Detailed Description
The term "station" (STA) hereafter includes, but is not limited to, a user equipment, a wireless transmit/receive unit, a fixed or mobile subscriber unit, a pager, or any other type of device capable of operating in a wireless environment. Hereinafter referred to as an "Access Point (AP)" includes, but is not limited to, a node B, a base station, a site controller, or any other interfacing device in a wireless environment.
The present invention will be described with reference to the drawings, wherein like reference numerals represent like elements throughout. It should be noted that the illustrations provided in this invention are high-level functional block diagrams, and the functions performed by the functional blocks may be performed somewhat by blocks. The features of the present invention may be incorporated into an Integrated Circuit (IC), or be configured in a circuit comprising a multitude of interconnecting components.
Embodiments of the present invention provide a transmitter and receiver matched filter that can implement space frequency block coding multiple-input multiple-output coding. Embodiments also provide transmitter channel precoding and receiver antenna processing and channel decomposition functions.
The system operation has two modes: closed loop and open loop. The closed loop is used when the transmitter has channel state information available. The open loop is used when channel state information is not available. The variation may be used for transmission to the native station that may provide diversity benefits.
In closed loop mode, channel state information is used to create virtual independent channels by decomposing and diagonalizing the channel matrix and precoding at the transmitter. Spread out over the eigenvalues of a given TGn channel, the present invention adds robustness using space-frequency multiple-input multiple-output coding in the transmitter at the input of the channel precoder to reduce the data rate penalty. Any coding scheme in mimo must handle diversity versus multiplexing gain permutation. It is expected to have the permutation scheme best suited for the particular channel statistics. Space frequency block coding is selected for low mobility and long channel coherence time. This scheme allows a simpler receiver implementation than a Minimum Mean Square Error (MMSE) receiver. The combined solution may contribute to higher yield over a wide range. Embodiments of the present invention allow per-carrier power/bit loading and maintain a continuous robust link through closed loop operation of channel state feedback. Another potential benefit is that it can easily measure any number of antennas at the transmitter and receiver.
Channel state information may be obtained from the transmitter by feedback from the receiver or by exploiting channel reciprocity. Channel reciprocity is useful in a primary fdd based system. In this case, the transmitter and receiver can independently estimate and resolve the channel. The channel update rate can be reduced when the feedback bandwidth loading is reduced due to a high signal-to-noise ratio. The inherent frequency non-selectivity of latency requirements and feedback data rates to the inherent values is generally not significant.
The closed loop mode requires calibration of the transmitter to compensate for estimated channel jitter and phase differences in the uplink and downlink directions. This is not often done during station association or under application control, for example, and channel reciprocity can be used to estimate the channels at both ends. In addition, a Channel Quality Indicator (CQI) (or signal-to-noise-signal ratio) for each eigenbeam is fed back to the transmitter to support adaptive rate control.
Fig. 1 is a block diagram of an ofdm-mimo system 100 implementing a closed loop mode. The system 100 includes a transmitter 110 and a receiver 130. The transmitter 110 includes a channel encoder 112, a multiplexer 114, a power loading unit 116, a plurality of space frequency block coding units 118, a plurality of serial to parallel (S/P) converters 120, a plurality of eigen-beamformers 122, a plurality of ifft units 124, and a plurality of transmit antennas (not shown). The channel encoder 112 may preferably encode data with a channel quality indicator transmitted from the receiver 130. The cqi is used to determine the coding rate and modulation scheme per carrier or sub-carrier group. The encoded data streams are multiplexed into two or more data streams by multiplexer 114. The transmit power level of each data stream is adjusted by the power loading unit 116 based on feedback. The power loading unit 116 adjusts the power level for each eigenbeam data rate to balance the total transmit power on all eigenbeams (or subcarriers), as will be explained in detail below.
The spatial frequency block coding unit 118 performs spatial frequency block coding on the data stream. Space frequency block coding is achieved on the native beams and subcarriers for the data rate being transmitted. The eigenbeam and sub-carrier pairs are selected to ensure independent frequency channels. The OFDM is performed on K sub-carriers. The bits accommodate space frequency block coding, and the subcarriers are divided into L pairs of subcarriers (subcarrier groups). The bandwidth of each set of subcarriers should be less than the channel coherence bandwidth. However, when combining eigen-beamforming, this limitation is relaxed due to the eigen-beam frequency slowness.
