AU731787B2

AU731787B2 - Method for direction estimation

Info

Publication number: AU731787B2
Application number: AU80077/98A
Authority: AU
Inventors: Paul Walter Baier; Josef Blanz; Martin Haardt; Ralph Schmalenberger
Original assignee: Siemens AG
Current assignee: Nokia Solutions and Networks GmbH and Co KG
Priority date: 1997-03-25
Filing date: 1998-03-25
Publication date: 2001-04-05
Anticipated expiration: 2018-03-25
Also published as: KR100532803B1; EP0970388A2; RU2195682C2; JP3935968B2; KR20010005580A; CN1201167C; CN1251656A; JP2001519033A; DE59802725D1; WO1998043106A3; AU8007798A; EP0970388B1; WO1998043106A2

Description

GR 97 P 8026 -1- Description Method for direction estimation The invention relates to a method for direction estimation of wave elements of a received signal, for example in base stations for mobile radio networks or in applications for radar or sonar systems and for seismic measurement systems.

One or more subscriber signals can be caused by transmitting from one or more communication subscribers to a common receiving station, by transmitting from one or more transmitters with superimposed measurement signals or by reflections of a measurement signal on obstructions or geological layers.

Methods for determining the directions of arrival of various signals are known from R. Roy and T.

Kailath, "ESPRIT Estimation of signal parameters via rotational invariance techniques", IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-37, pp.984-995, July 1989. For example, a direction estimation method which is known from DE 195 11 752 and used as the UNITARY-ESPRIT method leads to the direction of wave elements being determined directly from the received signals.

By way of example, the application of direction estimation in mobile radio will be discussed in the following text.

With mobile radio or methods similar to mobile radio, a new field of application has been opened up for direction estimation. When signals propagate in a propagation medium, they are subject to interference caused by noise. Owing to diffraction and reflections, signal components pass through different propagation paths and are superimposed at the receiver, where this leads to cancellation effects. Furthermore, if there is more than one signal source, this leads to these GR 97 P 8026 -l signals being superimposed. Frequency-division multiplex (F'DMA), time-division multiplex (TDMA) or a GR 97 P 8026 2 method which is known as code division multiplex

(CDMA)

are used to distinguish between the signal sources, and thus to evaluate the signals.

If, for example, a CDMA method is used for subscriber separatidn, then a plurality of subscriber signals can be transmitted in one frequency channel at the same time, and can be separated in the receiver.

Mathematical descriptions, the method of operation and the structure of CDMA (Code Division Multiple Access) radio transmission systems are known from P. Jung, J. Blanz, "Joint detection with coherent receiver antenna diversity in CDMA mobile radio systems", IEEE Transactions on Vehicular Technology, Volume VT-44, 1995, pages 76-88. When such systems are used for mobile communication, there is a radio interface between fixed-position base stations and moving mobile stations. The transmission path from a base station to a mobile station is called the downlink path, and the transmission path from a mobile station to a base station is called the uplink path.

P. Jung, J. Blanz, "Joint detection with coherent receiver antenna diversity in CDMA mobile radio systems", IEEE Transactions on Vehicular Technology, Volume VT-44, 1995, pages 76-88, furthermore shows that the transmission quality in such radio transmission systems can be improved since it is possible to use an arrangement of a plurality of receiving sensors instead of a single receiving sensor.

In accordance with the terminology used in the abovementioned document, K denotes the number of subscriber signals which are transmitted from a base station at the same time in the same frequency channel, for example of supplied mobile stations. Ka denotes the number of receiving sensors which are assigned to a receiving device, for example the base station. In one such scenario, there are, in consequence, K*Ka radio channels in the uplink path between the K mobile stations and the GR 97 P 8026 -3- Ka receiving sensors in the base station. Each of these radio channels is characterized by a discrete-time baseband equivalent of its channel impulse response g k) (ka) where k ka 1..Ka. These channel impulse responses guW(ka) are used for channel modelling in data detection. This method does not envisage any statements relating to the directions in which wave elements arrive.

The invention is based on the object of specifying an improved method for direction estimation, which results in a reduction in the influence of interference signals on the direction estimation, with little calculation complexity. This object is achieved by the method having the features of patent claim 1.

