GB2365695A

GB2365695A - Cellular communications systems

Info

Publication number: GB2365695A
Application number: GB0105366A
Authority: GB
Inventors: Stephen William Wales
Original assignee: Roke Manor Research Ltd
Current assignee: Roke Manor Research Ltd
Priority date: 2000-03-03
Filing date: 2001-03-05
Publication date: 2002-02-20
Anticipated expiration: 2021-03-05
Also published as: GB0105366D0; GB2365695B; GB0005120D0

Abstract

A cellular communication systems, in particular a system for third generation mobile communications, makes use of the application of adaptive antennas at base stations in the UMTS Terrestrial Radio Access system for obtaining capacity advantages. Intra-cell interference can be substantially eliminated via joint detection, whilst inter-cell interference can be eliminated though exploitation of the spatial interference of inter-cell interfering sources. Several receiver structures are presented.

Description

<Desc/Clms Page number 1> IMPROVEMENTS IN OR RELATING TO CELLULAR COMMUNICATIONS SYSTEMS Field of the Invention The present invention relates to cellular communications systems. In particular, the present invention relates an adaptive antenna arrangement with improved carrier to interference performance.

Backgound to the Invention Mobile telephony using cellular radio systems is now fairly well established new cellular systems are being developed to meet increasing capacity demands, not only through an increase in subscribers and the number of calls made, but also. because of the amount of data, with text messaging and internet access becoming common place. One third generation system which is presently being developed is, the UMTS Terrestrial Radio Access a (UTRA) system and comprises two modes of operation, distinguished by the duplex technique employed, either frequency division duplex (FDD) or the time division duplex mode (TDD). The TDD mode employs both time division and code division multiple access, and shares many common features with FDD mode. The two modes are intended to be complimentary in their application, whilst the FDD mode is primarily intended for macro-cell and micro-cell applications, the TDD mode is primarily intended for macro-cell and pico-cell applications. Inter-cell interference will, however, limit the ability to fully reuse time slots in neighbouring cells and intra-cell interference limits the extent of use of the orthogonol code space.

The access method used in the TDD mode is a hybrid of time division multiple access and code division multiple access (TD-CDMA). In the time domain lOms frames are defined (in common with FDD) and each frame divided into 15 time slots. The division of time slots between the uplink and downlink is arbitrary allowing a different proportion of the total capacity in each direction. With data services expected to dominate traffic the ability of the TDD mode to handle asymmetrical capacity demands efficiently is considered to be an important feature of the system. Within each time slot code division multiple access is also possible. Code division is effected by direct sequence spreading using orthogonal codes with varying spreading factor. Spreading factors of 1, 2, 4, 8 and 16 are possible and a cell unique scrambling sequence is also applied to the signal. An example of the time slot structure used in the TDD mode is shown in Figure 1 where the midamble is used to facilitate channel estimation in the receiver.

In the development of the TDD standard a multi-user detection technique was originally considered for reception of signals in both base and mobile station a receivers with zero-forcing or minimum mean square error (MMSE) equalisation" techniques applied to a single digitally modulated and distorted signal whereby to apply more than one direct sequence spread spectrum signal. The technique relies upon estimating the channel impulse response for each user, and the midamble structure has been designed to facilitate this. The channel estimate for all users can be obtained with little additional processing over that required to estimate the channel, assuming a single user is present. A block diagram of the multi-user detector is given in Figure 2.

One prior-art receiver structure for an uplink receiver comprises a diversity combining arrangement in conjunction with a multi-user detector. A block diagram

of this receiver is shown in Figure 3. With NA antenna elements, a channel estimate is obtained for each user signal, from each antenna (hk") n E [o, N,, - ilk E [o, x - 1]). The multi-user detector constructs a matrix representing the expected output of the channel in the absence of any data modulation for the k-th user from one antenna (B(") ). This is formed from the convolution of the channel impulse response with the spreading sequence of each user (Ck ),giving rise to the K Q+W-1 column vectors b (k") _ Ck * hk"> , where W is the duration of the channel impulse response in chip intervals. No particular use is made of the spatial separation of either inter-cell or intra-cell interfering signals.

f In the multi-user detector on the uplink, the knowledge of each spreading code used in the cell, together with an estimate of the channel impulse response th each mobile station facilitates the combating of intra-cell interference. On the ' downlink the mobile station can estimate the channel impulse response for alt signals present and, by determining which codes are active, configure itself to- detect only active codes. Object to the invention The present invention seeks to provide an improved wireless communications cellular base station. The present invention further seeks to provide an antenna arrangement operable to increase the capacity of a cellular communications base station.

