(12) United States Patent
`Gesbert et al.
`
`(10) Patent No.:
`(45) Date of Patent:
`
`US 6,760,882 B1
`*Jul. 6, 2004
`
`US006760882B1
`
`(54) MODE SELECTION FOR DATA
`TRANSMISSION IN WIRELESS
`COMMUNICATION CHANNELS BASED ON
`
`(75)
`
`Inventors: David J. Gesbert, Mountain View, CA
`(US); Severine E. Catreux, San Jose,
`CA (US); Robert W. Heath, Jr.,
`Mountain View, CA (US); Peroor K.
`Sebastian, Mountain View, CA (US);
`Amgyaswami J- Paulraj, Staflfmda CA
`(US)
`
`.
`(73) Assignee:
`
`Intel Corporation, Santa Clara, CA
`(US)
`
`6,064,662 A
`6,144,711 A
`6,167,031 A
`6,175,550 B1
`
`.............. .. 370/330
`5/2000 Gitlin et al.
`.. 375/347
`11/2000 Raleigh et al.
`
`.... .. 370/252
`12/2000 Olofsson et al.
`1/2001 Van N66 ................... .. 370/206
`
`EP
`W0
`
`FOREIGN PATENT DOCUMENTS
`0951091 A2
`10/1999
`.......... .. H01Q/3/26
`W0 98/09381
`3/1998
`
`OTHER PUBLICATIONS
`Paulraj, A., Taxonomy of space—time processing for wireless
`networks, IEE Proc—Radar Sonar Navig., vol. 145, No. 1,
`Feb. 1998.
`
`* cited by examiner
`
`( * ) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C.154b b 553d
`.
`0 y
`ays
`
`Primary Examiner—Emmanuel L. Moise
`(74) Attorney, Agent, or Firm—Michael A. Proksch
`(57)
`ABSTRACT
`
`This patent is subject to a terminal dis-
`claimer.
`
`A method and communication system for selecting a mode
`for encoding data for transmission in a Wireless communi-
`cation channel between a transmit unit and a receive unit.
`
`(22)
`(51)
`
`(21) App1. N0; 09/665,149
`.
`SeP- 19: 2000
`F1199:
`Int. Cl.7 ...................... .. H03M 13/00; G06F 11/00;
`H04Q 7/34
`..................... .. 714/774, 714/704, 370/252,
`(52) U.S. Cl.
`(58) Field of Search ................................. 714/774, 704,
`714/708’ 748’ 751; 370/252’ 345’ 206;
`375/261 259 267. 455/102 6711 6713
`’
`’
`’
`’
`’
`C't d
`R f
`1 e
`e erences
`U.S. PATENT DOCUMENTS
`1 er e a.
`............ ..
`,
`,
`714/704
`5559 810 A *
`9/1996 Glb t
`t
`1
`370/206
`5,815,488 A
`9/1998 Williams et al.
`
`370/330
`5,933,421 A
`8/1999 Alamouti et al.
`6,044,485 A *
`3/2000 Dent et al. ................ .. 714/774
`
`56
`
`)
`
`(
`
`The data is initially transmitted in an initial mode and the
`selection of the subse uent mode is based on a selection of
`q
`first-order and second-order statistical parameters of short-
`term. and 1°ng'term.quamy p.arameter.S' Suitable Sh°rt'te¥m
`quality parameters include signal-to-interference and noise
`‘a‘.it°b(1SH;’R)> fig“a1“°‘1‘¥fiSe ratio ESNR.) foyer level atnd
`S111 a
`C 0Ilg' CIIII qualy paraIIle CIS IIICU 6 error Ia CS
`Such as b“ em” rate (BER) and packet em” me (PER) The
`method of the invention can be employed in Multiple Input
`Multiple Output (MIMO), Multiple Input Single Output
`(MISO), Single Input Single Output (SISO) and Single Input
`Multiple Output (SIMO) communication systems to make
`.
`.
`subsequent mode selection faster and more efficient. Fur-
`thermore the method can be used in communication systems
`employing various transmission protocols including
`OFDMA’ FDMA’ CDMA’ TDMA'
`
`61 Claims, 4 Drawing Sheets
`
`(90
`
`4"‘
`J,’
`
`‘S-92
`
`k
`
`Signal
`
`100
`
`1
`Detection
`De°°dl“3
`
`...01101,..