Developments of the invention can be found in the dependent claims.

For the method according to the invention for direction estimation of wave elements of at least one subscriber signal, a number Ka of receiving sensors are assigned to one receiving device. Ka received signals are received via the receiving sensors, being caused by at least one subscriber signal which has a transmitter-specific fine structure impressed on it, in which case a k-t subscriber signal, k is transmitted by means of Kd wave elements whose directions of arrival at the receiving location differ.

Channel impulse responses assigned to the Ka receiving sensors are determined from the received signals, and the direction of arrival of at least one wave element is determined from the channel impulse responses.

The accuracy of the direction estimation is improved since the channel impulse responses already take account of the characteristics of the channels as raw information for the direction estimation. Knowledge relating to the received signal, in the form of the transmitter-specific fine structure, can be used for channel estimation in the receiving device. The 7)direction estimation is thus more accurate than when GR 97 P 8026 u 3a s using unknown data which are still to be detected.

GR 97 P 8026 4 According to an advantageous development of the invention, the channel impulse responses are determined from training sequences, which form the transmitter-specific fine structures, of the subscriber signals. Such training sequences are known from mobile radio, for example as midambles in GSM useful channels.

These training sequences can be used in a simple manner for the method according to the invention. The method according to the invention can thus be implemented in mobile radio networks with little complexity.

The subscriber signals (which can be separated by the transmitter-specific fine structures) from a plurality of transmitters or reflectors arrive at the receiving device and are superimposed to form the received signals, in which case these signals are transmitted at the same time in one frequency channel.

The fine structures can advantageously at the same time be used for direction estimation and for subscriber separation, when message transmission is taking place.

This means a further reduction in complexity at the receiver end.

According to a further advantageous refinement of the invention, the determination of the directions of arrival of the wave elements also takes account of information relating at least to one of the following values: direction of arrival, power, spectrum or a correlation matrix of interference signals. The greater the amount of knowledge that is available about interference sources, the better is it possible to evaluate the signals for which the directions of arrival are intended to be estimated. This measure also allows the direction estimation to be improved.

According to a development of the invention, the subscriber signals can be separated by despreading using individual subscriber codes. In this case, the directions of arrival of the wave elements can be assigned to the subscriber signals. This type of ssignment allows the advantageous use of the GR 97 P 8026 4a invention, for example in mobile radio systems in which the direction GR 97 P 8026 estimation represents additional information for data detection.

High-resolution direction estimation methods are advantageously used to determine the direction of arrival. For example, a MUSIC method (Multiple Signal Classification) or a single-dimensional or multi-dimensional UNITARY ESPRIT method (Estimation of Signal Parameters via Rotational Invariance Techniques) carry out a high-precision direction estimation, building on the channel impulse responses, with economic computation complexity. The MUSIC or ESPRIT methods use knowledge from the complex radiation polar diagram of the receiving sensors and/or specific geometric preconditions for the arrangement of the receiving sensors, in order to carry out an accurate direction estimation that involves little signal processing complexity.

In order to determine the directions of arrival for wave elements, the specific values are averaged over a time interval, according to a further refinement of the invention. The direction of arrival varies little within a time interval which may correspond to a multiple of the coherence time of the channel impulse responses. Averaging improves the direction estimation, since random errors are reduced. In the case of radio-block transmission of subscriber signals in TDMA systems, the averaging can be carried out for a radio block or else for a large number of radio blocks. The number of radio blocks for averaging, that is to say the time interval, may in this case be varied, with changes in the directions of arrival leading to a change in the time interval. If the channel conditions change quickly, for example if the movement of a mobile station speeds up, then the direction estimation can be limited to a shorter time interval.

In addition to the applications explained above in mobile radio systems, further advantageous applications are envisaged in a radar GR 97 P 8026 6 or sonar system or in a seismic measurement system. In the case of the latter applications, the at least one subscriber signal may also arrive at the receiving station in the form of one or more reflected wave elements.

