The invention relates generally to multiple-input multiple-output (MIMO) communication systems, and more particularly to generating training frames for MIMO systems.
It is well known that using multiple-input multiple-output (MIMO) techniques in a wireless communication system significantly increases capacity in a multipath environment. However, multiple antennas increase complexity and cost because each transmit antenna and each receive antenna requires a separate radio-frequency (RF) chain including a modulator and/ or demodulator, AD/DA converter, up/down converter, and a power amplifier.
Antenna and/or beam selection can reduce the number of RF chains, while still taking advantage of the capacity and diversity increase provided by multiple antennas and/or beams. With antenna selection, each input/output RF chain is associated with one selected antenna. Antenna selection depends on small-scale fading, which varies with frequency. Therefore, an antenna selected for one frequency is usually not appropriate for some other frequency when the two frequencies are separated by more than one coherence bandwidth.
Beam selection depends on the path angles of arrival, which are approximately the same for the entire frequency band of interest. Beam selection associates each input/output RF chain with a selected beam, which can be formed by a linear transformation of the signal vector including the received/transmitted signals at all antennas.
In antenna and/or beam selection, typically, a channel submatrix is selected from a complete channel matrix, or a transformed channel matrix for beam selection, according to some criterion. To implement antenna and beam selection, the channel matrix is estimated by sending training frames that enable the two stations to estimate characteristics of the channel completely.
In the case both the stations have the selection capability, by reciprocity of the channel, the estimated channel should be the same in both directions, and both stations can select the same submatrix independently, without an explicit exchange of selection results. Then, the selected submatrix can be used for coherent detection of transmitted data frames.
However, because of channel ambiguity caused by estimation error or channel variation caused by time difference between training frames in different directions, the observed channel is different when a station is operating in transmit mode or receive mode, and the independent antenna selection can cause the two stations to select different submatrices. If different submatrices are used by the two stations, then the performance of the system can be degraded severely.
To solve this problem, explicit signaling can be used to exchange the information about the selection in either the physical (PHY) layer or the media access layer (MAC) layer of the stations. However, the additional signaling information in the physical (PHY) layer or signaling delay in the MAC layer is undesirable due to practical limitations.
The invention provides a method and system to transmit training frames in a MIMO system.
A long sequence of training frames is transmitted from a station B to a station A via a channel of the MIMO system in which the station A includes NA antennas and NA
A number of frames in the long sequence of training frames is at least equal to
where
is a lower bound operation, i.e., a smallest integer larger or equal to (NB.NA/NA
A short sequence of training frames is transmitted from the station A to the station B via the channel in response to receiving the long sequence of training frames. A number of frames in the short training sequence is at least equal
As shown in
This switching process is commonly referred to as antenna or beam selection 251. The parameters NA=NB and NA
The station B, operating in a transmit mode, generates 111 a long sequence of training frames (TF) 200, see
where
denotes a lower bound, i.e., a smallest integer larger or equal to (NB.NA/NA
Therefore, as defined herein, the long sequence of training frames includes at least one training frame for each of the transmit antennas 201-204 and each of the receive antennas 301-304 as shown in
The long sequence of training frames 200 are transmitted to the station A operating in receive mode via the channel 130. The station A estimates 121 complete characteristics of the channel 130 from the long sequence of training frames. The complete channel characteristics can be used to select a subset of available antennas or beams in the station A. The subset has fewer members than the total number of available antennas or beams. For example, antennas 301-302 are selected. Typically, the size of the subset is NB
According to the complete channel characteristics, the station A, operating in the transmit mode, generates 122 a short sequence of training frames 300, see
The short training frames 300 are transmitted to the station B operating in the receive mode via the channel 130 via the selected antennas. The station B estimates 112 partial characteristics of the channel from the short sequence of training frames. The partial channel characteristics can be used to select a subset of antennas or beams in the station B. This subset has fewer members than the number of available antennas, e.g., antennas 201-202, see
The estimated channel characteristics and the selected antennas or beams can be used subsequently to transmit data frames (DT) from the station A, operating in transmit mode, 120 to the station B 110, operating in receive mode, via the channel 130 and vice versa.
Model for MIMO Systems
In a flat-fading MIMO system with NA antennas at the station A 120 and NB antennas at the station B 110, a relationship between transmitted and received signals can be expressed as:
rB=FBH(HA→BFAsA+n),
where rB is a NB
A matrix FA is a NA×NA
An equivalent channel matrix after antenna or beam selection is a NB
Antenna Selection
For antenna selection, the selection is simply done by switching an output signal from an RF modulator chain to the selected transmit antenna, or the input signal from the selected receive antenna to an RF demodulator chain.
Beam Selection
For beam selection, the transformation can be implemented in the RF domain using some or all of the following components: a Butler matrix, phase-shifters, attenuators/amplifiers, and linear combiners and switches, and a following switching circuit. Alternatively, transformation and selection can be done jointly, again employing phase-shifters, attenuators/amplifiers, and linear combiners. In both cases, the required number of RF chains to modulate/demodulate the transmitted/received signals is less than the total available number of transmit and receive antennas, and the complexity and cost of the system is reduced.
