φ Hj

Hj 0 tr tr PJ Hj φ O: t PJ rt Φ 0 0! CΛ cn ti φ ti Φ rt HS tr rt yQ PJ: rt tr τ> H vQ 0 Z Hj Q α Φ P- vQ P- 0 cn rt Cn

• rt Φ Φ - »Φ Hi P- P>: φ φ P- 0 0: Φ Ω φ 0 rt 0 P- Ό Φ

ZN ^y Q cn s Hj 0: Φ Hj><φ N tr Hj tr Z 0 0 ^y Q P- Hj Q. φ 0 Φ o <o 0 Hj φ rt s? Φ 3 Φ rt PJ: P. 0 O Φ P. 0

H 0 0 0 Φ tr cn 0 Φ rt Hj et J Hj S! tr yQ tr Ω Φ 0 yQ Hj 0 rt vQ Φ PJ NN 0 tr tr rt tr Φ yQ P ¹ PJ • - - Φ Φ tr H vQ

Φ 0 φ vQ Φ P ⁾ cn P- 0 PJ PJ P> Φ Φ Hj Φ Hj φ rt 3 H (0 Φ

PJ 0 Hj H 0 N 0 PJ tr tr CΛ Hj P- cn rt ti Q. 0 W α

0 α Φ O: α α φ Φ P "<! Tr P" CΛ tr X rt Φ o Z W PJ et PJ φ

PJ cn tP P »P- cn φ J: rt Φ Φ P- 3 φ 0 Hj J tr 0 H

0 rt K Φ &> N rt tr Ω Hj 0 p ^j : H cn 0 3 P- 3 Hj α cn 0 Φ J tr OHJ Hj z Φ Φ tr Hj yQ Φ rt rt rt 0 B Φ φ P- φ rt PJ Hj Φ M ⁾ 3 o 0 rt α P- 0 cn pi: P- P- H • P- P. rt H 0 φ Φ rt H f H

0 B rt PJ P- cn cn rt 0 yQ PJ 3 Φ 0 L P "H O Hi φ N Z φ Ω Φ Φ N yQ tr tr ö 3 0 ω tr Z φ I- 'B J P- Φ P-

0 P- tr tr P- Hj 0 φ <Φ yQ PJ rt Φ φ Hj tr Ό υq 0 Hi 0

P- • Φ Φ Hj ϊ φ 0 Φ 0 Hj O P- Φ tr tr PJ O ω Hj PJ 0 0 α Hi

Φ P ¹ rt α

3 P. α 0 yQ 0 ti Z Φ Φ

P- • N PJ Od • α 0 x Hj P "0 P- P- 0

HS o P- Λ Φ P- Φ PJ ^> cn iQ N 0

3 PJ 0 φ Z Φ ^y Q rt P- 3 0 0 PJ ω P- P- «P tr ^ ö tr Φ H ö Φ H Φ α Ω y. Φ PJ P- 0 ΪV φ

Φ rt 0 PJ • - • 0 0: Hj PJ 0 ti P ⁾ Z Φ tr rt P- 0 rt J 0 0 CΛ

P- 0 3 C Ω 3 W 0 0 Φ 0 1 P. Z 0 Φ rt P- yQ P- P. P- PΪ P- P ⁾ Z 0 Hj 0 0 Φ φ O pj: Φ 0 φ rt Φ Φ 1 rt P ⁾ 0 φ α Xi 3 i 0 φ 0 Hj 1 P> Hj cn rt φ φ

1 1 p- 1 0 ti φ

1 shows a block diagram of a mobile radio network,

2 shows a schematic representation of the frame structure of the radio interface,

3 shows a schematic illustration of the structure of a radio block,

4 shows a block diagram of the receiver of a radio station,

5 shows a block diagram of a digital signal processing cleaning agent,

6 shows a schematic representation of the channel estimation problem, and

FIG 7 a flowchart of channel estimation.

The structure of the radio communication system shown in FIG. 1 corresponds to a known GSM mobile radio network which consists of a multiplicity of mobile switching centers MSC which are networked with one another or which provide access to a fixed network PSTN. Furthermore, these mobile Switching centers MSC each connected to at least one base station controller BSC. Each base station con

C co MMP ^>

Cn o cn O cn o n

φ ^y Q £ P- Od P 0 0 p. Hi cn cn y- ^{y for} PP £ Hj σ OPPPM tr ö P 3 <S yi

P- 0 o 0 P 0 Φ PP 0: Φ O P- O Φ P- 0 o 0 P- P- P- P ¹ P- p- 0 P- φ 0 0 o PP

0 0 tr cn CΛ tr N tr tr Hj Ω 3 0 Hj Hj tr 0 Φ <Ω P rt φ 0 0 Φ 0 Ω 3 H (rt φ ^y Q P- φ P- ^• Q p fr 0 Hj rt Φ tr Ό Φ Ω P- ^y Q P- tr Ό Φ NP φ tr 3 • rt P-