Statement of the Invention In accordance with a first aspect of the present invention there is provided a TDD base station comprising an adaptive antenna array operable to provide vector

signals; a signal weighting unit operable to receive said vector signals and to apply adaptation weights to said vector signals; a beam former operable to receive signals from the weighting unit and provide channel impulse response signal to a multi- user detector, wherein the beam forming network is operable to suppress inter-cell interference by exploiting the angular distribution of this interference and to direct nulls in the antenna response toward the interfering signals; and wherein the multi- user detector is operable to receive the channel impulse response signals and the received signals to produce estimates of the transmitted signal data sequence.

Conveniently the weights of the beam forming networks are obtained by applying the Wiener-Hopf technique to the received signal samples of the mid- amble sequence.

Conveniently the weights of the beam-forming network (wnk' ) can by obtained from the following minimization:

where S; is the modulated version of the mid-amble sequence.

The beamforming can be performed upon the time delay component with the greatest magnitude, for example. Other beamforming networks can be synthesised that also take into account the time dispersion of a channel.

The channel impulse response estimates pertaining to the K signals from the beam forming networks can be obtained from the following transform:

Accordingly the present invention provides a system which can overcome the requirement in third generation systems to avoid frequency, time slot or code planning restrictions. The present invention allows the operation of an unplanned network with the restrictions imposed by inter-cell interference being overcome by the present invention where a dynamic channel allocation mechanism.

In accordance with a second aspect of the present invention there is provided

a method of operating a TDD base station comprising the steps of: receiving vecto-r signals from an adaptive antenna array; receiving said vector signals by a signal weighting unit and applying ada adaptation weights to said vector signals; receiving g signals from the weighting unit to a beam former and providing channel impulse response signal to a multi-user detector, whereby the beam forming network suppresses inter-cell interference by exploiting the angular distribution of thig r interference and directs nulls in the antenna response toward the interfering signals;; and, receiving the channel impulse response signals and the received signals by the multi-user detector whereby to produce estimates of the transmitted signal data sequence.

The present invention can thus overcome the problems arising in complexity due to the inclusion of inter-cell interference within a multi-user detector. Through improved signal to interference ratios, the present invention can allow a more efficient use of the orthogonal code space and can allow time slots to be reused over a greater proportion of a cell area.

Brief description of the Fi res The invention may be understood more readily, and various other aspects and features of the invention may become apparent from consideration of the following description and the Figures as shown in the accompanying drawing sheets, wherein: Figure 1 shows an example of a time slot structure used in a TDD protocol; Figure 2 shows a block diagram of a multi-user detector; Figure 3 shows a prior-art receiver; Figure 4 shows a receiver made in accordance with the invention; Figure 5 shows a detailed view of the multi-user detector arrangement;

Figure 6 shows comparative performance obtained by diversity combining ill . conjunction with multi-user detection relative to adaptive beamforming with` respect to intra-cell interference; Figure 7 shows comparative performance obtained by diversity combining in. . conjunction with multi-user detection relative to adaptive beamforming with respect to inter-cell interference; Figure 8 shows comparative performance obtained by single user detection and multi-user detection in conjunction with adaptive bearnforming with respect to intra-cell interference; and, Figure 9 shows comparative performance obtained by single user detection and multi-user detection in conjunction with adaptive beamforming with respect to inter-cell interference.

Detailed description of the invention There will now be described, by way of example, the best mode contemplated by the inventors for carrying out the invention. In the following

description, numerous specific details are set out in order to provide a complete understanding of the present invention. It will be apparent, however, to those skilled in the art, that the present invention may be put into practise with variations of this specific.

