WO2012043202A1

WO2012043202A1 - User equipment and method for pre-coding matrix index feedback

Info

Publication number: WO2012043202A1
Application number: PCT/JP2011/070674
Authority: WO
Inventors: Ming Ding; Renmao Liu; Zeng YANG; Yongming Liang
Original assignee: Sharp Kabushiki Kaisha
Priority date: 2010-09-30
Filing date: 2011-09-06
Publication date: 2012-04-05
Also published as: CN102447523A

Abstract

A novel scheme for Pre-coding Matrix Index (PMI) transmission over Physical Uplink Shared Channel (PUSCH) is provided for Channel State Information (CSI) feedback in dual codebook pre-coding architecture. According to the present invention, the PMI is decomposed into a frequency-selective portion and a non-frequency-selective portion, which are then jointly coded. The frequency-selective portion of the PMI can be one of a beam selection portion and a phase combination portion of the PMI while the non-frequency-selective portion of the PMI can be the other one of the beam selection portion and the phase combination portion of the PMI. The present invention has advantages such as simple implementation and low signaling overhead and is applicable in LTE-Advanced/4G cellular communication systems and future 5G cellular communication system.

Description

DESCRIPTION

TITLE OF INVENTION :

USER EQUIPMENT AND METHOD FOR PRE-CODING MATRIX INDEX FEEDBACK TECHNICAL FIELD

The invention relates to communication technology, and more particularly, to a technique for a User Equipment (UE) to transmit a Pre-coding Matrix Index (PMI) over a Physical Uplink Shared CHannel (PUSCH) in a cell of a multi-antenna multi-carrier base station.

BACKGROUND ART

Multi-antenna wireless transmission technique, or Multiple In Multiple Out (MIMO) , can achieve spatial multiplex gain and spatial diversity gain by deploying a plurality of antennas at both the transmitter and the receiver and utilizing the spatial resources in wireless transmission . Researches on information theory have shown that the capacity of a MIMO system grows linearly with the minimum of the number of transmitting antennas and the number of receiving antennas.

Fig. 1 shows a schematic diagram of a MIMO system. As shown in Fig. 1 , a plurality of antennas at the transmitter and a plurality of antennas at each of the receivers constitute a multi-antenna wireless channel containing spatial domain information . Further, Orthogonal Frequency Division Multiplexing (OFDM) technique has a strong anti-fading capability and high frequency utilization and is thus suitable for high speed data transmission in a multi-path and fading environment. The MIMO-OFDM technique, in which MIMO and OFDM are combined, has become a core technique for a new generation of mobile communication.

For instance, the 3^rd Generation Partnership Project (3GPP) organization is an international organization in mobile communication field which plays an important role in standardization of 3G cellular communication technologies . Since the second half of the year 2004 , the 3GPP organization has initiated a so-called Long Term Evolution (LTE) proj ect for designing Evolved Universal Terrestrial Radio Access (EUTRA) and Evolved Universal Terrestrial Radio Access Network (EUTRAN) . The MIMO-OFDM technique is employed in the downlink of the LTE system. In a conference held in Shenzhen, China in April 2008, the 3GPP organization started a discussion on the standardization of 4G cellular communication systems (currently referred to as LTE-A systems) . Again, the MIMO-OFDM technique becomes a key technique for air interface in the LTE-A system.

In the LTE-A system, Carrier Aggregation (CA) is a new concept. Fig. 2 illustrates the CA concept in which a base station is provided with a plurality of downlink carriers and a plurality of uplink carriers . A number of carriers can be virtually combined into one carrier, which is referred to as carrier aggregation . The LTE-A system can support continuous CA as well as intra-band or inter-band non-continuous CA, with a maximum aggregated bandwidth of 1 00MHz. In order to ensure effective utilization of the carriers at the initial stage of the commercial deployment of the LTE-A system, i. e . , to ensure that LTE UEs can acce ss the LTE-A system, each carrier should be configured to be backward compatible with the LTE system . However, it is also possible to design a carrier dedicated to the LTE-A system . At the research stage of the LTE-A system, related studies on CA focus on improvement of spectral utilization for continuous CA, design of control channels for asymmetric uplink/ downlink CA scenario , and the like . Herein , the design of control channels involves feedback of downlink Channel State Information (C SI) from a UE to a B S .

There are two feedback channels for downlink CSI feedback, a Physical Uplink Control CHannel (PUCCH) and a Physical Uplink Shared CHannel (PUSC H) . In general, the PUCCH is configured for transmission of synchronized, basic C SI with low payload ; while PUSCH is configured for transmission of bursty, extended C SI with high payload . For the PUCCH , a complete CSI is composed of different feedback contents which are transmitted in different sub-frames . For the PUSCH , on the other hand , a complete CSI is transmitted within one sub-frame . Such design principles will remain applicable in the LET-A system.

The feedback contents can be divided into three categories: Channel Quality Index (CQI) , Pre-coding Matrix Index (PMI) and Rand Index (RI) , all of which are bit quantized feedbacks. In the LTE-A system, these three categories of contents are still the primary feedback contents. For PMI , it is currently agreed that a PMI is collectively determined from two pre-coding matrix indices # 1 and #2 (W l and W2) (i. e . , a dual codebook design scheme) , where W l represents wideband/ long-term channel characteristics and W2 represents sub-band / short-term channel characteristics. In transmission of W l and W2 over the PUCCH , it is not necessary for simultaneous feedback of W l and W2 within the same sub-frame. Moreover, W l or W2 may be omitted in the feedback. This is described in 3GPP R l - 102579 , "Way forward on Rel. 10 feedback".

All the frequency ranges corresponding to the CSI feedback are referred to as Set S . In the LTE system where there are only single-carrier situations, the Set S is defined as equal to the carrier bandwidth of the system. In the LTE-A system where there are additionally multi-carrier situations, the Set S can be defined as equal to the bandwidth of one single carrier or equal to the summed bandwidth of multiple carriers.

In the LTE system, the following eight MIMO transmission approaches for downlink data are defined:

1 ) Single antenna transmission . This is used for signal transmission at a single antenna BS . This approach is a special instance of MIMO system and can only transmit a single layer of data.

2) Transmission diversity. In a MIMO system, diversity effects of time and/ or frequency can be utilized to transmit signals, so as to improve the reception quality of the signals. This approach can only transmit a single layer of data.

3) Open-loop space division multiplexing. This is a space division multiplexing without the need for PMI feedback from UE.

4) Closed-loop space division multiplexing. This is a space division multiplexing in which PMI feedback from UE is required.

5) Multi-user MIMO . There are multiple UEs simultaneously participating in the downlink communication of the MIMO system.

6) Closed-loop single layer pre-coding. Only one single layer of data is transmitted using the MIMO system. The PMI feedback from UE is required.