HS cn P- CΛ φ Φ o »» rt P Φ o 0? R tr cn <cn Φ rt Hj Od Φ φ 00 0 P- 0 O cn 0 CΛ ^ Q HS 3 3 p. 0 0 0 rt cn P- Φ P- P- 0 P 0 PP w ö 0 0 cn PJ rt φ rt P cn Ό P - φ rt φ to rt 0 Hj OH <P CΛ Λ ω Q. 0 P P- ^ Q CΛ

Ω 0 PJ 3 P 0 o P P- 0 INI 0 Φ 0 Φ P Hl tr 0 Φ Φ rt P- rt P- p- rt φ cn 1 tr PJ: rt rt ^y Q ^> Q 0 P φ φ rt H Hi rt O P- P- Hj P- CΛ Hj P- cn φ P- Hi

0 P- ^ P- Φ ti Φ <P- Φ cn> P- Hj 0 £ <l CΛ P P- cn P o TI 0 0

Φ φ o Hj O ti P 0 tr £ O rt 0 Ω 0 Φ O 3 0 s rt tr Φ rt M 0 0 0 0 0

P ^ 0 Φ 0 _^ rt Φ φ 0 Ό tr cn? 0 P 0 φ & PPP Hj P cn 0 f P. iQ Φ cn Φ P- tr 0 P- tr rt Φ rt 0 rt H- rt 0 rt φ N rt Hl φ cn? rt φ φ £ 0 0 0 Hj φ 0 Φ φ Hj P- 0 P- ^y Q 3 Φ Ω P- Hj • P- P- Hj * rt P- Sl

0 HS CΛ Φ P N α Z P- I ω α φ O O φ TS φ 0 tr O 03 O 0 Φ cn P- O P

0 03 tP φ - Φ φ 0 rt P- 0 £ 0 0 Hj 0 0 cn • 0 α rt o 0 HS

Od H P- N CΛ P- Φ H ^y Q φ Φ vQ rt Φ CΛ Φ φ cn Φ Φ rt 0 TI Φ 0 Φ rt

Φ φ cn cn rt Hj Φ H 0 Ω 0 0 0 3 φ 0 P- td HS P Od 0 0 3 P 0 0

Z P- rt PJ P P- P Φ P φ tr 0 0 P-> Ό - - CΛ P 0 ^y Q 0 P ^J 0 φ 0 3 Ω tr <3 0 Od tr p- Φ vQ tr 0 cn rt OP Φ tr cn SV 0 P- Hi vQ

> Q ti φ P tr tr Φ • d cn φ Φ Ω 0 CΛ Φ rt P- Ω H Φ <P- N 1 P rt 0: cn

0 φ cn W p: Hj Hl tr Z p- tr z P- Φ (Φ p- P- yQ φ rt Q. cn c; Ω pj: HS N

0 tr 3 0 cn P Hj φ Φ> φ cn Hj N Φ cn φ 0 Hj Hj H rt O tr rt φ Q 3 0: Ό P- ιQ Ω 0 Φ ^y Q α 0 vQ T3 Ξ 0 cn Hj vQ O: CΛ ^y Q Φ 3 α 0

Φ Φ tr HI Φ P- ^y Q P- 0 Φ cn P- P- - 0 Φ & a O Φ P- 3 M 3 P P. P rt

H P- φ P vQ P- Φ rt 0 H ι-3 tr tr Φ φ cn 3 0 cn Φ HS Hj P- P- 0 φ P- cn H φ cn 0 Hj 0 H3 ^' φ 0 0 vQ Φ H Φ P ^{1 y} Q Ω 0 φ υq φ H ^y Q P- W rt 0 0 H Φ p

H (P- φ P ^{1 y} Q φ 0: 0 φ P- φ Hj cn φ tr rt P- rt Φ rt Ω P- cn P- cn £ 3 vQ P CΛ P- tr Φ Φ vQ P- 3 P " P- <Z ti Φ N 0 0 N Φ tr 0 Φ PW Φ o