Referring now to Figure 4, there is shown an embodiment of the invention, wherein received signals r; and r, - rNA from an adaptive antenna system (not shown) are applied to a set of beamforming networks 401 and a set of weight calculation units 402. The input signals are combined in a multi-user detector 404 in the same manner as for diversity combining. The weight calculation units

provide the weights for the beamforming networks. The weights are calculated by-, minimising a mean squared error between the received version of a know41 sequence and a locally generated version supplied by the units 403. Channel impulse response estimates h;(') - h;(`) pertaining to each of the antennas are also, generated by the unit 405 and passed to each of the beamforming networks 404 for the purpose of providing a transformed set of channel impulse response estimates to the multi-user detector 404. Figure 5 shows a more detailed block diagram of thj unit 404. In Figure 5 the transformed channel impulse response estimates are applied to a set of convolution units 501 that convolve the channel impulse response estimates with each of the channelisation codes. A set of system matrices are then constructed in the units 503 that relate to the expected output of a matched filter in the absence of data modulation. System matrices are calculated that relate to the output of each of the beamforming networks and these are summed together to produce the overall system matrix as input to the unit 504. The received signal samples from each of the beamforming networks are applied to a set of matched filters 502. The outputs from each of the matched filters are summed to produce an input to an equation solver unit 505. Estimates of the data symbols for each user are

obtained from the second equation solver unit 506. By combining the outputs from several beamforming networks additional suppression of inter-cell interference is obtained by the exploitation of the spatial distribution of this interference over diversity combining. With regard to intra-cell interference suppression the combination of the outputs from each of the diversity combining eliminates most of the spatial suppression of this interference due to the presence of the beamforming networks; but a diversity gain is obtained, and multi-user detection provides suppression of the intra-cell interference.

By employing an adaptive antenna at the base station it is possible, through

improved signal to interference ratios, to allow more of the orthogonal code spaci; to be used and to allow time slots to be reused over a greater proportion of the cell` area than would otherwise be the case.. 4 It should be noted that any improvement in the uplink capacity through greater reuse of time slots, and greater use of the orthogonal code space can, benefit the downlink, particularly where there is an asymmetry in the traffic levels: If the uplink can carry more traffic per time slot, but the uplink capacity requirement is low; then in a TDD system the spare time-slots can be transferred to the downlink. Conveniently, the weights of the beam forming networks can be obtained by applying the Wiener-Hopf technique to the received signal samples of the mid-amble sequence. The weights of the beam-forming network (w,(,") can be obtained from the following minimization:

where S; is the modulated version of the mid-amble sequence. Beamforming would be performed upon the time delay component with the greatest magnitude.

The K signals obtained from the outputs of the beam forming networks are then applied to the multi-user detector. Associated with each of the K signals are a set of K channel impulse response estimates. These channel impulse response estimates are obtained by first estimating the channel impulse response pertaining to the signal from each user and each antenna element. Denoting this estimate as hnk', the channel impulse response estimates pertaining to the K signals from the, beam forming networks are obtained from the following transform:

The multi-user detector constructs a matrix representing the expected output of the channel in the absence of any data modulation for the k-th user, from one beam-forming network antenna (B("). This is formed from the convolution of the channel impulse response with the spreading sequence of each user (ck.) k' E= [o, K -1] ), giving rise to the K Q+ W-1 column vectors bkk' = c,, * b('', where W is the duration of the channel impulse response in chip intervals. The matrix B 'k' is then constructed in the following manner:

boko 0 ... 0 bold) 0 ... 0 b(k) ... b(k) ... . 0,0 0,1 b (k) ... 0 b(k) ... 0 Q+W-1,0 Q+W -1,1 0 b (k) ... 0 b (k) ... B (k) Q+W-1,0 Q+W-1,1 . 0 ... 0 0 ... 0 0 b0(k) 0 ... bokK-1 0 0 ... bQ+W-1,0 p 0 ... hQ+W-l,K-1 where o is a column vector of dimension Q and containing all zeros. The equation B(k'd + n = r (k' can be formed, where r (k' and n are both NQ+W-I dimensional vectors containing the received samples from the n-th beam forming network and noise plus residual inter-cell-interference samples respectively. Q I,.,- the spreading factor, and d is a K N dimensional vector containing the data symbol values for a block of N data symbols for all users. To enable the NA antenna system d to be represented the following column concatenations are performed on the matrix'

and vector

are formed. A minimum mean; square error (MMSE) solution for the linear data symbol estimates can be obtained from d = (BH R-'1'B + Raa rl B"R_,1, r where the matrix Rnn is the covariance matrix of the noise-plus-residual inter- cell-interference, and Rdd is the covariance matrix of the data symbols. If only additive white noise is present then the matrix kn becomes a 21, where a 2 is the noise variance. Alternatively, the zero-forcing solution is obtained from:

d - (BHRnrtBr' BHRnnr .