`
`p exmg
`Demoni-
`
`
`° 1‘
`
`Receive
`Processing
`
`E
`
`
`
`‘I,
`K
`«fr }{
`91N —.N94
`108
`.
`Window
`Adjustment
`106A
`10
`
`
`
`
`I’
`l‘
`Qua ity arameler
`Computation
`S‘““5“°.5
`
` 11 0
`
`Selection
`Block
`
`Feedback
`
`“Z
`118
`
`
`SPRINT 1005
`
`
`
`116
`
`PER Statistics
`
`

`
`U.S. Patent
`
`Jul. 6, 2004
`
`Sheet 1 of4
`
`US 6,760,882 B1
`
`FIG. 2
`
`

`
`Mm
`
`0
`
`US 6,760,882 B1
`
`Hmm
`
`om
`
`Momnvmom
`
`Sfimbxm
`
`cm
`
`8
`
`UM:
`
`ommnfimfl
`
`U.S. Patent
`
`J
`
`Mfikflv
`
`"\
`
`mm»
`
`aE,I§El
`
`m2s_|.|lMaEs,
`
`Ecmcfie
`
`wfimmoooi
`
`5::S...
`
`FIG. 3
`
`cm
`
`mm
`
`
`

`
`U.S. Patent
`
`Jul. 6, 2004
`
`Sheet 3 of 4
`
`US 6,760,882 B1
`
`90
`
`/
`
`2
`
`RF&A/ D
`
`4*"
`‘S 9
`‘G’?
`,———|
`9“ j
`{,’.}-
`91B
`
`RF&A/D
`
`f ~
`
`'
`:
`
`RS
`1
`
`‘In’.
`
`ff
`
`lifilllil
`91N E
`
`94
`
`108
`
`Adjustment
`
`1 10
`
`106A
`104
`
`PER Statistics
`
`mo
`
`52
`
`."01101m
`
`Signal
`Detection
`
`_
`Decodmg
`Demu1ti-
`
`Receive
`Processing
`
`
`
`k
`
`102
`
`Computation
`
`I‘
`
`1063
`
`
`Selection
`
`Block
`
`1 14
`
`j
`
`112
`
`118
`
`116
`
`Feedback
`
`FIG. 4
`
`

`
`U.S. Patent
`
`Jul. 6, 2004
`
`Sheet 4 of 4
`
`US 6,760,882 B1
`
`FIG. 5
`
`

`
`US 6,760,882 B1
`
`1
`MODE SELECTION FOR DATA
`TRANSMISSION IN WIRELESS
`COMMUNICATION CHANNELS BASED ON
`STATISTICAL PARAMETERS
`
`FIELD OF THE INVENTION
`
`The present invention relates generally to wireless com-
`munication systems and methods, and more particularly to
`mode selection for encoding data for transmission in a
`wireless communication channel based on statistical param-
`eters.
`
`BACKGROUND OF THE INVENTION
`
`Wireless communication systems serving stationary and
`mobile wireless subscribers are rapidly gaining popularity.
`Numerous system layouts and communications protocols
`have been developed to provide coverage in such wireless
`communication systems.
`Wireless communications channels between transmit and
`
`receive devices are inherently variable and their quality
`fluctuates. Specifically, the quality parameters of such com-
`munications channels vary in time. Under good conditions
`wireless channels exhibit good communication parameters,
`e.g., large data capacity, high signal quality, high spectral
`efficiency and throughput. At
`these times significant
`amounts of data can be transmitted via the channel reliably.
`However, as the channel changes in time, the communica-
`tion parameters also change. Under altered conditions
`former data rates, coding techniques and data formats may
`no longer be feasible. For example, when channel perfor-
`mance is degraded the transmitted data may experience
`excessive corruption yielding unacceptable communication
`parameters. For instance, transmitted data can exhibit exces-
`sive bit-error rates or packet error rates. The degradation of
`the channel can be due to a multitude of factors such as
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`loss of
`general noise in the channel, multi-path fading,
`line-of-sight path, excessive Co-Channel Interference (CCI)
`and other factors.