£ 0 ^• Q o P 'Φ 0 0 φ Ό 0 rt o φ 0 P- ^y QP rt 0 rt N P- rt P Hj Hj tr Q o PJ J φ Hj 0 Hj Φ NH Hh Φ 0 Hj φ 0 φ Ω & Φ Φ 0 N rt Hi P- £ tr P 0 Hj rt Φ £ 0 P cn tr ZN TI rt P- Φ P- P- (Λ P ^J O

P- φ PJ 0 tT PO cn £ 0 g ^" 0 ^y Q CΛ α Hi ^y Q φ φ ti Φ O φ Φ P 0 ω Hj Ϊ O 0 rt Hh

P- • N 3 ^y Q> tr P o ^y Q Φ cn Hj φ Hj Hj rt Φ Hj tr 0 P- rt O 0 CL H 0 H ~ J cn P. cn 0 φ Φ 0 P- 3 tr Φ Hj Z 0 PH 3 tsi P- CL P ¹ : τ 0 H s 3 cn 0 0 0 Φ rt 0 Ω> Hj Hj cn P "3 P- ti ω φ Hi 0 Od P rt 0 Φ NP cn 0 IV f P

P> Hj tr 0 Φ cn 0 tr cn φ p. P- ^y Q Φ £ P rt Φ 0 <Φ Φ tr rt Φ ^• <vQ rt 0 l- ¹ rt Ω p: P- P- • Hj rt 0 cn φ vQ Φ 0 cn φ cn P- Z 3 ^• » O P- Φ 0 cn 0 Φ P-

P- tr rt ^y Q 0 ^y Q φ P rt 0 0 0 φ tr P- O φ 0 0 Φ 0 rt ^y Q HS rt CΛ o NNP Φ ti SJ P- rt 3 PPZ tr Hj cn 0 Hj CΛ cn φ Φ Hj 0 Φ Φ N P-

0 0 0 tr 3 P φ rt P- P- rt CΛ P ¹ N φ p: Z cn Φ ö OM • 0 0 Φ P 3 ti S φ

Hi 0 Φ P '0 O rt P- rt cn 0 Hj 0 φ rt 0 Φ s P- 0 0 φ 0 P- tr H

£: vQ W φ rt 0 0 OP o α α ^< QP [j3 0 W CΛ tr Φ 0 rt N rt

CΛ tr Φ 3 Φ ^y QZ 0 rt N Hj Φ Φ φ rt 0 1 PN CΛ Φ Φ P- • 0: N

Hj α z

Hi P- Ό Hj cn £ φ Φ P- 0 α 0 P P- P- K <&> Φ P CΛ 0 tr rt W

P> Φ: 0 HI φ tr Z CΛ 0 o Φ • 0 OP φ y- ¹ ? rt rt Hj Φ -. O

0 0 H Φ P Hj Φ rt 0 0 0 N cn 0 0 0 Hj P cn O: O Φ P- Hj o Hi 0 cn Hj 0 cn ti ^y Q cn Φ £ 0 tr Ö 0 tr 0 0 Hi 0 rt 0 Hj tr. Ω rt CΛ 3 0: rt ti α vQ Ω Φ P- Hj CΛ Φ rt PP CΛ Hj 03 CL • P Ω P 0 rt tr P- HS HS

Φ P- Φ CΛ tr Z Ω Φ P- Φ HS 3 p: Φ CΛ tr tr tr Φ • rt P Φ Tl O

P- P- Φ 3 CΛ P- Φ tr 0 P 0 HS Φ P- rt P- CΛ Hj 0 Φ 0 ^y Q 0 H3 I— ¹

Ω φ Ό P- Φ Hj 0 cn rt N rt 0 P- Φ Φ φ Hj -l 0 tr P- 0 Φ 1— 'tr et HI ^y Q ti P. CΛ cn. , Ω tΛ 0 0 ^y QH (LJ. Φ vQ P cn f P- 1

Φ PJ tr P ti Φ φ rt - tr P- rt P- 0 α ti ti 0 Φ P. Hj rt 1 ¹

0 Φ 0 P 0 cn Φ O: Φ <Q HS Ω ^y QP. -. φ α 0 Φ Φ 0 α Ω Hj ^y Q Φ P Hj Φ 0 Φ tr - CΛ. -. o Hj H Φ L. o 0 tr rt φ HS <H-3 CΛ 0 PP P- P Hi Φ p- Ό Hj ^y Q Ω Ω Hj P- Φ 3 P α

Hj ti Hj o Φ p- 0 0 << 1 P ¹ Hi N P. ω P p: φ tr tr P 1 0 α 1 P tr PJ p. X ti vQ ^y Q 0 1 Φ c P- P- Ω 0 3 Φ o 1

