US007701919B2
`
`(12) Ulllted States Patent
`Ah Lee
`
`(10) Patent No.:
`(45) Date of Patent:
`
`US 7,701,919 B2
`Apr. 20, 2010
`
`(54) METHOD OF ASSIGNING UPLINK
`REFERENCE SIGNALS, AND TRANSMITTER
`AND RECEIVER THEREOF
`
`.................. .. 370/527
`2004/0246998 A1 * 12/2004 Ma et :11.
`
`.... .. 370/343
`2005/0237989 A1* 10/2005 Ahn et a1.
`.............. .. 455/450
`2006/0009227 A1*
`1/2006 Cudak etal.
`2006/0039296 A1*
`2/2006 Nakamata et al.
`......... .. 370/252
`
`(75)
`
`Inventor:
`
`Jung Ah Lee, Pittstown, NJ (US)
`
`(Continued)
`
`(73) Assignee: Alcatel-Lucent USA Inc., Murray Hill,
`NJ (US)
`
`EP
`
`FOREIGN PATENT DOCUMENTS
`0607755 A1 *
`7/1994
`
`(COI1tiI111ed)
`
`OTHER PUBLICATIONS
`3GPP TR 25.814 V1 .2.2 (Mar. 2006).
`(Continued)
`
`Primary Examiner—Rafae1 Perez-Gutierrez
`Assistant Exami/1er—Alla1iyar Kasraian
`(74) Attorney, Agent, or Firm—Hamess, Dickey & Pierce
`
`(57)
`
`ABSTRACT
`
`In an embodiment ofthe method, uplink reference signals are
`~
`~
`assigned to users in a group of cells. For example, a first
`constant amplitude sequence having low cyclic cross corre-
`lation is assigned to each user in a first cell of the group of
`Cells. Here, each user is assigned the first Sequence. A150,
`simultaneously transmitting users in the first cell are assigned
`to different Sub_earriers. A Seeend eenstam amplitude
`sequence having low cyclic cross correlation is assigned to
`eachuser in a second cell ofthe group ofcells. Here, each user
`in the second cell is assigned the second sequence. The first
`sequence and the second sequence are different sequences,
`and the first cell and the second cell are adjacent. Also, si1iiul-
`Ouslty tr1:‘nSm1.mng}11S1erS 13 the S?°°‘;d°e1111aff;SS1.g“e§‘t‘°
`1 ere“ 5“ '°amerS'
`3' 5“ ‘Camels ° W 1°
`3 51m“ a‘
`neously transmitting users of the second cell are assigned
`overlap in frequency with the sub-carriers to which the simul-
`taI1e011S1YtranSm1tt1I1g users Ofthe first 0611 are aSS1gI1ed~
`
`15 Claims, 8 Drawing Sheets
`
`( * ) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 1024 days.
`
`(21) Appl. No.: 11/414,402
`
`(22)
`
`Filed3
`
`May 1: 2006
`
`(65)
`
`Prlor Pubhcatlon Data
`US 2008/0123616 A1
`May 29, 2008
`
`(2006.01)
`
`(51)
`
`(56)
`
`Int‘ Cl‘
`H04B 7/208
`17//02016
`)
`’
`(
`(52) U.S. Cl.
`..................... .. 370/344; 370/485; 370/342;
`_
`_
`_
`370/335
`(58) Field of Classification Search ............... .. 370/344,
`370/342, 485: 335
`See appliealien file for eemplete Seareh hi5l01'Y~
`Referenees Cited
`
`U.S. PATENT DOCUMENTS
`6,536,024 B1 *
`3/2003 Hathaway ...................... 716/6
`7,379,741 B2 *
`5/2008 Ahn et a1.
`455/450
`2001/0040882 A1* 11/2001 Ichiyoshi
`370/342
`5/2002 Eiteletal.
`2002/0051436 A1*
`370/335
`370/349
`9/2002 Ha11 et a1.
`2002/0126650 A1*
`370/342
`2002/0172180 A1* 11/2002 Hall et al.
`
`.
`
`
`
`2004/0240400 A1* 12/2004 Khan ....................... .. 370/280
`
`Spread pilot (S=2)
`
`UE2
`
`UE3
`
`UE4
`
`1
`
`APPLE 1029
`
`APPLE 1029
`
`1
`
`

`
`US 7,701,919 B2
`Page 2
`
`U.S. PATENT DOCUMENTS
`
`3GPP TSG-RAN WG1 Meeting #44bis R1 -061094 Athens, Greece,
`Mar. 27-31, 2006.
