LONG-TERM-CSI-AIDED MU-MIMO SCHEDULING METHOD, BASE STATION AND USER EQUIPMENT BACKGROUND OF THE INVENTION

FIELD OF INVENTION

The present invention relates to the domain of wireless Communications, more particularly, to a long-term-Channel State Information (CSI)-aided Multi-User Multi-Input Multi-Output (MU-MIMO) scheduling method, a base station and a user equipment, and aims to reduce the feedback overhead for joint MU-MIMO and multi-user scheduling.

DESCRIPTION OF PRIOR ART

MU-MIMO has been shown to provide a great capacity advantage over single-user MIMO (SU-MIMO) in theory. It points out a promising way to solve the throughput bottleneck in wireless communication systems. The potential capacity benefit of MU-MIMO can be increased by orders through a joint operation with multiuse scheduling to exploit the multiuse diversity gain as well as the spatial diversity gain. In such a joint operation, the base station (BS) needs to know the full channel state Information (CSI) for all users in the cell, not only for active users selected for service. This will lead to a heavy feedback overhead in frequency division duplex (FDD). When the total user number is large, the feedback overhead becomes prohibitive, which greatly restricts the application of MU-MIMO in practical systems. The aim of invention is to provide a long-term-CSI-aided MU-MIMO scheduling strategy to solve this problem.

Most existing solutions to the above problem adopt the concept of codebook to reduce the feedback overhead. The basic idea is to use a pre-determined codebook containing a set of vectors with unit norm, which is known to both the BS and all terminals, and to provide the BS with a vector index, as well as a real-valued channel quality indicator (CQI), from each terminal. The codebook technique effectively restricts the feedback overhead for each single user, but the overall feedback overhead still grows linearity with the total user number K in a cell, which can be very large when K Is large.

SUMMARY OF THE INVENTION

The basic principle of this solution is to use long-term CSI of all users to reduce the feedback overhead required for a joint MU-MIMO and scheduling scheme.

A joint MU-MIMO and scheduling operation can be divided into the scheduling phase and multi-user preceding phase. In the scheduling phase, a set of active users is selected out of all users according to sum-capacity maximization or other criterion, which will be served over the same time and frequency resource. In the multi-user preceding phase, the BS generates preceding vectors for active users and then performs multi-user preceding. It is only in the scheduling phase that CSI is required for all users. In the multi-user preceding phase, on the other hand, only CSI for active users will be needed. Since the total user number is always much larger than active user number, the scheduling phase is usually responsible for unacceptable feedback overhead.

Motivated by the above discussion, we propose to use long-term statistical CSI, in particular the mean channel matrixes, instead of instantaneous CSI, in the scheduling phase. Since the long-term statistical CSI changes much slower, it can be updated in a much lower speed, which can effectively reduce the overhead of the scheduling phase. In the multi-user preceding phase, instantaneous CSI is still used since the performance of multi-user preceding is sensitive to the inaccuracy of CSI. Keeping in mind that the scheduling phase generally dominates the feedback overhead, the proposed idea is effective in feedback reduction.

According to a first aspect of the present invention, there is provided a joint Multi-User Multi-Input Multi-Output (MU-MIMO) recoding and scheduling method, which comprises steps of: a multi-user scheduling step of performing multi-user scheduling by using mean channel matrixes of respective user equipments; and a multi-user preceding step of performing multi-user preceding by using instantaneous channel matrixes of respective user equipments selected in the multi-user scheduling.

Preferably, tile multi-user scheduling step includes sub-steps of: informing all user equipments to feed back their mean channel matrixes; receiving the mean channel matrixes calculated and fed back< by all the user equipments; and performing the multi-user scheduling by using the fed back mean channel matrixes so as to generate an optimized scheduling result.

More preferably, the optimized scheduling result indicates at least the number of selected/active user equipments, the indexes of the selected/active user equipments, and the numbers of the selected/active user equipments' data streams.

More preferably, the optimized scheduling result is generated so as to maximize the throughput of a wireless communication system. Moreover, the optimized scheduling result is generated so as to maximize the capacity of the wireless communication system.