The invention will be explained in more detail in the following text with reference to two exemplary embodiments, in which direction estimation for data detection in mobile radio systems is used on the one hand, and a radar system carries out a direction estimation for a missile on the other hand, with reference to illustrations in the drawing, in which: FIG 1 shows a block diagram of a mobile radio network, FIG 2 shows a block diagram of a frame structure of the radio blocks for the radio interface, FIG 3 shows a block diagram of a receiving device having associated receiving sensors, FIG 4 shows a block diagram of a directional channel estimator, FIG 5 shows a block diagram of a detection device, FIG 6 shows a schematic illustration of a radar scenario, and FIG 7 shows a block diagram of a receiving device for the radar system.

The first exemplary embodiment will be explained with reference to FIGS 1 to GR 97 P 8026 7 The structure of the mobile communications system illustrated in FIG 1 corresponds to that of a known GSM mobile radio network which comprises a large number of mobile switching centers MSC, which are networked with one another and produce access to a fixed network PSTN.

Furthermore, these mobile switching centers MSC are connected to in each case at least one base station controller BSC. Each base station controller BSC in turn allows a connection to at least one base station BS. Such a base station BS is a radio station which can set up an information link to mobile stations MS via a radio interface. By way of example, FIG 1 shows three radio links between three mobile stations MS and one base station BS. An operation and maintenance center OMC provides monitoring and maintenance functions for the mobile radio network, or for parts of it. This structure can be transferred to other mobile radio networks in which the invention may be used.

The communication links between the base station BS and the mobile stations MS are subject to multipath propagation, which are caused by reflections, for example on buildings or vegetation, in addition to the direct propagation path. If it is assumed that the mobile stations MS are moving, then multipath propagation together with other. interference leads to the signal components of the various propagation paths of a subscriber signal being superimposed as a function of time at the receiving base station BS. It is also assumed that the subscriber signals from different mobile stations MS are superimposed at the receiving location to form a received signal e, em. The function of the receiving base station BS is to detect data d transmitted in the subscriber signals, and to assign such data d to individual, subscriber-specific communication links.

FIG 2 shows the transmission of the subscriber signals via the radio interface. The radio interface in this case has GR 97 P 8026 8 a frequency-division multiplex (FDMA), a time-division multiplex (TDMA) and a code-division multiplex component (CDMA). A number of frequency bands are provided along the frequency axis f for the mobile radio network. Furthermore, the time axis t is subdivided into a time frame comprising a plurality of time slots per time frame, in such a manner that transmission takes place in radio blocks. The subscriber signals from a plurality of mobile stations MS are assigned to a subscriber group Tlnl, Tln2 Tlnl20, that is to say, during the radio block of a subscriber group, for example the subscriber group Tln3 for the three mobile stations MS in FIG 1 that is to say K 3 in the exemplary embodiment, the subscriber signals which are denoted by different subscriber codes are superimposed to form a received signal e, em, which can be evaluated by a receiving device in the base station BS.

Within a radio block, a subscriber signal comprises two data-carrying sections with data d, in the center of which a subscriber-specific training sequence tseql to tseqK is introduced. The radio block is terminated by a guard time gp. The subscriber signals are distinguished by a subscriber code c, as a result of which [lacuna] within the data-carrying sections by subscriber-specific fine structures which are defined by the subscriber-specific CDMA codes c k, k The subscriber signals can be separated by means of these CDMA codes c (which will be referred to as subscriber codes in the following text), which are known at the receiving end.

FIG 3 shows a receiving device having associated receiving sensors A. This receiving device is part of the base station BS and receives received signals e, em from the transmitting mobile stations

MS

in the mobile radio network. The following text describes a situation when the base station BS is receiving although, nevertheless, there is normally a GR 97 P 8026 two-way communication link, that is to say the base station BS also has a transmitting device.

GR 97 P 8026 -9 Ka 4 receiving sensors A form an antenna device, which is designed as an intelligent antenna device, that is to say a plurality of receiving sensors A in this intelligent antenna device receive received signals e or em at the same time, which signals are combined with one another in such a way that the transmission quality is improved in comparison with systems having one receiving antenna.

Digital signals are produced from the received signals e, em, for example by conversion to baseband and subsequent analog/digital conversion, and these digital signals are evaluated in the receiving device.