`3GPP TSG RAN WG1 #44 R1-060373 Denver, Colorado, Feb.
`13-17, 2006.
`3GPP TSG RAN WG1 Ad Hoc on LTE R1-050822 London, United
`Kingdom, Aug. 29-Sep. 2, 2005.
`3GPP TSG RAN WG1 #44 R1-060388 Denver, Colorado, Feb.
`13-17, 2006.
`3GPP TSG RAN WG1 Ad Hoc on LTE R1-051062 San Diego,
`California, Oct. 10-14, 2005.
`3GPP TSG RAN WG1 #42 on LTE R1-050851 London, United
`Kingdom. Aug. 29-Sep. 2. 2005.
`3GPP TSG RAN1 #44 R1-060390 Denver, Colorado, Feb. 13-17,
`2006.
`John C. Ng et al., “Multi—phase Optimal Sequences for Fast Initial-
`ization ofChannel Estimation and Equalization”, IEEE International
`Conference on Communications, vol. 3, Jun. 8, 1997, pp. 1484-1487.
`John C. Ng et al., “Complex Optimal Sequences with ConstantMag—
`nitude for Fast Channel Estimation Initialization”, IEEE Transac-
`tions on Communications, vol. 46, No. 3, Mar. 1998.
`B.M. Popovic, “GCL Pob/phase Sequences with Minimum Alp/za-
`bets”, vol. 30, No. 2, Jan. 20, 1994.
`
`* cited by examiner
`
`3/2007 Anderson ................. .. 370/277
`2007/0047474 A1*
`5/2007 Tirkkonen et al.
`. 370/318
`2007/0097901 A1*
`
`7/2007 McCoy ..................... .. 370/344
`2007/0165588 A1*
`8/2007 Muharemovic et al.
`2007/0183386 A1
`2007/0258404 A1* 11/2007 Tirkkonen et al.
`2008/0095254 A1*
`4/2008 Muharemovic et al.
`2008/0298438 A1* 12/2008
`2009/0052470 A1*
`2/2009
`2009/0227261 A1*
`9/2009
`2009/0268695 A1* 10/2009 Zhao et al.
`
`. 370/329
`375/260
`375/145
`. 370/491
`. 455/450
`................ .. 370/336
`
`
`
`.
`
`FOREIGN PATENT DOCUMENTS
`
`W0
`W0
`
`WO 98/59450
`WO 2007/084988
`
`12/1998
`7/2007
`
`OTHER PUBLICATIONS
`
`3GPP TR 25.813 V0.8.3 (Apr. 2006).
`3GPP TR 25.912 V0.1.2 (Mar. 2006).
`3GPP TR 25.913 V7.3.0 (Mar. 2006).
`3GPP TSG RAN WG1 Meeting #44bis R1 -061066 Athens, Greece,
`Mar. 27-31, 2006.
`
`2
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 1 of8
`
`US 7,701,919 B2
`
`FIG. 1
`
`PRIOR ART
`
`
`
`EEIEEI
`
`FIG. 2
`
`PRIOR ART
`
`
`
`Co-channelInterferencePower[dB]
`
`
`
`
`
`3
`
`I
`
`(Tl
`
`L. C:
`
`L. 01
`
`2'}:8
`
`20
`
`40
`
`60
`
`80
`
`100
`
`120
`
`140
`
`Sequence Length [samples]
`
`3
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 2 of8
`
`US 7,701,919 B2
`
`
`
`§._$:_w\n_
`
`4
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 3 of8
`
`US 7,701,919 B2
`
`toIFFT
`UHHEJFLLH
`
`U)
`._
`C
`8
`"
`
`‘
`
`m
`L0
`.
`
`0
`
`LI.
`
`FIG.5A
`
`2
`S
`‘T
`
`‘
`
`
`
`EI-
`.25
`
`I-
`LI.
`'-_L
`
`.9
`
`from DFT
`
`5
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 4 of8
`
`US 7,701,919 B2
`
`FIG. 6A
`
`Spread pilot (S=2)
`
`“E4
`
`Frequency
`
`
`
`Frequency
`
`6
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 5 of8
`
`US 7,701,919 B2
`
`.o&ms_
`
`.m_._._8-n=w
`
`7
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 6 of8
`
`US 7,701,919 B2
`
`FIG. 8
`
`1, fort=0,1,...,P-1
`jg‘-s
`_
`e 3
`,fort— P,P+1,...,2P-1
`
`.2
`
`1?nS(s'1),fort=(S-1)P,(S-1)P+1,...,SP-1
`
`.2
`
`eI?"(3'1), fort: P,P+1,...,2P-1
`
`.2/'fg£(S'1)(S'1),fort=(S-1)P,(S-1)P+ 1,...,SP-1
`
`8
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 7 of8
`
`US 7,701,919 B2
`
`§m_e.oo
`
`._mE8-n=m
`
`m.0_n_
`
`x83roamuo>_8om
`
`9
`
`
`

`
`U.S. Patent
`
`Apr. 20, 2010
`
`Sheet 8 of8
`
`US 7,701,919 B2
`
`FIG. 10
`
`nI:5...»«>
`IIIA
`
`10
`
`10
`
`

`
`US 7,701,919 B2
`
`1
`METHOD OF ASSIGNING UPLINK
`REFERENCE SIGNALS, AND TRANSMITTER
`AND RECEIVER THEREOF
`
`BACKGROUND OF THE INVENTION
`
`Reference signal design is an important issue to fully
`exploit the potential gain of the single carrier frequency divi-
`sion multiple access (SC-FDMA) system considered in the
`uplink of the evolved-UTRA (E-UTRA). The uplink is from
`user equipment (UE) to a node B. The UE may also be
`referred to as a mobile unit, mobile station, etc. The UE may
`be a wireless phone, wireless equipped PDA, a wireless
`equipped computer, etc. The node B may also be referred to as
`a base station, base station controller, base station router, etc.
`The reference signal sent by a UE to the node B is needed
`for uplink CQI estimation as well as detection and coherent
`data demodulation. There has been on-going discussion on
`the uplink reference or pilot signal structure, and in particular,
`the pros and cons of code division multiplexing (CDM) and
`frequency division multiplexing (FDM) pilot structures.
`FDM involves dividing the bandwidth for sending reference
`signals into tones or sub-carriers and assigning, for example,
`different sub -carriers to different UEs. Each UE uses the same
`
`pilot sequence, but transmits that pilot sequence over differ-
`ent sub -carriers (i.e., different frequency or bandwidth). Code
`division multiplexing involves allowing each UE to use the
`entire bandwidth, but having each UE transmit using a differ-
`entiating pilot sequence. Well-known constant amplitude
`zero autocorrelation sequences (CAZAC) have been pro-
`posed as the codes. More specifically, the use of a generalized
`chirp like (GCL) sequence like a Zadoff-Chu sequence has
`been proposed. In these proposals, a GCL sequence spanning
`the bandwidth is chosen. A shifted version of the same GCL
`
`sequence is then assigned to each UE to differentiate between
`UE transmissions.
`
`Desired elements of the SC-FDMA pilot signal design
`include:
`
`1. Equal channel sounding in the frequency domain
`2. Immunity to co-channel interference
`3. Support of multiple user resource block sizes
`4. Support of both localized and distributed sub-carrier
`mapping with reliable channel estimation performance
`5. Efficient transmitter and receiver structures
`
`6. Large number of sequences with the desired character-
`istic to support n1ulti-cell deployment
`The FDM pilot has been advocated because it offers in-cell
`user orthogonality in the presence of fading. So far, comb-
`shaped pilots and staggered pilots are the proposed reference
`patterns for the FDM pilot. One of the major drawbacks of
`FDM pilots is the impact of dominant co-channel interfer-
`ence. When two users at the cell edge use the same pilot
`sub-carriers, the charmel carmot be estimated reliably due to
`collision.
`
`SUMMARY OF THE INVENTION
`
`In an embodiment of the method, uplink reference signals
`are assigned to users in a group of cells. For example, a first
`constant amplitude sequence having low cyclic cross corre-
`lation is assigned to each user in a first cell of the group of
`cells. Here, each user is assigned the first sequence. Also,
`simultaneously transmitting users in the first cell are assigned
`to different sub-caniers. A second constant amplitude
`sequence having low cyclic cross correlation is assigned to
`each user in a second cell ofthe group ofcells. Here, each user
`in the second cell is assigned the second sequence. The first
`
`2
`
`sequence and the second sequence are different sequences,
`and the first cell and the second cell are adjacent. Also, simul-
`taneously transmitting users in the second cell are assigned to
`different sub-carriers. The sub-carriers to which the simulta-
`
`neously transmitting users of the second cell are assigned
`may overlap in frequency with the sub-carriers to which the
`simultaneously transmitting users of the first cell are
`assigned.
`In one embodiment, the first and second sequences have a
`same length. For example, the length may be an odd number
`13.
`
`In one embodiment, the first and second sequences have a
`same length. For example, the length may be number 12. This
`may be generated by truncating a longer-length sequence
`such as length 13 sequence.
`In another embodiment, the first and second sequences
`have different lengths. For example, the first and second
`sequences may have lengths differing by one such as a length
`of 12 and a length of 13, respectively.
`In an embodiment, the first and second sequences are gen-
`eralized chirp like (GCL) sequences. For example, the first
`and second sequences may be Zadoff-Chu GCL sequences.
`In one embodiment, the assigning sub-carriers to simulta-
`neously transmitting users in the first cell step assigns at least
`a minimum number of sub-carriers to each simultaneously
`transmitting user; and the assigning sub-carriers to simulta-
`neously transmitting users in the second cell step assigns at
`least the minimum number of sub-carriers to each simulta-
`
`neously transmitting user. For example, the minimum num-
`ber may be 12.5. In one embodiment, the first and second
`sequences have lengths dependent on the minimum number.
`According to an aspect of the present invention, a trans-
`mitter is provided. In one embodiment, a transmitter at a
`device includes a spreader spreading, in the time domain, a
`constant amplitude sequence having low cyclic cross corre-
`lation by a number of resource blocks assigned to the device.
`For example, each resource block equals a set number of
`sub-carriers, and the spreader bit-wise spreads the sequence.
`A transformer transforms the spread sequence from the time
`domain to the frequency domain, and a mapper maps fre-
`quency domain samples of the transformed spread sequence
`to sub-carriers. An inverse transformer transforms the sub-
`carriers to the time domain for transmission.
`
`In one embodiment, the sequence has a length correspond-
`ing to the set number of sub-carriers.
`In one embodiment, the mapper performs localized map-
`ping, and in another embodiment, distributed mapping is
`performed.
`Another embodiment of the transmitter performs the
`spreading operation in the frequency domain.
`The present
`invention also provides a receiver. For
`example, one embodiment of a receiver includes a trans-
`former transforming a time domain signal received from a
`device to the frequency domain, a de-mapper mapping sub-
`carriers of the transformed signal to frequency samples, and
`an inverse transformer transforming the frequency samples to
`the time domain. A despreader despreads the time domain
`output of the inverse transformer by a number of resource
`blocks assigned to the device to obtain a sequence. Here, each
`resource block equals a set number of sub-carriers. A corr-
`elator correlates the obtained sequence with a reference
`sequence.
`
`In another embodiment, the dispreading operation is per-
`formed in the frequency domain.
`
`10
`
`15
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`11
`
`11
`
`

`
`3
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`US 7,701,919 B2
`
`The present invention will become more fully understood
`from the detailed description given herein below and the
`accompanying drawings which are given by way of illustra-
`tion only, wherein like reference numerals designate corre-
`sponding parts in the various drawings, and wherein:
`FIG. 1 illustrates an example of the proposed uplink trans-
`mission sub-frame structure;
`interference suppression
`FIG. 2 shows the co-charmel
`characteristic of a GCL sequence for sequence lengths rang-
`ing from 3 to 150;
`FIG. 3 illustrates a transmitter structure according to an
`embodiment of the present invention for transmitting a refer-
`ence signal according to the embodiments of the present
`invention;
`FIG. 4 illustrates the example of spreading a sequence of
`length P over two resource blocks;
`FIGS. 5A and 5B illustrate localized and distributed sub-
`
`carrier mapping according to embodiments of the present
`invention;
`FIGS. 6A and 6B illustrate the pilot structure in the fre-
`quency domain for the case of four simultaneously transmit-
`ting UEs;
`FIG. 7 illustrates another embodiment of a transmitter
`
`structure according to the present invention;
`FIG. 8 illustrates one example embodiment of the fre-
`quency domain spreader shown in FIG. 7;
`FIG. 9 illustrates an example embodiment of the receiver
`structure according to the present invention; and
`FIG. 10 illustrates a sequence assigmnent scheme accord-
`ing to an embodiment of the present invention.
`
`DETAILED DESCRIPTION OF THE EXAMPLE
`EMBODIMENTS
`
`The pilot structure of the present invention applies CDM
`concepts to FDM. For example, UEs in a same cell for simul-
`taneous transmission are assigned the same pilot or reference
`sequences, but are then transmitted over different sub-carri-
`ers; and UEs in a different, adjacent cell may transmit over the
`same sub -carriers, but are assigned different pilot or reference
`sequence. First, a discussion of the pilot sequences will be
`provided.
`According to an example embodiment ofthe present inven-
`tion, the pilot sequence should have the following properties:
`1. Unit magnitude in the transform domain
`2. Optimal cyclic auto-correlation
`3. Low, constant cyclic cross-correlation for odd-length
`sequences
`
`For example, constant amplitude sequences having low
`cyclic cross correlation such as CAZAC sequences have these
`properties. However, it will be understood that the present
`invention is not limited to CAZAC sequences. As discussed
`previously, a GCL sequence is a specific example of a
`CAZAC sequence. An odd-length GCL sequence of length P
`has optimal cyclic cross correlation of \/P and qualifies as a
`pilot sequence. For the purposes ofexample only, the embodi-
`ments ofthe present invention will be described using Zadoff-
`Chu GCL sequences. For example, the Zadoff-Chu sequence
`of length P in the time domain is generated as:
`
`n(n+ 1)
`2
`
`M for P odd
`for P even
`
`Ck(H) =
`
`J/2”/([
`
`H2]
`11+?
`
`(1)
`
`10
`
`15
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`Next, the up-link transmission frame structure win be
`described. FIG. 1 illustrates an example of the proposed
`uplink transmission sub-frame structure. Uplink transmis-
`sions are from a UE to a node B. The UE may also be referred
`to as a mobile station, mobile unit, etc. A UE may be a
`wireless phone, wireless equipped computer, wireless
`equipped PDA, etc.A node B may also be referred to as a base
`station, a base station controller, a base station router, etc.
`As shown in FIG. 1, the sub-frame structure includes sev-
`eral long block (LBs) and two short blocks (SBs) separated by
`cyclic prefixes (CPs). The long blocks carry data and the short
`blocks carry reference signals. The pilot structure according
`to an example embodiment of the present invention may be
`transmitted in one or both of the short blocks.
`
`FDM involves dividing the bandwidth of a SB into sub-
`carriers. For example, for a 5 MHz bandwidth signal, current
`proposals divide the SB into 150 sub-carriers. Also, in FDM,
`each UE is assigned a number of the sub-carriers. Current
`proposals limit the minimum number ofpilot sub -carriers that
`may be assigned to a UE as 12.5. This block of sub-carriers
`will be referred to as a resource block, and it will be under-
`stood that the present invention is not limited to a resource
`block size of 12.5 sub-carriers. Accordingly, the maximum
`number ofusers that can transmit simultaneously in a SB is 12
`(:INT(1 50/ 12.5). There are a few options to generate the
`reference or pilot signal of required length to support this
`maximum number ofusers while suppressing inter-cell inter-
`ference:
`
`Option 1 : Provide six different sequences oflength 13 (e.g.,
`P:13), and six different sequences of length 12 (e.g.,
`P:12) to fit in the SB of 150 pilot sub-carriers. For
`example, GCL sequences of length P:13 may be used,
`and the sequences of length 12 may be generated by
`truncating the GCL sequences of length 13. Unlike con-
`ventional differentiation using GCL sequences by phase
`shifting the same GCL sequence,
`this embodiment
`involves using different sequences, but of the same
`length. For example, in equation (1 ), P will be 13, but the
`value of k will change to obtain the different sequences
`(which are not shifted versions of one another).
`Option2: Use 12 sub-carriers for all resource blocks. Over-
`all, 12><12:144 pilot sub-carriers out of the available
`150 sub-carriers are used. While different sequences of
`length 12 may be used, instead, different sequences of
`length 13 truncated to length 12 may be used.
`Option 3: Use multiple sequences with the lengths corre-
`sponding to all possible number of pilot sequences.
`Option 4: Ifthe number of allowed pilot sub-carriers is 15 6,
`a length-13 sequence may be used.
`interference suppression
`FIG. 2 shows the co-charmel
`characteristic of a GCL sequence for sequence lengths rang-
`ing from 3 to 150. The full-length sequence (sequence of 150
`sub -carriers) has the capability to suppress the interference by
`up to 21.8 dB. Although the shorter sequences in the embodi-
`ments ofthe present invention are not as powerful, mo st ofthe
`gain is obtained with shorter sequences. For sequences with
`lengths of 13 and 25, that correspond to resource block sizes
`of 1 and 2, the interference may be suppressed by -11.1 dB
`and -14.0 dB, respectively.
`
`12
`
`12
`
`

`
`US 7,701,919 B2
`
`5
`As will be understood from the discussion above, accord-
`ing to embodiments of the present invention, the length ofthe
`reference or pilot signal sequence used by a UE corresponds
`to the size of the smallest resource block. However, as will be
`appreciated, more than one resource block may be assigned to
`a UE. In these instances, the embodiments of the present
`invention construct longer pilot sequences from the assigned
`resource block based sequence.
`To better understand this aspect of the present invention, a
`transmitter structure according to an embodiment of the
`present invention will now be described. FIG. 3 illustrates a
`transmitter structure according to an embodiment of the
`present invention for transmitting a reference signal accord-
`ing to the embodiments of the present invention. As shown, a
`time domain GCL sequence determined according to, for
`example, option 2 above is supplied to a spreader 10. The
`spreader 10 spreads the GCL sequence bit-by-bit in the time
`domain based on the number (S) of resource blocks assigned
`to the UE. For a user assigned S resource blocks, spreading by
`S results in a length (S><P) spread sequence. For example,
`FIG. 4 illustrates the example of spreading a sequence of
`length P over two resource blocks (i.e., S:2).
`The sequence output from the spreader 10 is discrete Fou-
`rier transformed (DFT) by a DFT unit 12 to generate a fre-
`quency-domain signal or samples. A sub-carrier mapper 14
`then maps the frequency domain samples to sub-carriers to
`generate localized or distributed FDM reference signals that
`correspond to the assigned resource block ofthe UE. Namely,
`each frequency domain sample is mapped to one of the N
`inputs of a N-point inverse fast Fourier transform (IFFT) unit
`16. The sub-carrier mapping determines which part of the
`spectrum is used for transmission by inserting a suitable
`number of zeros at the upper and/or lower end as shown in
`FIGS. 5A and SB. Between each frequency domain sample,
`L—1 zeroes are inserted. A mapping with L:1 corresponds to
`localized transmission, and the spread sequence is mapped to
`consecutive sub -carriers. This is shown in FIG. 5A. With L>1,
`distributed transmission results, as shown in FIG. 5B.
`FIGS. 6A and 6B illustrate the pilot structure in the fre-
`quency domain for the case of four simultaneously transmit-
`ting UEs. FIG. 6A shows localized pilot structures and FIG.
`6B shows the distributed pilot structures. Suppose UEs 1 and
`3 are assigned 2 resource blocks (S:2) and UEs 2 and 4 are
`assigned 1 resource blocks (S:1). For UEs 1 and 3, bit-by-bit
`spreading in the time domain generates a length-S><P refer-
`ence sequence. For localized data mapping, the sub-carrier
`mapping generates the localized FDM pilot structure shown
`in FIG. 6A. For distributed data mapping, the distributed
`FDM pilot structure shown in FIG. 6B is generated.
`Returning to FIG. 3, the resulting sequence of frequency
`domain samples received by the N-point IFFT 16 is converted
`to the time-domain. After parallel-to-serial conversion by a
`parallel-to-serial converter 18, a CP inserter 20 adds a cyclic
`prefix before the transmission of the reference signal in the
`SB.
`While the transmitter structure of FIG. 3 has been
`
`described with respect to the transmission of reference sig-
`nals such as pilot sequences, this structure as well as the other
`transmitter and receiver structures described below may be
`adapted for transmission of data trafiic such as voice or con-
`trol signaling. At the transmitter, the data would undergo
`encoding at an encoder prior to receipt by the spreader 10 and
`would undergo modulation (e.g., BPSK, QPSK, 16QAM,
`etc.) by a modulator after spreading by the spreader 10. This
`transmitter structure may be particularly applicable to
`CDMA-OFDAM systems. As will be appreciated, the reverse
`of the above operations are performed at the receiver.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`6
`Instead of spreading the sequences in the time domain and
`then performing DFT using the spreader 10 and the S><P DFT
`unit 12 described above with respect to FIG. 3, the spreading
`may be performed in the frequency domain. FIG. 7 illustrates
`another embodiment of a transmitter structure according to
`the present invention. As shown, this embodiment is the same
`as the embodiment of FIG. 3, except that the spreader 10 and
`S><P DFT unit 12 have been replaced by a P-point DFT unit 30
`and a frequency domain spreader 32. The P-point DFT unit 30
`converts the time domain sequence of length P into frequency
`domain samples. The frequency domain spreader 32 then
`spreads the frequency domain samples by the number (S) of
`resource blocks assigned to the UE to generate S><P frequency
`domain samples. Accordingly, the output of the frequency
`domain spreader 32 is the same as the output from the S><P
`DFT unit 12 in FIG. 3.
`
`FIG. 8 illustrates one example embodiment of the fre-
`quency domain spreader 32. As shown, the frequency domain
`spreader 32 includes S branches 50. Each branch 50 receives
`the frequency samples output from the P-point DFT unit 30.
`Each branch 50 includes a first multiplier 52 and a second
`multiplier 54. The first multiplier 52 phase shifts the fre-
`quency samples by the sub-carrier index t of the frequency
`sample (i.e., frequency sub-carrier index for output of the
`P-point DFT unit 30). For each branch s:0, 1,
`.
`.
`.
`, S-1, the
`phase shift
`is performed by multiplying the frequency
`samples by:
`
`(2)
`
`The second multiplier 54 multiples by a constant depending
`on the frequency sub-carrier range. This is performed by
`multiplying the output of the first multiplier 52 by:
`
`l,fort=O,l,...,P—l
`2n
`e’T,fort=P,P+l,
`
`,2P—l
`
`(3)
`
`2” S1
`eIT5"’,forz=(s—1)P, (S—1)P+1,... ,SP—l
`
`The output of the branches 50 is added to the output of the
`P-point DFT unit 30 by an adder 56 to produce the spread
`frequency domain samples.
`Having covered the transmitter structures above in detail,
`the receiver structure according to embodiments of the
`present invention will now be described. FIG. 9 illustrates an
`example embodiment of the receiver structure according to
`the present invention. As shown, a CP remover 60 removes
`the CP in the received signal, and a serial-to -parallel converter
`62 converts the serial time domain signal to parallel. An
`N-point FFT unit 64 then converts the parallel time-domain
`signal to the frequency domain. A sub-carrier demapper 66
`performs the inverse operation ofthe sub —carrier mapper 14 to
`obtain the original frequency domain samples. In the embodi-
`ment shown in FIG. 9, these samples are converted back to the
`time domain by the inverse DFT (IDFT) unit 68.A despreader
`70 then performs the inverse operation of the spreader 10 on
`the time domain sequence output from the IDFT 68. As will
`be appreciated, instead of performing the despreading opera-
`tion in the time domain, the despreading may be performed in
`the frequency domain. For example, this may be accom-
`
`13
`
`13
`
`

`
`US 7,701,919 B2
`
`7
`plished by performing the inverse of the operations per-
`formed by the frequency domain spreader 32 and the P-point
`DFT unit 30.
`
`After converting the received short block to a reference
`sequence, the channel is estimated in the code domain. The
`raw charmel estimate is obtained by correlating the received
`pilot sequence output from the despreader 70 with the refer-
`ence GCL pilot sequence at a correlator 72. Because the node
`B assigned the GCL sequence to the UE, the node B knows
`the GCL sequence to be received. The raw channel estimate is
`converted to frequency domain responses for each data sub-
`carrier. The N-point SB FFT may be used followed by fre-
`quency-domain upsampling and smoothing for the two adja-
`cent sub-carriers. Alternatively, a long-block (LB) FFT may
`be used to obtain frequency domain channel responses for all
`used data sub-carriers. Frequency-domain interpolation may
`be applied for sub-carriers within the channel coherence
`bandwidth. Time-domain interpolation may be applied in
`conjunction with frequency domain smoothing to improve
`the charmel estimation performance when the charmel
`is
`time-varying within the sub-frame. The interpolation may be
`done for each sub-carrier. Instead of 2 one-dimensional chan-
`
`nel interpolators (frequency and time domain), a single two-
`dimensional channel interpolator may be used. The interpo-
`lation is done on a time-frequency grid for the signals, after
`conversion to the frequency domain channel response.
`The estimated frequency-domain charmel response is used
`as the input to an equalizer. The equalizer may be a single-tap
`frequency-domain equalizer. Either a zero-forcing (ZF) or a
`minimum mean-squared error (MMSE) equalizer may be
`used.
`
`Next, assignment of sequences to node Bs and sequence
`reuse will be discussed. Typically when discussing the
`assignment of frequencies or sequences to node Bs, the
`assignment is discussed in terms of the cell (e.g., geographic
`area handled by the node B). Accordingly, this typical form
`for discussing such assignments will be used here.
`For in-cell users, the same GCL sequence may be reused
`without causing same-cell interference because the UEs have
`different sub-carrier assignments. A pilot or
`reference
`sequence can be allocated initially when a user is admitted to
`a cell, and does not need to be scheduled, which simplifies
`user configuration. FIG. 10 illustrates a pilot sequence assign-
`ment rule for a 3 sector system with clover-leaf cell shape. As
`an example, a GCL sequence of length 13, denoted as GCL
`(13) is used. There are 11 different GCL sequences of length
`13. As shown, the different GCL sequences of length 13 are
`indicated by the number k in FIG. 10, where k is in the form
`GCLk(13). The variable k may be the same variable k dis-
`cussed above with respect to equation (1 ). As shown, each cell
`assigns in-cell UEs the same GCL sequence, but the different
`cells in the reuse group assign different GCL sequences. "he
`reuse group is then repeated throughout the system. "he
`assignment scheme shown in FIG. 10 allows 1/7 reuse of a
`GCL sequence. As will be appreciated, different length GCL
`sequences will allow for different reuse schemes based on the
`number of different GCI, sequences of that length.
`One main difference between the frequency-domain
`CAZAC sequence and the time-domain CAZAC sequence is
`in the achievable processing gain. Since the channel can be
`assumed to be relatively constant over the SB, processing
`gain (PG) is obtained by accumulating over the length of the
`pilot sequence. For P:13, processing gain of PG:l0><log1O
`(l3):l1.1 dB is possible. For high Doppler, the coherence
`time may become smaller than the short block length, in
`
`10
`
`15
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`8
`which case, a coherent accumulation window for the correla-
`tor may need to be reduced. This will result in a reduction in
`processing gain.
`Note that equivalent gain is not obtained by using the
`CAZAC sequence in the frequency domain.
`For a GSM TU channel profile, coherence bandwidth
`defined as 1/ztmx is approximately 100 kHz. This corre-
`sponds to 3 pilot sub-carriers. For localized mapping
`using a conventional FDM pilot, processing gain of 10
`log10(3):4.8 dB is possible.
`For distributed mapping, in general, the pilot sub-carriers
`are separated by L-1 zeros and typically, uncorrelated.
`No processing gain may be obtained for distributed map-
`ping.
`It has been argued that FDM pilots suffer from co-channel
`interference at the cell edge. The problem exists when comb-
`shaped pilot tones are used or a frequency-domain CAZAC
`sequence is used. In this case, each pilot tone collides with the
`pilot of a user in a neighboring cell, who is sharing the same
`resource block. However, the low constant cyclic cross-cor-
`relation property of GCL sequences allows co-charmel inter-
`ference suppression in the code domain, thus allowing accu-
`rate channel estimation in the presence of strong interferer.
`This allows advanced receivers such as inter-cell pilot inter-
`ference cancellation or channel estimation for handover deci-
`
`sion without using the full-bandwidth CDM pilot structure.
`Embodiments of the present invention provide a ‘reduced-
`length DFT-precoded sequence’ as the FDM pilot signal. A
`GCL sequence may be suitable as the basis for generating the
`FDM pilot sequences. Compared with the comb-shaped
`FDM pilot
`tones, or
`the
`frequency-domain CAZAC
`sequence, the pilot structure of the present invention solves
`the problem of pilot collision due to dominant interferers by
`exploiting the cyclic cross-correlation property. Some of the
`advantages of using the proposed sequence include:
`1. By using the DFT-precoded pilot sequence, the channel
`may be estimated in the time domain a11d converted to
`frequency domain. For most UE velocities, coherence
`time of the channel does not exceed the pilot block
`length. Thus, processing gain corresponding to the
`sequence length is possible by coherently accumulating
`raw sarnple-rate channel estimates over the length ofthe
`pilot sequence.
`2. By using bit-by-bit spreading ofthe sequence to generate
`the FDM reference to cover users assigned multiple
`resource blocks, the same sequence may be reused for all
`in-cell users,
`irrespective of the number of resource
`blocks.
`
`3. By using the low cyclic cross-correlation property of
`GCL sequences, dominant co-charmel interferers can be
`suppressed. Such gain is not possible with other fre-
`quency-domain CAZAC pilot structure, since the fre-
`quency-domain CAZAC sequence does not have the
`desired cross-correlation property in the time domain.
`4. Although the proposed pilot is a FDM pilot, it offers th