`
`40
`
`In mobile systems, signal degradation and corruption is
`chiefly due to interference from other cellular users within or
`near a given cell and multipath fading, in which the received
`amplitude and phase of a signal varies over time. The fading
`rate can reach as much as 200 Hz for a mobile user traveling
`at 60 mph at PCS frequencies of about 1.9 GHZ. In such
`environments, the problem is to cleanly extract the signal of
`the user being tracked from the collection of received noise,
`CCI, and desired signal portions.
`In Fixed Wireless Access (FWA) systems, e.g., where the
`receiver remains stationary, signal fading rate is less than in
`mobile systems. In this case, the channel coherence time or
`the time during which the channel estimate remains stable is
`longer since the receiver does not move.
`Prior art wireless systems have employed adaptive modu-
`lation of the transmitted signals with the use of feedback
`from the receiver as well as adaptive coding and receiver
`feedback to adapt data transmission to changing channel
`conditions. Such adaptive modulation is applied to Single
`Input Single Output (SISO) and Multiple Input Multiple
`Output (MIMO) systems, e.g., systems with antenna arrays
`at the transmit and receive ends.
`
`In both SISO and MIMO systems, however, the funda-
`mental problem of efficient choice of the mode to be applied
`to the transmitted data remains. For general prior art on the
`subject the reader is referred to A. J. Goldsmith et al.,
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`“Variable-rate variable power MQAM for fading channels”,
`IEEE Transactions of Communications, Vol. 45, No. 10,
`October 1997, pp. 1218-1230; P. Schramm et al., “Radio
`Interface of EDGE, a proposal for enhanced data rates in
`existing digital cellular systems”, Proceedings IEEE 48th
`Vehicular Technology Conference (VTC’ 1998), pp.
`1064-1068; and Van Noblen et al., “An adaptive link
`protocol with enhanced data rates for GSM evolution”,
`IEEE Personal Communications, February 1999, pp. 54-63.
`U.S. Pat. No. 6,044,485 to Dent et al. teaches a transmis-
`sion method and system which adapts the coding of data
`based on channel characteristics. The channel characteristics
`are obtained either from a channel estimation circuit or from
`
`an error feedback signal. In particular, when the signal-to-
`noise (SNR) characteristic is chosen as an indicator of the
`state of the channel, then a different coding is applied to the
`data being transmitted for high and low SNR states of the
`channel. In addition, the encoding also employs information
`derived from the cyclic redundancy check (CRC).
`The method taught by Dent varies the coding rate only
`and not the modulation rate. This method does not permit
`one to select rapidly and efficiently from a large number of
`encoding modes to adapt to varying channel conditions.
`U.S. Pat. No. 5,559,810 to Gilbert et al. teaches a com-
`munication system using data reception history for selecting
`a modulation technique from among a plurality of modula-
`tion techniques to thus optimize the use of communication
`resources. At least one block of data is transmitted with a
`
`particular modulation technique and a data reception history
`is maintained to indicate transmission errors, e.g., by keep-
`ing a value of how many blocks had errors. The data
`reception history is updated and used to determine an
`estimate of transmission signal quality for each modulation
`technique. This value is then used in selecting the particular
`modulation technique.
`The system and method taught by Gilbert rely on tracking
`errors in the reception of entire blocks of data. In fact, signal
`quality statistics, e.g., signal-to-interference and noise ratio
`(SINR) are derived from the error numbers for entire blocks
`of data, which requires a significant amount of time. Thus,
`this method does not permit one to select rapidly and
`efficiently from a large number of encoding modes to adapt
`to varying channel conditions, especially in the event of
`rapid fades as encountered in mobile wireless systems.
`It would be an advance to provide a mode selection
`technique which allows the system to rapidly and efficiently
`select the appropriate mode for encoding data in a quickly
`changing channel. It is important that such technique be
`efficient in all wireless systems, including Multiple Input
`Multiple Output (MIMO), Multiple Input Single Output
`(MISO), Single Input Single Output (SISO) and Single Input
`Multiple Output (SIMO) systems as well as systems using
`multiple carrier frequencies, e.g., OFDM systems.
