Adaptive feedback for mimo communication systems

[0001] The present invention relates to methods of operating communication networks, and in particular to the layout and method of matching transmitter and receiver configurations depending on the condition of channels and other factors. [0002] Most conventional second generation (2G) and third generation (3G) wireless communication systems ), both the transmitter and the receiver are equipped with a single antenna. Such systems are known as single-input systems, single output (SISO - single input, single output). In recent years, experts have proposed the use of multiple antennas in the transmitter and / or receiver to improve communication system performance. One of the examples of using the multi-antenna transmission method is the multi-path transmission which has been proposed as a way to improve the signal-to-interference ratio (SINR) in the receiver. The SINR parameter mentioned in this document includes noise interference. Multi-input systems are also known as multi-path transmitting systems. single output (MISO - muitiple input, single output) Systems with multi-path transmission, i.e. MISO systems, use multiple antennas at the transmitter to send a signal to the receiver. The signals from each transmitting antenna reach the receiver through various propagation channels. The receiver can choose the "best" signal (usually the one with the highest SINR), or can combine multiple signals to achieve a higher SINR by adding up the signal value. In addition, the improvement of SINR with multi-path transmission is influenced by the coding of the transmission signal before transmission, using space-time codes. While the methods of working with multi-path transmission provide improved SINR, this improvement in maximum transmission speed with single-antenna modulation and coding options is limited. [0003] Systems with multiple input and multiple output (MIMO - muitiple input, muitiple output) use multiple antennas at both the receiver and the transmitter. MIMO systems use spatial signal parameters for achieving higher spectral efficiency and higher data transmission speeds, without increasing bandwidth The results of this compromise result in greater transmitter and receiver complexity. MIMO systems can be loosely classified as systems with limited feedback (also known as open loop systems) and systems with strong feedback (also known as closed loop systems) In MIMO systems with limited feedback, from the receiver to the transmitter only limited information about the communication channel is provided. Systems with limited feedback improve the achieved peak transmission speed, and they are more favorable in environments with high SINR. In systems with strong feedback the receiver sends back channel coefficients or other information to the transmitter. on the basis of which it is possible to estimate these channel coefficients. The transmitter can use this knowledge of the channel to filter the transmission signal before its transmission. Systems with strong feedback provide an additional advantage due to the gain of the antenna system from multiple transmission antennas, which can be significant for certain antenna configurations.

[0004] Unfortunately, I have not one configuration that works much better than others in the whole range of foreseeable operating conditions. Some of these conditions

0 operational that affect performance include the expected operational SINR at the receiver; propagation channel and interference environment observed at the receiver;

allowed value; and the allowable number of desired transmit and receive antennas It is desirable to have reliable access that works well with different antenna configurations and in a wide range of operational conditions [0005] Document EP 0 966 125 A relates to methods for improving the bandwidth of a duplex system having some first and second frequency bands. a variable amount of feedback data is sent to the frequency band, corresponding to a determined bandwidth value in said band, and is used to improve the quality of communication in the second band. By using feedback data, if available, transmissions in the second band are optimized. A variety of optimization schemes with assigned bandwidth values ​​are stored in the base station, referring to the required bandwidth value for feedback data transmission on the uplink to the base station. [0006] the invention relates to an adaptive system with at least two levels of adaptation At level one, Feedback mode is selected based on a certain quality measure that indicates receiver performance. The measure can be, for example, one or more SINRs at the output of the terminal receiver or the baud rate supported by the receiver. This quality measure may depend on a number of measurable quantities, including the number receiving antennas, channel quality, travel speed, propagation channel condition, and / or the available bandwidth between the mobile station and the base station. The terminal can calculate this measure of quality

0 for closed loop modes with open loop and switch to a mode that meets the switching criterion In the embodiment, Feedback modes include limited feedback mode (open loop) and strong feedback mode (closed loop) Transmitter configurations for limited feedback include individual transmission speed control of individual antennas (PARC - per antenna ratata control) and spatial multiplexing (SM) ) in combination with the repeated use of code In these approaches, the data stream is split into multiple streams, and each sub stream is transmitted from a different antenna. Transmitter configurations for closed loop mode include on-site matched field transmission diversity (MFTD) at which the data stream is split into one or more sub streams and each the sub stream is

0 pre-filtered to match the channel to the selected receiving antenna.

[0007] At the second level, the multi-path mode is selected. When the open loop mode is selected, the multi-path mode selection corresponds to the choice of the number of antennas. When all antennas are not used, the selection may include the "best" subset of antennas. [0008] When the closed loop mode is selected. the multi-path mode corresponds to the selection of the number of data streams When the number of data streams is less than the number of receiving antennas, the selection may include a subset of the receiving antennas, to which the pre-filters are matched The number of selection levels could also be greater than two, for example, when the MIMO open loop scheme was chosen, multiple schemes may be available for selection [0009]

0 Fig. 1 shows a multi-input, multi-output communication system

Fig 2-4 show a comparison of SISO, MISO, and MIMO systems with different antenna configurations

Fig 5 shows an exemplary configuration of an open loop transmitter using individual control of antenna transmission speed

Fig 6 shows an example configuration of an open loop transmitter using a spatial multiplex with code reuse

Fig 7 shows an exemplary configuration of a closed loop transmitter using on-site multi-path transmission [0010] Fig 1 shows a radiocommunication system 10 with multi-input / multi-output (MIMO multiple rnput / multiple output) comprising first station 20 and second station 30 First station 20 includes transmitter 22 for transmitting signals to the second station 30, while the second station includes a receiver 32 for receiving signals transmitted by the first station 20 It is obvious to a person skilled in the art that both the first station 20 and the second station 30 may include both transmitter 22 and receiver 32 as shown in Fig. 1 for bidirectional communication W one embodiment, the first station 20 is a base station in a radiocommunication network, and the second station 30 is a mobile station.

[0011] In the first station 20, an information signal is input into the transmitter 22 in the form of a binary data stream. The transmitter comprises a controller 24, a processing circuit of the transmitted signal 26 and a set of transmission antennas 28. The controller 24 manages the operation of the transmitter 22 Broadcast signal processing circuit 26 encodes error correction coding and mapping input bits in the form of complex modulation symbols. Broadcast signal processing circuit 26 generates multiple transmission signals that can be independent, partially redundant, or fully redundant After increasing frequency conversion , filtration and gain, the transmission signals are transmitted from the corresponding transmission antennas 28, via communication channel 12, to the second station 30 [0012] The receiver 32 in the second station 30 comprises a controller 34, a received signal processing circuit 36, and a certain set of antennas 38 The controller 34 manages the operation of the receiver 32 The received signal processing circuit 36 ​​demodulates and decodes the signals received in each antenna 38 In the absence of bit errors, the output signal from the receiver 32 will be the same, as the original identification signal in the transmitter 22 Because many data streams are transmitted in parallel from different antennas 28, there is a linear increase in bandwidth with each pair of antennas 28, 38 added to the system, without increasing the bandwidth requirements. [0013] MIMO systems are around the world is the subject of extensive research activity related to radiocommunication networks because of their potential possibility of achieving high spectral efficiency, and therefore high data transmission speeds. Many types of MIMO systems have been proposed, which can be classified as closed-loop systems or open-loop systems. In MIMO closed-loop systems, the transmitter receives the feedback channel coefficients from the receiver, what is referred to here as strong feedback In open-loop MIMO systems, the transmitter receives limited feed status feedback. This limited feedback typically does not contain channel coefficients, but may contain channel status information such as the communication channel SINR. [0014] Because systems with a closed loop have more knowledge about the communication channel, such systems usually have better performance compared to systems with open loop but require feedback of channel coefficients. Feedback is proportional to the number of transmitting antennas, number of receiving antennas, channel delay spreading. Fig. 2-4 show the comparison of efficiency of multi-antenna closed-loop and open-loop systems loop for various antenna configurations 2 shows a comparison of open-loop and closed-loop capacity in a multi-input, single-output (MISO) system Fig 3 shows a comparison of open-loop and closed-loop capacity in a MIMO 4x2 system Fig 4 shows a comparison of open-loop and closed-loop capacity in a MIMO 4x4 system [0015] In a cellular environment, the channels are random variables; therefore, the mutual information corresponding to each transmission scheme is also a random variable with a certain cumulative distribution function (CDF). Fig. 2-4 are SI NR plots relative to a point of 1% of CDF mutual information for different system configurations. This point of the CDF function is sometimes called mutual information that causes a probability of 1% decay, that is, the mutual information in each scheme exceeds the corresponding value of 1 percentile with a probability of 0.99 The continuous curves in these figures correspond to the closed loop capacity, the curves dashed with stars correspond to open-loop throughput, curves drawn in dashed lines with "x" correspond to a single-input system, single output (SISO) with one receiving antenna and one transmitting antenna [0016] In the MISO 4x1 system (fig 2), closed-loop bandwidth gives a gain of 6 dB compared to open-loop bandwidth due to gains of multi-path and antenna system In addition, open-loop bandwidth is not much greater than the SISO channel capacity in the wide operational range of SI NR. This indicates that some form of multi-path transmission is beneficial for receiving increased transmission speeds. In the case of the MIMO 4x2 system (fig 3), there is still a large gap of approximately 4.5 dB between closed loop and open loop bandwidth. This is somewhat less than in a 4x1 system, but still shows a significant benefit in closed-loop MIMO system performance over an open-loop MIMO system. In the case of the MIMO 4x4 system (fig 4), this distance decreases to about 3 dB, occurs mainly at small SINRs [0017] Exemplary MIMO systems with open loop MIMO use methods of individual antenna control (PARC) or spatial multiplex (SM) combined with multiple code usage In an open loop configuration, a single input data stream is split into multiple data streams that are mapped to different transmit antennas. With the PARC approach, described in more detail below, each sub stream is independently coded and modulated. In the SM approach, the input data stream is coded before it is separated and then the individual sub streams are modestly modulated and transmitted. field transmission diversity), also known as transmission diversity with rich feedback (TDRF), which is described in more detail below. In a closed-loop MIMO system using MFTD, the input data stream is transmitted from multiple transmit antennas 28 to a receiver 32 with one or more antennas receiving 38 The input stream is pre-filtered to match the channel between the transmitting and receiving antennas 28, 38. [0018] Fig. 5 shows an example transmitter configuration 100 for an open loop mode using an approach with individual control of individual antenna data transmission rate (PARC). It is obvious to the skilled person that PARC is a multi-stream method used in radiocommunication systems to increase system capacity. The transmitter 100 in this embodiment is configured for WCDMA (HSDPA High-Speed ​​Downlink Packet Access).

[0019] The transmitter 100 in Fig. 5 includes a demultipiekser 102 for dividing the original information bit stream and into M bit streams (tbbit). b_M(t)}, coding and modulation circuits 104 for each data stream, and a certain set of antennas 120 Coding and modulation circuit 104 for each bit stream (b ^ t / b ^ t), b_M(t)} includes an encoder 106 for encoding each bit stream {b₁(T), b₂(t), b_M(t)}, a certain set of demultipieksers 108 for further splitting of each stream {bi (t), b₂(t),. b_M(t)} to sub streams, a certain set of symbol mappers 110 for mapping each sub stream at a certain point in the signaling constellation, a certain set of signal splitters 112 for applying the selected spreading code to each sub stream, and an adder 114 for reconnecting the sub streams to generate the transmitted signal (x₁(T), x₂(t), x_M(t)} for transmission Broadcast signals {x₁(T), x₂(T)_l. x_M(t)} can be additionally combined by adder 116 with one or more other simultaneously transmitted signals d_m(t) which contain a number of specialized and control channels as well as a common pilot channel. Encoders 106 for each bit stream {b₁(T), b₂(t), b_M(t)} encode original information bits at different speeds. These speeds depend on the channel quality indicator (CQI) returned from the receiver. 32 Coded output signal

0 by each encoder 106 is then green by the 108 demultiplexers on the K sub streams. Each sub stream is mapped to symbols by one of the K symbol mappers 110, and scattered with one of the K spreading codes by the signal spreaders 112 These K spreading codes can be reused on other antennas 120 The combiner returns the K distributed signals from each spreader 112 signals fig 5, the number of encoded signals and the number

5 transmitting antennas 120 are the same However, in other embodiments, the number of encoded signals could be equal to or less than the number of transmit antennas 120 [0020] Fig 6 shows an exemplary transmitter configuration 50 for open loop spatial multiplex (SM) mode with multiple code use The transmitter configuration 150 shown in fig 6 includes encoder 152 for coding original information stream and generating a coded bit stream, demultipiekser 154 for dividing this encoded bit stream into N bit streams (b₁(T), b₂(t), b_M(t)}, modulation circuits 156 for each bit stream, and a certain set of antennas 168 The modulation circuit 156 for each bit stream includes a demultipiekser 158 for further dividing each bit stream into more and fewer streams, a number of symbol mappers

160 for mapping each sub-stream to a signaling constellation point, a set of diffusers

162 for applying the selected diffuse code to the sub streams, and the combiner 164 to recombine the sub streams to form the transmitted signal {Xi (t), x₂(t),. x_M(t)} for transmission to the second station 30 Broadcast signals (x₁(T)_ANDx₂(t),. x_M(t)} can be additionally combined by adder 166 with one or more than one of the other simultaneously transmitted d signals_m(t) which

0 include a number of specialized channels, control channels, as well as a common pilot channel. [0021] Fig. 7 shows an example transmitter configuration 200 for a system for operating in a closed loop mode. Transmitter 200 of Fig. 7 uses the method here called MFTD (matched field transmission diversity). Transmitter configuration 200 shown in Fig. 7 includes a demultipiekser 202 for dividing information bit stream into a number of bit streams, coding-modulation circuit 204 for each bit stream, pre-filter 212 matched to the communication channel between transmitter and receiver, and a certain set of transmit antennas 214 Each coding-modulation circuit 204 includes an encoder 206 for coding a respective bit stream, a symbol mapper 208 for mapping the coded bits into modulation symbols, and a signal spreader 210 for applying the selected spreading code to each sub-stream. Signals output from the modulation-coding circuits 204 are transmission signals s ^ t), p_N(t) for transmission to the second station 30, where N is less than or equal to the number of antennas at receiver 32. Transmission signals s ^ t), s_N(t) are introduced to the pre-filter 212 The pre-filter filters the transmission signals based on the knowledge of the communication channel between the transmitter 22 and the receiver 32. The pre-filter gives the output filtered signals (^ ί), x_M(t) to antennas 214 for transmission to second station 30, where M is the number of transmit antennas 214 [0022] The following discussion explains how the filters for pre-filter 212 are calculated. It is assumed that the signals s -, (t), s_N(t) introduced into pre-filter 212 are mutually independent, and each has a certain power spectrum A (f) inside the channel bandwidth Power spectrum of the transmission signal vector 5 ^) = ^ (t), s₂(t), p_N(t)] is denoted by P_ss (f) The power spectrum of the x (t) filtered signal can be expressed in terms of Pxx (f) according to:

«-I where A (f) is a real function selected to ensure that the total emitted power of the transmitter is fixed and G (f; n, m) is the frequency response of the downlink from the mth transmit antenna to the nth receive antenna [ [0023] As seen in Fig. 7, the MFTD 200 transmitter consists of N parallel data streams, i.e. subchannels and M antennas 214. Each of the N parallel data streams has two parts: 1) modulation-coding circuit 204, which maps identification bits to scalar waveforms, and 2) single-input, multi-output matched filter bank, which is part of pre-filter 212. When designating these H-matched filters (f, n, m), these filters can be represented by (2). [0024] The filters defined by Eq 2 have several interesting properties. First, their frequency response is the product of two words. The first word G * (f; n, m) in the meter is a filter matched to the downlink channel between the nth transmitting antenna and the nth receiving antenna^Dru9^ands^ra2 7ΣΪΣΪΐ^σ(ή ">, in the denominator can be considered as a whitening filter for the information signal. Secondly, Eq 2 results in a certain overflow effect in the following sense If one of the downlink channels on a specific frequency is strongly attenuated, then the output power from the corresponding transmission antenna 214 will also be proportionally attenuated on this frequency. With constant output power from the antenna 214, the transmitter 200 then transfer more power to these frequencies, where downlink channels are not impaired [0025] Some simplification can be achieved by neglecting the word dependent frequency iowo ^ -ι in the denominator This approximation leads to the following form of pre-filters 212:

// (/; w, / w) = ZG '(/; «, /«) Eq (3) where λ is a scale factor chosen to ensure that the total transmitted power is crj, i.e.

Eq. (4) [0026] The implementation of the filters given by equation Eq 3 is very simple; because the impulse response G * (f; n, m) is simply g * (- t; n, m). This implies that these filters can be implemented in the transmitter 200 as a simple FIR filter matched to the original downlink channel. Based on this observation, the downlink channel from the mth transmit antenna 214 to the nth receive antenna 38 can be modeled as jf g (t; n, m) = Xa_K(n, m) S (/ -r_t) Eq (5)

Asi [0027] Then the impulse response of the mth filter in the nth sub channel is in the form κ

Λχ <(/ ί,? Η) <5 (ί + ί ·_Α) _eq ,_θ) ł = 1 [0028] The base station A with the transmitter structure of fig. 7 only requires the shaping of these filters matched based on the knowledge of the downlink channels available in the base station. [0029] To reduce the load due to feedback in the case of the MFTD scheme, filters can be used h (t; n, m) adjusted on the basis of only partial information about the state of the channel The basis of the principle of the MFTD scheme is the use of suboptimal FIR initial filters with a limited number of branches. The coefficients a ^ (n, m) for these pre-filters can be selected as equal to L by the strongest channel coefficients using the fixed grid principle, where the grid of equidistant L fingers is arranged in the "region" of signal energy indicated by power / delay, and where L is less than or equal to the number of channel coefficients. With this simplified approach, the impulse response of the mth filter in the nth subchannel is postać (ζ; λι₃/ τι) ~ λ ^ α_κ -tr_x.)

Eq (7)

[0030] This MFTD scheme allows reducing the value of feedback and at the same time achieves most of the bandwidth with a closed loop of the scatter channels. The proposed configuration of the transmitter MFTD can also work with NC multicodes, which are repeatedly used in each transmitting antenna. The multicodes are repeatedly used to avoid excessive consuming a code resource when transmitting with multiple antennas [0031] According to the present invention, transmitter 20 can be configured for closed loop operation, i.e. for strong feedback mode, and for limited feedback mode, i.e. for open loop operation at different times, based on the expected receiver efficiency in each feedback mode. Factors related to to determine the expected efficiency include channel parameters, code and power availability, receiver antenna configuration, operating conditions In general, closed-loop mode is selected, when the expected increase in efficiency in closed-loop mode compared to open-loop operation justifies the additional load overhead required for feedback of channel coefficients In the open-loop mode the transmitter configurations shown in Fig. 5 and / or 6 can be used the loop configuration of the transmitter shown in fig. 7 may be used. [0032] When the conditions change during transmission, the feedback mode can be change, based on one or more quality measures indicating the expected performance of the receiver. For example, one of the quality measures may be the SINR at the output of the receiver 32 or the transfer rate of the receiver 32 that can be ensured. The measure of quality may depend on a number of factors including the number of receiving antennas 38, scalar or vector speed, and channel state. Thus, according to the present invention, to maximize efficiency, there is an adaptive feedback change as operational conditions change, [0033] In addition to adaptation of the feedback mode feedback levels additional adaptation levels can be used For example, a selection can be made from two different feedback modes shown in Figs. 5 and 6. In addition, a multi-path mode selection can be made after the feedback mode has been selected. The selection of a multi-path mode may depend, for example, on channel parameters, code availability and power availability.

[0034] When open loop mode is selected, different multi-path modes may correspond to different antenna configurations at transmitter 22 Thus, the choice of multi-path mode means selection of the number of antennas 28 and selection of which antennas 28 are to be used for transmission to mobile station 30 starting work in the scheme with selective PARC (S-PARC) (fig. 5) or in a scheme with selective spatial multiplex coding (S-SM) (fig 6) When choosing a closed loop mode, different multi-path modes correspond to different pre-filter configurations. So, the choice of multi-path mode means the choice of the number of pre-filters 212 or data streams in the transmitter and the choice of the number of receiving antennas 38 at the receiver 32 to which the pre-filters 212 are matched when starting work in the selective scheme MFTD (S-MFTD) [0035] In the following embodiment . multi-path mode in the case of open-loop feedback mode is selected on the basis of the expected SINR ratio or the achievable data transmission speed at the receiver output 32 Both S-PARC, as well as S-SM are described additionally for the conditions of high speed packet access to the downlink (HSDPA - High-Speed ​​Downiink Packet Access) in the standard wideband multiple access with code division (W-CDMA - Wideband Coda Division Muitiple Access) as an example In the following the case of two transmitting antennas was considered 28 Possible antenna configurations are {A}, {B}, or {A, B} We will also introduce an additional case {B, A} for S-PARC Here, ordering indicates the order of decoding operations in the operation of the receiver with S-PARC. In the case of S-PARC, the decoding order may have little effect on the supported data transmission speed in dispersion channels. Therefore, it can be beneficial as far as attention to maximizing the achieved data transmission speed, considering a different order in the decoder for a given combination of antennas. However, in most channel conditions, the difference between the achievable data rates for {A, B} and {B, A} is very small, and therefore, to minimize complexity, it may be beneficial to consider only one decoder ordering for a given antenna combination [ 0036] In the case of S-PARC, active antenna 28 transmits a self-decoding data stream The data stream transmission speed is determined by the signal-to-interference-noise ratio in the symbol signal (SINR signal-to-interference-plus-noise), characterizing the performance of the receiver In the following discussion, we will use the output SINR ratio as an example for SIC-GRAKE, SICGRAKE is described in pending US Patent Application No. Kol 00 / 795,101 filed March 5, 2004. [0037] It is assumed that all available power is evenly distributed among all active transmit antennas 28 For antenna configuration {A}, all available P power is allocated to transmit antenna A, and as a result the power per channel code is P / K, where K is the number of channel codes Since there is only one active antenna 28, there is no need to carry out successive interference cancellation after G-RAKE. In this case, the SINR output for G-RAKE can be represented as

<img file="PL1716655T3_D0001.tif" />

Eq (8) where h (A) is the network response corresponding to the channel assigned to the antenna A, and R {a} (A) is the noise covariance The lower noise covariance index indicates that the noise covariance depends on the antenna configuration. In practice, in this case, both network response and noise covariance can be determined from measurements of Common Pilot Channel. R relationship_{AND} (A) = Rcpich (A) means that the noise covariance can be determined by measuring the common pilot channel. Then you can determine the serviceable data transmission speed at the SINR output from G-RAKE by viewing the table or using the SINR mapping function for data transmission speed,

Raie ({A}) ^ f (SlNR?))

Eq (9) [0038] Similarly, for the {B} antenna configuration, the output SINR for G-RAKE can be represented

5 as,

Eq (10) where h (B) is the network response for the channel assigned to the antenna B and R_{B} (B) is the noise covariance Also in this case both the network response and the noise covariance

SINR ,,, (2?) = 2 h (B)^H R7 '_s. (B) h (J3),

Λ can be obtained from the measurement in the common pilot channel Dependency R_{B} (B) = Rcpich (B) means that the noise covariance can be determined by measuring the common pilot channel. There are cases in which Rcpich (A) = Rcpich (B) The serviceable data rate for this case is

Rate ({B}) = f (SINR,_li}(B)) [0039] For the antenna configuration {A, B}, power P / 2 is allocated to each antenna; antenna A signal will be decoded first and antenna B signal will be decoded last. In this case, the SINR at the G-RAKE output for antenna signal A is = Eq (12) '2X

R_AND (Λ) = ĄH (4) + - (β) where the noise covariance is '' 2K The additional word to the right of the equation explains the disruption of code reuse occurring on the same channel used to transmit the antenna signal B [0040] For antenna signal B SIC will be used because S / NR at the G-RAKE 5 output reflecting SIC is

SJNR., = _Eq 13) p

where the noise covariance is ^ ah) (B) - R_ct,_{fi H} (E) The second word to the right of the equation explains the interference from the antenna signal A, which is removed by the SIC process. The allowable data rate for this antenna configuration can be determined by

Rate ({A, B}} = / (S / NR_[AND ", (/ ()) + / (SM? ,,," (£)) Eq. (14) [0041] Similarly, for the {B, A} antenna configuration, the SINR at the G-RAKE output for the antenna signal B is SJNR._e (B) =, where the noise covariance is

R., / Β ·) = + Zh (A) h^HIA) Eq (15) [0042] For antenna signal A, SINR at the G-RAKE output, reflecting the SIC is p

SINR_{h A}. (A) = —R7 ©_}(A) h (A) ·, where the noise covariance is '' 2 / l

P

KfriCH (^) ~ 2 "^ C #) The serviceable baud rate for this antenna configuration can be determined by

Rate ({B, 4}) =, / (S / NR ,,, "(β)) + / (SINR ,,,,, (A))

Eq (16) [0043] Having acceptable data transmission rates for individual antenna configurations, the base station controller 24 or the mobile station controller 34 can choose the configuration that provides the highest permissible data transmission speed. The above example shows that the calculation of the permissible data transmission speed depends on the allocated power (P), the assigned code (K), and the channel parameters h (A), h (B), R_{AND} (A), R_{B&gt; (B), R {ab} (A), R_{AB)(B), R {ba} (A), R_{Ba} (B)) It is obvious to the skilled person that this can easily be generalized to a case involving more than two transmit antennas 28 [0044] The same procedure applies to selective spatial multiplexing code as described previously for S-PARC. The only difference is SINR formulation and mapping function 10 SINR-to-speed-data-transmission for antenna configurations {A, B} Achievable data transmission speeds for antenna configurations {A}, and {B} for S-SM are exactly the same as for S-PARC. So we will only discuss the case, {A, B} [0045] The SINR values ​​at the G-RAKE output in this case are SINR ,, "(A) = A-HFA)^M and

SINR ^ "(S) = XA (Ą" Λ, -J, "(S) A (S)

Eq (17)

Eq (18) where the noise covariances are R_{l / t} (/) = R_{(f / (H} (A) + —— h (B ') h^at (5) ip

^ Cr / c / Λ ^ ⁺(^AND) The basis for one of the methods of determining the achievable data transmission speed is the sum of individual SINR I values ​​is given as Rale ({A, B}) = / (SIFR ^^ A) ® f (SłNR._alt](B)) Eq (19) [0046] Other methods may also be considered, such as those based on the geometric mean (instead of the arithmetic one) Having acceptable data transmission speeds for individual antenna configurations, the base 24 controller or the 34 station controller Mobile can choose the configuration that provides the highest permissible data transmission speed. The above example shows that the calculation of the permissible data transmission speed depends on

5 allocated power (P), assigned code (K), and channel parameters (h (A), h (B), R_{AND} (A), P {B} (B), R {abj (A), R {ab) (B), R {ba} (A), R {ba} (B)) [0047] Decision making for determining the multi-path mode for an open loop transmitter configuration can take place at mobile station 30, base station 20, or in the division between base station 20 and mobile station 30 In the first approach (all decisions made at mobile station 30), mobile station 30 estimates SINR values ​​for all possible antenna combinations and select this particular combination, which as a result gives the maximum total data transmission speed in the sub streams. Then it gives back the baud rate indicator, i.e. CQI (Channel Quality Indicator) for each stream, i.e. if N-mode is selected, where N is the selected number of antennas, then N CQI indicators are signaled back, where N &lt;M The CQI indicator is typically a mapping from the size of the SINR ratio to the data transmission rate. It also requires selection of antenna selection signaling from 2^m-1 possibilities, requiring M bits [0048] At the second approach (all decisions made at base station 20), mobile station 10 estimates the SINR values ​​for all possible combinations of modes and antennas selection

However, since no decision is made at the mobile station 30 as to which is best, the mobile station 30 must provide CQI values ​​for all possible combinations back, which leads to a very high load on the feedback. For example in the case of s-parc with 4 with transmitting antennas 28, mobile station 30 would have to feed back one CQI indicator for each of the 4 modes-1 antenna, 2 CQI indicators for each of the 6 modes-2 antenna selection, 3 CGI indicators of each choice 4 modes-3 antennas, and 4 CQI indicators for each of the choices single mode -4, which would result in a total of 32 CQI values ​​This is compared with the first approach, where the feedback load is 4 or less CGI indicators and one antenna selection indicator As such, providing a full base station

0 freedom in choosing both the mode and the antenna, requires a large amount of feedback.

[0049] According to the third type of approach (decision making separation between mobile station 30 base station 20), mobile station 30 selects antennas for each possible mode and base station 20 selects mode. Thus, the size of the feedback is significantly reduced compared to the second approach Actually, if the antenna choices are limited to providing a certain property (called a subset property), the feedback can be reduced to a level just slightly higher than in the first approach. Although this approach is suboptimal, the loss of performance is small. However, without using the subset properties, the load for the example with M = 4 transmitting antennas would be 1 CQI for mode-1, 2 CQI for mode-2, 3 CGI for mode-3, and 4 CQI for mode-4, resulting in 10 indicators CGI

In addition, an antenna selection would be required for each mode Undoubtedly, the required feedback load is much smaller than the second approach, which requires 32 CQI indicators. [0050] As explained in British Patent (UK) No. 2358071 issued April 2, 2003, if selected property of a subset, this feedback load can be reduced to just M = 4 CQI and one decoding sequence. It's just a bit more, than the minimum feedback load for the first approach, which requires N = 4 CQI indicators (when mode-4 is selected) and one antenna selection indicator. The decoding sequence is the permutation of M integers, and indicates in what order the streams should be decoded for each possible mode Because M exists! possible

0 decoding sequence, reverse feed of the decoding sequence requires slightly more bits than just the antenna selection indicator. For example, with M - 4, the decoding sequence requires ceil (log2 (Ml)) = 5 bits as opposed to the antenna selection indicator which only requires ceil (! HC2 (2^M13

1)) = 4 bits. SINR estimation and feedback can be done in a similar way in the case of SM, with the only difference being that one CQI indicator is estimated and returned for each antenna combination. The concept of multi-path mode selection can also be used for Closed loop transmitter configuration of Fig. 7 According to the present invention, when the conditions change, the number of data streams is adaptively changed Each data stream uses a set of suboptimal filters given by the equation Eq 7 In addition, depending on the channel conditions, the "best" set of subsets of streams that are intended for transmission is selected, which is a type of channel selection One interpretation the term "best" is the subset of streams transmitted that provides the highest transmission speed.

[0052] The selection of the multi-path mode can be performed either by the mobile station controller 34 or by the base station controller 24. In the first case, the mobile station estimates the Si NR values ​​for all possible combinations of transmitted data streams and selects the specific combination that results in the maximum total transmission speed. Then it returns the CQI indicator for each stream and the corresponding set of channel coefficients. For example, if the mode is selected -N, then N CQI indicators and (Μ x L χ N) channel coefficients are transmitted back, where M is the number of transmitting antennas 28 and L is the number of transmitter pre-branches 212 of the transmitter It should be mentioned that N &lt;N_RX, where N_RX is the number of receiving antennas 38 In the second case, the mobile station gives back (Μ x L x N_RX) channel coefficients in some situations, you may need interference power from another cell to transfer to the base station.

[0053] The present invention can, of course, be implemented by other specific routes than those shown without departing from the scope of the essential characteristics of the invention. These embodiments should therefore be considered in all respects as illustrative and not

<img file="PL1716655T3_D0002.tif" />