Training Frames
The station A 120 estimates a complete channel matrix HA→B representing the complete channel characteristics, from station A to station B, to determine the selection matrix FA. This is achieved by transmitting the long sequence of training frames 111 from the station B to the station A. According to a reciprocity of the channel 130, the estimated channel matrix HB→A, from the station B to the station A, at the station A 120 is a transpose of the matrix HA→B, i.e., HB→A=HA→BT.
A similar process can be performed by the station B 110 using the short sequence of training frames. Note that the determination of selection matrices FA and FB is done independently at both stations based on their estimated channel matrices. The selected antennas or beams can be different at the two stations because of channel estimation errors or channel variations in the different directions. This discrepancy can sometimes cause severe performance degradation.
For an example MIMO system, let both stations A and B operate according to a “two-out-of-four” antenna selection policy. The estimated channel at the station A 120 is
and the selected transmit/receive antennas at station A 120 are {1,2} and {1,2}, respectively, so that a desired equivalent channel is
At the station B, the estimated channel matrix is
due to some channel ambiguity. Thus, the selected transmit/receive antennas at the station B are {3,4} and {3,4}, respectively, so that the desired equivalent channel is
Without some exchange of information regarding the selected antennas at both stations, station A uses transmit antennas {1,2}, while station B uses receive antennas {3,4}, and the resultant channel matrix is
In the prior art, this problem is solved by providing some information exchange mechanism in either the PHY layer or the MAC layer so that both stations use the same set of selected antennas.
However, for the PHY layer, the addition of a field in the preamble of each frame for this purpose is undesirable because this field is only used in the training frames during the selection process. The field is not needed when data frames are transmitted subsequently. In the MAC layer, the delay caused by the additional information exchange is also a concern. Given the reasoning above, we determine the antenna/beam selection sequentially in the two stations: first in the station that will receive data frames, and then in the receiver transmitting the data frames.
The long sequence of training frames 200 are transmitted from the station B 110 to the station A 120 for complete channel estimation. Antennas/beams at the station A are selected based on the estimated complete channel matrix ĤA→B using the long sequence of training frames.
The short sequence of training frames 300 are transmitted from the station A to the station B for estimating a partial channel matrix corresponding to the number of selected antennas/beams at the station A. Antennas/beams to be used at the station B are determined based on the estimated partial channel matrix.
Note that transmission of any data frames subsequently is independent of the selected channel submatrix. A straightforward solution is to use a fixed channel submatrix for all stations. For example, by transmitting a complete channel training frame from the station B to the station A, the estimated complete channel matrix is
and the selected transmit/receive antennas determined at the station A are {1,2} and {1,2}, respectively.
By sending the short sequence of training frames using the selected transmit antennas 1 and 2 at station A, the estimated partial channel matrix is
and the selected antennas at station B are 1 and 2. The equivalent channel matrix becomes
Structure of Training Frames
Due to the limitations on the number of available RF chains for data modulation and demodulation, the training frames are organized in an alternating manner, in where only a subset of antennas are used concurrently for signal transmission and reception.
In
As shown in
Due to the transition time needed for switches to settle to a stable output, some additional protection interval is inserted between consecutive training frames. This additional interval can be taken into account in the guard interval in OFDM systems when the transition period plus the maximum delay spread of the channel is less than the guard interval.
For the case when only one station has antenna/beam selection capability, the training scheme simplifies to a one-step process that only includes antenna/beam selection at station B.
The method described above can also be applied to the cases where the system is frequency-selective, such as an OFDM systems designed according to the IEEE 802.11n standard, because antenna selection and RF-band processing can be implemented independent of frequency. The RF-baseband processing has the advantage that the performance gain is independent of frequency-selectivity, while the gain by antenna selection tends to be averaged out by frequency-selectivity.
In OFDM systems, it is also possible to frequency-interleave training frames, so that different tones of training frames are transmitted simultaneously from several transmit antennas,
Although the invention has been described by way of examples of preferred embodiments, it is to be understood that various other adaptations and modifications may be made within the spirit and scope of the invention. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the invention.
Number | Name | Date | Kind |
---|---|---|---|
7327983 | Mehta et al. | Feb 2008 | B2 |
7336727 | Mukkavilli et al. | Feb 2008 | B2 |
7366245 | Li et al. | Apr 2008 | B2 |
7369511 | Utsunomiya et al. | May 2008 | B2 |
20040131011 | Sandell et al. | Jul 2004 | A1 |
20050003863 | Gorokhov | Jan 2005 | A1 |
20050152387 | Utsunomiya et al. | Jul 2005 | A1 |
20060171482 | Trachewsky | Aug 2006 | A1 |
20060270343 | Cha et al. | Nov 2006 | A1 |
Number | Date | Country |
---|---|---|
1414177 | Apr 2004 | EP |
Number | Date | Country | |
---|---|---|---|
20060274847 A1 | Dec 2006 | US |