The receiving device comprises a plurality of channel estimators JCE, a plurality of direction estimators DOAE, a directional channel estimator JDCE, and a detection device JDD. In addition to the received signals e, em, the receiving device has a-priori information about the number K of subscribers, their training sequences tseql,..,tseqK and their subscriber code c, and information relating to interference signals may also possibly be available.

The already digitized received signals em from the receiving sensors A are supplied to the channel estimators JCE. The channel estimators JCE are used to determine the non-directional channel impulse responses g by means of a Gauss-Markov estimate, or a maximum-likelihood estimate. The received signal from a receiving sensor A is evaluated per channel estimator JCE, with K non-directional channel impulse responses g being produced respectively at the outputs of the channel estimators JCE. These non-directional channel impulse responses g are calculated from the received signals ka) signals emka), ka 1..Ka, which are caused by the training sequences tseql to tseqK of the K subscriber signals.

GR 97 P 8026 i0 The non-directional channel impulse responses g are each supplied to the K direction estimators

DOAE

which carry out a direction estimation related to subscribers, based on these non-directional channel impulse responses g. The number of directions of arrival determined per subscriber signal is denoted by Kd. This number Kd may differ from subscriber signal to subscriber signal. The single-dimensional or multi-dimensional UNITARY ESPRIT algorithm is used to determine the directions of arrival DOA (also called the incidence directions). The direction estimation according to the invention is carried out in the direction estimators

DOAE.

In the directional channel estimator JDCE, the received signals em ka caused by the training sequences tseql to tseqK in the receiving sensors A and the specific directions of arrival DOA of the wave elements are processed, and directional channel impulse responses h are determined from them. This channel estimation is based on the maximum-likelihood estimation method.

(ka) Finally, the Ka received signals e(ka) ka 1..Ka, the specific directional channel impulse responses h and the specific directions of arrival DOA are supplied to the detection device JDD, which also processes the subscriber codes c and additional a-priori information known about the direction of arrival of interference signals in the form of Rn or the geographical position of mobile stations MS with respect to the base station BS.

The detection of the data d is carried out in this detection device JDD, based on the received signals e(ka) caused by the sections which carry data. A zero-forcing method is used for this purpose.

Alternative advantageous methods are the maximum-likelihood estimation or an MMSE method. The data detection results in the GR 97 P 8026 11 detected data d for the K subscriber signals for a radio block being present at the outputs of the detection device JDD.

For analysis (according to the method) of the data detection, channel estimation of channel impulse responses g without taking into account direction inhomogeneities is carried out in a first method step.

In a second step, the directions of arrival DOA of one or more wave elements are determined from the determined channel impulse responses g and then, in a third step, directional channel impulse responses h, that is to say channel impulse responses which can be associated with different directions of arrival, are determined from the received signals, taking account of the directions of arrival DOA. This step is based on the knowledge that each of the conventional, non-directional channel impulse responses g(k (ka) are obtained by superimposition of Kd directional channel impulse responses h(k (ka) where k 1..K and kk 1..Ka.

It can thus be said that: Kd 2 hkd l where k K and ka 1..Ka.

(1) In this case, a(k (ka) (kd) are complex weighting factors for superimposition of the directional channel impulse responses h(k)ka) on the non-directional channel impulse responses g(k) (ka) Knowledge about the directions of arrival or correlation matrices of interference wave elements can also be used, if appropriate, to determine the directional channel impulse responses.

The total number W*K*Ka of parameters to be estimated in gK k ka 1..Ka, in the case of multi-antenna systems is normally considerably greater than the total number W*K*Kd of parameters to OrS z GR 97 P 8026 -h be estimated in h W) (kd) ,k 1. kd 1 .Kd, since Ka Kcl. The computation complexity for GR 97 P 8026 12 estimation of parameters using the method according to the invention can thus be reduced.