The set of sub-carrier groups used for block coding is considered separately. The following is an example of Alamouti type space frequency block coding applied to ofdm:
img id="idf0001" file="A20058002738800091.GIF" wi="104" he="51" img-content="drawing" img-format="GIF"/
once the space frequency block coding unit 118 constructs the ofdm symbols for all subcarriers, the coded blocks are multiplexed and input to the eigenbeamformer 122 by the serializer 120. The native beamformer 122 distributes native beams to transmit antennas. The ifft unit 124 transforms the data in the frequency domain into data in the time domain.
The receiver 130 includes a plurality of receiving antennas (not shown), a plurality of fast fourier transform units 132, a eigen-beamformer 134, a spatial frequency block codec 136, a combiner 138, a channel decoder 144, a channel estimator 140, a channel state information generator 142 and a channel quality indicator generator 146.
The fft unit 132 converts the received samples to the frequency domain, and the eigenbeamformer 134, the space frequency block codec unit 136 and the channel decoder 144 perform the inverse operations that are performed at the transmitter 110. The combiner 138 combines the results of the space frequency block coding decoding using Maximal Ratio Combining (MRC).
The channel estimator 140 uses the training queues transmitted from the transmitter to generate a channel matrix, and decomposes the channel matrix into two beamforming unitary matrices U and V (U for transmission and V for reception) per carrier (or per carrier group) and an diagonal matrix D by Single Value Decomposition (SVD) or eigenvalue decomposition. The channel state information generator 142 generates channel state information from the channel estimation result, and the channel quality indicator generator generates a channel quality indicator based on the decoding result. The channel state information and the channel quality indicator are transmitted back to the transmitter 110.
The channel matrix H between nT transmit antennas and nR receive antennas can be written as follows:
the channel matrix H is decomposed by single value decomposition as follows:
H＝UDVH，
where U and V are a single matrix and D is a diagonal matrix. U is belonged to CnR×nRAnd V ∈ CnT×nT. Then, for the transmitted symbol vector s, transmit precoding is simply performed as follows:
x is Vs (transmitted signal).
The received signal becomes as follows:
y＝HVs+n，
where n is the noise signal introduced into the channel. The receiver uses a matched filter to perform the decomposition:
VHHH＝VHVDHUH＝DHUH。
after generalizing the channel gain of the eigenbeam, the estimate of the transmitted symbol s becomes
＝αDHUHVs+η
＝s+η
s is detected without performing successive interference cancellation or a minimum mean square error type detector.
DHD is a diagonal matrix formed by H eigenvalues across the diagonals. Thus, the generalization factor α ═ D-2. U is HHHIs an eigenvector of, V is HHHAnd D is a single value diagonal matrix of H (HH)HThe square root of the eigenvalue of).
FIG. 2 is a block diagram of a system 200 implementing open loop mode in accordance with the present invention. The system 200 includes a transmitter 210 and a receiver 230. In open loop mode, space frequency coding and space spreading in the transmitter 210 are combined to provide diversity without the need for channel state information. This variant of the scheme can be used when operating the native 802.11a/g station.
The transmitter 210 includes a channel encoder 212, a multiplexer 214, a power loading unit 216, a plurality of space frequency block coding units 218, a plurality of serial to link converters 220, a beam shaper network (BFN)222, a plurality of inverse fast fourier transform units 224, and a plurality of transmit antennas 226. As in the closed loop mode, the channel encoder 212 uses the channel quality indicator to determine the coding rate and modulation for each sub-carrier or sub-carrier group. The encoded data streams are multiplexed into two or more data streams by a multiplexer 24.
In the open loop, the native beamformer is replaced by a beamformer network 222. Beamformer network 222 forms N beams in space, where N is the number of antennas 226. The beams are constructed pseudo-randomly by matrix manipulation of the beamformer network. Groups of independent sub-carriers used for space frequency block coding are transmitted on the respective beams.