To examine the performance of the receiver structures described previously a software model has been developed where K-user signals are generated that are randomly distributed in azimuth. To model the effects of the propagation environment the signals are applied to a space-time channel model. The signal is received by an antenna array and Gaussian noise added to the received signal. Inter-cell interference is modelled by generating a second signal composed of K- user signals that are randomly distributed in azimuth. The space-time channel model disperses the signal in both the time domain and in angle of arrival. The model is based on that developed in the TSUNAMI research programme and is purely statistical. The model describes the average power as a function of angle of arrival (azimuth only) and time delay. The average power profile is given by

as and a, are the azimuth spread and delay spread respectively. The channel model is developed by generating randomly phased f components at delays equal to multiples of the chip interval and with angles of arrival that are generated randomly with probability

where a is the mean angle of arrival that is itself uniformly distributed in [o,2f ]. The power of each component is set according to the average power delay profile. Simulation results have been obtained with a circular array of 8 elements, spaced by half a wavelength. The spreading factor used was set to 4, and in all the simulations 4 users with equal power are present in both the wanted signal and any inter-cell interfering signal. Consequently the .entire orthogonal code space is used. Different randomly generated scrambling codes are applied to the wanted and inter-

cell interfering signals, and these change on a frame by frame basis. The azimuthal spread is set to 1 degree and the time delay spread to 0.362s.

In the results of Figure 6 the performance obtained by diversity combining in conjunction with multi-user detection (curves labelled C-MUD) and adaptive beamforming in conjunction with a single multi-user detector (curves labelled SA- SMUD) is compared. For comparison purposes, results for the diversity combining multi-user detector a receiver with two widely separated antennas are also shown. This receiver is considered as the standard base station receiver for UTRA TDD. The results of Figure 6 show that with closely coupled antennas and a low value for

the azimuth spread a 4dB improvement results for an 8-element array with diversity combining over the 2-element receiver with widely separated antennas. Adaptive L beamforming offers little advantage under these circumstances. The results of Figure 7 where inter-cell interference in addition to intra-cell interference is, S present shows that adaptive beamforming offers significant performance' improvements under these circumstances over diversity combining. Here the spatial, a separation of inter-cell interferers is being exploited by the array. Adaptive beamforming provides an additional 7.5dB tolerance to inter-cell interference over a receiver employing diversity combining.

The use of single user detection and multi-user detection in conjunction with adaptive beamforming has also been examined. In the results of Figure 8 adaptive beamforming is combined with single user-detection (SA-SUD), a single-multi- user detector and.also multiple multi-user detectors (SA-MMUD). In the multiple multi-user detector the receiver architecture each beam-forming network provides inputs to a multi-user detector, thus no combining of the outputs of the K beam- forming networks takes place. Post detection combining of the K sets of K outputs

could also be performed in this structure, but only one output is taken from each multi-user detector. Practically this architecture is complex, but the simulated results demonstrate the performance loss experienced by the single user detector, where spatial separation of intra-cell interfering signals does not occur. The results of Figure 8 show the situation where only intra-cell interference is present, and the adaptive beamforming single multi-user detector performs better than the single user detector with adaptive beamforming, by about 5dB at a bit error rate of 5%. The single multi-user detector with adaptive beamforming also performs better than the multiple multi-user detector with adaptive beamforming by about 1 dB at a bit error rate of 5%. This is believed to be due to the combining used in the single multi-user detector. The results of Figure 9 show the situation where inter-cell interference is present, and there is little difference in the performance of the three receiver structures, because essentially the dominant source of degradation is being dealt with in a very similar way in each of the receiver structures. If the signal to noise ratio of the wanted signal were lowered, then there would be a greater performance difference between the different receiver structures. Simulations have shown that adaptive beamforming on the uplink can lead to significant improvements in the inter-cell interference performance of UTRA TDD systems, and also allow the complete orthogonal code space to be used. Results shown here indicate that when only intra-cell interference is present, simple diversity combining of the received signal in conjunction with multi-user detection provides similar performance to the case where individual beams are formed for each user and the outputs supplied to a multi-user detector. This is achieved even with relatively low azimuth spreads where the correlation between antenna outputs would be relatively high. With inter-cell interference present the process of adaptive beamforming offers significant advantages over diversity combining.

