GB2353181A

GB2353181A - Channel estimation in mobile communications networks

Info

Publication number: GB2353181A
Application number: GB9918858A
Authority: GB
Inventors: Behzad Mohebbi; Majid Boloorian
Original assignee: Fujitsu Ltd
Current assignee: Fujitsu Ltd
Priority date: 1999-08-10
Filing date: 1999-08-10
Publication date: 2001-02-14
Also published as: GB9918858D0

Abstract

In a cellular mobile communications network a group of mobile stations (20) transmit respective predetermined transmit signals to a base station (10) via different respective uplink channels such that those signals are received substantially simultaneously by the base station. The predetermined transmit signals all embody the same predetermined sequence of symbols but the embodied sequences are shifted in phase one from the next in the different signals by predetermined amounts. The base station includes a channel impulse response estimating unit (14) which employs the predetermined transmit signals as received by the base station and the predetermined sequence to produce estimates (UPEST<SB>1</SB> UPEST<SB>n</SB>) of the respective channel impulse responses of the different uplink channels. An estimation control unit (16) controls the operation of the channel impulse response estimating unit (14) such that a range of delay time values (e.g. W<SB>2</SB>) covered by the estimate produced for at least one channel is made greater than a range of delay time values (e.g. W<SB>1</SB>) covered by the estimate produced for another of the said channels. The range (W) for each channel may be set in dependence upon an expected maximum delay time for the channel concerned. In this way, more accurate estimates can be produced for "longer" channels and unnecessary calculations of insignificant elements of the estimates for "short" channels can be avoided. In another embodiment the ranges are set equally but dynamically in dependence upon the downlink-channel maximum delay times of the group measured in the mobile stations and reported back to the base station.

Description

2353181 CHANNEL ESTIMATION IN MOBILE COMMUNICATIONS NETWORKS The present

invention relates to channel estimation in mobile communications networks.

In a mobile communications network the characteristics of the transmission medium (channel) through which a radio signal is transmitted change continually. As a result, if the same signal is transmitted at different times from a particular transmitter to a particular receiver, the received signals will be different due to differences in the channel characteristics between the transmitter and the receiver over time. However, if the channel characteristics can somehow be measured or estimated at the receiver on a regular basis, information regarding the channel characteristics can be used to estimate the original transmitted signal. Clearly, the accuracy of the transmitted signal estimation depends on the accuracy of the measurement/estimation of the channel characteristics (channel estimation process).

when a signal is transmitted from a transmitter TX to a receiver RX in a mobile communications network the signal propagates along a number of different paths before reaching the receiver. For example, Figure 1 of the accompanying drawings shows a signal propagating along three different paths P1 to P3 between the transmitter TX and the receiver RX. The path PI is a direct (line of sight) path from the transmitter TX to the receiver RX. The paths P2 and P3, on the other hand, are indirect paths along which the signal is reflected by signal reflectors such as buildings or other objects 0 located between the transmitter TX and the receiver RX.

Because of the multipath propagation of the transmitted signal, the receiver RX detects replicas of the transmitted signal. These replicas add up vectorially at the receiver in such a way that the received signal may bear no resemblance to the original transmitted signal.

Each path P1 to P3 in Figure 1 may be described by three parameters, namely amplitude, phase and delay.

These parameters are random variables and change according to the relative motion between the transmitter, the receiver and the environment. When the transmitted signal is a single impulse TI as shown in Figure 2, three impulses RIj, R12 and R131 corresponding respectively to the different paths Pi to P3, are detected at the receiver RX. These impulses together represent the impulse response of the channel between the transmitter TX and the receiver RX.

is The second impulse R12 detected at the receiver RX is delayed by a delay time To relative to the first detected impulse RI,, and the third detected impulse R13 is delayed by a delay time. -c, (Tj > TO) relative to the first detected impulse RIj. Incidentally, only the amplitudes And delay times of the received impulses RI, to R13 are represented in Figure 2; phase is not shown for the sake of simplicity. The separation between the received impulses RI is dependent on the differences in length between the different paths P1 to P3 between the transmitter TX and receiver RX. The range of delay times is measured by a parameter referred to as "excess delay". In Figure 2, the excess delay is represented by the delay time T, between receipt of the f irst impulse RI, and the last impulse R13 - The aim of channel estimation is to obtain a good estimate of the channel impulse response. This is generally achieved at the receiver RX by detecting how the channel affects a transmitted signal known to the receiver. Conventional channel estimation techniques include correlation and matrix inversion techniques.

In both cases, the transmitted signal may be a known binary sequence. Also, both techniques are based on the fact that the received signal at the receiver is obtained by the convolution of the transmitted signal and the channel impulse response.

In mobile communications networks, a single receiver, for example in a base station of the network, may be required to process signals transmitted to it from a large number of different transmitters (mobile stations). Such a receiver must therefore be capable of carrying out channel estimation for the channel between it and.each different transmitter.. It is possible to estimate, in one timeslot, the channel impulse responses of two or more channels by arranging for the different transmitters to transmit phase shifted versions of the same known sequence to the receiver in the timeslot concerned.

For example, one mobile communications network currently under development by the European Telecommunications Standards Institute (ETSI) is referred to.as a UTRA TD-CDMA mode network. UTRA stands for UMTS Terrestrial Radio Access, UMTS stands for Universal Mobile Telecommunications System (a third generation mobile communications system) and TD stands for Time Division. The proposed UTRA TD-CDMA mode network is a CDMA/TDMA network using time division duplexing (TDD). TDMA stands for time-division multiple access. In such a TDD network, the available timeslots for two-way communication between a base station and a mobile station are divided into downlink and uplink timeslots. The downlink timeslots are used for transmission of user data and/or control information from the base station to the mobile station, and the uplink timeslots are used for transmission of user data and/or control information from the mobile station to the base station.

Effectively, such a TDD mode of operation still has respective uplink and downlink channels.

It is presently proposed that, in the UTRA TD-CDMA network, channel estimation be performed for up to sixteen users per timeslot per frequency in the uplink direction. The different transmitters (mobile stations) whose channel estimates are to be determined in the same timeslot are synchronised such that respective predetermined transmit signals thereof arrive substantially simultaneously at the receiver (base station). Within their respective predetermined transmit.signals the different transmitters are allocated respective phase-shifted versions of the same known binary sequence. The phase-shifts between the different transmitters (mobile stations) are equal to is one another and are set large enough to cater for the maximum expected excess delay of the "worst-case" channel. In practice, for example, the worst-case channel may be a channel whose transmitter (mobile station) is located at a position far away from the receiver (base station) such that there are a large number of relatively weak multipaths having a high excess delay.

The allocation to the different transmitters of equal phase-shifts determined by the worst-case maximum expected excess delay causes several problems. In particular, because it is designed to cater for the worst-case communication situation, the phase-shift allocated must be relatively high, which in turn limits the number of channels that can be estimated together in a particular timeslot. A portion of each timeslot allocated to channel estimation should be kept as small as possible, as otherwise the remaining portion of the timeslot available for transmission of user data and/or control information may be insufficient to maintain the required throughput for the channel concerned.

-s- According to a first aspect of the present invention, there is provided communications apparatus including: receiver means; and transmitter means for transmitting a plurality of predetermined transmit signals to the receiver means via different respective channels such that those signals are received substantially simultaneously by the receiver means, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts; wherein the receiver means include: channel impulse response estimating means for employing the predetermined transmit signals as received by the receiver means and the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and estimation control means operable to control the operation of the channel impulse response estimating means such that a range of delay time values covered by the estimate produced for at least one channel is made greater than a range of delay time values covered by the estimate produced for another of the said channels.