For inherent support, space frequency block coding may not be performed. Alternate diversity via beam permutation is performed which improves the performance of the native 802.11a/g device.
The receiver 230 includes a receiving antenna 231, a fast fourier transform unit 232, a beam former network 234, a space frequency block coding decoding and combining unit 236 and a channel decoder 238. The fast fourier transform unit 232 transforms the received signal in the time domain into a signal in the frequency domain. The space frequency block coding decoding and combining unit 236 decodes and combines the symbols received from the sub-carrier groups/eigenbeams and converts them from parallel to serial using the prior knowledge of constellation size. Symbols are combined using maximal ratio combining. The channel decoder 238 decodes the combined symbols and generates a channel quality indicator.
The first embodiment of the power load is explained as follows. Spatial processing is a combination of space frequency block coding and eigen-beamforming. This is performed to give the best compromise between the redundant gain suffered by the space frequency block coding and the spatial multiplexing provided by the inherent beamformer. The power loading scheme operates across the channel matrix eigenmodes. However, space frequency block coding also introduces the limitation that the encoder output has the same power load regardless of the input power load due to inter-encoder interleaving.
Fig. 3 is a block diagram of a transmitter 110 depicting a power load. Fig. 3 depicts the 4 × 4 case as an example, and the first embodiment of the power load will be explained with reference to the 4 × 4 case. It should be noted, however, that the 4 x 4 case can be extended to any other case.
For a particular subcarrier k, four data streams are mapped to 2 pairs of power loading/Adaptive Modulation and Coding (AMC) modes. That is, the modulation order of each pair of inputs is selected to be the same. This is later mapped to the eigenmode pairing. The output of the power loading unit 116 is applied to a dual 2 x 2 space frequency block coding unit 118 and then sent to the native beamformer 122. The eigenbeamformer 122 maps the inputs to the eigenmodes of the channels through pre-processing.
For all k subcarriers, the inherent values of the channel matrix are known to be located at the transmitter. The channel energy for each eigenmode is defined as follows:
img id="idf0003" file="A20058002738800111.GIF" wi="99" he="42" img-content="drawing" img-format="GIF"/
wherein λi，kIs the ith eigenvalue of the channel of the kth subcarrier. The two signal-to-noise-plus-interference ratio (SNIR) is defined for two coupled eigenmodes as follows:
img id="idf0004" file="A20058002738800112.GIF" wi="106" he="42" img-content="drawing" img-format="GIF"/andimg id="idf0005" file="A20058002738800113.GIF" wi="117" he="43" img-content="drawing" img-format="GIF"/
where M is the number of eigenmodes. That is, the eigenmodes are grouped such that half of the eigenmodes with the largest channel energy (or signal-to-noise-plus-interference ratio) are in one group and the other half with the weakest channel energy are in the other group. Thus, the harmonic signal-to-noise plus interference ratio represents the total channel energy of the stronger and weaker eigenmodes. The channel energy is an indicator of how strong the eigenmodes are, and therefore the signals carried on these eigenmodes will be. This information is used to apply different adaptive modulation and coding and/or different power loading to the halves as explained in more detail later. The separation of the noise plus interference ratio of the coupled signal is defined as follows:
Δβ＝βmod1-βmod2
during closed loop operation, the transmitter 110 has knowledge of the current channel state information from which it can extract the eigenvalues and the preprocessing matrix. The transmitter 110 also deduces from the channel state information the data rate that can be supported in the link, Rb. Then, the power loading given the acceptable channel state information is an optimization between the modulation types to be used by each mode for the number of bits that can be transmitted per OFDM symbol.
As described above, using the channel energy calculated for eigenmode i, the maximum bit rate that can support the channel conditions is determined. Thus, using the above mode independent calculation, it is possible to determine how the bit rate must be allocated between the two pairs of modes. Fig. 4 is a diagram of power loading and adaptive modulation and coding mapping between two pairs of modes. In this example, the bit rate at which a particular sub-carrier can be supported is 24 bits per OFDM symbol. The lowest modulation order that satisfies the bit rate is found as indicated by the dashed arrow in fig. 4. In this example, the first and second modes (the first pair of coupled modes) would use 16 Quadrature Amplitude Modulation (QAM) and the fourth mode (the second pair of coupled modes) would use 256 quadrature amplitude modulation.