During reception of a combined received signal em, which is advantageously caused by the training sequences of the subscriber signals and contains the receied sinals ka) received signals em ka 1..Ka of the Ka receiving sensors, this received signal em is in the form: em G-h nm (2) where G is the known matrix (L*Ka)x(W*K*Kd), with L denoting the number of samples of the received signal em at discrete times, and W denoting the length of the channel impulse responses. This matrix G results from the geometrical arrangement and the complex characteristics of the Ka receiving sensors, from the transmitted training sequences, and from the Kd directions of arrival DOA. The vector h includes the time-discrete baseband equivalent of the K*Kd directional channel impulse responses h(K) (Kd) nm denotes an unknown (L*Ka) column vector of a time-discrete interference signal.

G and em are known from equation so that the directional channel impulse responses h can be determined.

During the sections which carry data, the combined received signal e of the received signals e(ka) of the receiving sensors is in the form: e A-d n. (3) where A is an (M*Ka)x(N*K) matrix in which M is the number of discrete times at which the received signal is sampled and N is the number of data symbols transmitted per subscriber. n is once again an unknown (M*Ka) column vector of a time-discrete interference signal.

GR 97 P 8026 13 In equation A from the K*Kd directions of arrival, the directional channel impulse responses h, the geometric arrangement and the complex characteristics of the receiving sensors and, when using CDMA subscriber separation, from the subscriber code that is used and e are known, so that the data d can be detected.

In a fourth method step, received signals e caused by those sections of the K subscriber signals which carry data are used to detect the data d, using the previously determined directions of arrival DOA and the directional channel impulse responses h. Knowledge relating to directions of arrival, the power, the spectrum or the covariance matrix of interference signals may also be used, if required, in this step.

The directional channel impulse responses h are advantageously determined using the Gauss-Markov estimation method, in which case an estimated value for the directional channel impulse responses h can be calculated from:

(G

T

Rn-) G*T Rn em (4) Rn denotes the covariance matrix of the interference signal nm, which is governed by the directions of arrival and relative power levels of the interference wave elements, the spectra of the interference signals, the geometric arrangement, and the complex radiation polar diagram of the receiving sensors. This method corresponds to the maximum-likelihood estimation of the directional channel impulse responses h, and can be achieved with reasonable complexity by recursive resolution of (4) Relationships between the direction estimation and the determination of the directional channel impulse responses and the data detection are used as follows. The K subscriber GR 97 P 8026 14 signals comprise sections carrying data and training sequences, in which case the received signals caused by the training sequences of the K subscriber signals are used to determine the directional channel impulse responses, and the data are detected from the received signals caused by the sections which carry data.

The complexity can also be reduced by once again determining the directions of arrival DOA and/or the directional channel impulse responses h using a follow-up method of the one period which is longer than a frame structure related to radio blocks.

FIG 4 shows a directional channel estimator JDCE, which contains beamformers BF which weight each of the Ka received signals em(ka) by beamformer-specific weighting factors wl to w4 and w5 to w8, and add up the signal components in an addition device S to form a signal for which the signal-to-noise ratio is maximized, with this signal subsequently being supplied to a decorrelating filter DMF matched to the signal.

The self-interference SI and cross-interference CI is compensated for in an interference-cancellation device IC, and directional channel impulse responses h are obtained.

The information relating to the directions of arrival DOA of the wave elements and the directions and relative power levels of the interference wave elements are also processed in the beamformers BF. These directions influence the weighting factors wl to w4 and to w8 for each beamformer BF individually. The beamformers BF and the decorrelating filters DMF which are matched to the signal act like a spatially resolving decorrelating filter which is matched to the signal, and are each applied to one wave element that is to say K*Kd.

FIG 5 shows the detection device JDD. This detection device JDD processes those GR 97 P 8026 15 sections of the received signals e which carry data, in which case a spatially resolving decorrelating filter which is matched to the signal superimposes the K*Kd wave elements of the received signals e, in a manner corresponding to the described procedure, in the directional channel estimator JDCE, in order to maximize the signal-to-noise ratio. This process of maximizing the signal-to-noise ratio is carried out for each direction of arrival DOA of each subscriber signal, with the Kd signal components of the individual wave elements of a subscriber signal being superimposed using the maximum ratio combining method in addition devices S1 to SK.