Such communications apparatus can take advantage of the fact that in practice the channel impulse responses of the different channels to be estimated are not of the same length. For example, a mobile station that is far away from its current serving base station will tend to have a channel impulse response longer than that of a mobile station that is near to the base station. By recognising this difference, communications apparatus embodying the present embodiment can set a range of delay times to be covered by the channel impulse response estimate for one channel that is different from the:range, of delay times to be covered by the channel impulse response estimate for at least one other channel. In this way, it is possible to provide more accurate estimates of the channel impulse responses for channels with relatively long channel impulse responses whilst, at the same time, avoiding performing unnecessary calculations of insignificant elements of the channel impulse responses of channels having relatively short channel impulse responses. As a result of the more accurate estimation of the channel impulse responses of the "longer" channels, a better link performance for those channels can be achieved. In addition, the valuable processing power at the receiver can be utilised to process useful rather than redundant information.

According to a second aspect of the present invention, there is provided a mobile communications network including communications apparatus embodying the first aspect of the invention, wherein: the said receiver means are provided in a base station of the network; the said transmitter means are provided collectively by a plurality of individual mobile stations; and the said different channels are respective uplink channels between the said mobile stations and the said base station.

According to the third aspect of the present invention, there is provided receiving apparatus including: receiving means operable to receive a plurality of predetermined transmit signals transmitted to the apparatus via different respective channels such that those signals are received substantially simultaneously by the apparatus, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts; channel impulse response estimating means operable;to employ the predetermined transmit signals as received by the apparatus and the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and estimation control means operable to control the operation of the channel impulse response estimation means such that a range of delay time values covered by the estimate produced for at least one of the said different channels is made greater than a range of delay time values covered by the estimate produced for another of the said channels.

According to a fourth aspect of the present invention, there is provided a channel impulse response estimation method for use in a receiver to which a plurality of predetermined transmit signals are transmitted via different respective channels such that those signals are received substantially simultaneously by the receiver, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts, in which method: the predetermined transmit signals as received by the receiver are employed with the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and a range of delay time values covered by the estimate produced for at least one of the said different channels is made greater than a range of delay time values covered by the estimate produced for another of those channels.

According to a fifth aspect of the present invention, there is provided a base station for use in a cellular communications network, including receiving apparatus embodying the third aspect of the invention.

According to a sixth aspect of the present invention, there is provided a mobile station, for use in a mobile communications network, including: signal transmission means for transmitting a predetermined transmit signal, embodying a predetermined sequence of symbols, to a base station of the network via an uplink channel provided by the network for the mobile station concerned, the transmission of the predetermined transmit signal being synchronised with the transmission of a further such predetermined transmit signal, also embodying the said predetermined sequence of symbols, to the receiver by another mobile station of the network such that-the respective signals of the two mobile stations are received substantially simultaneously by the receiver; and phase shift setting means operable to selectively adjust a predetermined is amount by which the embodied predetermined sequence included in the predetermined transmit signal transmitted by the said signal transmission means is shifted in phase relative to the predetermined sequence embodied in the further predetermined transmit signal transmitted by the other mobile station.

According to a seventh aspect of the present invention there is provided a cellular communications network including a base station and a plurality of mobile stations, each mobile station having at least one uplink channel and at least one downlink channel for communication with the said base station, wherein:

two or more of said mobile stations form a channel estimation group and transmit respective predetermined transmit signals to the base station such that those signals are received substantially simultaneously by the base station; each mobile station of the said group includes: downlink channel assessment means operable to produce a measure of at least one predetermined characteristic of the said downlink channel of the mobile station concerned; and control signal transmitting means operable to cause the mobile station to transmit to the base station a predetermined control signal dependent on the said measure produced; and the said base station includes: receiver means for receiving such predetermined control signals and s such predetermined transmit signals from the mobile stations of the said group; channel impulse response estimating means for employing the received predetermined transmit signals to produce estimates of the respective channel impulse responses of the uplink channels of the mobile stations of the said group; and estimation control means operable to employ the received predetermined control signals to control the operation of the said channel impulse response estimating means such that a range of delay time values is covered by the estimate produced for each mobile station of the group is set in dependence upon the said measures produced by the mobile stations of the group.

The ranges set for the mobile station of the group can be equal to one another, as well as unequal.

Reference will now be made, by way of example, to the accompanying drawings, in which:

Figure 1, discussed hereinbefore, is a schematic diagram for use in explaining multipath propagation in a mobile communications network; Figure 2, also discussed hereinbefore, is a schematic diagram for use in explaining a.channel impulse response in the Figure I network; Figure 3 is a schematic diagram for use in explaining a correlation method of performing channel estimation for a single channel; Figure 4 is a schematic diagram for illustrating the results of the correlation method of Figure 3; Figure 5 shows a schematic diagram, corresponding to Figure 4, relating to a correlation method of channel estimation for estimating plural channels; Figure 6 is a schematic diagram illustrating desired outputs of a correlator in a correlation method for estimating a single channel; Figure 7 is a schematic diagram for use in explaining how a predetermined transmit signal may be constituted to achieve the correlator outputs shown in Figure 6; Figure 8 is a table for use in explaining how the correlator outputs shown in Figure 6 are produced; Figure 9 is a schematic diagram showing desired outputs of a correlator in a correlation method for estimating two channels; Figure 10 is a schematic diagram for use in explaining how two predetermined transmit signals may be constituted to produce the desired correlator outputs shown in Figure 9; Figure 11 is a schematic diagram illustrating desired outputs of a correlator in a first embodiment of the present invention; Figure 12 is a schematic diagram for use in explaining how two predetermined transmit signals may be constituted to produce the desired correlator outputs shown in Figure 11; Figure 13 shows parts of a base station and a mobile station embodying the present invention; Figure 14 is a schematic diagram illustrating uplink and downlink signals transmitted by the Figure 13 base station and mobile station; and Figure 15 is a schematic diagram for use in explaining operation of a second embodiment of the present invention.

A first embodiment of the present invention carries out channel estimation based on a correlation technique. Before describing the first embodiment in detail, an overview of the basic principles of the correlation technique will first be given with reference to Figures 3 to S.

Firstly, referring to Figure 3, the correlation technique will he explained for a single channel between a transmitter and a receiver. A predetermined transmit signal p(t), made up of a sequence of symbols (bits) known to the receiver, is formed at the transmitter by cyclically concatenating the whole, and segments of, a predetermined sequence, m(t), of symbols. This predetermined sequence must have good auto-correlation properties, i.e. the result of the correlation of the sequence with phase-shifted versions of itself must be insignificant compared to the result of the correlation of the sequence with the aligned (non-phase-shifted) version of itself. For example, as shown in Figure 3, if the predetermined sequence m(t) is made up of f our symbols m, to M41 the transmit signal sequence p may be made up of one complete version [ml, M21 M31 M41 of the sequence m with cyclically concatenated segments IM3, mj and IMI, M21 of that sequence as.its prefix and suffix.

The transmit signal p(t) is transmitted via the channel to the receiver, the channel having an impulse response h(t).

At the receiver, the received signal is correlated by a periodic signal, q(t), which has the predetermined sequence m(t) as its period. The correlation process is shown in Figure 3 schematically as being carried out by a multiplier and an integrator, which constitute a correlator.

Incidentally, it will be appreciated that the signals p(t) and q(t), the sequence m(t), the symbols m, to m, and the impulse response h(t) are complex parameters having real and imaginary components corresponding to amplitude and phase respectively.

When the predetermined transmit signal p(t) including.the sequence m(t) is inserted periodically I (at intervals T) into an information signal being transmitted by the transmitter, for example in a pre or midamble portion of each timeslot of the information signal, the correlator output will provide a periodic estimate of the channel impulse response h(t), as illustrated in Figure 4. The reason for this will be explained later with reference to Equations 1 to 5 and Figures G to 8.

It is also possible to perform the correlation process for two or more transmitters in the period T between successive transmissions of the predetermined transmit signal p(t) by the transmitters concerned.