`SUMMARY
`
`The present invention provides a method for selecting a
`mode for encoding data for transmission in a wireless
`communication channel between a transmit unit and a
`receive unit. The data is first encoded in accordance with an
`initial mode and transmitted from the transmit unit to the
`
`receive unit. One or more quality parameters are sampled in
`the data received by the receive unit. Then, a first-order
`statistical parameter and a second-order statistical parameter
`of the quality parameter are computed and used for selecting
`a subsequent mode for encoding the data.
`The one or more quality parameters can include a short-
`term quality parameter or several short-term quality param-
`
`

`
`US 6,760,882 B1
`
`3
`eters and be selected among parameters such as signal-to-
`interference and noise ratio (SINR), signal-to-noise ratio
`(SNR) and power level. Conveniently, a first sampling time
`or window is set during which the short-term quality param-
`eter is sampled. In one embodiment, the length of the first
`sampling window is based on a coherence time of the
`wireless communication channel. In another embodiment,
`the duration of the first sampling window is based on a delay
`time required to apply the subsequent mode to encode the
`data at the transmit unit. In yet another embodiment, the
`second-order statistical parameter is a variance of the short-
`term quality parameter and the length of the first sampling
`window is selected on the order of the variance computation
`time.
`
`The one or more quality parameters can also include a
`long-term quality parameter or several long-term quality
`parameters. The long-term quality parameter can be an error
`rate of the data, such as a bit error rate (BER) or a packet
`error rate (PER) at the receive unit. Again, it is convenient
`to set a second sampling time or window during which the
`long-term quality parameter is sampled. In one embodiment,
`the first-order statistical parameter is a mean of the long-
`term quality parameter and the length of the second sam-
`pling window is set on the order of the-mean computation
`time. In another embodiment,
`the length of the second
`sampling window is set on the order of an error rate
`computation time.
`In many instances, it is convenient when the first-order
`statistical parameter is a mean of the quality parameter and
`the second-order statistical parameter is a variance of the
`quality parameter. The variance can include two variance
`types: a temporal variance and a frequency variance. The
`latter is useful in systems employing multiple frequencies
`for
`transmitting the data. Specifically,
`it
`is particularly
`convenient to compute both temporal and frequency vari-
`ances when the data is transmitted in accordance with a
`multi-carrier scheme.
`
`The initial mode for encoding the data can be selected
`from a set of modes. The set of modes can be made up of a
`number of modes which are likely to work based on a
`preliminary analysis of the channel. The set of modes can be
`organized in accordance with the at least one quality param-
`eter whose first-order and second-order statistics are used in
`
`subsequent mode selection.
`Conveniently, the subsequent mode is communicated to
`the transmit unit and applied to the data to maximize a
`communication parameter in the channel. For example, the
`subsequent mode can maximize data capacity, signal quality,
`spectral efficiency or throughput of the channel or any other
`communication parameter or parameters as desired.
`The method of the invention can be used in Multiple Input
`Multiple Output (MIMO), Multiple Input Single Output
`(MISO), Single Input Single Output (SISO) and Single Input
`Multiple Output (SIMO) communication systems, e.g.,
`receive and transmit units equipped with multiple antennas.
`Furthermore the method can be used:
`in communication
`
`systems employing various transmission protocols including
`OFDMA, FDMA, CDMA, TDMA.
`The method of invention can also be used for selecting the
`mode from a set of modes and adjusting the selection. For
`this purpose data encoded in an initial mode selected from
`the set of modes is received by the receive unit. The
`short-term quality parameter is then sampled to determine a
`statistical parameter of the short-term quality parameter. Of
`course, the statistical parameter can include any combina-
`tion of first-order and second-order statistical parameters.
`
`4
`The subsequent mode is selected based on the short-term
`statistical parameter.
`In addition,
`the long-term quality
`parameter of the data received by the receive unit is also
`sampled. The subsequent mode selected based on the short-
`term statistical parameter is then adjusted based on the
`long-term quality parameter.
`The set of modes can be arranged in any suitable manner,
`e.g., it can be arranged in a lookup table and ordered by the
`short-term quality parameter and specifically the first-order
`and/or second-order statistics of the short-term quality
`parameter for easy selection. In fact, the lookup table can be
`modified based on the short-term quality parameter.
`The invention also encompasses a system for assigning a
`subsequent mode for encoding data. The system has a
`transmit unit equipped with a transmit processing block for
`encoding the data in a mode. A receive unit is provided for
`receiving the data transmitted from the transmit unit. The
`receive unit has a statistics computation block for sampling
`at least one quality parameter of the received data and
`computing the first-order and second-order statistical param-
`eters of the data. The receive unit also has a mode selection
`
`block for assigning the subsequent mode based on the
`first-order and second-order statistical parameters.
`Conveniently, the system has at least one database con-
`taining the set of modes from which the mode, e.g., the
`initial mode, and the subsequent mode are selected. In one
`case, the receive has a first database containing the modes
`and the transmit unit has a second such database.
`
`The system also has a feedback mechanism for commu-
`nicating the subsequent mode from the receive unit to the
`transmit unit. This feedback mechanism can be a separate
`mechanism or comprise the time-division duplexing (TDD)
`mechanism.
`
`A detailed description of the invention and the preferred
`and alternative embodiments is presented below in reference
`to the attached drawing figures.
`
`BRIEF DESCRIPTION OF THE FIGURES
`
`FIG. 1 is a simplified diagram illustrating a communica-
`tion system in which the method of the invention is applied.
`FIG. 2 is a graph illustrating the effects of channel
`variation in time and frequency.
`FIG. 3 is a block diagram of an exemplary transmit unit
`in accordance with the invention.
`
`FIG. 4 is a block diagram of an exemplary receive unit in
`accordance with the invention.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`FIG. 5 is a schematic diagram illustrating data transmitted
`in a wireless communication channel.
`
`DETAILED DESCRIPTION
`
`55
`
`60
`
`65
`
`The method and systems of the invention will be best
`understood after first considering the simplified diagram of
`FIG. 1 illustrating a portion of a wireless communication
`system 10, e.g., a cellular wireless system in which the
`method of invention can be employed. For explanation
`purposes, downlink communication will be considered
`where a transmit unit 12 is a Base Transceiver Station (BTS)
`and a receive unit 14 is a mobile or stationary wireless user
`device. Of course, the method can be used in uplink com-
`munication from receive unit 14 to BTS 12.
`
`Exemplary user devices 14 include mobile receive units
`such as a portable telephone 14A, a car phone 14B and a
`stationary receive unit 14C. Receive unit 14C can be a
`wireless modem used at a residence or any other fixed
`
`

`
`US 6,760,882 B1
`
`5
`wireless unit. Receive units 14A and 14C are equipped with
`multiple antennas or antenna arrays 20. These receive units
`can be used in Multiple
`Input Multiple Output (MIMO) communications taking
`advantage of techniques such as spatial multiplexing or
`antenna diversity. Receive unit 14B has a single antenna 11
`and can be used in Single Input Single Output (SISO)
`communications. It will be understood by those skilled in the
`art that receive units 14A, 14B, 14C, could be equipped in
`SISO, MISO (Multiple Input Single Output), SIMO (Single
`Input Multiple Output), or MIMO configurations. For
`example, in FIG. 1 receive unit 14B is shown having a single
`antenna therefore it can be employed in SISO or MISO
`configurations. MISO configuration can be realized in the
`case of 14B for example by receiving signals from the
`antenna array at BTS 12A or from distinct BTS such as 12B,
`or any combination thereof. With the addition of multiple
`receive antennas 14B, as 14A and 14C, could also be used
`in SIMO or MIMO configurations. In any of the configu-
`rations discussed above, the communications techniques can
`employ single-carrier or multi-carrier communications tech-
`niques.
`Afirst exemplary transmit unit 12 is a BTS 12A equipped
`with an antenna array 16 consisting of a number of transmit
`antennas 18A, 18B, .
`.
`.
`, 8M for MIMO communications.
`Another exemplary transmit unit 12 is a BTS 12B equipped
`with a single omnidirectional antenna 13. BTSs 12A, 12B
`send data in the form of transmit signals TS to receive units
`14A, 14B, 14C via wireless communications channels 22.
`For simplicity, only channel 22A between BTS 12A and
`receive unit 14A and channel 22B between BTS 12B and
`receive unit 14C are indicated.
`
`The time variation of channels 22A, 22B causes trans-
`mitted signal TS to experience fluctuating levels of
`attenuation, interference, multi-path fading and other del-
`eterious effects.
`
`Therefore, communication parameters of channel 22A
`such as data capacity, signal quality, spectral efficiency and
`throughput undergo temporal changes. The cumulative
`effects of these variations of channel 22A between BTS 12A
`
`and receive unit 14A are shown for illustrative purposes in
`FIG. 2. In particular, this graph shows the variation of a
`particular quality parameter, in this case signal strength of
`receive signal RS at receive unit 14A in dB as a function of
`transmission time t and frequency f of transmit signal TS
`sent from transmit unit 12 A. Similar graphs can be obtained
`for other quality parameters, such as signal-to-interference
`and noise ratio (SINR), signal-to-noise ratio (SNR) as well
`as any other quality parameters known in the art. Of the
`various quality parameters signal strength (power level),
`SINR and SNR are generally convenient to use because they
`can be easily and rapidly derived from receive signals RS as
`is known in the art.
`
`In accordance with the invention, a mode for encoding
`data at transmit units 14 is selected based on a first order
`
`statistical parameter and a second order statistical parameter
`of the quality parameter. The details of the method will now
`be explained by referring to the operation of a transmit unit
`50, as illustrated in FIG. 3 and a receive unit 90 as illustrated
`in FIG. 4.
`
`Transmit unit 50 receives data 52 to be transmitted; in this
`case a stream of binary data. Data 52 is delivered to a
`transmit processing block 56. Transmit processing block 56
`subdivides data 52 into a number k of parallel streams. Then,
`processing block 56 applies an encoding mode to each of the
`k streams to thus encode data 52. It should be noted, that
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`6
`before transmission data 52 may be interleaved and pre-
`coded by an interleaver and a pre-coder (not shown). The
`purpose of interleaving and pre-coding is to render the data
`more robust against errors. Both of these techniques are
`well-known in the art
`
`The mode is determined by a modulation during which
`data 52 is mapped into a constellation at a given modulation
`rate, and a coding rate at which this translation is performed.
`For example, data 52 can be converted into symbols through
`modulation in a constellation selected from among PSK,
`QAM, GMSK, FSK, PAM, PPM, CAP, CPM or other
`suitable constellations. The transmission rate or throughput
`of data 52 will vary depending on the modulation and coding
`rates used in each of the k data streams.
`
`TABLE 1
`
`Mode
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`
`Modulation Rate
`(bits/symbol)
`2
`2
`2
`2
`4
`4
`4
`4
`5
`5
`5
`5
`6
`6
`6
`6
`
`Coding Rate
`3/4
`2/3
`1/2
`1/3
`3/4
`2/3
`1/2
`1/3
`3/4
`2/3
`1/2
`1/3
`3/4
`2/3
`1/2
`1/3
`
`Throughput
`(bits/s/Hz)
`3/2
`4/3
`1
`2/3
`3
`8/3
`2
`4/3
`15/4
`10/3
`5/2
`5/3
`9/2
`4
`3
`2
`
`Table 1 illustrates some typical modes with their modu-
`lation rates and coding rates and the corresponding through-
`puts for data 52. The modes are indexed by a mode number
`so as to conveniently identify the modulation and coding
`rates which are to be applied to data 52 in each mode.
`Lookup tables analogous to Table 1 for other coding and
`modulation rates can be easily derived as these techniques
`are well-known in the art.
`
`Referring back to FIG. 3, a set of modes, arranged
`conveniently in the form of lookup table indexed as
`described above, is stored in a database 78 of transmit unit
`50. Database 78 is connected to a controller 66, which is also
`connected to transmit processing block 56 and spatial map-
`ping unit 58. Controller 66 controls which mode from
`database 78 is applied to each of the k streams and spatial
`mapping to be performed by spatial mapping unit 58.
`In addition to encoding the k streams, transmit processing
`block 56 adds training information into training tones T (see
`FIG. 5) and any other control information, as is known in the
`art. Thus processed,
`the k streams are sent
`to an
`up-conversion and RF amplification stage 70 having indi-
`vidual digital-to-analog converters and up-conversion/RF
`amplification blocks 74 through the spatial mapping unit 58.