Φ 1 φ 1 φ 0 3 φ φ Φ B p: HS Ω

P- 0 1 0 1 1 dr

CΛ 0 yQ s cn Φ cn ö 3 cn rt yQ

Ω P 0 φ Φ P- P- P- P P- Hj φ tr P iP 0 Ω Φ rt Φ P 0

H rt 0 φ tr p- Hj ^ Q P

P Φ Φ CΛ rt Λ o 0 Φ 0 rt 0 Hj 0 P 0 0 0

N Hh tr N tr P- P; Φ yQ Φ 0

Φ tsi Φ p- Hj Ω 3 0 cn 0

Φ Hj 0 rt φ tr Φ P- α rt P- cn P- • 0 N 0 P

CΛ rt ö P- Φ rt cn 0 Hi rt P

Φ P CΛ XX tr cn rt o Φ f

<0 0 rt P Φ Hj i 0 rt

O Φ tr 0 3 0 Φ 3 CΛ 0

Hj P <Hj P- Φ p: fr rt P ^ Φ u3 Φ 0 cn P- tP rt Φ rt 3 H- ¹

Φ H H-3 0 rt vQ 0 P- o ^* cn vQ Φ Φ Φ Φ H rt O o

Φ <Hj tr 0 P- P- 0 tr Φ N tr rt 0 Φ Φ φ cn

Φ Hj 0 P TI Φ Φ Hj 0 rt

0 tr Hj P H Hj H Φ

• <

P- tr 0 Φ 0 ω o

0 X CΛ iP H3 TI P- 0 o α 0 ö 0 0

0 3 Φ tr ¹ Φ 0 0 3 CL

0 Ό Λ 0 P t ^ 0 0

<Q Φ π N 1 cn α rt P- φ 0 tsi tr tr W Ω N 0 ti cn φ Φ o tr Ö P-

P P- Ό H 3 ti P 0 P.

P rt rt cn Φ V rt Hl φ

0 P- n _^ P- o rt Φ O 0

Hi o Ω Ω rt 0 Hf 00

Tr 0 tr Φ tr Φ cn 3 H3

P- tr Φ 0 rt <P φ

0 0 Φ CΛ rt Φ O rt P-

P 0 rt 0 Φ ti P-

0 rt N Hh O ti

P. Φ Φ P tr P- φ O 0 φ φ Hj CΛ Φ Cn Hj Φ tr

H CΛ P- rt yQ 0 3

Hh Ω rt cn CΛ cn P Φ o tr cn Ό Φ rt 0 0 HS y- ^> P- α P- P- P- P- cn yQ φ Φ φ φ 0 P cn • * P- φ P- H P- ¹ Φ 0 P v ti 0 cn cn rt (Λ 0

Q> φ ö Z> P- P- P φ Ω P φ 0 ^ O vQ P-

Hj tr CΛ 0 p- Hi PH 0 0 φ

Φ Ω φ cn rt O cn P 0 tsi Hj tr H φ Φ P- P "φ 0 P-) 0 P- 0:

P- CΛ rt H P. Hh 1 tr rt P- N Ω Φ 0 Φ O Φ

1 vQ N • H 0 Hj HS Hj 1 φ vQ 1 1 1

P- rt

Within a broadband frequency range B, the successive time slots ts divided by a frame structural ^¬ structure. Eight time slots ts are thus combined into a frame, a specific time slot of the frame forming a frequency channel for the transmission of useful data and being used repeatedly by a group of connections. Additional frequency channels, for example for frequency or time synchronization of the mobile stations MS, are not introduced in every frame, but at predetermined times within a multi-frame.

The parameters of the radio interface are e.g. as follows:

Duration of a radio block 577 μs number of chips per midamble m 243

Protection time Tg 32 μs

Data symbols per data part N 33

Symbol duration Ts 6.46 μs

Chips per symbol Q 14 chip duration Tc 6/13 μs

The parameters can also be set differently in the upward (MS -> BS) and downward direction (BS -> MS).

The receiver according to FIG. 4 relates to radio stations, which can be either a base station BS or a mobile station MS. The receiving device according to the invention is used in the receiver for channel estimation. A possible implementation of the corresponding transmitter can be found, for example, in German patent DE 197 34 936.

The reception path of the device is shown in detail in FIG. In the sub-module E1, the received signals rx are converted from the transmit frequency band into the low-pass range and split into a real and an imaginary component. An analog low-pass filter takes place in submodule E2. and finally in the submodule E3 a double oversampling of the received signal with 13/3 MHz and a word length of 12 bits.