Figure 5 illustrates the correlation process for two different transmitters, the first transmitter transmitting its information signal (Information Signal 1) via a first channel (Channel 1) and the second transmitter transmitting its information signal (Information Signal 2) via a second channel (Channel 2). Information Signal 1 includes periodically a predetermined first transmit signal p, (t) and Information Signal 2 includes periodically a second transmit signal P2(t). The transmission times by the two different transmitters of their respective predetermined transmit signals p, (t) and P2 (t) are controlled such that, on arrival at the receiver, the signals are synchronised. For example, the transmitters may be controlled remotely by the receiver so that the respective maximum energy centres of the signals as received coincide temporally.

For example, in a code-division multiple access (CDMA) mobile communications network the synchronisation between different transmitters may be set to half a "chip period". In such a CDMA network, symbols in the information signals are "chipped" using a predetermined binary spreading sequence, and each bit in the spreading sequence is a "chip". Typically, there are sixteen chips per symbol period of the information signal. In one proposal, the chip period is 244 nanoseconds (corresponding to a chip rate of 4.096 Mchips/sec). In another proposal, there are fifteen chips per symbol period, in which case the chip period is increased slightly to 260 nanoseconds (3.84 Mchips/sec). Incidentally, the or each predetermined transmit signal p(t) used for channel estimation purposes is not itself chipped using a spreading sequence as it is generated at the chip rate in the first place and is intended to have good autocorrelation properties in its own right.

As the respective predetermined transmit signals in the Information Signals 1 and 2 arrive in the same is time window at the receiver, in order to enable the impulse responses for the two channels to be separated at the receiver, a sequence phase shift, or time-shift, T is applied to the predetermined transmit signal p2(t) for the second transmitter compared to the predetermined transmit signal p, (t) for the first transmitter such that p2 (t) = p, (t- T). This phase-shif t T must be greater than or equal to the excess delay of Channel 1. The phase-shift T is applied in a circular manner so that for 0_.t<T p2(t) takes the values of pl(t) for L-i:t<L. When p,(t) is a periodic signal such that p,(t L)=p,(t) the signal p2(t) can be defined as pl(-T-+t) for 0-.<t<T and as pl(t-T) for T:!gt<L.

At the receiver, the predetermined transmit signals pl(t) and p2(t) included in Information Signals 1 and 2 are received and correlated in the same manner as in Figure 4. However, in this case, in each period Tthe channel impulse response hl(t) for channel 1 is obtained, followed a time T later by the channel impulse response h2(t) for channel 2.

It will of course be appreciated that channel impulse responses for further channels can be estimated I within the period T between transmissions of the transmit signals p, (t) and P2 (t) - Conventionally, the phase-shift i used to separate the respective impulse responses of the channels at the receiver is set to a worst-case value representative of the maximum expected excess delay of any of the channels being received by the receiver.

The correlation method will now be explained in more detail with reference to Figures 6 to 12.

If a known signal s(t) is transmitted through a channel whose impulse response is h(t), and the received signal is correlated with the known signal at the receiver, the following result will be obtained f (t) [s(t) h(t)] s(- t) .. (1) In this equation represents convolution and s(-t) denotes the mirror image of s(t) about the time origin. Convolution with s(-t) is equivalent to correlation with s(t). Incidentally, it is assumed here that s(t) is an infinite-length continuous signal, as opposed to a discrete signal.

Equation 1 above may be rearranged as f (t) = [s(t) s(- t)] h(t) (2) If s(t) has good auto-correlation properties, the terms in the square brackets of Equation 2 may be approximated by the delta function 5(t), resulting in f (t) =_ [AS(t)] h(t) Ah(t) where A represents the size of the correlation peak. It follows from Equation 3 that a correlator may be used to obtain the channel impulse response h(t).

The accuracy of the obtained channel impulse response depends on the auto-correlation properties of the known signal s(t:), and therefore the auto-correlation of the signal should be as close as possible to the delta function for a useful result to be obtained.

Referring back to Figure 3 again, in a practical implementation a discrete known signal, p, as shown in Figure 1 is used in place of the infinite-length continuous known signal s(t) of Equation 1. Also, at the receiver, the periodic signal q is used in place of the known signal at the receiver.

The channel impulse response may be represented by h= {hl,h2,...,hw} (4) where W is the number of elements or coefficients in the channel impulse response.

Equation'2 above may be approximated by f (n) = [p(n) q(- n)] h(n) with n representing time in the discrete time domain.

The correlation of the receiver known sequence (periodic signal) q(-n) and the transmitter known sequence (predetermined transmit signal) p(n) should result in an approximation of the equivalent, in the discrete domain, of 5(t) if the channel is to be successfully estimated without being affected by the data transmitted before and after the transmitter known sequence p(n). This equivalent of <5(t) may be referred to as the sampled impulse sequence (unit sample sequence) and is defined as.... 0, 0, 1, 0, 0, Incidentally, in the discrete time domain the elements (samples or coefficients) of p(n), q(n), and h(n) are complex parameters having real (inphase) and imaginary (quadrature) components corresponding respectively to amplitude and phase.

Taking, by way of example, a case in which W in Equation 4 is four, the desired correlation result p(n) q(-n) is shown in Figure G. In this case, the correlator output, represented by Equation 5, would be equal to the channel impulse response coefficient h, at time n=O, the channel impulse response coefficient h2 at time n=1, channel impulse response coefficient h3 at time n=2, and channel impulse response coefficient h3 at time n=3 (=W-1). In Figure 6, X denotes "don't care".

The length of the approximation of the sampled impulse sequence (0001000) within the correlation result p(n) q(-n) is important. This sequence should be at least.2W-1 samples long with at least W-1 leading and W-1 trailing zeros.

Figure 7 shows how the transmitter known sequence (predetermined transmit signal) p may be constructed to cause the correlation result p(n) q(-n) to contain the sampled impulse sequence 0001000 as shown in Figure 6. Correlation is performed over one period of q, starting with the element ml.

The actual correlation process resulting in the sequence 0001000 is shown in Figure 8. In Figure 8, the correlation result for time n=7 is much less than one because the sample of the transmitted signal that is combined with the element M4 of the receiver periodic sequence (q) is in this case the first data sample following the transmitter known sequence p. In Figure 8 this first data sample is denoted by x. It follows of course that x must not be equal to M4- The transmitter known sequence p comprises PE+2(W-1) samples of m arranged in a circular fashion, where PE is one full period of the receiver periodic sequence q. L=PE+2(W-1) is the minimum length of the transmitter known sequence. In practice, it is generally desired that PE >> W and L >> PE+2(W-1) and that the transmitter known sequence p should contain only one full period of m to avoid the possibility of data samples (transmitted before and after the transmitter known sequence p) corrupting the channel impulse response detection at thereceiver. As PE is increased, larger auto-correlation peaks are obtained.

When more than one channel is to be estimated using the correlation method, the period PE of m for plural-channel estimation must satisfy PE > KW, where K is the number of channels whose impulse responses are to be estimated at the receiver. As the excess delays of the different channels are assumed to be the same, K can be considered to represent the number of distinct equal-length channel impulse responses to be detected at the receiver.

In the following example, it is assumed that two channels a and b are to have their impulse responses h,, and hb estimated. Thus, K=2. The excess delay (maximum expected delay time) of each channel is assumed to be W=2. The sampled impulse sequences corresponding to channels a and b should be as shown in Figure 9. In this case, the correlator output would be equal to the channel impulse response coefficient hai of channel a at time n=O, channel impulse response coefficient ha2 Of channel a at time n=1, channel impulse response coefficient hb, of channel b at time n=2 and channel impulse response coefficient hb2 Of channel b at time n=3.

The transmitter known sequences (predetermined transmit signals) p,, and pb suitable for producing the sampled impusle sequences shown in Figure 9 are shown in Figure 10. Each transmitter known sequence contains at least PE+3W-2 elements. Each transmitter known sequence can contain only one whole period of m.

Next, the first embodiment of the present invention, using the correlation method, will be explained with reference to Figures 11 and 12. The first embodiment is applicable for example to channel estimation used for joint detection purposes.

This embodiment is intended to take advantage of the fact that in practice the channel impulse responses of the different channels to be estimated are not of the same length. For example, a mobile station that is far away from its current serving base station will tend to have a channel impulse response longer than that of a mobile station that is near to the -base station. By recognising this difference, the present embodiment can set a range of delay times to be covered by the channel impulse response estimate for one channel that is different from the range of delay times to be covered by the channel impulse response estimate for at least one other channel. In this way, it is possible to provide more accurate estimates of the channel impulse responses for channels with relatively long channel impulse responses whilst, at the same time, avoiding performing unnecessary calculations of insignificant elements of the channel impulse responses of channels having relatively short channel impulse responses. As a result of the more accurate estimation of the channel impulse responses of the "longer" channels, a better link performance for those channels can be achieved. In addition, the valuable processing power at the receiver can be utilised to process useful rather than redundant information.

By way of example, it will be assumed that two channels,.channel a and channel b, are to have their channel impulse responses estimated, and that the estimate of channel impulse response h,, f or channel a can be represented by W2, = 1 element and that the estimate of channel impulse response hb f or channel b requires Wb = 3 elements to represent it usefully.

The desired correlator outputs in this example are shown in Figure 11. As Figure 11 shows, at time n=O the correlator output is the element hal of the channel impulse response h,, for channel a (which in this example is the only element of that response), and at subsequent times n=l, 2 and 3 the elements hbl, hb2 and hb3 Of the channel impulse response hb are output respective from the correlator.

The way in which the transmitter known sequences (predetermined transmit signals) p,, and pb are made up to produce the correlator outputs shown in Figure 11 is illustrated in Figure 12.

To ensure that a particular transmitter known sequence p does not generate unwanted copies of the same channel impulse response at the receiver, the maximum numbers of leading and trailing zeros of the approximations of the sampled impulse sequences must each be less than PE. Furthermore, PE must at least be equal to the largest W which in this case is Wb = 3.

Looking at Figure 11 it can be seen that the maximum number of leading zeros is three (the three zeros in the sampled impulse sequence for pb(n) q(-n)) and the maximum number of trailing zeros is five (the five zeros following the Ill 11 in the sampled impulse sequence of p,,(n) q(-n)). Accordingly, PE=6 is chosen in this case.

In practice, PE >> 6 should be chosen to avoid the possibility of data samples corrupting the channel estimation process. In the case of Figure 12, for example, if M6 is transmitted as part of the data stream of channel a after the last element of the sequence p,.,, it will result in the detection of an unwanted correlation peak at the receiver. However, if PE is much larger than the number of trailing zeros, the possibility of the data stream causing a correlation peak will be greatly reduced. The same applies to the leading zeros.

In general, if there are K channels to be dealt with, the minimum length of the transmitter known sequences (predetermined transmit signals) p is given by L=W,+W2+ - - - +WK+PE+WK-2, with PE determined as set out above. However, in.practiqe, to avoid the above mentioned problem of data samples corrupting the channel estimation process, much larger values of L may be necessary.

It has also been assumed in this example that the required spacing between the detected adjacent channel impulse responses is zero. In practice, however, it may be necessary to use a larger spacing than this between adjacent channel impulse responses. This would also involve a requirement for a larger value of L.

In the embodiment described above, the different ranges of delay time values (different W values) to be covered by the channel impulse response estimates for the different channels could be fixed or predetermined if, for example, it is known in advance what maximum delay time each channel will have. For example, if a particular mobile station is known to be stationary (and the environment is relatively static, too) at a certain location the number of elements required to form its channel impulse response can be known in advance. Alternatively, each particular mobile station could be allocated a particular number of elements for its channel impulse response estimate according to the result of a negotiation between the user of the mobile station and the network. For example, a user with a low quality of service requirement could decide to accept a small number of elements in the channel impulse response estimate in return for a reduced tariff.

In practice, however, it will be preferable to make the allocation of the ranges of delay time values to the different channels a dynamic allocation based on the particular requirements of each particular channel in use of the network. Thus, if a mobile station is moving away from a base station over time, it will tend to be necessary to allocate it progressively more elements in its channel impulse response estimate.

A method of achieving such dynamic all ocation will be explained next with reference to Figures 13 and 14.

Figure 13 shows parts of a base station 10 and a mobile station 20 for use in a UTRA TD-CDMA mode mobile communications network.

The base station 10 includes a control signal receiving unit 12, an uplink channel estimation unit 14, an estimation control unit 16 and an estimation data storage unit 18.

The control signal receiving unit 12 is connected to receiver circuitry (not shown) of the base station for receiving therefrom control signals CS. The control signal receiving unit 12 is also connected to the estimation control unit 16 for applying thereto delay time values W. The estimation data storage unit 18 has a plurality of entries corresponding respectively to the different uplink channels CH1, CH2, .. in use at the base station concerned. Each entry holds a delay time value W:L, W21 for its corresponding channel. Entries in the estimation data storage unit 18 are accessible both by the estimation control unit 16 and by the uplink channel estimation unit 14.

The uplink channel estimation,unit 14 is also connected to the base station receiver circuitry (not shown) for receiving therefrom uplink signals US,, US2,... corresponding respectively to the in-use channels and for applying thereto uplink channel estimation data UPESTI, UPEST21... corresponding respectively to the in-use channels.

The estimation control unit 16 is also connected to transmitter circuitry (not shown) of the base station for applying thereto delay setting signals DELAYSET.

The mobile station 20 includes a downlink channel estimation unit 22, a excess delay monitoring unit 24, a control signal generating unit 26, and a delay setting unit 28.

The downlink channel estimation unit 22 is connected to receiver circuitry (not shown) of the mobile station 20 for receiving therefrom one or more downlink signals DS sent to the mobile station 20 by the base station 10. The downlink channel estimation unit 22 is also connected to the excess delay monitoring unit 24 for applying thereto downlink channel estimation data DOWNEST. The excess delay monitoring unit 24 is connected to the control signal generating unit for applying thereto an activation signal ACT and a delay time value W. The control signal generating unit 26 is connected to transmitter circuitry (not shown) of the mobile station for applying thereto a control signal CS. The delay setting unit 28 is connected to the mobile station receiver circuitry for receiving therefrom a delay setting signal DELAYSET.

operation of the Figure 13 base station and mobile station will now be described. As mentioned in the introductory portion of the present specification, the

UTRA TD-CDMA mode network is a CDMA/TDMA network using time division duplexing (TDD). In.such a TDD network, the available timeslots for two-way communication between the base station 10 and the mobile station 20 are divided on a preselected time-division basis into downlink and uplink timeslots. The downlink timeslots are used for transmission of user data and/or control information from the base station 10 to the mobile station 20, and the uplink timeslots are used for transmission of user data and/or control information from the mobile station 20 to the base station 10. The alternate downlink and uplink timeslots still effectively provide the base station and mobile station with"respective uplink and downlink channels. In such a TDD mode of operation, particularly when operating in an indoor environment (e.g. with Doppler spread of around 20Hz leading to 50msec channel coherence time), the variations of the uplink and downlink channels over a period of one or several frames (each frame is lomsec) are relatively small. Furthermore, it can be assumed, to a reasonable approximation, that the channel characteristics and variations for the uplink channel wil.1 be the same as for the downlink channel.

It is presently proposed that, in the UTRA TD-CDMA network, channel estimation should be performed both for the uplink channel and for the downlink channel.

For this purpose, a known sequence of symbols is transmitted at intervals in both the uplink and downlink directions. In the Figure 13 embodiment the sequence of symbols for the uplink, which is a multipoint-to-point communication scenario, may be a sequence p as described previously in relation to Figure 12. A number n of mobile stations, for example eight or sixteen mobile stations, transmit respective phase-shifted versions Pf the same sequence p during a predetermined portion of a particular timeslot. This predetermined portion may be an initial, or "preamble", portion or preferably a middle, or."midamble", portion of the timeslot, where there is less scope for transmission errors. The amount of the phase-shift of the sequence p transmitted by each mobile station is set by the delay setting unit 28 in the station concerned, as described later in more detail.