Preferably the multiuser precoding step includes sub-steps of: informing all selected/active user equipments to feed back their instantaneous channel matrixes; receiving the instantaneous channel matrixes fed back by all the selected/active user equipments; calculating precoding matrixes for the respective selected/active user equipments based on their instantaneous channel matrixes; multiplying data vectors of the respective selected/active user equipments with their respective precoding matrixes to generated the respective precoded messages for each selected/active user equipments; and sending the respective precoded messages to each selected/active user equipments.

Preferably, the instantaneous channel matrixes are measured every first time interval T|, and the mean channel matrixes are calculated every second time interval TL by using the measured instantaneous channel matrixes during the immediate second time interval TL, the first time interval T| is smaller than the second time interval TL. More preferably, the second time interval TL is an integral multiple of the first time intervals Ti.

According to a second aspect of the present invention, there is provided a base station, which comprises: a sending unit for sending a message to inform user equipments to feed back their mean channel matrixes and/or instantaneous channel matrixes; a receiving unit for receiving the mean channel matrixes and the instantaneous channel matrixes fed back from user equipments; a multiuser scheduling unit for performing multiuser scheduling by using the mean channel matrixes of the respective user equipments; and a multiuser precoding unit for performing multiuser precoding by using the instantaneous channel matrixes of respective user equipments selected by the multiuser scheduling unit.

Preferably, the multiuser scheduling unit informs all user equipments to feed back their mean channel matrixes by means of the sending unit, receives the mean channel matrixes calculated and fed back by all the user equipments by means of the receiving unit, and performs the multiuser scheduling by using the fed back mean channel matrixes so as to generate an optimized scheduling result.

More preferably, the optimized scheduling result indicates at least the number of selected/active user equipments, the indexes of the selected/active user equipments, and the numbers of the selected/active user equipments' data streams.

More preferably, the multiuser scheduling unit generates the optimized scheduling result so as to maximize the throughput of the wireless communication system in which the base station is located. Moreover, the multiuser scheduling unit generates the optimized scheduling result so as to maximize the capacity of the wireless communication system.

Preferably, the sending unit is also used for sending precoded data to the user equipments, and the multiuser precoding unit informs all active user equipments selected by the multiuser scheduling unit to feed back their instantaneous channel matrixes by means of the sending unit, receives the instantaneous channel matrixes fed back by all the active user equipments by means of the receiving unit, calculates precoding matrixes for the respective active user equipments based on their instantaneous channel matrixes, multiplies data vectors of the respective active user equipments with their respective precoding matrixes to generated the respective precoded messages for each active user equipments; and sends the respective precoded messages to each active user equipments by means of the sending unit.
Preferably, the multiuser precoding unit calculates the multiuser precoding matrixes for active user equipments every first time interval T|, and the multiuser scheduling unit performs the multiuser scheduling every second time interval TL, wherein the first time interval T| is smaller than the second time interval TL. More preferably, the second time interval TL is an integral multiple of the first time intervals Ti.
According to a third aspect of the present invention, there is provided a user equipment, which comprises: a receiving unit for receiving a message requesting to feed back instantaneous channel matrix and/or mean channel matrix from a base station; an instantaneous channel matrix measuring unit for measuring its instantaneous channel matrix every first time interval Ti; a mean channel matrix calculating unit for calculafing its mean channel matrix every second time interval TL by using the instantaneous channel matrixes measured by the instantaneous channel matrix measuring unit during the immediate second time interval TL; and a sending unit for sending the measured instantaneous channel matrix and/or the calculated mean channel matrix to the base station, wherein the first time interval T| is smaller than the second time interval TL.