In the sub-module E4, digital low-pass filtering is carried out using a filter with a bandwidth of 13/6 MHz with the highest possible slope for channel separation. This is followed by a 2: 1 decimation of the double oversampled signal in sub-module E4.

The received signal e obtained in this way essentially consists of two parts, namely a part em for channel estimation (training sequence tseq with known data symbols t) and parts el and e2 for data estimation. in the

(k) submodule E5, the channel coefficients h of the channel impulse responses are estimated using a known midamble basic code m of all transmitted in the respective time slot

Data channels.

In sub-module E6, parameters b (k) for matched filters are determined for each data channel using the CDMA codes c (k). The sub-module E7 eliminates the interference originating from the midambles m (k) in the reception blocks el / 2 used for data estimation. This is possible by knowing h (k) and m (k).

In the sub-modules E8 to E12, the data symbols d are determined using a pseudo-inverse of the combined channel matrix A. An alternative solution method is a singular value decomposition of the combined channel matrix A. Further solution methods relate to another optimization criterion, e.g. the minimum ean square error criterion (MMSE) instead of the zero forcing (ZF) criterion. Feedback structures are also possible. These solution methods can also be combined with one another. In sub-module E8, the cross-correlation matrix A A is calculated. Since AA has a Töplitz structure, only a small part of the matrix has to be calculated, which can then be used to expand to the full size. When the mobile station MS moves slowly, this matrix A * τ A is large because large sub-blocks are selected. Only a small part of this matrix A * τ A is then calculated. When moving quickly, the matrix A * τ A shrinks, so that a complete calculation may be selected to achieve a lower-noise solution. In the sub-module E9, a Cholesky decomposition of A takes place H, where H is an upper triangular matrix. Due to the töplitz structure of AA, H also approximately has a töplitz structure and does not have to be fully calculated. A vector s represents the reciprocal of the diagonal elements of H, which can be used advantageously for solving equations.

Adapted filtering of the received symbol sequences el / 2 with b (k) takes place in sub-module E10. Sub-module Eil realizes the system solver 1 for H * τ * zl / 2 = el / 2, and

Submodule E12 the system solver 2 for H * dl / 2 = zl / 2. in the

Sub-module E13, the estimated data dl / 2 are demodulated, descrambled and finally fold-decoded using a Viterbi decoder. The decoded data blocks ^β _E (k ₁ ) ₃ are optionally fed to a first data sink D1 or via the source decoder E14 to a second data sink D2. In addition, the decoded data blocks ^β _E (k _l ) ₃ are fed back to the sub-module E5, which uses them to track the estimated channel coefficients h.

At the receiving end (see FIG 5) takes place by an analog processing, instead ie amplification, filtering, conversion to the baseband in the RF section, a digital low-pass filtering the ^Empfangss' ignale rx in a digital low pass filter. A part of the digitized "received signal e, which is represented by a vector em of length L = M * W and does not contain any interference of the data part dt, becomes one a digital signal processing means containing channel estimator KS. The data estimation in the joint detection data estimator is carried out jointly for all connections, and a detailed description of the German patent DE 197 34 936 can be found.

The channel estimation that is carried out in submodule E5 is explained in more detail below. The sub-module E5 contains a channel estimator KS, a memory device ¹ SP and a control device SE, between which an information exchange is possible. Data symbols t of the known training sequences tseq from the middle area 1 m, as well as already detected data symbols d for tracking the estimate, are stored in the memory device SP. The control device SE can switch subscriber-individually between different modes of channel estimation, the channel estimation being carried out in the channel estimator KS. With the aid of the channel estimation, the channel coefficients h are determined from the sample values el..el6 of the received signal e.

The channel estimation problem is shown in FIG. 6. Transmitted data symbols d are modified by the transmission method, represented by the combined channel matrix A and are available to the receiving radio station as samples e. The receiving radio station tries to determine estimated values d for the transmitted data symbols d by determining an estimated combined channel matrix A during the channel estimation. The optimization criterion is, for example, the minimization of the quadratic error of the TL deviation of sample values e to the estimated received signal e. In the channel estimation from the training sequence tseq, d = t and in the tracking, already estimated d are fed back and d = d.