In the downlink direction, which is a point-to multipoint communication scenario, each mobile station only needs to perform channel estimation for one channel, i.e. its own particular downlink channel. To this end, in the downlink direction as well, a known sequence of symbols is transmitted in each timeslot, typically again in a I'midamble" portion. At each mobile station, the impulse response of just one channel needs to be estimated using this midamble portion, so in the downlink direction the maximum is excess delay that can be handled is large compared to the uplink case.

Because the respective excess delays of the uplink and downlink channels are similar when operating in the TDD mode, an estimate of the downlink-channel excess delay, produced in the mobile station 20, can be used reliably in the base station 10 as an estimate of the uplink-channel excess delay. Thus, the mobile station in the Figure 13 embodiment is designed to estimate its downlink-channel excess delay and report this information at appropriate intervals to the base station 10 for use by the base station in carrying out channel estimation for the uplink channels.

As shown in Figure 14, in the TDD mode of operation, timeslots are allocated alternately to downlink and uplink transmissions. only one downlink timeslot, followed by one uplink timeslot, is shown in Figure 14. Each timeslot is a multiple access timeslot shared amongst a plurality of different users (in this example, eight users). In the downlink timeslot, for example, the base station transmits to each of the different users a downlink signal DS containing an initial block of data followed by a midamble portion MA followed by a further block of data. The downlink transmission signals DS, to DS8 for the different users are spread using different short (or channelisation) codes, so as to enable the transmission signals to he distinguished from one another.

Similarly, in the uplink timeslot, the eight different users transmit respective uplink signals US, to US8 to the base station. Each uplink signal US includes an initial block of data followed by a midamble portion MA followed by a final block of data.

Again, the uplink transmission signals for the different users are distinguished from one another by using different short (or channelisation) codes to is spread those signals.

In the mobile station 20 of one of the users, the receiver circuitry receives the downlink signal DS for that user and applies it to the downlink channel estimation unit 22. The downlink channel estimation unit 22 operates, for example as described previously with reference to Figures 3 and 4, to produce channel impulse response data DOWNEST corresponding to the downlink channel impulse response h. This data DOWNEST is supplied to the receiver circuitry in the mobile station for use in receiving subsequent downlink signals. In addition, the data DOWNEST is supplied to the excess delay monitoring unit 24 which, based on that data, determines the excess delay (maximum delay time) W of the downlink channel. The excess delay monitoring unit 24 compares the latest W value determined thereby with one or more W values held therein obtained from previously-received downlink signals DS. When, for example, a preselected change in the series of W values determined for the downlink channel is detected, the excess delay monitoring unit 24 outputs an activation signal ACT, together with the latest-determined W value, to the control signal generating unit 26. For example, the preselected change required to trigger the ACT signal could be a simple change of more than a predetermined percentage or amount from one W value to the next, or could be based on a more sophisticated analysis such as when a moving average of the W values changes by more than a predetermined percentage or amount.

In response to the ACT signal, the control signal generating unit 26 applies the control signal CS to transmitter circuitry of the mobile station 20. The control signal CS includes the latest-determined W value for the downlink channel.

The control signal can take any number of forms but in this embodiment so-called "blank-and-burst" in band control signalling is used to transmit the control signal to the base station 10. In this case, as shown by way of example in Figure 14, the predetermined control signal CS replaces one or both of the blocks of data in the next uplink signal US of the user concerned.

In the base station 10, the blank-and-burst in band control signals, including any control signals CS containing W values, are passed by the receiver circuitry to the control signal receiving unit 12. The control signal receiving unit 12 is responsive to any control signal CS containing a W value to extract therefrom the W value contained therein. This extracted W value, together with information from the receiver circuitry as to the identity of the channel (user) to which that W value relates, is employed by the estimation control unit 16 to update the W value held in the entry of the estimation data storage unit 18 corresponding to the identified.channel.

In parallel with such updating of the estimation data storage unit 18, the uplink channel estimation unit 14 operates to perform channel estimation for the uplink channels. Uplink channel estimation for the eight different users shown in Figure 14, for example, is performed in the same uplink timeslot. The uplink channel estimation unit 14 uses the W values held in the estimation data storage unit 18 for the different uplink channels (users) concerned to determine individually how many elements are to make up the channel impulse response estimate h for each of those different channels.

In general the W values held in the estimation data storage unit 18 for the different uplink channels will differ from one another, in accordance with the different W values determined for their respective corresponding downlink channels by the users concerned.

Accordingly, in general the uplink channel estimation unit allocates different numbers of constituent elements to the estimates for the different uplink channels, for example as described previously with reference to Figures 11 and 12.

Data UPESTI, UPEST21... representing the impulse response estimates h of the different uplink-channels is output from the uplink channel estimation unit 14 to receiver circuitry of the base station for use in receiving subsequent uplink signals from the mobile stations concerned.

The estimation control unit 16 in the base station also serves to set the individual phase-shift values of the known sequences (predetermined transmit signals) p transmitted in the midamble portions MA of the uplink timeslots. To this end, the estimation control unit 16 monitors the W values held in the estimation data storage unit 18 and, when the W value of any one of the channels to be estimated in the same timeslot is updated,,transmits the delay setting signal DELAYSET to each of the users concerned containing the new required phase-shift value for the user concerned.

The DELAYSET signals could be broadcast on one of the common control channels (CCH), or alternatively blank-and-burst in-band control signalling could be used for the transmission of individual DELAYSET signals to the different users.

In the mobile station 20, the delay setting unit 28 receives the DELAYSET signals transmitted from the base station under the control of the estimation control unit 16 and extracts therefrom the phase-shift value applicable to its own mobile station.

Thereafter, the known sequence of symbols p is included in the midamble portion MA with a phase-shift (relative to a predetermined temporal reference point) corresponding to the phase-shift value specified for that mobile station by the estimation control unit 16.

Fo.r example, referring to Figure 12, the phase shift value for channel a would be set to 0, indicating that the predetermined transmit signal p,, has the predetermined sequence of symbols m at the very beginning, whereas for channel b the phase-shift value would be set to 3, indicating that in the predetermined transmit signal pb the predetermined sequence of symbols m starts after the third symbol. If, at any time, the W value for channel a increases, the phase-shift value for channel b would need to be increased by sending a DELAYSET signal to the delay setting unit 28 in the mobile station using channel b.

The DELAYSET signals can take any number of forms.

For example, the channels to be estimated in the same timeslot may be assigned channel numbers from 1 to n in ascending order of phase-shift value. Then, when the spread Wi of channel i (1:5i:n) is determined to have changed by an amount LWj, the phase-shift values of all the subsequent channels i+1, i+2,... will need to be I increased by the same amount AWj. Accordingly, a preferred form of DELAYSET signal suitable for broadcast to all mobile stations would specify the channel (i) whose W value had changed and the amount ZWj of the change. Then, each delay setting unit would only respond to those DELAYSET signals whose specified channel number was less than its own channel number.

This would reduce the volume of signalling required and enable the phase-shift values to be updated simultaneously in the mobile stations affected by the change.

In the Figure 13 embodiment, the excess delay monitoring unit 24 causes the latest-determined W value to be transmitted to the base station only when a is preselected change in the series of W values is detected. However, it will be appreciated that in other embodiments, the latest-determined W value could be transmitted in each timeslot, although in this case the amount of data throughput in the uplink direction will be reduced due to the increase in control signalling.

It is also not necessary'for the control signal to embody the W value itself. For example, if desired, the downlink-channel estimation data DOWNEST as a whole could be transmitted from the mobile station to the base station. In this case, the estimation control unit 16 could employ the DOWNEST data to determine the W value appropriate for uplink channel estimation for the user concerned.

