Patent Application
20090124290
This invention relates generally to a spatial division multiple access (SDMA), and more particularly to selecting antennas for SDMA transmissions in OFDMA networks.
OFDMA
Orthogonal frequency-division multiplexing (OFDM) is a modulation technique used at the physical layer (PHY) of a number of wireless networks, e.g., networks designed according to the well known IEEE 802.11a/g, and IEEE 802.16/16e standards. OFDMA is a multiple access scheme based on OFDM. In OFDMA, separate sets of orthogonal tones (subchannels) and time slots are allocated to multiple transceivers (users or mobile stations) so that the transceivers can communicate concurrently. The IEEE 802.16 standard defines an air interface, while WiMAX includes both the IEEE 802.16 air interface and the networking aspect of the system. Terms and definitions used herein are included in the Appendix.
Antenna Selection
According to the IEEE 802.16 standards, multiple antenna elements and radio frequency (RF) chains can be supported in base stations (BS) and mobile stations (MS). Due to the high cost of RF chains and relatively low cost of antennas, the number of RF chains (N) is usually less than the number of antennas (M) at the MS, that is, N≦M.
It is known that each antenna provides a different propagation path that experiences a distinct channel gain. Therefore, it is important to selectively connect a subset N of the M available antennas to N RF chains so that the transmission and reception performance is optimized. This function is known as antenna selection (AS). Antenna selection improves system performance in terms of bit error rate (BER), signal to noise ratio (SNR) and throughput (TH).
Antenna selection at mobile station for non-SDMA transmission in an OFDMA network, where a base station communicates to one mobile station at a time, is described in U.S. patent application Ser. No. 11/777356, “Method and system for selecting antennas adaptively in OFDMA networks,” filed by Tao et al on Jul. 13, 2007, incorporated herein by reference. In U.S. patent application Ser. No. 11/777356, however, Tao et al. have not described antenna selection at mobile station for SDMA transmission, where a base station communicates with multiple mobile stations concurrently using a single piece of time and frequency resource.
It is desired to provide antenna selection at mobile station for SDMA transmission in an OFDMA network.
A method selects subsets of antennas at mobile stations for SDMA transmission in an OFDMA network. Furthermore, it is desired that the selected subsets of antennas are globally optimal for the network. A mobile station measures the channels state information (CSI) for a downlink (DL) channel from a base station to the mobile station, and feeds back the CSI to the base station. The base station measures the CSI for the uplink (UL) channel from the mobile station to the base station.
The method determines whether the downlink and the uplink channels are reciprocal based on the CSI of the DL and UL channels. The method also determines whether the antennas are to be selected locally by each mobile station or globally by the base station for SDMA transmissions.
The downlink receive antenna selection at the mobile station can be done by either the mobile station or the base station, depending on the channel reciprocity and the selection criteria. The uplink transmit antenna selection at the mobile station can be determined by the base station.
Spatial Division Multiple Access (SDMA)
From time to time, antenna selection is performed for selected mobile stations. The designated mobile stations can be one or more of the mobile stations in the set that have more than one antennas. The designated mobile station can be a station that is entering the network, or a mobile station is experiencing a degradation of performance due to its location, and perhaps can improve its channel by having if a different subset of its antennas were used. Either the base station or a mobile station itself can determine whether it should be a designated mobile station for the purpose of antenna selection, see
Spatial division multiple access (SDMA) can be achieved by using multiple-input multiple-output (MIMO) techniques with precoding and multi-user scheduling. SDMA exploits information of the locations of the MSs 102 within a cell. With SDMA, the radiation patterns of the RF signals 104 are adapted to obtain a high gain in a particular direction towards the location of the MSs, and low gain in the direction of interferers (other MSs operating on the same resource). This is often called beam forming or beam steering. BSs that support SDMA transmit signals to multiple MSs concurrently using the same resources (e.g., OFDMA symbol and frequency subchannel), but perhaps, directed in different directions. SDMA can increase network capacity, as multiple transmissions can occur in parallel. MSs with multiple antennas can also employ SDMA. The precoding is performed at the BS before downlink transmission, using channel state information (CSI) acquired by the BS through feedback from the MS or obtained based upon channel reciprocity.
In IEEE 802.16e, SDMA can be performed in either AAS or STC zone, Linder various permutation zones. The related definitions are given in the Appendix.