The channel estimate in the mobile radio system must take into account the movement of the mobile stations MS. The movement leads to a Doppler shift, which is caused by an amplitude and LO co to K) P "cn o Cn o cn o Cn

3 Hj fr CL P tr INI P CΛ P Φ CΛ P P z vQ CL tr P? Od P- CΛ t P p <a X 'T <V

P- 0: φ o Φ φ H- ¹ 0 P- p- 0 P- w O 0 Φ Φ p- Φ φ Φ O Φ 0: P- P 0 o Φ Φ PO tr φ Ω M φ Hj Hj O: Hj < ^y rt 0 P- Hj H HS HS cn N φ P- tr ιQ 3 0 rt Φ Hj Hj 0 3 P

H fr ω Hi 0 Ω P 0 Z Φ P- ⁽ Q Ω rt P Hi o Hi HS Pt o 'P z Hi z CL P n cn

Φ ^y Q Hi H p ^ ≥; 0 PO HS 0 rt tr CL Φ HS TI vQ H, P "rt Φ OO φ Φ Φ

0 φ P- P- Φ φ P Ω y- ^y Hj Q 0 ti P. Φ HS Φ P- P y-- H P- 0 0 3 Φ 0

Hi 0 N σ P- Ω Φ y- ^y rt vQ P- φ HS P- XX Φ 0 N 0 Z rt φ O rt N CL OX <

0: P- P- ^• - • 0 tr - j Φ Φ er 0 Hi Ω Φ φ 0 iP Φ P- vQ Φ Φ Φ P- φ <CL Φ φ

O: tr TI Φ Φ tr 0 Hi • 0 0 3 tr 0: tr P- Φ 0 0 φ cn HS 0 α Hh P ^ φ rt P- φ H

0 Hj P- 0 rt 0: Hi Φ p: tr 0 £ 0 Φ 0 P α Hi O 0 Φ h ^ a p:

0 rt vQ rt u O: Φ t N o cn 0 φ t L Φ P- N 0 rt 3 φ P s: P- 0 rt Φ 0

Φ • • Φ P Ω Hj Hj P- P. cn iQ Hj rt P- rt N P Φ φ 0 Φ p: N CΛ φ cn φ Hj

0 0 rt f rt 0 Φ Φ Φ P- • φ ≥; rt tr tr 0 0 0 • Hj tr P- rt 0 O rt Φ

Φ Φ Φ 0 rt Hj 0 vQ M P Φ Φ o P Hi Φ P K ti Φ H cn 3 tr 0 P- vQ φ fr 3 M 1- Ω HS Ό tr cn £ Φ 0 0 tr <P - - - 0

P- cn Φ 0 Φ ^• b P- HS tr ti 0: Ό cn Φ O 0 rt rt - φ 0 Q, 0

Ω Φ φ

P- 0 Φ P- rt zo P- P- Hi 0 P Hh ti Ω P- Ω P CΛ tr α Φ • HS P φ tr vQ tr P- 0 rt Ω »φ H 0 rt P P-: ^y Q p ^y 0 tr 0 Ω P- 0 LJ. p: Hj Φ

0 0 α Φ tr Hj H Φ 0 0 g 0 tr CΛ vQ p: Hi tr Φ ö φ 0 cn <

CL φ φ f cn CL Φ Hj α <! vQ 3 P- H 0 P- cn rt Z cn P- tr P P CL ei CL Ω O

P- cn P- rt CL O Φ Cn Φ O cn cn 0 0 0 Ω N φ P- rt CL o Φ φ Φ tr ti φ <P- P- 0 Ό HS 0 Φ Ω ^y Q 0 rt tr Z 0 P - 0 P Φ 0 Ω Hj Hi 0 Hj φ ι-3 φ o φ o ^ 0 P- tr cn vQ φ rt Φ 0 0 CL rt CΛ 0 Hj tr 0 Hh rt P- Φ

1-3 Hj Φ 0 P CL 0 0 P. 0 P CΛ Hj P- HS ^y Q φ P- P- Hj Ω P- ω Φ p- φ N p- CL x Go P- P. Φ ^y Q Φ Hj φ a P. vQ CL Hj vQ O ^ Ω tr tr CΛ N Ω tr 0

P- O: 1— 'Φ tr P φ P- Hj cn H P- rt XX Φ Hj rt Φ 0: ^ 0 0 tr P- P- tr Φ N vQ tr rt φ ^' 0 P Φ 0 ^y Q Ω Φ 0 H 0: • 0 tr £ Φ 0 p φ Ω Φ Φ 0 Φ tr Φ Φ 0 P 0 σ H 0 φ Q tr 0 Φ Ω Φ P- P- £> P tr 0 P "

Hj O fr- 0 ^y QP Φ ^y Q cn Φ rt 0 tx ^ - Od 0 H rt rt CΛ tr 0 0 tr rt

O: 0 Ω z 0 rt rt tr Φ rt 0 cn Ω 0 0 α N P rt φ 0 rt φ CΛ 0 rt α Φ <v Φ Φ

Ω 0 for P- N Φ ti Φ P. vQ tr P 0 φ Φ 0 P- HS 3 Φ 0 O z P- 0 P Hj 0 fr- vQ cn Φ rt CL O 0 Φ Φ z 0 vQ Hj 0 P 0 Φ P- Ω φ 3 P. φ rt CL