The subscriber signals are then supplied to an interference-cancellation device IC, which compensates for the intersymbol interference ISI and multiple access interference MAI. The information relating to the subscriber codes c, the directions of arrival

DOA,

the directional channel impulse responses h and, if appropriate, a-priori knowledge relating to the interference sources in the form Rn is also processed in this case. The detected data d for the subscriber signals are present in separated form at one output of the interference-cancellation device IC. A so-called

JD

(Joint Detection) method is used for interference cancellation.

The receiving device reduces the time dispersion and the variance of the received signals.

Furthermore, spatial resolution allows a greater number of mobile stations MS to be supplied in a radio area of a base station BS, and the radio area can be shaped by the directional effect such that the transmission power levels of the mobile stations MS are also considerably reduced.

FIG 6 shows a second exemplary embodiment, using a radar system scenario.

GR 97 P 8026 15a A radar system comprises a transmitting device and a receiving device, which are connected via a coupling element K to an GR 97 P 8026 16 antenna device A having Ka receiving sensors. During a transmission period, a transmission signal which is provided with a transmitter-specific fine structure is transmitted via the antenna device. A portion of the transmitted energy reaches a missile, where it is reflected.

The reflected signal, called the subscriber signal in the following text, comprises a large number of wave elements, with Kd wave elements reaching the antenna device A via different propagation paths. The wave elements of the subscriber signal are thrown back to the radar system by refraction, diffraction and reflection on clouds or buildings, and are received and evaluated in the receiving device.

FIG 7 shows, schematically, the receiving device with which the Ka receiving sensors A are associated. A channel impulse response g is estimated, related to sensors, in Ka channel estimators KS. For this purpose, the correlation of the known transmitter-specific fine structure with the received signal is determined in the radar system.

A direction estimator DOAS is supplied with the Ka channel impulse responses g determined in this way, and uses them to determine these Kd directions of arrival DOA for the subscriber signal. The methods used for channel and direction estimation correspond to those in the first exemplary embodiment.

The channel impulse responses g and the directions of arrival DOA of the wave elements of the subscriber signal are evaluated in a signal combiner SC in the radar system in order to determine the position from the directions of arrival DOA, the signal delay time and the received field strength of the wave elements and the speed of flight from the Doppler frequency.

Claims

2. The method as claimed in claim i, in which the channel impulse responses are determined from training sequences (tseql, tseq2, tseqK) of the subscriber signals, which training sequences (tseql, tseq2, tseqK) form the transmitter-specific fine structures.
3. The method as claimed in claim 1 or 2, in which subscriber signals from a plurality of transmitters (MS) or reflectors (P1, P2) arrive superimposed, to form the received signals, at the receiving device, in which case these signals are transmitted at the same time in one frequency channel.
4. The method as claimed in one of the preceding claims, in which the determination of the directions of arrival (DOA) of the wave elements also takes account of information relating at least to one of the following values: direction of arrival (DOA), power, spectrum or a correlation matrix of interference signals. 1 7 ;3S -18- The method as claimed in claim 3 or 4, in which the subscriber signals can be separated by despreading using individual subscriber codes (tcl, tc2, tcK).
6. The method as claimed in one of the preceding claims, in which high-resolution direction estimation methods are used to determine the direction of arrival (DOA).
7. The method as claimed in claim 6, in which a MUSIC method is used to determine the direction of arrival (DOA).
8. The method as claimed in claim 6, in which a single-dimensional or multi- dimensional UNITARY-ESPRIT method is used to determine the direction of arrival (DOA).
9. The method as claimed in one of the preceding claims, in which the specific values are averaged over a time interval in order to determine the directions of arrival (DOA) of the wave elements.
10. The method as claimed in one of the preceding claims, for use in a mobile radio system.
11. The method as claimed in one of the preceding claims, for use in radar or sonar system.
12. The method as claimed in one of the preceding claims, for use in a seismic measurement system.
13. A method for direction estimation for wave elements of at least one subscriber signal, substantially as herein described with reference to any one of the embodiments as illustrated in Figs. 1 to 7. DATED this Eighth Day of September 1999 Siemens Aktiengesellschaft N Patent Attorneys for the Applicant n SPRUSON FERGUSON [R:\LIBPP]01708.doc:iad