Frame Structure
The basic unit of resource for allocation in OFDMA is a slot 200. The size of the slot is based on the permutation modes that the MS and the BS use for transmissions in downlink (DL) and uplink (UL). A permutation mode defines the type of resource allocation in time and frequency domains. Different modes are defined for the downlink and uplink. By using a specific permutation, a given number of OFDMA symbols 201 and subchannels are included in each slot.
The slot has an associated time (k) and subchannel (s). Each slot can carry one or more OFDMA symbols. The base station partitions the time domain into contiguous frames 210 including a DL subframe 251 and a UL subframe 252. During the DL subframe, all traffic is from the base stations to the mobile stations. During the UL subframe, all traffic is from the mobile stations to the base stations.
The DL subframe starts with a preamble 220 on all subchannels. The preamble enables the mobile stations to perform synchronization and channel estimation. The first subchannel in the first two OFDMA symbols in the downlink is the frame control header (FCH) 202. The FCH is transmitted using QPSK rate ½ with four repetitions. The FCH specifies a length of the immediately succeeding downlink MAP (.DL-MAP) 260 and the repetition coding used for the DL-MAP.
The BS uses the downlink MAP (DL-MAP) and an uplink MAP (UL-MAP) 261 to notify the MSs of the resources allocated to data bursts 204 in the downlink and uplink direction, respectively, within the current frame. The bursts are associated with connection identifiers (CID). Based upon a schedule received from the BS, each MS can determine when (i.e., OFDMA symbols) and where (i.e., subchannels) the MS should transceive (transmit or receive) with the BS. The first two subchannels 203 in the UL subframe are used for ranging.
A receive/transmit gap (RTG) separates the frames, and a transmit transition gap (TTG) separates the subframes within a frame. This enables the transceivers to switch between transmit and receive modes.
Data signals (OFDMA symbols) are transmitted in bursts 204 comprising one or more slots. Each channel or wireless connection 104 between the BS and the MS is allocated a time domain and frequency domain resource, which contains a two dimensional block of slots, i.e., time duration and frequency subchannels. With OFDMA, stations can communicate with each other on the connections 104 by using the allocated two-dimensional resources.
OFDM Symbol
The current IEEE 802.16e standard, which uses OFDMA for both downlink and uplink for multiple access, does not support antenna selection at mobile stations.
Antenna Switching
During antenna selection, the mobile station tests which subset of antennas is optimal. In the OFDMA network, a station can switch antennas during the cyclic prefix (CP) period of an OFDMA symbol, and use the selected set of antennas to transmit or receive the OFDMA symbol. The antenna switching is usually completed in tens or hundreds of nanoseconds, while the CP duration is several microseconds. Therefore, the CP prefix is sufficiently long to support antenna switching without any loss of data.
Antenna selection can be performed for transmission, i.e., transmit antenna selection (TAS), or for reception, i.e., receive antenna selection (RAS), or both. For example, in an IEEE 802.16 network, antenna selection can be used at the BS or the MS, or both, in either the downlink or the uplink or both. However, due to cost or complexity considerations, it is usually the MS that has fewer RF chains than antennas.
Under most circumstances, antenna selection is only used when the number of antennas is greater than the number of RF chains. The following description thus focuses on antenna selection for the MS, which essentially refers to TAS in the uplink case and RAS in the downlink case. Nevertheless, the method described below can also apply to the BS.
We will concentrate on the antenna selection for SDMA transmissions, wherein same allocated resource is used for communications between the base station and multiple mobile stations. In this scenario, preceding and scheduling are performed across the multi-user channels, which correlates the antenna selection on one MS with that on other MSs.
The selection can be locally or globally optimal. Local selection is based only on information available at a particular mobile station. Global selection considers information not only related to the MS where antenna selection occurs, but also that associated with other MSs involved in the same SDMA transmission. Thus, local selection is locally optimal and can be performed directly at the MS, while global selection is globally optimal and can only be performed by the BS.
Channel Reciprocity
The MS can be stationary, nomadic or mobile and can be located at any location in the coverage area of the BS 101. In a time division duplexing (TDD) system, DL and UL subframes share the same frequency band in each frame. The frame duration of the IEEE 802.16 network is of the order of 10 ms, which is typically less than the channel coherence time. Hence, it is reasonable to assume that the UL and DL channels are reciprocal. By reciprocal, we mean that the channel states and qualities are substantially the same on the downlink and the uplink.