Φ H 3 rt 0 cn P- 0 3 TI tr 0 tr Φ> H P- P tr Φ Φ

≤ rt P. ti 3 Φ Ω ti vQ CL £ rt ?? CL P- 0 Hj 0 P CΛ rt rt CΛ a 0 0

0: Φ Φ Φ P- ^y Q φ P- <tr LP ^ 0 o CΛ o rt 0 Φ vQ cn W 0 Φ 3 rt g ω tr Hj H 0 Φ 0 0 P ¹ o Z P- Φ Hj P <Φ φ tr rt ^ H P- PP 0 Φ ⁱ i HS P- cn fr P

Φ L 0 Φ Hj ^y Q Ω P- XX Hh Φ tr Φ φ 0 0 vQ P- Hj rt rt Ω P

H Φ tr N ^y Q ≤ tr 0 rt tr Hi P- Φ P- Φ HS P Ω 0 o Φ tr P ^

P 0 φ 0 Φ φ α P Φ Q. ^y Q P- HS O: ^ H φ y- ^> tr P. Φ Hi vQ Hj 0 O

P 0 fr- <! 0 HS φ P * P- CL CL Φ Φ N tr Φ Ω 0 0 cn Φ H P-

PN α P "P- Φ ^y QP o CL H rt XX rt Φ P- HS tr P- O: 0 ^y 0 P Ω Hi rt Φ Φ Hi Hi φ 0 0 Hj 0 Φ N ^ - Hj φ P- Φ tr Φ p - 0 CL tr 0: Φ ≥; Hj rt cn Hi

Hj cn Hj 0 0 0 TI Φ Φ WP ¹ 0 Φ cn tr Hi P- p: Hj - PP Ω P- P-

* 0 _* Ω rt tr H 0 P- Φ CΛ CΛ P CL rt Hj Φ φ rt rt Ω φ φ tr Ό φ N tr tX Φ Φ frT 0 rt rt Hj Ω rt 0 Φ Φ Φ cn Φ O: CL N tr tr Hj P- Φ M tr P- pj: P 0 cn 0 P fr tr Φ P rt 0 0 Ω HS Ω P- a 0 O: Φ Hi 0 rt 3 Φ Φ cn rt 0 0 0 tr ^ - _^ Φ pj : 0 Φ tr p ^ - Φ Φ 0 tr Hj 0: t N 0

ONP σ P- tr 0 rt P- P rt Φ cn 0 PQZ XX Φ HS ^y Q Φ rt tr P φ D α rt

0 P 3 Φ -> O- P- 0 0 N Hj Ω 0 Φ • £ rt Hj Φ Hj 0 e (Tι P Φ

0 cn rt 3 α Ω 3 Φ P 0 Φ tr £ Ω CΛ ti φ P- Φ

Ω Z Φ Hj 0 P ti cn N 0

P vQ Φ rt Φ for P- Hj 0 P- p: P- tr Ω CL rt φ 0 φ rt ω 0 ξ tr 0 Φ 0 P Φ 0 ^y Q 0 rt rt tr p- Ω 3 rt α cn σ ^y Q ti P n tr

N pj: cn 0 rt tP ^y Q 3 3 Hj N rt P z vQ tr P Φ rt Φ P 3 0 CL tr 0 rt v <^ Φ P- Ό P- 0 P 0 P ^ Φ Hd H Φ CΛ rt CL Ό X CL> P

Φ H (N K 0 z ti rt 0 <! Ω 0 3 tP ti Φ 0 P & Hj Ω Φ φ 0 P cn P

P- 0 CT P CL p: h- ¹ O tr vQ ^" φ CL P- Φ w ^y Q tr 0 HS 0 P ^»

1 3 0 o 0 tr ι-3 α cn 0 et Φ H P- rt H3 P- ¹ P Φ z 1 cn P.

P- vQ P p: HS Φ Φ 1 0 Φ tr vQ Φ φ 1 P- l

0 Φ 0 Φ p- H 0 HS 0 P 1 0 p: 3 P 0 1

P- 1 1 0 y- ^{> y} Q 1 1 1

1 CL P- 1 3 tr 1

are written, the combined channel matrix A being the influence of the spread by the CDMA codes c and the modulation by the rtra un scanal mi n

After the data has been estimated, if necessary in the case of fast-moving mobile stations MS (step 3 in FIG. 7), a re-evaluation of sub-blocks of the part of the received signal e containing useful data can be carried out to track the channel coefficients h:

e (t) = ∑c _k (t) χh _k (t, r) χd _k (t) + n (t),

where c and h are combined as a combined chip sequence g and a matrix notation:

e = Gh + n,

results, so that the channel coefficients h by means of

can be estimated (step 6 in Fig. 7).

Many transmission channels can be described with sufficient accuracy with a few relatively powerful channel coefficients h, which may be distributed over a large time interval caused by the signal delays.

The delays (positions) of these significant channel coefficients h are determined in a step 4 (see FIG. 7). In the following, only this reduced number of channel coefficients h is then determined when evaluating the sub-blocks.

An example of a system of equations to be solved with three subscriber signals A, B, C and 16 samples el..elβ of the received signal e has the following form:

e = h + n.

This system of equations is only slightly overdetermined and therefore leads to a noise-disturbed estimation result. The chip sequence matrix G and the vector of the channel coefficients h can be punctured so that only the significant channel coefficients h remain:

e _p +

If the dotted elements are removed, a shortened system of equations results:

= Gs h _s π n.

This system of equations is strongly over-determined for the reduced number of channel coefficients h = hs, so that the accuracy of the estimate (step 5 in FIG. 7) is in accordance with

h _s = G ⁺ e

is improved.

This procedure can also be used to increase the signal delays that can be represented by the channel impulse responses. The number of channel coefficients h remains the same, but their spacing is increased.

Claims

claims

1. Method for channel estimation in communication systems with transmission channels between base stations (BS) and mobile stations (MS), whereby both base stations (BS) and mobile stations (MS) can send and receive, in which

a received signal (e), which contains data symbols modified by a transmission method, is received in a receiving station (MS, BS),

the reception signal (s) is broken down into individual samples (el .. el6) on the reception side,

- Individual known data symbols (d, t) of the received signal (e) are stored in the receiving station (MS, BS), - the sample values for determining channel coefficients (h)

(el .. el6) and the known data symbols (d, t) are compared,

- A reduced number of channel coefficients (h) is determined for fast-moving mobile stations (MS).

2. The method according to claim 1, in which the known data symbols (t) are compared with samples (el..el6) of data symbols in the transmitted training sequence (tseq).

3. The method according to claim 1, in which the known data symbols (d) are stored after data detection and are compared with the originally received samples (el .. el6).

4. The method according to any one of the preceding claims, in which a reduced number of values of a chip sequence matrix (G _s ) is taken into account when determining the channel coefficients.

5. The method according to claim 4, wherein the selected values of the reduced number of channel ^¬ coefficients (h) on the average performance are, as a non-selected values.

6. The method according to claim 5, wherein the selected values have been taken from a previously determined larger number of channel coefficients (h).

7. The method according to any one of the preceding claims, wherein the distances between the selected values of the reduced chip sequence matrix (G _s ) are not equidistant.

8. The method according to any one of the preceding claims, in which a greatly over-determined system of equations e = Gh + n, with n as the noise component and G as the chip sequence matrix, is established by reducing the number of the determined channel coefficients (h).

9. The method as claimed in one of the preceding claims, in which the number of sampled values (el. Elβ) taken into account in the channel estimation is reduced depending on the channel conditions of the transmission channels.

10. The method according to any one of the preceding claims, wherein a distance between the channel coefficients (h) based on signal delay times is increased without the number of channel coefficients (h) increasing to the same extent.

11. Receiving device for a communication system, with a storage means (SP) for storing samples (el .. elβ) of a received signal (e), which contains data symbols modified by a transmission method, and for storing known data symbols (d, t) of the Receive signal (s), and with a channel estimator (KS) for channel estimation for transmission channels, - wherein to determine channel coefficients (h) the samples (el .. elβ) and the known data symbols (d, t) are compared, and

12. Receiving device according to claim 11, in which the selected values of the reduced number of channel coefficients (h) are on average more powerful than unselected values.

13. Receiving device according to claim 11 or 12, in which the number of sampled values (el .. el6) taken into account in the channel estimation is reduced depending on the channel conditions of the transmission channels.

14. Receiving device according to one of claims 11 to 13, in which a plurality of subscriber signals are transmitted in a frequency channel via different transmission channels, which they superimpose on the receiving device (EE) to form a received signal (e).

15. Receiving device according to one of claims 11 to 14, in which a training sequence (tseq) containing the known data symbols (t) and subscriber signals are transmitted in the same frequency channel.