Preferably, the second time interval TL is an integral multiple of the first time intervals T|.
This invention has the following benefits:
1. It leads to a considerable reduction in the feedback overhead of a joint MU-MIMO and multiuser scheduling system.
2. It greatly simplifies the scheduling complexity, since the scheduling operation is performed only when the statistical CSI is updated.
3. It simplifies the terminal operation related to channel estimation and CSI feedback.
It does not bring any additional complexity at the BS.
BRIEF DESCRIPTION OF THE DRAWINGS
The above and other objects, features and advantages of the present invention will be clearer from the following detailed description about the non-limited embodiments of the present invention taken in conjunction with the accompanied drawings, in which:
Fig. 1 shows the flowchart of the scheduling phase according to the present invention.
Rg. 2 shows the flowchart of the multiuser precoding phase according to the present invention.
Fig. 3 shows the comparison results between the inventive long-term-CSI-aided scheme with the existing instantaneous-CSI-aided scheme based on the MET algorithm and the greedy scheduling strategy.
Fig. 4 shows a schematic block diagram of a base station 400 to implement the long-term-CSI-aided MU-MIMO scheduling method according to the present invention.
Fig. 5 shows a schematic block diagram of a user equipment 500 to implement the

long-term-CSI-aided MU-MIMO scheduling method according to the present invention.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
Hereunder, the present invention will be described in accordance with the drawings. In the following description, some particular embodiments are used for the purpose of description only, which shall not be understood as any limitation to the present Invention but the examples thereof. VVhlIe it may blur the understanding of the present invention, the conventional structure or construction will be omitted.
The basic idea of this invention is to use mean channel matrixes for user scheduling and instantaneous channel matrixes for precoding matrix generation. In the following, we will use the multiuser eigenmode transmission (MET) algorithm and full search strategy as an example to explain this idea. Please note that the proposed idea also applies directly to other algorithms and scheduling strategies.
MET alQorithm
Denote by NT the number of transmit antennas at the BS, NR the number of receive antennas per user, N the number of active users, H„ the channel matrix for user-n.
Define s„, T„=G„A„ and R„=C„B„as the number of data streams, the
transmit (precoding) matrix, and the receive matrix for user-n. The MET algorithm can be divided into three steps:

Thus, H„ is changed from a NRXNT matrix into a s„x/S/r matrix. " is used together with C„ (will be described later in Step 3) to construct a diagonal equivalent channel matrix for user-n, so that lts different data streams will not

interfere with each other.
Step 2: Calculate G„ for n =1~N.

Step 3 is an SVD operation of the equivalent channel H„, to help constructing a diagonal equivalent channel matrix for user-n.
Thereby, T„=G„A„ and R„=C„B„ (the transmit (precoding) matrix and the receive matrix) for user-n can be obtained.
The above generates the precoding matrix Tn. A complete precoding operation also involves the multiplication of data vector with the precoding matrix, as shown later.


Fuli search schedulinq
Denote by K the number of users in a cell vying for service. The function of the scheduler is to decide which and how many users out of K users will be active, as well as the number of data streams assigned to each active user. Each possibility of the scheduling result, denoted by 2C. can be expressed by three elements: the
active user number N(k) the indexes of active users k(k)n=1-N(k)with
1≤k(k)≤k, and the numbers of their data streams {s^,n=l~N(k). With the full
search scheduling, the scheduler searches over the set of all possibilities of x,
denoted by R, and selects an optimized 't as

where C(k)is the throughput achievable by 3c; y is the value of yin (A) in Step 3 corresponding to K; y(k)/σ2 is the signal-to-noise-plus-interference ratio (SINR) for the s'" data stream of user-n in k > and ƒ (.) is a function representing the relationship between SINR and the throughput. If the theoretical capacity is

considered, ƒ (.) is the Shannon capacity formula, otherwise, it depends on the modulation coding schemes (MCSs) used. In the existing scheduling strategies, instantaneous channel matrixes are used for scheduling, so y'(k) is calculated following Steps 1 - 3 by setting

The object of formula (1) is to find an optimized 5c so that the throughput of the system is maximized. In this case, if ƒ (.) is the Shannon capacity formula, then it means to find an optimized :^ so that the capacity of the system is maximized.
Based on the above discussion, the proposed long-term-CSI-aided joint MU-MIMO and scheduling scheme can be described in two phases as follows:
Scheduling phase: In this phase, the scheduler performs multiuser scheduling according to (1), using the mean channel matrixes instead of the instantaneous channel matrixes. In particular, for each %,, the value of y^^ in (1) is estimated
following Steps 1 - 3 listed above by setting

where H^ denotes the mean channel matrix of user-/f.
Each user calculates its mean channel matrix {H^} by averaging over several
observations of its physical channel during a period TA. By using the mean channel matrixes instead, we can greatly reduce the feedback overhead in this phase and

also the scheduling complexity, since {Hk} can be updated in a much lower speed than {Hk}, and the scheduling operation is performed only when {Hk}is updated.
In detail, Fig. 1 shows the flowchart of the scheduling phase according to the present invention. At step 101, BS informs all user equipments to feed back their mean channel matrixes. At step 102, all user equipments calculate and feed back their mean channel matrixes. At step 103, BS performs the scheduling by using the fed back mean channel matrixes so as to generate the optimized scheduling result
%, which determines the active user equipment number N(k)), the indexes of active
user equipments , and the numbers of the active
user equipments' data streams . Then, at step 104, BS goes into
the multiuser precoding phase which will be described in detail later.
In addition, the whole scheduling procedure (steps 101 - 103) is repeated each time interval TL.
Multiuser precoding phase: In this phase, the BS generates precoding matrixes for users selected in , following Steps 1 - 3. Here the instantaneous channel
matrixes {H^(^,,«=1~JV^'^'} are still used to insure good performance of multiuser precoding.
In detail, Fig. 2 shows the flowchart of the multiuser precoding phase according to the present invention. At step 201, BS informs the selected user equipments (active user equipments) to feed back their instantaneous channel matrixes. At step 202, the selected user equipments feed back their instantaneous channel matrixes. At step 203, BS calculates precoding matrixes for the selected user equipments based on their instantaneous channel matrixes.
This multiuser precoding operation (steps 201 - 203) is repeated each time interval

T|. It shall be noted that Ti is much smaller than TL defined above, since the instantaneous channel matrix changes much faster than the mean channel matrix, so it nas to be updated more frequently. It shall be also noted that during the whole operation, the users should keep measuring their physical channel matrixes each time interval Ti, so that they can generate the mean channel matrixes for the time duration TL. Preferably, for simplifying the calculation of the mean channel matrixes, the time interval TL is an integral multiple of the time intervals T|.
At step 204, BS multiplies the data vectors of selected user equipments with their precoding matrixes so as to generate the respective precoded messages for each selected user equipment. Finally, at step 205, BS sends the respective precoded messages to each selected user equipment.
In Fig. 3, we compare the proposed long-term-CSI-aided scheme with the existing instantaneous-CSI-aided scheme based on the MET algorithm and the greedy scheduling strategy. The performance of the MET algorithm with random scheduling strategy is also included for reference, in which NT active users are selected randomly from K users for each transmission. In this simulation, we employ 4 transmit antennas at the base station, two receive antennas per user, and a total user number of 20. The Jacks Model is used to generate frequency-flat time-selective channels for all users with parameters: carrier frequency = 2.5GHz, velocity = 3km/h and sampling frequency = 11.2MHz. The mean channel matrix of each user is generated by averaging its instantaneous channel matrixes over T consecutive symbols. The mean channel matrixes are updated once in a T-symbol period, so the feedback overhead of the proposed scheme is reduced to about 1/T that of the instantaneous-CSI-aided scheme. In our simulation, T is set to 50, 100, 150 and 250. From this figure, we can see that when 7=50, the proposed scheme can achieve almost the same performance as the instantaneous-CSI-aided scheme with a feedback overhead of about 1/50. Even with a quite large 7=250, the proposed scheme can still achieve a considerable gain over the random scheduling scheme.

Fig. 4 shows a schematic block diagram of a base station 400 to implement the long-term-CSI-aided MU-MIMO scheduling method according to the present invention.
As shown in Fig. 4, the base station 400 includes: a sending unit 410 for sending a message to inform user equipments to feed bacl< their mean channel matrixes and/or instantaneous channel matrixes; a receiving unit 420 for receiving the mean channel matrixes and the instantaneous channel matrixes fed back from user equipments; a multiuser scheduling unit 430 for performing multiuser scheduling by using the mean channel matrixes of the respective user equipments; and a multiuser precoding unit 440 for performing multiuser precoding by using the instantaneous channel matrixes of respective user equipments selected by the multiuser scheduling unit 430.
In particularly, the multiuser scheduling unit 430 informs all user equipments to feed back their mean channel matrixes by means of the sending unit 410, receives the mean channel matrixes calculated and fed back by all the user equipments by means of the receiving unit 420, and performs the multiuser scheduling by using the fed back mean channel matrixes so as to generate an optimized scheduling
result it which determines the active user number the indexes of active
users , and the numbers of the active users' data
streams
The sending unit 410 is also used for sending precoded data to the user equipments. The multiuser precoding unit 440 informs all active user equipments selected by the multiuser scheduling unit 430 to feed back their instantaneous channel matrixes by means of the sending unit 410, receives the instantaneous channel matrixes fed back by all the active user equipments by means of the receiving unit 420, calculates precoding matrixes for the respective active user equipments based on their instantaneous channel matrixes, multiplies data vectors of the respective active user equipments with their respective precoding matrixes to

generate the respective precoded messages for each active user equipments; and sends the respective precoded messages to each active user equipments by means of the sending unit 410.
The multiuser precoding unit 440 calculates the multiuser precoding matrixes for active user equipments every time interval T|, and the multiuser scheduling unit 430 performs the multiuser scheduling every time interval TL. The time interval T1 is much smaller than the time interval TL. Preferably, for simplifying the calculation of the mean channel matrixes, the time interval TL is an integral multiple of the time intervals T|.
Fig. 5 shows a schematic block diagram of a user equipment 500 to implement the long-term-CSI-aided MU-MIMO scheduling method according to the present invention.
As shown in Fig. 5, the user equipment 500 includes a receiving unit 510, an instantaneous channel matrix measuring unit 520, a mean channel matrix calculating unit 530 and a sending unit 540. In particularly, the receiving unit 510 receives a message requesting to feed back instantaneous channel matrix and/or mean channel matrix from the base station 400. The instantaneous channel matrix measuring unit 520 measures the user equipment 500's instantaneous channel matrix every time interval T|. The mean channel matrix calculating unit 530 calculates the user equipment 500's mean channel matrix every time interval TL by using the instantaneous channel matrixes measured by the instantaneous channel matrix measuring unit 520 during the immediate time interval TL. The sending unit sends the measured instantaneous channel matrix from the instantaneous channel matrix measuring unit 520 and/or the calculated mean channel matrix from the mean channel matrix calculating unit 530 to the base station 400. The time interval T| is much smaller than the time interval TL. Preferably, for simplifying the calculation of the mean channel matrixes, the time interval TL is an integral multiple of the time intervals T|.

The above embodiments are provided for the purpose of example only, and are not intended to limit the present invention. It is to be understood by those skilled in the art that there may be various modifications or replacements to the embodiments without departing from the scope and the spirit of the present invention, and they shall fall into the scope defined by the appended claims.

What is claimed is:

1. A multiuser multi-input multi-output scheduling method, comprising steps of:
a multiuser scheduling step of performing multiuser scheduling by using mean channel matrixes of respective user equipments; and
a multiuser precoding step of performing multiuser precoding by using instantaneous channel matrixes of respective user equipments selected in the multiuser scheduling.

2. The multiuser multi-input multi-output scheduling method according to claim 1,
wherein the multiuser scheduling step includes sub-steps of:
informing all user equipments to feed back their mean channel matrixes;
receiving the mean channel matrixes calculated and fed back by all the user equipments; and
performing the multiuser scheduling by using the fed back mean channel matrixes so as to generate an optimized scheduling result.

3. The multiuser multi-input multi-output scheduling method according to claim 2, wherein the optimized scheduling result indicates at least the number of selected/active user equipments, the indexes of the selected/active user equipments, and the numbers of the selected/active user equipments' data streams.

4. The multiuser multi-input multi-output scheduling method according to claim 2, wherein the optimized scheduling result is generated so as to maximize the throughput of a wireless communication system.

5. The multiuser multi-input multi-output scheduling method according to claim 4, wherein the optimized scheduling result is generated so as to maximize the capacity of the wireless communication system.

6. The multiuser multi-input multi-output scheduling method according to claim 1, wherein the multiuser precoding step includes sub-steps of: informing all selected/active user equipments to feed back their instantaneous channel matrixes; receiving the instantaneous channel matrixes fed back by all the selected/active user equipments; calculating precoding matrixes for the respective selected/active user equipments based on their instantaneous channel matrixes; multiplying data vectors of the respective selected/active user equipments with their respective precoding matrixes to generate the respective precoded messages for each selected/active user equipments; and sending the respective precoded messages to each selected/active user equipments.

7. The multiuser multi-input multi-output scheduling method according to claim 1, wherein the instantaneous channel matrixes are measured every first time interval Ti, and the mean channel matrixes are calculated every second time interval TL by using the measured instantaneous channel matrixes during the immediate second time interval TL, the first time interval Ti is smaller than the second time interval TL.

8. A base station, comprising:
a sending unit for sending a message to inform user equipments to feed back their mean channel matrixes and/or instantaneous channel matrixes; a receiving unit for receiving the mean channel matrixes and the instantaneous channel matrixes fed back from user equipments; a multiuser scheduling unit for performing multiuser scheduling by using the mean channel matrixes of the respective user equipments; and a multiuser precoding unit for performing multiuser precoding by using the instantaneous channel matrixes of respective user equipments selected by the multiuser scheduling unit.

9. The base station according to claim 8, wherein the multiuser scheduling unit informs all user equipments to feed back their mean channel matrixes by means of the sending unit, receives the mean channel matrixes calculated and fed back by all the user equipments by means of the receiving unit, and performs the multiuser scheduling by using the fed back mean channel matrixes so as to generate an optimized scheduling result.

10. The base station according to claim 9, wherein the optimized scheduling result indicates at least the number of selected/active user equipments, the indexes of the selected/active user equipments, and the numbers of the selected/active user equipments' data streams.

11. The base station according to claim 9, wherein the multiuser scheduling unit generates the optimized scheduling result so as to maximize the throughput of the wireless communication system in which the base station is located.

12. The base station according to claim 11, wherein the multiuser scheduling unit generates the optimized scheduling result so as to maximize the capacity of the wireless communication system.

13. The base station according to claim 8, wherein the sending unit is also used for sending precoded data to the user equipments, and the multiuser precoding unit informs all active user equipments selected by the multiuser scheduling unit to feed back their instantaneous channel matrixes by means of the sending unit, receives the instantaneous channel matrixes fed back by all the active user equipments by means of the receiving unit, calculates precoding matrixes for the respective active user equipments based on their instantaneous channel matrixes, multiplies data vectors of the respective active user equipments with their respective precoding matrixes to generate the respective precoded messages for each active user equipment; and sends the respective precoded messages to each active user equipment by means of the sending unit.

14. The base station according to claim 8, wherein the multiuser precoding unit calculates the multiuser precoding matrixes for active user equipments every first time interval Ti, and the multiuser scheduling unit performs the multiuser scheduling every second time interval TL, wherein the first time interval TI is smaller than the second time interval TL.

15. A user equipment, comprising:
a receiving unit for receiving a message requesting to feed back instantaneous channel matrix and/or mean channel matrix from a base station;
an instantaneous channel matrix measuring unit for measuring its instantaneous channel matrix every first time interval Ti;
a mean channel matrix calculating unit for calculating its mean channel matrix every second time interval TL by using the instantaneous channel matrixes measured by the instantaneous channel matrix measuring unit during the immediate second time interval TL; and
a sending unit for sending the measured instantaneous channel matrix and/or the calculated mean channel matrix to the base station,
wherein the first time interval TI is smaller than the second time interval TL.