Another possibility would be for the control signal CS to be a W-value increment/decrement signal such that, in response to each control signal received at the base station from a particular mobile station, the W-value registered in the estimation date control unit 18 for the uplink channel of that mobile station would be,incremented or decremented by a fixed amount, e. g. 1.

Furthermore, the W-value transmitted need not be the latest-determined W-value. For example, the W value transmitted to the base station could be the highest value of W determined over the most recent monitoring period, or an average value over that period. Any suitable measure of excess delay can be transmitted.

As mentioned previously, another channel estimation technique is based on a matrix inversion method, and an embodiment of the present invention employing an adapted version of the matrix inversion method will now be explained.

To facilitate an understanding of the matrix inversion-based embodiment, first a description of the basic principles of the matrix inversion method will be given.

Firstly, take the case of a single user in a CDMA network, who transmits signals via a channel whose impulse response, when sampled at the chip rate of the CDMA network, is modelled as:

HT [h,h2... 5hW1 25... (6) where HT and h. denote the transpose of H and the ith coefficient of the channel impulse response, respectively.

This indicates that the channel is of length (i.e.

excess delay) W chips. For a known transmit sequence (predetermined transmit signal) of length L chips, the convolution of the known sequence and the channel impulse response will be L+W-1 chips long.

Of the L+W-1 convolution elements, W-1 elements are affected by the data transmitted prior to the known sequence, and a further W-1 samples are affected by the data transmitted after the known sequence (assuming L>W). It follows that the portion of the convolution result which is only affected by the known sequence is obtained by discarding the first W-1 samples and the last W-1 samples during the channel estimation process.

The number of samples of interest is hence found as:

D= L+W-1-2(W-1)= L-(W-1) (7) Representing the known sequence of length L by GT IMI,---,ML-1,MLI 15... (8) the wanted portion of the convolution between the channel impulse response and the known sequence can be found using the following expressions:

e2 = Mw+l-hl+ mw.h2+ "'+ M2.hw e3=Mw+2-hl+mw+,-h2 + ''' + M3-hw eD=ML-h,+ML-l.h2+ "-+ MD-hw (9) with D being determined according to Equation 7 and ej representing the samples of the result.

When W=D and the expressions in Equation 9 are independent of each other, unique solutions for the coefficients of the channel impulse response may be obtained.

Equation 9 may be rewritten in matrix form as E M H el MW Mw- I MI h I e2 Mw+ 1 Mw M2 h2 e3 Mw+ 2 Mw+ I M3 x h3 ...........

eD-Dxl ML ML- 1 MD-Dx W L hWJWXI .. (10) is where E, M and H are matrices containing the received signal samples, the elements of the known sequence and the channel impulse response coefficients, respectively.

When W=D, as stated above, the matrix M is a circular square complex matrix containing the known sequence and shifted versions of the known sequence. Being square, M is capable of having an inverse M-1. Not all square matrices have inverses, so the elements of M, i.e. the elements of the known sequence GI, must be selected appropriately. Assuming that the inverse M-1 does exist, the channel impulse response vector H is found from H= M-ix E The matrix inversion method is also applicable to estimation of the respective channel impulse responses of plural channels "simultaneously". When, for example, channel estimations are to be performed for two channels simultaneously at a common receiver, the following equation may be formed:

E = El + E2 = (Ml x HI) + (M2 x H2) 10... (12) where Hi is the channel impulse response of the first channel to be estimated and R2 is the channel impulse response of the second channel to be estimated; M1 denotes a matrix containing the known transmit sequence (predetermined transmit signal) for channel 1 and M2 denotes a matrix containing the known transmit sequence (predetermined transmit signal) for channel 2; El denotes the received signal vector for channel 1 and E2 denotes the received signal vector for channel 2. E is the sum of the received signal vectors.

Equation 12 may be rewritten in terms of three components E, M, and 11, as:

E=MxH where E= E1+E2 M = [M2, M11 H = [H2] H1 (14) A unique solution for Equation 13 can be obtained if M is a square matrix and has an inverse.

Extending the example from two channels to K channels, and assuming that the excess delay of each channel is W chips, the dimensions of E, M and H are Dx1, DxKW and KWx1 respectively. For the two-channel case, these dimensions are Dx1, Dx2W and 2Wxl.

For M to be a square matrix, D=KW. If this condition is satisfied, and the known sequences are chosen such that the inverse of M exists, H can be determined using Equation 11 above.

In general, it is possible to use a unique and independent known sequence per channel. However, to simplify the process of matrix inversion, it is preferable to use a single known sequence but allocate different respective cyclically-shifted versions of the sequence to the different transmitters whose channels are to be estimated simultaneously.

The first known sequence (first predetermined transmit signal) in this case is formed by concatenating a predetermined sequence of length P with a segment W-1 bits long of the same sequ. ence in a cyclic fashion. Theamount of the cyclic shift required to generate the known sequence (predetermined transmit signal) for the next channel is W, i.e. the excess delay which here is assumed to be the same for all channels. Again, the length of the known sequence is made equal to P+W-1 by cyclically concatenating segments of the predetermined sequence. The additional W-1 symbols are added to each known sequence to ensure that the received signal contains P symbols which are determined by the known sequence alone.

The dimensions of the matrices in Equation 14 are selected to satisfy D=KW=P, and the length of the known sequence is found from Equation 7 above as L = P+ (W- 1) (15) As a specific example, consider the case in which there are two channels to be estimated, each of excess delay W=3. In this case, K=2 and P=KW=6. L=P+(W-1)=8.

In this case, the known sequences for the two channels are:

GT I = [MPM2,M3,M4,M5,M6,Ml,M21 GT 2 = [M4,M5,M6,MPM2,M3,M4,M51 (16) The following matrix equations may then be formed:

El Mi A ell M3 M2 MI H1 e12 M4 M3 M2 e13 M5 M4 M3 hl, e14 M6 M5 M4 x hl2 e15 MI. M6 M5 h13- -el6-J LM2 Ml M6, (17) E2 M2 e2l M6 M5 M4 H2 e22 M1 M6 M5 e23 M2 M1 M6 h2l e24 M3 M2 x h22 ml h23- e25 M4 M3 M2 -e26- -M5 M4 M3(18) E E M H A el ell + e2l M6 M5 M4 M3 M2 Ml h2l e2 e12 + e22 MI M6 M5 M4 M3 M2 h22 e3 e13 + e23 M2 MI M6 M5 M4 M3 X h23 e4 e14 + e24 M3 M2 MI M6 M5 M4 hll e5 el5 + e25 M4 M3 M2 Ml M6 M5 hl2 _e6j _e16 + e26- M4 M3 M2 MI M6_ _hl3- .. (19) Equations 17 to 19 represent respectively the channel estimation equations for the first channel only, the second channel only, and the combined first and second channels. When the channels are of equal excess delay W, the known sequence matrix M and the channel impulse response vector H may be divided up into K equa-l-size sub-matrices. The equation below shows the subdivision into sub-matrices in the twochannel case, i.e. K=2.

E M H el M6 M5 M4 M3 M2 Ml h2l e2 Ml M6 M5 M4 M3 M2 h22 e3 M2 Ml M6 M5 M4 M3 h23 X e4 M3 M2 M1 M6 M5 M4 hll e5 M4 M3 M2 M1 M6 M5 h12 _e6j M5 M4 M3 M2 Ml M6 L hl3 25... (20) In an embodiment of the present invention, utilising the matrix inversion method, channel estimation is performed such that the range of delay time values covered by the estimate for one channel is different from the range of delay time values covered by the estimate for at least another one of the channels. The ranges may be set individually, for example based on the expected excess delay (length) of the channel concerned.

A two-channel example will be given, in which the expected excess delay (channel length) of channel 1 is W1=2 and the expected excess delay of channel 2 is W2=4. In this case, the period of q(t) is P=WI+W2=6, and the length of the known sequence (predetermined transmit signal) is L=P+(W,,,-l)=9, where Wmax is the maximum expected channel length which in this case is the expected length of channel 2, which is 4. In this case, the known sequences (predetermined transmit signals) for the two channels may be:

GT I = [M1,M2,M3,M4,M5,M6,M1,M2,M31 T 2 = [M3,M41M5,M61Ml,M2,M3,M4,M51 (21) 25 The channel estimation for channel 1 is:

E Ml A ell M2 Ml Hl e12 M3 M2 r- & el3 M4 M3 11 e14 M5 M4 h12 X1, I e15 M6 M5 el6j Ml M6_ .. (22) and for channel 2 is:

E2 M2 -5 e21 M6 MS M4 M3 H2---N e22 M1 M6 M5 M4 h21 e23 M2 M1 M6 M5 h22 e24 M3 M2 M1 M6 X h23 e25 M4 M3 M2 mi h24 -e26- -M5 M4 M3 M2- .. (23) and for channels 1 and 2 combined is:

E M H el M6 M5 M4 M3 M2 M1 h2l e2 MI M6 M5 M4 M3 M2 h2 2 e3 M2 M1 M6 M5: M4 M3 h23 e4 M3 M2 MI M6 M5 M4 h24 e5 M4 M3 M2 M1 M6 M5 h1i e6. 5 M4 M3 M2 MI M6 h12 25... (24) The matrix operation of equation 24 corresponds to the convolution and summation operations shown in Figure 15.

It can be seen from Equation 24 that in this case the known sequence matrix M and the channel estimation vector H are each partitioned unequally into sub matrices. Nonetheless, the basic form of Equation 24 is unchanged from Equation 19 and consequently the inverse of M will be the same as in the equal partitioning case described above. As a result, it is possible to use Equation 11 above to evaluate the elements of H. The only difference between the equal partitioning case of Equation 19 and the unequal partitioning case of the present embodiment is that, in the equal partitioning case, the first three elements of H belong to channel 2 with the remaining elements corresponding to channel 1, whereas in the present embodiment the first four elements of H correspond to channel 2 with the remaining two elements belonging to channel 1.

Although the above example was given for the case in which the number of channels was 2, it will be appreciated that more than two channels can be estimated using the unequal partitioning approach of the present embodiment.

The partitioning of the known sequence matrix M and the channel impulse response vector H, together with the cyclic shifts (phase-shifts) of the known sequences at the transmitters, can be determined dynamically, in the same basic manner as described above with reference to Figures 13 and 14. Because the matrix M itself is unchanged irrespective of the partitioning applied, the inverse M-1 can be pre calculated and kept at the receiver for the (or each) particular known sequence used by the transmitters.

Although the present invention has been described above in relation to a UTRA network, it will be appreciated that it can also be applied to other networks in which it is desirable to perform channel estimation for more than one channel in a particular timeslot or time period. These networks could be, or could be adapted from, other CDMA networks such as a wideband CDMA (W-CDMA) network or an IS95 network.

These networks could also be, or be adapted from, other mobile communication networks not using CDMA, for example networks using one or more of the following multiple-access techniques: time-division multiple access (TDMA), wavelength-division multiple access (WDMA), frequency-division multiple access (FDMA) and space-division multiple access (SDMA).

Similarly, although the Figure 13/14 embodiment, in which the excess delay of the downlink channel is reported in the uplink direction for uplink channel estimation purposes, was described in relation to a UTRA TDD mode network, it will be appreciated that the Figure 13/14 embodiment can also be applied to any other networks in which it is possible to assume, to a reasonable approximation, that the uplink and downlink channel excess delays will be the same as or similar to one another, for example any network which has a TDD mode of operation.

Also, although embodiments of the present invention have been described as having distinct "units" such as the uplink channel estimation unit, those skilled in the art will appreciate that a microprocessor or digital signal processor (DSP) may be used in practice to implement some or all of the functions of the base station and/or mobile station in embodiments of the present invention.

It will also he understood that the implementation details of the correlation and inverse matrix channel estimation techniques given hereinbefore are only given by way of example, and the present invention is not limited to those particular details. For example, if variations of the described correlation and/or inverse matrix techniques are used the values of minimum length L of the transmitter known sequence and of the full period PE of the receiver periodic sequence may be less than the values specified above. It may also be possible in some cases for the transmitter known sequence-to contain more than one whole period of m.

It will also be understood that embodiments of the present invention are applicable to channel estimation not using the correlation technique or the inverse matrix technique but some other technique.

Those skilled in the art will also appreciate that, in an embodiment of another aspect of the present invention, measures of the downlink channel characteristics (in particular a maximum delay time of each downlink channel) can he reported in the uplink direction by mobile stations to the base station. The estimation control function in the base station (e.g.

corresponding to the estimation control unit 16 in the Figure 13 embodiment) can then employ the reported measures to set equal ranges of delay time values for all of the uplink channels being estimated "simultaneously". For example, the range of delay time values set equally for the uplink channels concerned could be made greater than or equal to the maximum delay time (excess delay) of the downlink channel having the highest maximum delay time amongst the downlink channels corresponding respectively to those uplink channels.

Where it is possible to assume, to a reasonable approximation, that the uplink and downlink channel excess delays will be the same as or similar to one another, for example a network which has a TDD mode of operation, this can enable the equal range set to be adjusted dynamically, in use of the network, to provide channel estimation resources appropriate for the particular network conditions at a given time, rather than having to be preset in advance by a network operator to cope with an expected "worst case" excess delay that could apply to the base station in question.

An embodiment of this aspect of the invention could be based.on the Figures 13/14 embodiment but in this case the estimation control unit (16 in Figure 13) would control the uplink channel estimation unit (14 in Figure 13) so as to set equal ranges of delay times for the uplink channels being estimated together. In this case, the phase-shiftS T allocated to the mobile stations would be equal to one another, potentially simplifying the transmission of the DELAYSET signals to the mobile stations.

The reported measure from each mobile station is not limited to being a measure of maximum delay time for the downlink channel. Any other suitable measure of downlink channel performance can be reported to the base station for use by the estimation control function to set the equal range of delay time values to be covered by the estimates.

Claims

CLAIMS:

1. Communications apparatus including:

receiver means; and transmitter means for transmitting a plurality of predetermined transmit signals to the receiver means via different respective channels such that those signals are received substantially simultaneously by the receiver means, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts; wherein the receiver means include:

channel impulse response estimating means for employing the predetermined transmit signals as received by the receiver means and the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and estimation control means operable to control the operation of the channel impulse response estimating means such that a range of delay time values covered by the estimate produced for at least one channel is made greater than a range of delay time values covered by the estimate produced for another of the said channels.

2. Apparatus as claimed in claim 1, wherein the said estimation control means are operable to set such a range of delay time values to be covered by each of the said different channels.

3. Apparatus as claimed in claim 2, wherein the said range for each said channel is set in dependence upon an expected maximum delay time for the channel concerned.

4. Apparatus as claimed in claim 3, wherein the said range is set to he greater than or equal to the said expected maximum delay time for the channel concerned.

5. Apparatus as claimed in any preceding claim, wherein the said transmitter means are operable to transmit to the said receiver means control information for use by the estimation control means to set the range of delay time values to be covered by the estimate for at least one of the said channels.

6. Apparatus as claimed in claim 5, wherein the said control information includes a measure of the expected maximum delay time for at least one of the said channels.

7. Apparatus as claimed in claim 5 or 6, wherein the said control information is only transmitted following detection of a change in expected maximum delay time of one of the said channels.

8. Apparatus as claimed in any preceding claim, wherein each said estimate is made up of a series of one or more digital values corresponding respectively to successive times, and the said range of each estimate is set by setting the number of digital values to be included in the said series.

9. Apparatus as claimed in claim 8, wherein each said digital value is a signal-pair made up of an inphase value and a quadrature value.

10. Apparatus as claimed in claim 8 or 9, wherein the said successive times are at equal intervals.

11. Apparatus as claimed in any preceding claim, wherein the said channel impulse response estimating means include correlation means operable to correlate the predetermined transmit signals as received by the receiver with the said predetermined sequence to produce the said estimates.

12., Apparatus as claimed in claim 11, wherein the said correlation means are operable to perform a series of individual correlation operations in successive operation periods such that the estimates for the different channels are produced sequentially and, for each estimate, the elements of the estimate are produced sequentially, one such element being produced per correlation operation.

13. Apparatus as claimed in claim 11 or 12, wherein the number PE of symbols in the said predetermined sequence is set such that respective sampled impulse sequences for the said different channels have a maximum number of leading and trailing zeros less than PE.

14. Apparatus as claimed in claim 13, wherein, in each of the said predetermined transmit signals, a prefix sequence and/or suffix sequence of symbols, made up of a subset of the said predetermined sequence, is/are cyclically concatenated with the said predetermined sequence.

15. Apparatus as claimed in claim 13 or 14, wherein each said predetermined transmit signal has a minimum length L = W, + W2 + - - - + WK + PE + WK - 2, where K is the number of said different channels and W, to WK are the respective expected maximum delay times of the channels.

16. Apparatus as claimed in any one of claims 1 to 10, wherein the said estimation means include matrix operation means operable to perform a matrix multiplication operation H = M-1 x E to produce the said estimates for the said different channels, where M-1 is the inverse matrix of a circular square matrix M made up of the said predetermined sequen6eand shifted versions thereof, E is a vector having as its elements samples of the received signals, and H is a vector having as its elements the elements of the said estimates.

17. Apparatus as claimed in claim 16, wherein the said matrix operation means employ.the same said inverse matrix M-1 in the said matrix multiplication operation irrespective of the ranges set for the estimates by the estimation control means, and the estimation control means cause the elements of the vector H to be allocated to the different estimates in accordance with the ranges set.

18. Apparatus as claimed in any preceding claim, wherein the said estimation control means are operable to set the said predetermined amounts by which the said embodied sequences are shifted in phase one from the next in the different predetermined transmit signals in dependence upon the ranges of the estimates for the said different channels.

19. A mobile communications network including communications apparatus as claimed in any preceding claim, wherein:

the said receiver means are provided in a base station of the network; the said transmitter means are provided collectively by a plurality of individual mobile stations; and the said different channels are respective uplink channels between the said mobile stations and the said base station.

20. A network as claimed in claim 19, wherein:

at least one of the said mobile stations includes downlink channel assessment means operable to produce a measure of at least one predetermined characteristic of a downlink channel from the base station to the mobile station concerned, and also includes control signal transmitting means operable to cause the mobile station to transmit to the base station a predetermined control signal dependent upon the measure produced; and the said estimation control means in the base station are operable to receive the said predetermined control signal and to set the range of delay time values to be covered by the said estimate for the uplink channel of the mobile station concerned in dependence upon the received control signal.

21. A network as claimed in claim 19 or 20 wherein the said uplink and downlink channels for each said mobile station are provided on a time-division duplexing basis and the said range of the estimate for the uplink channel of each said mobile station is set in dependence upon a measure of the maximum delay time of the downlink channel for the mobile station concerned produced in that mobile station.

22. Receiving apparatus including:

receiving means operable to receive a plurality of predetermined transmit signals transmitted to the apparatus via different respective channels such that those signals are received substantially simultaneously by the apparatus, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts; channel impulse response estimating means operable to employ the predetermined transmit signals as received by the apparatus and the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and estimation control means operable to control the operation of the channel impulse response estimation means such that a range of delay time values covered by the estimate produced for at least.one of the said different channels is made greater than a range of delay time values covered by the estimate produced for another of the said channels.

23. A channel impulse response estimation method for use in a receiver to which a plurality of predetermined transmit signals are transmitted via different respective channels such that those signals are received substantially simultaneously by the receiver, the predetermined transmit signals all embodying the same predetermined sequence of symbols but the embodied sequences being shifted in phase one from the next in the different signals by predetermined amounts, in which method:

the predetermined transmit signals as received by the receiver are employed with the said predetermined sequence to produce estimates of the respective channel impulse responses of the said different channels; and a range of delay time values covered by the estimate produced for at least one of the said different channels is made greater than a range of delay time values covered by the estimate produced for another of those channels.

24. A base station for use in a cellular communications network, including receiving apparatus as claimed in claim 22.

25. A mobile station, for use in a mobile communications network, including:

signal transmission means for transmitting a predetermined transmit signal, embodying a predetermined sequence of symbols, to a base station of the network via an uplink channel provided by the network for the mobile station concerned, the transmission of the predetermined transmit signal being synchronised with the transmission of a further such predetermined transmit signal, also embodying the said predetermined sequence of symbols,.to the receiver by another mobile station of the network such that the respective signals of the two mobile stations are received substantially simultaneously by the receiver; and phase shift setting means operable to selectively adjust a predetermined amount by which the embodied predetermined sequence included in the predetermined transmit signal transmitted by the said signal transmission means is shifted in phase relative to the predetermined sequence embodied in the further predetermined transmit signal transmitted by the other mobile station.

26. A mobile station as claimed in claim 25, wherein the phase shift setting means are operable to adjust the said predetermined amount dynamically in use is of the network.

27. A mobile station as claimed in claim 25 or 26, wherein the said phase shift setting means are operable to adjust the said predetermined amount so that it is greater than or equal to a maximum expected delay time of an uplink channel provided by the network for the said other mobile station.

28. A mobile station as claimed in any one of claims 25 to 27, wherein the said phase shift setting means are operable to adjust the said predetermined amount in dependence upon a control signal sent to the mobile station by the base station.

A cellular communications network including a base station and a plurality of mobile stations, each mobile station having at least one uplink channel and at least one downlink channel for communication with the said base station, wherein:

two or more of said mobile stations form a channel estimation group and transmit respective predetermined transmit signals to the base station such that those signals are received substantially.simultaneously by the base station; -so- each mobile station of the said group includes:

downlink channel assessment means operable to produce a measure of at least one predetermined characteristic of the said downlink channel of the mobile station concerned; and control signal transmitting means operable to cause the mobile station to transmit to the base station a predetermined control signal dependent on the said measure produced; and the said base station includes:

receiver means for receiving such predetermined control signals and such predetermined transmit signals from the mobile stations of the said group; is channel impulse response estimating means for employing the received predetermined transmit signals to produce estimates of the respective channel impulse responses of the uplink channels of the mobile stations of the said group; and estimation control means operable to employ the received predetermined control signals to control the operation of the said channel impulse response estimating means such that a range of delay time values covered by the estimate produced for each mobile station of the group is set in dependence upon the said measures produced by the mobile stations of the group.

30. A network as claimed in claim 29, wherein the ranges set for the mobile stations of the said group are equal to one another.

31. A network as claimed in claim 30, wherein the said measure produced by each said mobile station of the group represents a maximum delay time of the said downlink channel of that mobile station, and the said estimation control means are operable to set the said range of delay time values to be greater than or equal to the maximum delay time of that one of the downlink channels of the mobile stations of the group having the highest maximum delay time.

32. A network as claimed in any one of claims 29 to 31, wherein the said uplink and downlink channels of each mobile station of the said group are provided on a time-division-duplexing basis.

33. Communications apparatus substantially as hereinbefore described with reference to the accompanying drawings.

34. Receiving apparatus substantially as hereinbefore described with reference to the accompanying drawings.

35. A mobile communications network substantially as hereinbefore described with reference to the is accompanying drawings.

36. A base station for use in a mobile communications network substantially as hereinbefore described with reference to the accompanying drawings.

37. A mobile station for use in a mobile communications network substantially as hereinbefore described with reference to the accompanying drawings.

38. A channel impulse response estimation method substantially as hereinbefore described with reference to the accompanying drawings.