Note that this mapping is described for acceptable one cqi and for one sub-carrier. In an alternative multiple input multiple output configuration, such as 2 x 4, 2 x 2, etc., the same power loading scheme is acceptable except that the total number of bits in the table entry is drawn down to represent the transmission capability and the power loading can be achieved in a single pair mode.
The power loading scheme according to the second embodiment is explained as follows. Intrinsic value of each sub-carrier (λ)1(k)＞λ2(k)＞...＞λnT(k) Are ordered, and the eigenbeams (E)1，E2，...，EnT) The following is created by grouping the same ordering eigenvalues for all sub-carriers:
Ei＝{λ1(1)，λ2(2)...，λi(k)}，i＝1，2，...，nT，
where K is the number of subcarriers, nT is the number of transmit antennas, and λi(j) The ith eigenvalue of the frequency channel of the j sub-carriers. NT is an even number.
The eigenvalue average per eigenbeam is calculated as follows:
img id="idf0006" file="A20058002738800121.GIF" wi="252" he="45" img-content="drawing" img-format="GIF"/
the eigenbeams are paired to create an Almouti spatial frequency code, e.g. { E }1，E2}1，{E3，E4}2，...，{E2i-1，E2i}i...{EnT-1，EnT}nT/2. However, if the signal-to-noise-signal ratio is greater than the SNRmaxThe pair of second eigenwaveforms is replaced by an eigenbeam having a next lowest eigenvalue average until its signal-to-noise-signal ratio is less than or equal to the SNRminUntil now.
SNR(i)=(λavi+λavi+1)/σn2,
Wherein sigman2For noise signal variation, SNRminThe minimum required signal-to-noise-signal ratio for the highest data rate required for quality of service. This step is repeated until all of the eigenbeams are paired. Fig. 5 shows an example of power/bit loading subcarrier group pairing.
The data rate for each pair of eigenbeams is determined by mapping a pair of signal-to-noise-signal ratios to the data rate for a given quality. The desired signal-to-noise signal ratio can be adjusted for all eigenbeam pairs to compensate for measurement errors and make the total transmit power fixed.
The weight vector for each pair of eigenbeams per carrier can be calculated as follows:
img id="idf0007" file="A20058002738800132.GIF" wi="167" he="51" img-content="drawing" img-format="GIF"/
where i is the ith pair of eigenbeams and j is the jth subcarrier.
In addition to the first or second embodiment, according to a third embodiment, another power load is applied across the subcarriers or subcarrier groups for weak eigenmodes. That is, in addition to being applied to the power loads of all the natural modes, it may be applied only to the weaker one, and thus may benefit most from the power loads. In this case, these eigenmodes not loaded by power may still have space frequency block coding or may have different individual settings of adaptive modulation and coding, while these eigenmodes loaded by power share the same adaptive modulation and coding settings. At the same time, the eigenmodes of the channel are always ordered from strongest to weakest power. By matching the eigenmodes of similar power, the power loading of the channels can be improved.
The spatial processing scheme may be configured for any combination of receive and transmit antennas. Depending on the number of antennas on each side, a combination of space frequency block coding and eigen-beamforming selection is used. The following table summarizes the spatial processing and power loading that can be applied to each case for the various configurations supported.
| Antenna configuration (TxX Rx) | Space frequency block coding | Intrinsic beamforming |
| M×N(M，N≠1) | M/2 block coding | N-beam at M-beam receiver at transmitter |
| 1×N(N≠1) | Is not used | By receiver to determine |
| M×1(M≠1) | M/2 block coding | M-beams at transmitter |
TABLE 1
Although the features and elements of the present invention are described in the preferred embodiments in particular combinations, each feature or element can be used alone without the other features and elements of the preferred embodiments or in various combinations with or without other features and elements of the present invention. Furthermore, the present invention may be implemented in any type of wireless communication system.