When channel reciprocity is assumed, the same subset of antennas is to be used for both uplink transmission and downlink reception by the mobile station. This simplifies the antenna selection for the MS, as the antenna subset selected for the downlink can be directly reused for the uplink, and vice versa. In an actual implementation, the RF chains for upconversion and downconversion in a device might be non-reciprocal. Such a non-reciprocity can be eliminated by appropriate calibration.
Nevertheless, it is not always valid to assume channel reciprocity: even if the channel appears reciprocal, because inter-cell interference is not reciprocal. Moreover, in networks that use frequency division duplexing (FDD), different frequency resources may be allocated to the MS in the downlink and the uplink. Then, the antenna subset selected for one direction may not be the best set for the reverse direction. In this case, it becomes necessary to perform antenna selection separately for the downlink and the uplink.
The adaptive antenna selection method according to embodiments of the invention can accommodate both reciprocal and non-reciprocal channels as described above.
Antenna Selection Method
A globally optimal subsets 122 of the sets of antennas is selected 120 for each designated mobile station based on all of the available CSI 111 for all of the mobile stations, and optionally the cost factors 112. The selection can minimize a cost function. There is one globally optimal subset for each designated mobile station.
Each designated mobile station is notified of the globally optimal subset 122 of the antennas to be used for subsequent communication between the base station and each designated mobile station.
Local and Global Optimizing During Antenna Selection
The optimization can be local or global. Local optimization only considers information available for the mobile station for which an optimal subset of antennas is being selected. Global optimization considers information available for all mobile stations when selecting optimal subsets of antennas.
As shown in
In the uplink case, the BS has the CSI of the uplink channel and can decide which antenna set each MS should use. Again, BS can either only consider the uplink channel information when it selects antenna subset for the MS, or it can also take information related to other MSs involved in the same SDMA transmission into consideration for antenna selection decision. The former uplink transmit antenna selection is locally optimal (i.e., LS UL TAS 422).
The IEEE 802.16e standard defines the protocol, MAC message and PHY support for SDMA. Our invention enables antenna selection at the mobile stations that are engaged in SDMA transmission. Based on the selection criteria 111-112, we describe how antenna selection is performed under various scenarios. In addition, we describe MAC signaling protocols to support the antenna selection of our invention.
Local Selection
For the locally optimal selection 420 at a designated mobile station (MSs), the antenna selection is based on the CSI between only this MS and the BS. In this case, antenna selection on one MS does not affect the selected antennas on other MSs. That is, stations other than the designated station(s) do not have to switch antennas. The procedure is the same for all of the MSs. Therefore, it is sufficient to describe the local antenna selection for just one MS, say MSk in
For LS DL RAS 421, once the MS decides to initiate antenna selection, the designated station uses different subsets of antennas to receive different OFDMA symbols that contain pilot subcarriers 801, and estimates the CSI 111 of the channel associated with each different subset of antennas.
As shown in
Different from the method described by Tao et al., the MS reports to BS the channel quality information (CQI) 802 associated with the channel between BS and the selected antenna subset at MS. Then, appropriate precoding can be chosen and applied for the SDMA DL data 803 in the subsequent frames.
If the channel is reciprocal, the same antenna subset selected for reception on the downlink channel can then be used for transmission on the uplink. Otherwise, when the channel is not reciprocal, the BS transmits an ASC UL IE 901 to MS to initiate the UL TAS process 422, as shown in
The designated MS transmits pilots 902 using different antenna subsets at different OFDMA symbol to the BS. The BS estimates the channel between it and various antenna subsets at the MS, and selects a subset. The BS then informs MS of this antenna selection result 920.
The manner in which pilot subcarriers are embedded in the allocated resource block has an impact on the antenna switching during the selection process. The pilot subcarriers, also known as the pilot, has different locations in different permutation schemes, e.g., FUSC, PUSC, AMC, TUSC1, TUSC2, etc.
As an example, we describe the PUSC permutation, which is mandatory for UL transmission. Notice that TUSC 1 is an optional permutation mode defined in AAS zone for the 802.16e DL and that the slot structure of TUSC 1 is the same as that for UL PUSC mode.
UL Partially Used Subcarrier (PUSC) Mode
As shown in
As defined in the Appendix, SDMA system allocates different mobile stations the same time and frequency resource but orthogonal pilot patterns (A-D) 601-604, as shown in
This antenna switching process continues until the MS finishes testing all possible antenna sets, or decides to terminate otherwise. Then, the BS can select the appropriate antenna subset based on the selection criteria (CSI and cost factors) 111-112 such as CSI, SINR or capacity using the cost function 121, and notify 130 MS of the selected antenna subset to the MS. The cost factors can include the cost of switching antennas, precoding and modulation costs, and the like.
DL/UL Adaptive Modulation Coding (AMC) Mode
The DL/UL AMC mode is an optional permutation mode for 802.16e. The subcarriers are grouped in each sub-channel and are physically adjacent. The slot structure is shown in
The pilots in an AMC AAS zone are regarded as part of the allocation, and as such are beamformed in a way that is consistent with the transmission of the allocated data subcarriers. In an SDMA region, the pilots of each allocation may correspond to a different pilot pattern, which consists of location and polarity. The pilot patterns 611,-614 for a 3×2 AMC slot 550 are depicted in
Global Selection
For globally optimal antenna selection (GS) 430 as shown in
We describe an example that shows why local selection is not optimal for antenna selection in SDMA. Consider an uplink SDMA between the base station (BS) and two mobile stations (MSs), each mobile station has two antennas and one RF chain. Define P1 (P2) to be MS, (MS2)'s transmit power, h1 (h2) to be the channel vector from MS1 (MS2) to the BS, θ(h1, h2) to be the angle between h1 and h2, and N0 to be the noise power at the BS.
To select one transmit antenna from each MS, in general, the performance of this SDMA system should be measured by its capacity region as
R
1
≦C
1
, R
2
≦C
2, and R1+R2≦:Csum, with
C
1=½×log [1+∥h1∥2×P1/N0];
C
2=½×log [1+∥h2∥2×P2/N0]; and
C
sum=½×log [1+∥h1∥2×P1/N0+∥h2∥2×P2/N0+∥h1∥2×∥h2∥2×P1×P2/N02×sin2θ(h1, h2)],
where
sin2θ(h1, h2)=1−″h1H×h2″2/(∥h1∥2×∥h2∥2),
from B. Sutard, G. Xu, H. Liu, and T. Kailath, “Uplink Channel Capacity of Space-Division-Multiple-Access Schemes” IEEE Transactions on Information Theory, vol. 44, no. 4, pp. 1468-1476, July 1998, incorporate herein by reference. The above can be cost factors to be considered during the optimization for the antenna selection.
Suppose now MS, has selected the antenna with channel vector h1, and it is the time to select antenna on MS2. If local selection is used at MS2, the antenna which gives a larger C2, i.e., a larger ∥h2∥2, is selected.
However, when global selection is used to select an antenna subset for MS2, we should consider not only C2 but also Csum. This may result in a different choice of antenna with channel vector h2′, because although ∥h2∥2>∥h′2∥2, h′2 may give a larger sin2θ(h1, h′2) than sin2θ(h1, h2), which could result in a larger Csum.
Using global selection in SDMA wireless network, the selection decision 410 is made at the BS, which has access to the measured CSI for all the related channels 104.
The uplink transmit antenna selection is normally initiated (triggered) by the BS. The MSs use different antenna subsets to transmit pilots for the BS to acquire the CSI for each antenna subset and the BS. After the BS collects the CSI from all MSs, global optimal criteria can be used to select antenna sets for each MS.
For the downlink, both MS and BS can initiate the DL receive antenna selection. Each MS measures the CSI between the BS and different antenna subsets, and feeds back the CSI to the BS. Again, the decision of which antenna subset to use is made by the BS, after globally considering the CSI from all MSs.
We next describe the global antenna selection for SDMA in a WiMAX network.
WiMAX Implementation
For the uplink 432, the base station directs a designated MS to perform global antenna selection by transmitting the ASC UL IE 901. In response, the designated MS transmits the OFDMA symbol with pilots using the different antenna subsets 902. The BS estimates the CQI from the MS, and determines an antenna subset that is globally optimal for the network. The BS indicates the selection in the ASC UWL IE 920, which can be used for subsequent data transmissions 930.
It is also possible that the BS sends additional pilot(s) 1005 prior to the CQICH, allocation E or after the CQICH allocation IE, in case the pilots in the CQICH allocation IE are not sufficient for MS to complete testing all its antenna sets.
Based upon the CQI report 1020 received from the MS, BS can select the subset of antennas for MS to use in downlink for reception 1040. BS indicates the selected antenna set in ASC_DL_IE 1030 to MS.
In this case, MSk 102 can initiate the downlink receive antenna selection 431, and request 1110 bandwidth from BS 101 by sending a CQICH allocation request 1120. In order to send CQICH allocation request, however, the MS has to first acquire proper uplink resource. MS can use various bandwidth request (BR) schemes and contention-resolution protocol defined in current IEEE 802.16 standard to acquire such uplink resource. The MS then can use the allocated uplink bandwidth to request a CQICH allocation, if the MS does not have a CQICH and its current subset of antennas in use do not give a satisfactory performance. The BS 102 allocates the CQICH to the MS 1130, and transmits 1135 the pilot signals in the DL. The MS receives the pilot signals using different subset of antennas, and estimates the DL channel associated with each antenna subset based upon the pilot. The MS informs 1140 the BS of the CQI for the channel associated with each antenna subset.
At some time, the CQICH has been allocated to the MS, but the last CQI report of the MS is no longer appropriate for the duration remaining until the next periodic CQI feedback. In this case, the MS can transmit the unsolicited CQI report 1150 to BS through a REP-RSP message. The MS ensures in advance that there is enough bandwidth for the REP-RSP message. Otherwise, the MS performs contention-based BR first.
As long as BS obtains the CQI from MS, either through CQICH or REP-RSP, the BS can select the appropriate subset of antennas for the MS by taking into account other MSs' CQI, which is stored at the BS.
The BS notifies tie MSs by transmitting the ASC DL IE 1160 to each MS before SDMA downlink transmission 1170. The ASC_DL_IE is an extended DL-MAP IE to support antenna selection signaling, as shown in Table 2 in the Appendix, where each field has a similar meaning to the corresponding field in Table 1 in the Appendix.
As a side note, after receiving the unsolicited CQI report 1150 from an MS, the BS can modify the Period (p) in the CQICH according to MS's request. Also the BS can disable the “triggered update” in the CQICH_Allocation_IE for a period of time so that the MS cannot transmit unsolicited CQI update.
Additional Issues
The antenna selection method described above should take care of some other issues when the invention is implemented in an SDMA network.
First, to perform precoding and antenna selection in an SDMA network, the BS needs to have a period update of all the CQI and response to the unsolicited CQI report from any MS. As a consequence, the BS stores the most currently acquired CQI for each MS and updates whenever there is a pilot or CQI feedback transmitted from some MS, and whenever a new decision is made on antenna selection.
Second, during SDMA communication in a wireless network, there may be some network update due to re-grouping of the users, re-allocation of the resources, mobility and handover of some users, and the change of traffic demand. Any update can initiate antenna selection at the mobile stations. To avoid too many CQI exchanges (which may be redundant) between MSs and BS, the BS may use the stored CSI to precode and select antennas at the MSs after an update. Antenna switching can be performed later when the configuration of network stabilizes.
Third, the MSs and the BS agree on the type of CQI feedback for each antenna subset. The CQI feedback type includes the channel matrix, average carrier-to-interference-and-noise ratio (CINR) and the maximum CINR. The BS needs the entire channel matrix to perform beam forming, and the average CINR to select the antenna subsets. The MS can select the antennas locally and feed back the maximum CINR to the BS.
Notification
Before the designated MS performs transmit or receive antenna selection, the MS informs the BS of its capability of supporting such functionality by exchanging Subscriber Station Basic Capability Request (SBC-REQ) and SS Basic Capability Response (SBC-RSP) messages with the BS.
Specifically, when the designated MS performs initialization and joins the network, the MS transmits an SS Basic Capability Request (SBC-REQ) message to indicate to the BS that the MS can perform antenna selection in SDMA, with the TLV Encoded Information shown in Table 3 in the Appendix.
In addition, MS can inform the BS of the number of possible antenna set combinations the MS tests. Therefore, we define two new TLVs in Tables 4 and 5 in the Appendix to support this signaling. By knowing the number of sets that are tested, the BS can signal the index to the order in which the sets are tested to indicate the best set by ASC IE.
When MS transmits unsolicited CQI report, indication of the corresponding antenna set should be notified to the BS. Therefore, we define a new TLV in Table 6 in the Appendix to support this signaling.
After BS receives the unsolicited report from MS, the BS indicates the action performed upon receiving this kind of trigger, which makes the update of the Type/function/action description of Trigger TLV, as shown in Table 7 in the Appendix.
Recall that we have defined the Antenna_Selection_Control_UL_IE and Antenna_Selection_Control_DL_IE in Extended-2 UIUC and Extended-2 DFUC codes respectively. So the Extended-2 UIUC and Extended-2 DIUC code assignments are updated as shown in Tables 8 and 9 in the Appendix. 1001141 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.