`The spatial mapping unit 58 maps the k streams to M inputs
`of the up-conversion and RF amplification stage 70. The M
`outputs of amplification stage 70 lead to corresponding M
`transmit antennas 72 of an antenna array 76.
`A person skilled in the art will recognize that the number
`M of transmit antennas 72 does not have to be equal to the
`number of streams k. That is because various spatial map-
`pings can be employed in assigning streams k to transmit
`antennas 72. In one mapping, a certain transmit antenna 72B
`transmits one of the k streams.
`In another mapping, a
`
`

`
`US 6,760,882 B1
`
`7
`number of transmit antennas 72 transmit the same stream k.
`
`In yet another embodiment, the k streams are assigned to M
`antennas 72 or a subset thereof via the spatial mapping unit
`58 and the unit 70. In fact, any kind of mapping involving
`the use of spatial multiplexing (SM) and antenna diversity
`can be used in the method and system of the invention.
`Transmit antennas 72 transmit data 52 in the form of
`
`transmit signals TS. FIG. 5 illustrates, as will be recognized
`by those skilled in the art, a multicarrier transmission
`scheme with n frequency carriers (tones). The vertical axis
`illustrates frequency carriers while the horizontal axis illus-
`trates OFDM symbol periods. Each block corresponds to
`one of n frequency carriers, during an OFDM symbol. The
`blocks marked with D correspond to data and the blocks
`marked with T correspond to training.
`FIG. 5 indicates that training is performed on all tones
`during an OFDM training symbol, it will be clear to a person
`skilled in the art that a subset of these tones could be used
`
`for training and the corresponding frequency response could
`be computed at the receiver by interpolating.
`Transmit signals TS propagate through channel 22 and
`there experience the effects of changing conditions of chan-
`nel 22, as described above. Transmit signals TS are received
`in the form of receive signals RS by a receive antenna 91A
`belonging to an antenna array 92 of a receive unit 90, shown
`in FIG. 4.
`
`Again referring to FIG. 4, receive unit 90 has N receive
`antennas 91A, 91B, .
`.
`.
`, 91N for receiving receive signals
`RS from transmit unit 50. Receive unit 90 can be any
`suitable receiver capable of receiving receive signals RS via
`the N receive antennas 92. Exemplary receivers include
`linear equalizer receivers, decision feedback equalizer
`receivers, successive cancellation receivers, and maximum
`likelihood receivers.
`
`Receive unit 90 has an RF amplification and down-
`conversion stage 94 having individual RF amplification/
`down-conversion/and analog-to-digital converter blocks 96
`associated with each of the N receive antennas 91A,
`91B, .
`.
`.
`, 91N. The N outputs of stage 94 are connected to
`a receive processing block 98 which performs receive pro-
`cessing to recover the k streams encoded by transmit pro-
`cessing block 56 of transmit unit 50. The recovered k
`streams are passed on to a signal detection, decoding and
`demultiplexing block 100 for recovering data 52. In the case
`of antenna diversity processing it should be understood that
`k is equal to one thus there is only a single stream recovered.
`The receive processing block 98 computes the quality
`parameters for each of k streams and sends this information
`to a statistics computation block for computing statistical
`parameters of the one or more quality parameters. The
`method of the invention can recognize slow and rapid
`channel variations and allows for efficient mode selection by
`taking both types of variations into account. This is accom-
`plished by taking into account at least two statistics of one
`or more quality parameters. This may include either or both
`short-term and long-term quality parameters. Suitable short-
`term quality parameters include signal-to-interference and
`noise ratio (SINR), signal-to-noise ratio (SNR) and power
`level. Suitable long-term quality parameters include error
`rates such as bit error rate (BER) and packet error rate
`(PER).
`the first-order and
`in one embodiment,
`For example,
`second-order statistics are derived from a short-term quality
`parameter such as the SINR. In another embodiment statis-
`tics of both a short-term and a long-term quality parameter
`are used.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`8
`In the present embodiment the short-term quality param-
`eter used is SINR. Statistics computation block 102 com-
`putes a first-order statistical parameter 104 and a second-
`order statistical parameter 106 of SINR. Conveniently, first-
`order statistical parameter 104 is mean SINR and second-
`order statistical parameter is a variance SINR. Variance 106
`of SINR actually consists of two values, SINR temporal
`variance 106A and SINR frequency variance 106B.
`In
`systems which do not employ multi-carrier transmission
`schemes frequency variance 106B does not have to be
`computed. It should be noted that each data stream of k
`streams will have an associated statistical parameter 104,
`106A, 106B.
`A window adjustment 108 such as a timing circuit is
`connected to statistics computation block 102. Window
`adjustment 108 sets a first time interval or first sampling
`window '51 (see FIG. 5) during which the SINR is sampled.
`Conveniently, SINR is sampled during training tones T
`occurring during sampling window '51. The present embodi-
`ment uses multiple carrier frequencies fc and thus the SINR
`is sampled and computed by block 102 for data 52 trans-
`mitted at each of the n carrier frequencies fc. By buffering
`the SINR values for all the training tones T during time
`window '51 statistics computation block 102 constructs the
`following matrix:
`
`SINRL1 SINRL2
`SINR2,1
`
`SINRLW
`
`SINRM
`
`SINR,W
`
`where SINR”. is the SINR at the i-th carrier frequency fa.
`during training phase j. There are thus 1 to n carrier
`frequencies fc and 1 to w training phases.
`First-order statistical parameter 104 of short-term quality
`parameter, in this case mean SINR, can be expressed as:
`
`SINRmean =
`
`W
`n
`1
`EZZSINR,-J.
`z:1 j:l
`
`Second-order statistical parameters 106A, 106B of short-
`term quality parameter, in this case SINR frequency vari-
`ance and SINR time variance can be expressed as:
`
`n
`1
`SINR]-J — 2;‘ SINR,”-
`
`
`
`
`
`,and
`
`2
`
`2
`
`1
`n
`W
`S”VRvar(freq>=E% %
`z:1
`j:l
`w
`1
`SINRvar(time) = 3 E
`k:l
`
`1
`"
`[22 SINR.-,7 — (SIA/Rmean)
`[:1
`
`
`
`In general, the duration of first sampling window '51 takes
`into account general parameters of the communication sys-
`tem and/or channel 22. For example, channel 22 has a
`coherence time during which the condition of channel 22 is
`stable. Of course, s the coherence time will vary depending
`on the motion of receive unit 90, as is known in the art. In
`one embodiment, window adjustment 108 sets first sampling
`window '51 based on the coherence time. Specifically, first
`sampling window '51 can be set on the order of or shorter
`than the coherence time. Thus, the first- and second-order
`statistical parameters 104, 106A, 106B computed during
`time window '51 are minimally affected by loss of coherence.
`In another embodiment window adjustment 108 sets first
`sampling window '51 to be much larger than the coherence
`time.
`
`

`
`US 6,760,882 B1
`
`10
`and a corresponding transmitter 118 for transmission of the
`feedback to transmit unit 50. Here the convenience of
`indexing modes becomes clear, since feedback of an index
`number to transmit unit 50 does not require much band-
`width. It should be noted, that in the present embodiment a
`mode selection is made for each of the k streams. In other
`words, a mode index indicating the mode to be used for each
`of the k streams is fed back to transmit unit 50. In another
`embodiment it may be appropriate to send a mode difference
`indicating how to modify the current mode for subsequent
`transmission. For example if the current transmission is
`mode 1, the mode index of the subsequent mode is 3, the
`mode difference would be 2. In yet another embodiment, it
`may be suitable to send the channel characteristics back to
`the transmitter. In this case the computation of statistics of
`the quality parameter, the mode selection are performed at
`the transmitter.
`
`transmit unit 50 receives
`Referring back to FIG. 3,
`feedback from receive unit 90 via a feedback extractor 80.
`
`10
`
`15
`
`20
`
`Feedback extractor 80 detects the mode index or any other
`designation of the selected modes for each of the k streams
`and forwards this information to controller 66. Controller 66
`
`looks up the mode by mode index in database 78 and thus
`determines the modulation, coding rate and any other
`parameters to be used for each of the k streams. In the event
`of using time-division duplexing (TDD) which is a tech-
`nique known in the art,
`the quality parameters can be
`extracted during the reverse transmission from receive unit
`90 or rem