7) Beam forming transmission. The beam forming technique is employed in the MIMO system. A dedicated reference signal is used for data demodulation at UE . Only one single layer of data is transmitted using the MIMO system . The PMI feedback from UE is not required.

8) Two-layer beam forming transmission. The UE can be configured to feed back PMI and RI , or not to feed back PMI and RI .

In the LTE-A system, the above eight transmission approaches may be retained and/ or canceled, and / or a new transmission approach, dynamic MIMO switching, can be added, by which the BS can dynamically adjust the MIMO mode in which the UE operates.

In order to support the above MIMO transmission approaches , a variety of CSI feedback modes are defined in the LTE system. Each MIMO transmission approach corresponds to a number of CSI feedback modes, as detailed below.

There are four CSI feedback modes for the PUCCH , Mode 1 -0, Mode 1 - 1 , Mode 2-0 and Mode 2 - 1 . These modes are combinations of four feedback types, including:

1 ) Type 1 : one preferred sub-band location in a Band Part (BP, which is a subset of the Set S and has its size dependent on the size of the Set S) and a CQI for the sub-band. The respective overheads are L bits for the sub-band location, 4 bits for the CQI of the first codeword and 3 bits for the CQI of the possible second codeword which is differentially coded with respect to the CQI of the first codeword .

2) Type 2 : wideband CQI and PMI . The respective overheads are 4 bits for the CQI of the first codeword, 3 bits for the CQI of the possible second codeword which is differentially coded with respect to the CQI of the first codeword and 1 , 2 or 4 bits for PMI depending on the antenna configuration at BS .

3) Type 3 : RI . The overhead for RI is 1 bit for two antennas, or 2 bits for four antennas, depending on the antenna configuration at BS .

4) Type 4 : wideband CQI . The overhead is constantly 4 bits .

The UE feeds back different information to the BS in correspondence with the above different types.

The Mode 1 -0 is a combination of Type 3 and Type 4. That is, the feedbacks of Type 3 and Type 4 are carried out at different periods and/ or with different sub-frame offsets . In the Mode 1 -0, the wideband CQI of the first codeword in the Set S and possibly the RI information are fed back.

The Mode 1 - 1 is a combination of Type 3 and Type 2. That is, the feedbacks of Type 3 and Type 2 are carried out at different periods and/ or with different sub-frame offsets . In the Mode 1 - 1 , the wideband PMI of the Set S , the wideband CQI s for the individual codewords and possibly the RI information are fed back.

The Mode 2-0 is a combination of Type 3 , Type 4 and Type 1 . That is, the feedbacks of Type 3 , Type 4 and Type 1 are carried out at different periods and/ or with different sub-frame offsets. In the Mode 2-0, the wideband CQI of the first codeword in the Set S, possibly the RI information as well as one preferred sub-band location in the BP and the CQI for the sub-band are fed back.

The Mode 2-1 is a combination of Type 3, Type 2 and Type 1. That is, the feedbacks of Type 3, Type 2 and Type 1 are carried out at different periods and/ or with different sub-frame offsets. In the Mode 2-1, the wideband PMI of the Set S, the wideband CQIs for the individual codewords and possibly the RI information, as well as one preferred sub-band location in the BP and the CQI for the sub-band are fed back.

There are thus the following correspondence between the MIMO transmission approaches and the CSI feedback modes:

MIMO transmission approach 1): Mode 1-0 and Mode 2-0;

MIMO transmission approach 2): Mode 1-0 and Mode 2-0; MIMO transmission approach 3): Mode 1-0 and Mode 2-0;

MIMO transmission approach 4): Mode 1-1 and Mode 2-1;

MIMO transmission approach 5): Mode 1-1 and Mode 2-1;

MIMO transmission approach 6): Mode 1-1 and Mode 2-1;

MIMO transmission approach 7): Mode 1-0 and Mode 2-0; and

MIMO transmission approach 8): Mode 1-1 and Mode 2-1, with PMI/RI feedback from UE; or

Mode 1-0 and Mode 2-0, without PMI/RI feedback from UE.

On the other hand, there are five CSI feedback modes for the PUSCH, Mode 1-2, Mode 3-0, Mode 3-1, Mode 2-0 and Mode 2-2.

In the Mode 1 -2 , the PMIs of the individual sub-bands in the Set S , the wideband CQIs of the individual sub-bands in the Set S and possibly the RI information are fed back.

In the Mode 3-0 , the CQI for the first codeword of each sub-band in the Set S , the wideband CQI of the first codeword in the Set S and possibly the RI information are fed back. Herein, the sub-band CQIs are differentially coded with respect to the wideband CQI , so as to reduce feedback overhead .

In the Mode 3- 1 , the CQIs for the individual codewords of each sub-band in the Set S , the wideband CQI s of the individual codewords in the Set S , the wideband PMI of the Set S and possibly the RI information are fed back. Herein, the sub-band CQIs are differentially coded with respect to the wideband CQIs, so as to reduce feedback overhead.

In the Mode 2-0 , the locations of the preferred M sub-bands in the Set S , the wideband CQI for the first codeword in each of the M sub-bands, the wideband CQI of the first codeword in the Set S and possibly the RI information are fed back.

In the Mode 2-2 , the locations of the preferred M sub-bands in the Set S, the wideband PMIs for the M sub-bands, the wideband CQIs for the individual codewords in each of the M sub-bands, the wideband PMI of the Set S , the wideband CQIs of the individual codewords in the Set S and possibly the RI information are fed back.

MIMO transmission approach 1): Mode 2-0 and Mode 3-0; MIMO transmission approach 2): Mode 2-0 and Mode 3-0;

MIMO transmission approach 3): Mode 2-0 and Mode 3-0;

MIMO transmission approach 4): Mode 1-2, Mode 2-2 and Mode 3-1;

MIMO transmission approach 5): Mode 3-1;

MIMO transmission approach 6): Mode 1-2, Mode 2-2 and

Mode 3-1;

MIMO transmission approach 7): Mode 2-0 and Mode 3-0; and

MIMO transmission approach 8): Mode 1-2, Mode 2-2 and Mode 3-1, with PMI/RI feedback from UE; or

Mode 2-0 and Mode 3-0, without PMI/RI feedback from UE.

There are currently few references available for the CSI feedback over PUSCH in the LTE-A system, as this has not been extensively discussed in the standardization process. The only existing documents mainly focus on the general design of the feedback, including:

1) Codebook design for Wl (PMI #1) and W2 (PMI #2). Wl represents angular domain information of a channel and is composed of a set of adjacent beams, while W2 is used for beam selection and phase combination. Reference can be made to 3GPP Rl-105011, "Way Forward on 8Tx Codebook for Rel.10 DL MIMO", ALCATEL-LUCENT, et al. However, this document does not mention the CSI feedback over PUSCH.

2) Design principle for CSI feedback over PUSCH. Wl and W2 can be fed back together in one PUSCH transmission. It is possible to support on PUSCH a new feedback mode 3-2 in which the sub-band PMI and the sub-band CQI are fed back. The feedback mode 3-2 and/or the original feedback mode 2-2 will be finally supported. Reference can be made to 3GPP Rl-105010, "WF on Aperiodic PUSCH CQI Modes in Rel.10", ALCATEL-LUCENT, et al.

3) Principle for transmitting Wl and W2 over PUCCH. When Wl and W2 are transmitted separately in different sub-frames, Wl and RI are jointly coded (the total overhead for feedback of Wl and RI is less than or equal to 5 bits) and transmitted together in one sub-frame. When Wl and W2 are transmitted in one single sub-frame, the codebook is down-sampled such that the total overhead for feedback of Wl and W2 is less than or equal to 4 bits. Reference can be made to 3GPP Rl-104234, "Way Forward on CSI Feedback for Rel.10 DL MIMO", Texas Instruments, et al.

In the above approach 2), no research has been made in the prior art on the design for sub-band PMI feedback over PUSCH, though it is closely related to the codebook design for Wl and W2. Therefore, in the codebook design of the above approach 1 ) , how to design the sub-band PMI feedback over PUSCH is a problem to be solved . However, there is currently no reference available for this aspect. SUMMARY OF INVENTION

It is an object of the present invention to solve the problem in the prior art of impossibility for sub-band PMI feedback over PUSCH in a dual codebook design by providing a novel method in which a PMI #2 is decomposed into a frequency- selective portion and a non-frequency-selective portion. According to the present invention, for each sub-band, the frequency-selective portion is fed back with a higher overhead while the non-frequency-selective portion is fed back with a lower overhead, such that the overall overhead can be reduced while achieving good performance .

According to a first aspect of the present invention, a method for Pre-coding Matrix Index (PMI) feedback is provided, which comprises the following steps of: receiving a downlink transmission approach, a feedback mode and feedback resources configured by a Base Station (BS) ; decomposing the PMI into a frequency-selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode, and jointly coding the frequency-selective and non-frequency-selective portions of the PMI ; and feeding the jointly coded PMI back to the BS over the feedback resources .

Preferably, the non-frequency-selective portion of the PMI is one of a beam selection portion and a phase combination portion of the PMI ; and the frequency- selective portion of the PMI is the other one of the beam selection portion and the phase combination portion of the PMI .

Preferably, the method for PMI feedback further comprises down-sampling a codebook of the PMI prior to the decomposing step; wherein the decomposing step comprises: decomposing the PMI into the frequency-selective portion and the non-frequency-selective portion based on the down-sampled codebook.

Preferably, the non-frequency-selective portion of the PMI is directly coded based on wideband . Alternatively, the non-frequency-selective portion of the PMI is differentially coded for feedback based on each sub-band with respect to wideband.

Preferably, the frequency-selective portion of the PMI is directly coded based on each sub-band. Alternatively, the frequency-selective portion of the PMI is differentially coded for feedback based on each sub-band with respect to wideband.

Preferably, the PMI is PMI #2 , i. e . , W2.

According to a second aspect of the present invention, a User Equipment (UE) is provided, which comprises: a receiving unit configured for receiving a downlink transmission approach, a feedback mode and feedback resources configured by a Base Station (BS) ; a coding unit configured for decomposing the PMI into a frequency-selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode and j ointly coding the frequency-selective and non-frequency-selective portions of the PMI ; and a transmitting unit configured for feeding the jointly coded PMI back to the BS over the feedback resources.

Preferably, the non-frequency-selective portion of the PMI is one of a beam selection portion and a phase combination portion of the PMI; and the frequency- selective portion of the PMI is the other one of the beam selection portion and the phase combination portion of the PMI .

Preferably, the UE further comprises a down-sampling unit configured for down-sampling a codebook of the PMI ; wherein the coding unit is configured for decomposing the PMI into the frequency-selective portion and the non-frequency- selective portion based on the down-sampled codebook.

Preferably, the coding unit is configured for directly coding the non-frequency- selective portion of the PMI based on wideband. Alternatively, the coding unit is configured for differentially coding the non-frequency- selective portion of the PMI for feedback based on each sub-band with respect to wideband . Preferably, the coding unit is configured for directly coding the frequency- selective portion of the PMI based on each sub-band. Alternatively, the coding unit is configured for differentially coding the frequency-selective portion of the PMI for feedback based on each sub-band with respect to wideband.

Preferably, the PMI is PMI #2 , i. e. , W2.

Additionally, in an actual system, the BS and / or the UE can adaptively employ one of the following Solution 1 and Solution 2 (i. e . , change the decomposition manner for W2 and swap the frequency-selective and non-frequency-selective portions of W2) according to variation of communication scenarios (e . g. , when the channel RI value and thus the applicable solution changes according to a correspondence table between channel RI values and applicable solutions; when the user changes from median-speed movement to non-median-speed movement; or when the number of scattering objects in a channel varies due to channel environment variation) , configuration of the UE by the BS, or autonomous selection made by the UE (which needs to be notified to the BS by means of feedback) .

The present invention has advantages such as simple implementation and low signaling overhead and is applicable in LTE-Advanced/ 4G cellular communication systems and future 5G cellular communication system. BRIEF DESCRIPTION OF DRAWINGS

The above and other obj ects, features and advantages of the present invention will be more apparent from the following preferred embodiments illustrated with reference to the figures, in which:

Fig. 1 is a schematic diagram of a MIMO system;

Fig. 2 is a schematic diagram of carrier aggregation;

Fig. 3 is a schematic diagram of a multi-cell cellular communication system;

Fig. 4 is a flowchart illustrating the method for PMI transmission over PUSCH according to the present invention; and

Fig. 5 is a schematic block diagram of the UE implementing the above method for PMI transmission over PUSCH according to the present invention.

DESCRIPTION OF EMBODIMENTS

Preferred embodiments of the present invention will be detailed with reference to the drawings. In the following description, details and functions unnecessary to the present invention are omitted so as not to obscure the concept of the invention.

For clear and detailed explanation of the implementation steps of the present invention, some specific examples applicable to the LTE-A cellular communication system are given below. Herein, it is to be noted that the present invention is not limited to the application exemplified in the embodiments. Rather, it is applicable to other communication systems, such as the future 5G cellular system.

Fig. 3 is a schematic diagram of a multi-cell cellular communication system. The cellular system divides a service coverage area into a number of adjacent wireless coverage areas, i. e. , cells. In Fig. 3 , the entire service area is formed by cells 100, 102 and 104, each being illustratively shown as a hexagon. Base Stations (BSs) 200 , 202 and 204 are associated with the cells 100 , 102 and 104, respectively. As known to those skilled in the art, each of the BSs 200-204 comprises at least a transmitter and a receiver. Herein, it is to be noted that a BS, which is generally a serving node in a cell, can be an independent BS having a function of resource scheduling, a transmitting node belonging to an independent BS , a relay node (which is generally configured for further enlarging the coverage of a cell) , or the like . As illustratively shown in Fig. 3 , each of the BSs 200-204 is located in a particular area of the corresponding one of the cells 100- 104 and is equipped with an omni-directional antenna. However, in a cell arrangement for the cellular communication system, each of the BSs 200-204 can also be equipped with a directional antenna for directionally covering a partial area of the corresponding one of the cells 100- 104 , which is commonly referred to as a sector. Thus, the diagram of the multi-cell cellular communication system as shown in Fig. 3 is illustrative only and does not imply that the implementation of the cellular system according to the present invention is limited to the above particular constraints .

As shown in Fig. 3, the BSs 200-204 are connected with each other via X2 interfaces 300, 302 and 304. In a LTE system, a three-layer node network architecture including base station, radio network control unit and core network is simplified into a two-layer node architecture in which the function of the radio network control unit is assigned to the base station and a wired interface named "X2" is defined for coordination and communication between base stations.

In Fig. 3, the BS s 200-204 are also connected with each other via air interfaces, A l interfaces, 3 10 , 312 and 3 14. In a future communication system, it is possible to introduce a concept of relay node . Relay nodes are connected with each other via wireless interfaces and a base station can be considered as a special relay node. Thus, a wireless interface named "A l " can then be used for coordination and communication between base stations.

Additionally, an upper layer entity 220 of the BSs 200-204 is also shown in Fig. 3 , which can be a gateway or another network entity such as mobility management entity. The upper layer entity 220 is connected to the BSs 200-204 via S I interfaces 320, 322 and 324 , respectively. In a LTE system, a wired interface named "S I " is defined for coordination and communication between the upper layer entity and the base station .

A number of User Equipments (UEs) 400-430 are distributed over the cells 100- 104 , as shown in Fig. 3. As known to those skilled in the art, each of the UEs 400-430 comprises a transmitter, a receiver and a mobile terminal control unit. Each of the UEs 400-430 can access the cellular communication system via its serving BS (one of the BSs 200-204) . It should be understood that while only 16 UEs are illustratively shown in Fig. 3 , there may be a large number of UEs in practice. In this sense, the description of the UEs in Fig. 3 is also for illustrative purpose only. Each of the UEs 400-430 can access the cellular communication network via its serving BS . The BS directly providing communication service to a certain UE is referred to as the serving BS of that UE, while other BSs are referred to non-serving BSs of that UE . The non-serving BSs can function as cooperative BSs of the serving BS and provide communication service to the UE along with the serving BS.

For explanation of this embodiment, the UE 4 16 equipped with 2 receiving antennas is considered . The UE 4 16 has BS 202 as its serving BS and has BSs 200 and 204 as its non-serving BSs. It is to be noted that this embodiment focuses on the UE 4 16, which does not imply that the present invention is only applicable to one UE scenario. Rather, the present invention is fully applicable to multi-UE scenario . For example, the inventive method can be applied to the UEs 408 , 4 10 , 430 and the like as shown in Fig. 3.

Moreover, according to 3GPP document TR36.2 13 V9. 1 .0, "Physical layer procedures", for a downlink LTE system with a bandwidth of 20MHz, there are around 96 spectral resource blocks in the frequency domain, in addition to a control signaling area. Each spectral resource block is composed of 12 sub-carriers and 14 OFDM symbols. According to a definition, these spectral resource blocks are sorted in an ascending order in terms of frequency. Every eight consecutive spectral resource blocks are referred to as a sub-band. Thus, there are around 12 sub-bands. It is to be noted that the above definition of sub-band, which is compliant with standardized protocols, is exemplified for explaining the embodiments of the present invention. The application of the present invention is not limited to the above definition and is fully applicable to other definitions. By reading the embodiments of the present invention, those skilled in the art can understand that the solution of the present invention is applicable to a general definition of sub-band .

Fig. 4 is a flowchart illustrating the method for PMI transmission over PUSCH according to the present invention. According to the present invention, a UE receives a downlink transmission approach, a feedback mode and feedback resources configured by a BS (step S400) . Then, the UE decomposes the PMI (W2) into a frequency- selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode, and j ointly codes the frequency-selective and non-frequency-selective portions of W2 (step S4 10) . Finally, the UE feeds the jointly coded PMI (W2) back to the BS over the feedback resources .

Herein, the non-frequency- selective portion of W2 can be a beam selection portion of W2 and the frequency-selective portion of W2 can be a phase combination portion of W2. Alternatively, the non-frequency-selective portion of W2 can be a phase combination portion of W2 and the frequency- selective portion of W2 can be a beam selection portion of W2.

The non-frequency-selective portion of W2 can be directly coded based on wideband. Alternatively, the non-frequency-selective portion of W2 can be differentially coded for feedback based on each sub-band with respect to wideband.

The frequency- selective portion of W2 can be directly coded based on each sub-band . Alternatively, the frequency- selective portion of W2 can be differentially coded for feedback based on each sub-band with respect to wideband.

Fig. 5 is a schematic block diagram of the UE implementing the above method for PMI transmission over PUSCH according to the present invention . As shown in Fig. 5 , the UE 5000 comprises a receiving unit 500 configured for receiving a downlink transmission approach, a feedback mode and feedback resources configured by a BS ; a coding unit 5 10 configured for decomposing the PMI (W2) into a frequency-selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode and jointly coding the frequency-selective and non-frequency-selective portions of W2 ; and a transmitting unit 520 configured for feeding the jointly coded PMI (W2) back to the BS over the feedback resources.

Herein, the non-frequency-selective portion of W2 can be a beam selection portion of W2 and the frequency-selective portion of W2 can be a phase combination portion of W2. Alternatively, the non-frequency-selective portion of W2 can be a phase combination portion of W2 and the frequency-selective portion of W2 can be a beam selection portion of W2.

The UE 5000 can further comprises a down-sampling unit 530 (shown in dashed line in Fig. 5) configured for down-sampling a codebook of W2. In this case, the coding unit 510 can be configured for decomposing W2 into the frequency-selective portion and the non-frequency-selective portion based on the down-sampled codebook.

The coding unit 5 10 can be configured for directly coding the non-frequency- selective portion of W2 based on wideband. Alternatively, the coding unit 5 10 can be configured for differentially coding the non-frequency- selective portion of W2 for feedback based on each sub-band with respect to wideband .

The coding unit 5 10 can be configured for directly coding the frequency-selective portion of W2 based on each sub-band. Alternatively, the coding unit 5 10 can be configured for differentially coding the frequency-selective portion of W2 for feedback based on each sub-band with respect to wideband.

In the following, ten specific application examples will be given for illustrating the method for PMI transmission over PUSCH according to the present invention, such that the present invention will be better understood by those skilled in the art.

By way of examples, the above scheme 1 ) (Codebook Design for W l and W2 , proposed by Alcatel-Lucent et al.) is assumed in the following for illustration of the present invention. It is to be noted that the application of the present invention is not limited to the codebook design of this scheme 1 ) and can be applied to other codebook designs. This scheme 1 ) proposed by Alcatel-Lucent et al. is as follows.

When RI= 1 , the codebook of W l contains 16 codewords:

W^ C, = {W₁ ⁽⁰⁾,W₁ ⁽¹⁾,W₁ ⁽²>,-,W₁^} ;

the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Ye{e,,e₂,e₃,e₄}.

When RI = 2, the codebook of Wl contains 16 codewords:

;

the codebook of W2 contains 16 codewords and has two codeword forms:

where each of the codeword forms has eight codeword values:

(Y Y₂)e{(e₁,e₁),(e₂,e₂),(e₃,e₃),(e₄,e₄),(e₁,e₂),(e₂,e3),(e^

When RI = 3, the codebook of Wl contains four codewords: the codebook of W2 contains 16 codewords and has one codeword form:

where the codeword form has 16 codeword values:

«i «₅])_>(«2>Κ e₆]),(e₃,[e₃ e₇]),(e₄,[e₄

*i e₅]),(e₆,[e₂ e₆]),(e₇,[e₃ e₇]),(e₈,[e₄

e₅]₅e₅),([^e2 ^e6]'^e ₆)'([^e3 e₇]₅e₇),([e₄ ei]'^ei)'([^{e e}2]'^e ₂)'([^e7 ^e3]'^e ₃)₅(h ^

When RI=4, the codebook of Wl contains four codewords: W^C, ={W₁ ⁽⁰⁾,W/¹⁾,W²⁾,W₁ ⁽³⁾}; the codebook of W2 contains eight codewords and has two codeword forms:

where each of the codeword forms has four codeword values:

Y€{ [e, e₅],[e₂ e₆],[e₃ e₇],[e₄ e₈] }.

When RI = 5, the codebook of Wl contains four codewords:

-_X(0) 0 ^" -_X(i) 0 ^" -_X(2) 0 -_X(3) 0 ^"

0 x⁽⁰⁾_ 0 x⁽¹⁾_ 0 x⁽²⁾_ 0 x⁽³⁾_ where

diag{\,e^\e ⁱ\e^J^^li} ^{0\ ^{i)=diag{\^ the codebook of W2 contains one codeword:

When RI = 6, the codebook of Wl contains four codewords:

-_X(0) 0 ^" -_X(i) 0 ^" -_X(2) 0 ^" -_X(3) 0 ^"

0 x⁽⁰⁾_ 0 x⁽¹⁾_ 0 x⁽²⁾_ 0 x⁽³⁾_ where 1 1 1 1

1 j -1 -j

Χ .1 14 ,· „;3ff/4_xv(0)

(⁰⁾ =-χ

1 -1 1 -1

1 -j -1 j

X⁽²⁾→ag{\,e^\e^j2^\e^JM&}X⁰ X⁽³⁾=d^ the codebook of W2 contains one codeword:

When RI = 7, the codebook of Wl contains four codewords:

-_X(0) 0 ^" -_X(i) 0 ^" -_X(2) 0 ^" -_X(3) 0 ^"

0 x⁽⁰⁾_ 0 x⁽¹⁾_ 0 x⁽²⁾_ 0 x⁽³⁾_ where

x⁽²⁾=^{l,e^^/ ^²^ e^^/8}χ^(o χ^<3)=Λ·flg{l,e^'^Γ/ _< the codebook of W2 contains one codeword

When RI = 8, the codebook of Wl contains one codeword:

where

1 1 1 1

1 j -1 -j

X⁽⁰⁾ =-x

1 -1 1 -1

1 -j -1 j the codebook of W2 contains one codeword:

Example 1

According to the codebook design of the above scheme 1), Wl represents a non-frequency-selectivity and thus needs to be fed back based on wideband. That is, only one Wl is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g., where the number of scattering objects distributed in a channel is less than 10, such as in a rural communication environment; or a UE is moving at low or high speed, such as lower than or equal to lOkm/h, or higher than 30km/h), the non-frequency-selective portion of W2 is a beam selection portion, i.e., the codeword value(s) of W2, while the frequency- selective portion of W2 is a phase combination portion, i.e., the codeword form(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 1 of the present invention. For example, when RI=1, the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Y e {e, , e₂ , e₃,e₄ } .

In this case, the codeword values Y e {e, , e₂ ,e₃ ,e₄ | of W2 are non-frequency- selective , while the codeword forms

1

of W2 are

frequency- selective.

The non-frequency-selective portion of W2 is fed back based on wideband. That is, only one non-frequency-selective portion of W2 is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, the frequency-selective portion of W2 is fed back by direct coding based on each sub-band.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion needs to be coded with 2 bits so as to represent the codeword values of W2. An example of such code table is given in Table 1 .

Table 1 - Code Table for Non-Frequency-Selective Portion of W2

On the other hand, the frequency- selective portion of the 12 sub-bands needs to be coded with 2 bits per sub-band, so as to represent the codeword forms of W2. An example of such code table is given in Table 2.

Table 2 - Code Table for Frequency-Selective Portion of W2

In this way, the overall feedback overhead for W2 is 2 + 2 x 12 = 26 bits. When compared with the overall feedback overhead of 4 * 12=48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention.

Example 2

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency-selectivity and thus needs to be fed back based on wideband. That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g. , where the number of scattering objects distributed in a channel is less than 10 , such as in a rural communication environment; or a UE is moving at low or high speed, such as lower than or equal to l Okm/ h, or higher than 30km/ h) , the non-frequency-selective portion of W2 is a beam selection portion, i. e . , the codeword value(s) of W2 , while the frequency- selective portion of W2 is a phase combination portion, i. e . , the codeword form(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 1 of the present invention. For example, when RI = 3 , the codebook of W2 contains 16 codewords and has a single codeword form:

where the codeword form has 16 codeword values :

In this case , the codeword values

frequency-selective while the codeword form of W2 is frequency-selective .

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion needs to be coded with 4 bits so as to represent the codeword values of W2. An example of such code table is given in Table 3.

Table 3 - Code Table for Non-Frequency-Selective Portion of W2

On the other hand, since W2 has only one codeword form, the frequency- selective portion of the 12 sub-bands needs to be coded with 0 bit per sub-band (i. e . , no code is required for identification) , so as to represent the codeword forms of W2.

In this way, the overall feedback overhead for W2 is 4 + 0 x 12=4 bits . When compared with the overall feedback overhead of 4 x 12=48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention .

Example 3 According to the codebook design of the above scheme 1), Wl represents a non-frequency-selectivity and thus needs to be fed back based on wideband. That is, only one Wl is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g., where the number of scattering objects distributed in a channel is less than 10, such as in a rural communication environment; or a UE is moving at low or high speed, such as lower than or equal to lOkm/h, or higher than 30km/h), the non-frequency-selective portion of W2 is a beam selection portion, i.e., the codeword value(s) of W2, while the frequency- selective portion of W2 is a phase combination portion, i.e., the codeword form(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 1 of the present invention. For example, when RI=1, the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Ye{e₁₅e₂,e₃,e₄}.

In this case, the codeword values Ye{e,,e₂,e₃,e₄} of W2 are non-frequency- selective while the codeword forms

frequency- selective .

The non-frequency- selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband . That is , one non-frequency- selective portion of W2 is fed back over the system bandwidth, which needs 2 bits . In addition , the non-frequency- selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit. On the other hand, the frequency- selective portion of W2 is fed back by direct coding based on each sub-band .

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion is composed of a 2 -bit feedback based on wideband (for which an example code table is given in above Table 1 ) and a differentially coded 1 -bit feedback based on sub-band for which an example differential code table is given in Table 4.

Table 4 - Differential Code Table Based on Sub-Band Feedback for Non- Frequency- Selective Portion of W2

On the other hand, the frequency- selective portion of the 1 2 sub-bands needs to be coded with 2 bits per sub-band , so as to represent the codeword forms of W2. An example of such code table is given in the above Table 2.

In this way, the overall feedback overhead for W2 is 2 + 1 x 1 2 + 2 x 1 2 = 38 bits. When compared with the overall feedback overhead of 4 x 1 2=48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention.

Example 4

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency- selectivity and thus needs to be fed back based on wideband. That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e .g. , where the number of scattering obj ects distributed in a channel is less than 10, such as in a rural communication environment; or a UE is moving at low or high speed, such as lower than or equal to l Okm/ h, or higher than 30km/ h) , the non-frequency-selective portion of W2 is a beam selection portion, i.e . , the codeword value(s) of W2 , while the frequency-selective portion of W2 is a phase combination portion, i. e. , the codeword form(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 1 of the present invention. For example, when RI= 1 , the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Y e {e, ,e₂,e₃,e₄} .

In this case, the codeword values Y€ je, ,e₂,e₃,e₄| of W2 are non-frequency- selective while the codeword forms

1

frequency- selective.

The non-frequency-selective portion of W2 is fed back based on wideband. That is, only one non-frequency- selective portion of W2 is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, the frequency-selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one frequency-selective portion of W2 is fed back over the system bandwidth, which needs 2 bits. In addition, the frequency-selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit.

In this example, 12 sub-bands are considered, of which the non-frequency-selective portion needs to be coded with 2 bits so as to represent the codeword values of W2. An example of such code table is given in above Table 1 .

On the other hand, the frequency-selective portion of the 12 sub-bands is composed of a 2 -bit feedback based on wideband (for which an example code table is given in above Table 2) and a differentially coded 1 -bit feedback based on sub-band for which an example differential code table is given in Table 5.

Table 5 - Differential Code Table Based on Sub-Band Feedback

Frequency-Selective Portion of W2

In this way, the overall feedback overhead for W2 is 2 + 2 + 1 x 12= 16 bits. When compared with the overall feedback overhead of 4 > 12 = 48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention .

Example 5

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency-selectivity and thus needs to be fed back based on wideband. That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion . In some communication scenarios (e .g. , where the number of scattering objects distributed in a channel is less than 10, such as in a rural communication environment; or a UE is moving at low or high speed, such as lower than or equal to lOkm/h, or higher than 30km/h), the non-frequency-selective portion of W2 is a beam selection portion, i.e., the codeword value(s) of W2, while the frequency- selective portion of W2 is a phase combination portion, i.e., the codeword form(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 1 of the present invention. For example, when RI=1, the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Ye{e₁₅e₂,e₃,e₄}.

In this case, the codeword values Ye{e,,e₂,e₃,e₄J of W2 are non-frequency- selective while the codeword forms

1

of W2 are

frequency- selective.

The non-frequency-selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one non-frequency-selective portion of W2 is fed back over the system bandwidth, which needs 2 bits. In addition, the non-frequency-selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit. On the other hand, the frequency- selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one frequency- selective portion of W2 is fed back over the system bandwidth, which needs 2 bits. In addition, the frequency- selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion is composed of a 2 -bit feedback based on wideband (for which an example code table is given in above Table 1 ) and a differentially coded 1 -bit feedback based on sub-band (for which an example differential code table is given in above Table 4) .

On the other hand, the frequency-selective portion of the 12 sub-bands is composed of a 2-bit feedback based on wideband (for which an example code table is given in above Table 2) and a differentially coded 1 -bit feedback based on sub-band (for which an example differential code table is given in above Table 5.

In this way, the overall feedback overhead for W2 is 2 + 1 x 12 + 2 + 1 x 12 = 28 bits. When compared with the overall feedback overhead of 4 1 2=48 bits for W2 without optimization , the feedback overhead is effectively reduced according to the present invention.

Example 6

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency- selectivity and thus needs to be fed back based on wideband . That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g. , where the number of scattering obj ects distributed in a channel is larger than 10, such as in an urban communication environment; or a UE is moving at median speed, such as higher than l Okm/ h and lower than or equal to 30km/ h) , the non-frequency-selective portion of W2 is a phase combination portion, i. e . , the codeword form(s) of W2 , while the frequency-selective portion of W2 is a beam selection portion, i. e . , the codeword value(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 2 of the present invention. For example, when RI= 1 , the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword In this case, the codeword forms of W2 are

non-frequency-selective while the codeword values Y e je, ,e₂,e₃,e₄j of W2 are frequency-selective .

The non-frequency-selective portion of W2 is fed back based on wideband. That is, only one non-frequency-selective portion of W2 is fed back over the system bandwidth, without the need for feedback for each sub-band . On the other hand, the frequency- selective portion of W2 is fed back by direct coding based on each sub-band.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion needs to be coded with 2 bits so as to represent the codeword forms of W2. An example of such code table is given in Table 6.

Table 6 - Code Table for Non-Frequency-Selective Portion of W2

On the other hand, the frequency- selective portion of the 12 sub-bands needs to be coded with 2 bits per sub-band, so as to represent the codeword values of W2. An example of such code table is given in Table 7.

Table 7 - Code Table for Frequency-Selective Portion of W2

In this way, the overall feedback overhead for W2 is 2 + 2 x 12 = 26 bits. When compared with the overall feedback overhead of 4 * 12=48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention .

Example 7

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency- selectivity and thus needs to be fed back based on wideband. That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion . In some communication scenarios (e .g. , where the number of scattering objects distributed in a channel is larger than 10, such as in an urban communication environment; or a UE is moving at median speed, such as higher than l Okm/ h and lower than or equal to 30km/ h) , the non-frequency-selective portion of W2 is a phase combination portion, i. e . , the codeword form(s) of W2 , while the frequency- selective portion of W2 is a beam selection portion, i. e . , the codeword value(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 2 of the present invention . For example , when RI = 4 , the codebook of W2 contains eight codewords and has two codeword forms:

where each of the codeword forms has four codeword values:

Y e { [e, e₅],[e₂ e₆],[e₃ e₇],[e₄ e₈] } .

In this case , the codeword forms

W2 are non-frequency- selective

while the codeword values Y e | [e, e₅],[e₂ e₆],[e₃ e₇ ],[e₄ e₈] } of W2 are frequency-selective .

The non-frequency-selective portion of W2 is fed back based on wideband. That is, only one non-frequency-selective portion of W2 is fed back over the system bandwidth, without the need for feedback for each sub-band . On the other hand, the frequency-selective portion of W2 is fed back by direct coding based on each sub-band.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion needs to be coded with 1 bit so as to represent the codeword forms of W2. An example of such code table is given in Table 8. Table 8 - Code Table for Non-Frequency-Selective Portion of W2

On the other hand, the frequency-selective portion of the 12 sub-bands needs to be coded with 2 bits per sub-band, so as to represent the codeword values of W2. An example of such code table is given in Table 9.

Table 9 - Code Table for Frequency-Selective Portion of W2

In this way, the overall feedback overhead for W2 is 1 + 2^12=25 bits. When compared with the overall feedback overhead of 4* 12=48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention.

Example 8

According to the codebook design of the above scheme 1), Wl represents a non-frequency-selectivity and thus needs to be fed back based on wideband. That is, only one Wl is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g., where the number of scattering objects distributed in a channel is larger than 10, such as in an urban communication environment; or a UE is moving at median speed, such as higher than lOkm/h and lower than or equal to 30km/h), the non-frequency-selective portion of W2 is a phase combination portion, i.e., the codeword form(s) of W2, while the frequency-selective portion of W2 is a beam selection portion, i.e., the codeword value(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 2 of the present invention. For example, when RI=1, the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Ye{e₁,e₂,e₃,e₄}.

In this case, the codeword forms

1

of W2 are

non-frequency-selective while the codeword values Y€|e_!,e₂,e₃,e₄| of W2 are frequency-selective.

The non-frequency-selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one non-frequency- selective portion of W2 is fed back over the system bandwidth, which needs 2 bits . In addition, the non-frequency- selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit. On the other hand, the frequency-selective portion of W2 is fed back by direct coding based on each sub-band.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion is composed of a 2-bit feedback based on wideband (for which an example code table is given in above Table 6) and a differentially coded 1 -bit feedback based on sub-band for which an example differential code table is given in Table 10.

Table 1 0 - Differential Code Table Based on Sub-Band Feedback for Non-Frequency-Selective Portion of W2

On the other hand, the frequency-selective portion of the

1 2 sub-bands needs to be coded with 2 bits per sub-band, so as to represent the codeword values of W2. An example of such code table is given in the above Table 7.

In this way, the overall feedback overhead for W2 is 2 + 1 x 12 + 2 x 1 2 = 38 bits. When compared with the overall feedback overhead of 4 >< 12 =48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention.

Example 9

According to the codebook design of the above scheme 1 ) ,

W l represents a non-frequency- selectivity and thus needs to be fed back based on wideband. That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e .g. , where the number of scattering objects distributed in a channel is larger than 10 , such as in an urban communication environment; or a UE is moving at median speed, such as higher than l Okm/ h and lower than or equal to 30km/ h) , the non-frequency-selective portion of W2 is a phase combination portion, i. e . , the codeword form(s) of W2 , while the frequency-selective portion of W2 is a beam selection portion, i. e. , the codeword value(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 2 of the present invention. For example, when RI= 1 , the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Y e {e, ,e₂,e₃,e₄} .

In this case , the codeword forms

non-fre uency-selective while the codeword values

of W2 are frequency-selective .

The non-frequency-selective portion of W2 is fed back based on wideband. That is, only one non-frequency-selective portion of W2 is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, the frequency- selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one frequency-selective portion of W2 is fed back over the system bandwidth, which needs 2 bits. In addition, the frequency-selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion needs to be coded with 2 bits so as to represent the codeword forms of W2. An example of such code table is given in above Table 6.

On the other hand, the non-frequency- selective portion of the 12 sub-bands is composed of a 2 -bit feedback based on wideband (for which an example code table is given in above Table 7) and a differentially coded 1 -bit feedback based on sub-band for which an example differential code table is given in Table 1 1 .

Table 1 1 - Differential Code Table Based on Sub-Band Feedback for Frequency-Selective Portion of W2

In this way, the overall feedback overhead for W2 is 2 + 2

+ 1 x 12= 16 bits. When compared with the overall feedback overhead of 4 < 12 = 48 bits for W2 without optimization, the feedback overhead is effectively reduced according to the present invention.

Example 10

According to the codebook design of the above scheme 1 ) , W l represents a non-frequency-selectivity and thus needs to be fed back based on wideband . That is, only one W l is fed back over the system bandwidth, without the need for feedback for each sub-band. On the other hand, W2 can be decomposed into a frequency-selective portion and a non-frequency-selective portion. In some communication scenarios (e.g. , where the number of scattering objects distributed in a channel is larger than 10, such as in an urban communication environment; or a UE is moving at median speed, such as higher than l Okm/ h and lower than or equal to 30km/ h) , the non-frequency-selective portion of W2 is a phase combination portion, i.e . , the codeword form(s) of W2 , while the frequency-selective portion of W2 is a beam selection portion, i.e. , the codeword value(s) of W2. Such an approach for decomposition of W2 is referred to as Solution 2 of the present invention . For example , when RI= 1 , the codebook of W2 contains 16 codewords and has four codeword forms:

where each of the codeword forms has four codeword values:

Y e {e, ,e₂,e₃,e₄} .

In this case , the codeword forms

1

of W2 are

non-frequency-selective while the codeword values Y e {e_{1 5}e₂,e₃,e₄} of W2 are frequency-selective .

The non-frequency-selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one non-frequency- selective portion of W2 is fed back over the system bandwidth, which needs 2 bits . In addition, the non-frequency-selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit. On the other hand, the frequency- selective portion of W2 is differentially coded for feedback based on each sub-band with respect to wideband. That is, one frequency- selective portion of W2 is fed back over the system bandwidth, which needs 2 bits . In addition, the frequency- selective portion of W2 is differentially coded and then fed back for each sub-band, which needs 1 bit.

In this example, 12 sub-bands are considered, of which the non-frequency- selective portion is composed of a 2 -bit feedback based on wideband (for which an example code table is given in above Table 6) and a differentially coded 1 -bit feedback based on sub-band (for which an example differential code table is given in above Table 1 0) .

On the other hand, the frequency-selective portion of the 12 sub-bands is composed of a 2-bit feedback based on wideband (for which an example code table is given in above Table 7) and a differentially coded 1 -bit feedback based on sub-band (for which an example differential code table is given in above Table 1 1 .

In this way, the overall feedback overhead for W2 is 2 + 1 x 12 + 2 + 1 x 12 = 28 bits. When compared with the overall feedback overhead of 4 12=48 bits for W2 without optimization , the feedback overhead is effectively reduced according to the present invention.

It is to be noted that the present invention does not exclude the possibility in combination with the codebook down-sampling approach according to the above scheme 3) . That is, the method for sub-band PMI transmission over PUSCH according to the present invention can be applied to sub-codebooks of W l and W2 as obtained by using the codebook down-sampling approach according to the above scheme 3) .

It is also to be noted that the above embodiments are applicable to scenarios such as PUSCH feedback mode 2 -2 and mode 3-2 in which sub-band PMI feedback is necessary. While the description of the embodiments is not given with respect to any specific mode, it should be appreciated by those skilled in the art that, from the teaching of the above embodiments, the present invention can be readily implemented in other modes in which sub-band PMI feedback is necessary.

Further, it is to be noted that the present invention does not exclude the possibility in which the Solution 1 and Solution 2 as described above co-exist. For example, in an actual system, the BS and/ or the UE can adaptively employ one of the above Solution 1 and Solution 2 (i.e. , change the decomposition manner for W2 and swap the frequency-selective and non-frequency-selective portions of W2) according to variation of communication scenarios (e.g. , when the channel RI value and thus the applicable solution changes according to a correspondence table between channel RI values and applicable solutions; when the user changes from median-speed movement to non-median-speed movement; or when the number of scattering objects in a channel varies due to channel environment variation) , configuration of the UE by the BS , or autonomous selection made by the UE (which needs to be notified to the BS by means of feedback) .

A number of examples have been illustrated in the above description. While the inventor has tried to list the examples in association with each other, it does not imply that it is required for the listed examples to have such correspondence as described. A number of solutions can be achieved by selecting examples having no correspondence as long as the conditions underlying the selected examples do not conflict with each other. Such solutions are encompassed by the scope of the present invention .

The present invention has been described above with reference to the preferred embodiments thereof. It should be understood that various modifications, alternations and additions can be made by those skilled in the art without departing from the spirits and scope of the present invention. Therefore, the scope of the present invention is not limited to the above particular embodiments but only defined by the claims as attached.

Claims

1 . A method for Pre-coding Matrix Index (PMI) feedback, comprising the following steps of:

-receiving a downlink transmission approach, a feedback mode and feedback resources configured by a Base Station (BS) ;

-decomposing the PMI into a frequency-selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode, and jointly coding the frequency-selective and non-frequency-selective portions of the PMI ; and

-feeding the jointly coded PMI back to the BS over the feedback resources.

2. The method for PMI feedback according to claim 1 , wherein

the non-frequency-selective portion of the PMI is one of a beam selection portion and a phase combination portion of the PMI ; and

the frequency- selective portion of the PMI is the other one of the beam selection portion and the phase combination portion of the PMI .

3. The method for PMI feedback according to claim 1 or 2 , further comprising:

-down-sampling a codebook of the PMI prior to the decomposing step;

wherein the decomposing step comprises: decomposing the PMI into the frequency-selective portion and the non-frequency-selective portion based on the down-sampled codebook.

4. The method for PMI feedback according to any one of claims 1 -3 , wherein

the non-frequency-selective portion of the PMI is directly coded based on wideband .

5. The method for PMI feedback according to any one of claims 1 -3 , wherein

the non-frequency- selective portion of the PMI is differentially coded for feedback based on each sub-band with respect to wideband.

6. The method for PMI feedback according to any one of claims 1 -3 , wherein

the frequency-selective portion of the PMI is directly coded based on each sub-band.

7. The method for PMI feedback according to any one of claims 1 -3 , wherein

the frequency- selective portion of the PMI is differentially coded for feedback based on each sub-band with respect to wideband .

8. The method for PMI feedback according to any one of claims 1 -7, wherein

the PMI is PMI #2 , i. e . , W2 , the W2 represents sub-band / short-term channel characteristics.

9. A User Equipment (UE) , comprising:

-a receiving unit configured for receiving a downlink transmission approach, a feedback mode and feedback resources configured by a Base Station (BS) ;

-a coding unit configured for decomposing the PMI into a frequency-selective portion and a non-frequency-selective portion based on the downlink transmission approach and the feedback mode and jointly coding the frequency- selective and non-frequency-selective portions of the PMI ; and

-a transmitting unit configured for feeding the jointly coded PMI back to the BS over the feedback resources.

10. The UE according to claim 9 , wherein

1 1 . The UE according to claim 9 or 10 , further comprising: -a down-sampling unit configured for down-sampling a codebook of the PMI ;

wherein the coding unit is configured for decomposing the PMI into the frequency- selective portion and the non-frequency-selective portion based on the down- sampled codebook.

12. The UE according to any one of claims 9 - 1 1 , wherein the coding unit is configured for directly coding the non-frequency-selective portion of the PMI based on wideband.

13. The UE according to any one of claims 9- 1 1 , wherein the coding unit is configured for differentially coding the non-frequency-selective portion of the PMI for feedback based on each sub-band with respect to wideband.

14. The UE according to any one of claims 9- 1 1 , wherein the coding unit is configured for directly coding the frequency-selective portion of the PMI based on each sub-band .

15. The UE according to any one of claims 9-11, wherein the coding unit is configured for differentially coding the frequency-selective portion of the PMI for feedback based on each sub-band with respect to wideband.

16. The UE according to any one of claims 9-15, wherein the PMI is PMI #2, i.e., W2, the W2 represents sub-band / short-term channel characteristics.