Inter-cell interference generated by mobile stations in neighbouring cells inhibits the ability to reuse time slots in neighbouring cells. With power control applied to both uplink and downlink transmissions time slots could be reused in neighbouring cells provided all mobiles transmitting in a time slot are a certain fraction of the cell radius from the base station. The precise fraction of the cell radius over which time slot reuse can occur is dependent upon the propagation environment, but could easily be 50% of the cell radius. In third generation systems there is a requirement to avoid frequency, time slot or code planning. To allow an unplanned network, taking into account the restrictions imposed by inter-cell interference, the present invention provides a dynamic channel allocation mechanism whereby use is made of the mobile station interference measurements that are reported to the base station. This allows time slots to be reused freely when interference conditions permit, although capacity can be lost through restrictions on time slot reuse.

Claims

Claims. 1. A TDD base station arrangement comprising an adaptive antenna array operable to provide vector signals; a signal weighting unit operable to receive said vector signals and to apply adaptation weights to said vector signals; a beam former operable to receive signals from the weighting unit and provide channel impulse response signal to a multi-user detector, wherein the beam forming network is operable to suppress inter-cell interference by exploiting the angular distribution of this interference and to direct nulls in the antenna response toward the interfering signals; and wherein the multi-user detector is operable to receive the channel impulse response signals and the received signals to produce estimates of the transmitted signal data sequence.
2 A base station arrangement according to claim 1, wherein the weights of the beam forming networks are obtained by applying the Wiener-Hopf technique to the received signal samples of the mid-amble sequence.
3 A base station arrangement according to claim 1, wherein the weights of the beam-forming network (wnk') can be obtained from the following minimization:

where S; is the modulated version of the mid-amble sequence.

<Desc/Clms Page number 16>
4 A base station arrangement according to claim 1 wherein the K signals obtained from the outputs of the beam forming networks are then applied to the multi-user detector.
5 A base station arrangement according to any one of claims 2 - 4, wherein the beamforming is performed upon a time delay component with the greatest magnitude.
6 A base station arrangement according to claim 1 wherein the multi- user detector operates upon a combined version of the outputs from the K beamforming networks.
7 A base station arrangement according to claim 1, wherein the channel impulse response estimates pertaining to the K signals from the K beam forming networks are obtained from the following transform:
8 A method of operating a TDD base station arrangement comprising the steps of: receiving vector signals from an adaptive antenna array; receiving said vector signals by a signal weighting unit and applying adaptation weights to said vector signals; receiving signals from the weighting unit to a beam former and providing channel impulse response signal to a multi-user detector, whereby the beam forming network suppresses inter-cell interference by exploiting the angular distribution of this interference and directs nulls in the antenna response toward the interfering signals; and,

<Desc/Clms Page number 17>

receiving the channel impulse response signals and the received signals by the multi-user detector whereby to produce estimates of the transmitted signal data sequence.
9 A method according to claim 8, wherein the weights of the beam forming networks are obtained by applying the Wiener-Hopf technique to the received signal samples of the mid-amble sequence.
10 A method according to claim 8, wherein the weights of the beam- forming network (w")) can be obtained from the following minimization:

where S; is the modulated version of the mid-amble sequence.
11 A method according to claim 8 wherein the K signals obtained from the outputs of the beam forming networks are then applied to the multi-user detector.
12 A method according to any one of claims 9 - 11, wherein the beamforming is performed upon the time delay component with the greatest magnitude.
13 A method according to claim 8 wherein the multi-user detector operates upon a combined version of the outputs from the K beamforming networks.

<Desc/Clms Page number 18>
14 A method according to claim 8, wherein the channel impulse response estimates pertaining to the K signals from the beam forming networks are obtained from the following transform: