CN109004963B - Wireless communication user optimal scheduling method based on opportunistic interference alignment - Google Patents

Abstract

The invention discloses a wireless communication user optimal scheduling method based on opportunistic interference alignment. The communication users are divided into two types of active users and passive users, and the algorithm has the main idea that: for a TDD communication system with G cells and K users in each cell, S users are selected from each cell to communicate in the same time slot and the same frequency, firstly, W users with the best current channel state are obtained through singular value decomposition as active users, and then, a useful signal space matrix U of each base station is designed on the principle of completely eliminating active user interference from other cells _g Then, the rest is selected based on the principle of minimizing the interference to other cellsS‑WA 'passive user', and finally designing a decoding vector f of each user_g,s. The method of the invention not only greatly improves the time fairness, but also obviously improves the system capacity.

Description

Wireless communication user optimal scheduling method based on opportunistic interference alignment

Technical Field

The invention belongs to the technical field of communication, and particularly relates to a wireless communication user optimal scheduling method.

Background

With the rapid development of mobile intelligent terminals and mobile internet, mobile services have been increased explosively, people have made higher requirements on data access rate and communication quality, and future mobile communication systems need to provide high-speed access rate for users and also meet special personalized requirements (such as rapid access) of some users. In the process of rapid development of mobile communication systems, interference is always a main factor influencing the communication quality of users and the improvement of system capacity, and how to avoid interference between users is always a research hotspot of the academic world. Interference alignment techniques proposed in recent years are widely studied for effectively eliminating interference, significantly improving the degree of Freedom (DOF) and increasing the system capacity by multiples, and design precoding at a transmitting end enables interference signals received at a receiving end to be compressed into a lower dimensional space, thereby using more dimensions for receiving useful signals.

In the prior art, the research related to the present invention mainly includes:

[1]Cadambe V R, Jafar S A. Interference alignment and degrees offreedom of the-user interference channel[J]. IEEE Transactions on InformationTheory, 2008, 54(8): 3425-3441.

[2]Cadambe V R, Jafar S A. Interference alignment and the degrees offreedom of wireless X networks[J]. IEEE Transactions on Information Theory ,2009, 55(9):3893-3908.

[3]Nan Zhao, F. Richard Yu, MingluJin, Qiao Yan and Victor C.M.Leung, Interference Alignment and its Applications: A Survey, Research Issuesand Challenges[J], IEEE Communications Surveys&Tutorials, 2016, 18(3):1779-1803.

[4]YU H. A review on interference alignment in multiuser interferencechannels[J]. Wireless Personal Communications, 2015, 83(3): 1751-1764.

[5]LEE J H, CHOI W. Opportunistic interference aligned user selectionin multiuser MIMO interference channels[C]//IEEE Global TelecommunicationsConference (GLOBECOM), Miami, USA, 2010: 1-5.

[6]LEE J H, CHOI W. Interference alignment by opportunistic userselection in 3-user MIMO interference channels[C]//IEEE InternationalConference on Communications (ICC). Kyoto, Japan, 2011: 1-5.

[7]Lee J H, Wan C. On the Achievable DoF and User Scaling Law ofOpportunistic Interference Alignment in 3-Transmitter MIMO InterferenceChannels[J]. IEEE Transactions on Wireless Communications, 2013, 12(6):2743-2753.

[8]Jung B C, Shin W Y. Opportunistic interference alignment forinterference-limited cellular TDD uplink[J]. IEEE Communications Letters,2011, 15(2): 148-150.

[9]Bang C J, Park D, Shin W Y. Opportunistic Interference MitigationAchieves Optimal Degrees-of-Freedom in Wireless Multi-Cell Uplink Networks[J]. IEEE Transactions on Communications, 2012, 60(7):1935-1944.

[10]WANG L F, LI Q, LI S Q, et al. A general algorithm for uplinkopportunistic interference alignment in cellular network[C]//IEEE GlobalTelecommunications Conference (GLOBECOM). Houston, USA, 2011: 436-440.

[11]Yang H J, Shin W Y, Jung B C, et al. Opportunistic interferencealignment for MIMO interfering multiple-access channels[J]. IEEE Transactionson Wireless Communications, 2013, 12(5): 2180-2192.

[12]YANG H J, JUNG B C, SHIN W Y, et al. Codebook-based opportunisticinterference alignment[J]. IEEE Transactions on Signal Processing, 2014, 62(11): 2922-2937.

[13]YOON J, SHIN W Y, LEE H S. Energy-efficient opportunisticinterference alignment[J]. IEEE Communications Letters, 2014, 18(1): 30-33.

[14]JIN H, JEON S W, JUNG B C. Opportunistic interference alignmentfor random access networks[J]. IEEE Transactions on Vehicular Technology,2015, 64(12): 5947-5954.

[15]Yoon J, Shin W Y, Lee H S. Opportunistic Interference Alignmentin Poor Scattering Channels[J]. IEEE Transactions on Vehicular Technology,2016, 65(2):768-779.

[16]LIU G Q, SHENG M, WANG X J, et al. Opportunistic interferencealignment and cancellation for the uplink of cellular networks[J]. IEEECommunications Letters, 2015, 19(4): 645-648.

[17]Zhao N, Yu F R, Leung V C M. Opportunistic communications ininterference alignment networks with wireless power transfer[J]. IEEEWireless Communications, 2015, 22(1):88-95.

[18]Ren Y, Lv T J, Gao H, et al. Wireless Information and EnergyTransfer in Multi-Cluster MIMO Uplink Networks Through OpportunisticInterference Alignment[J]. IEEE Access, 2016, 4: 3100-3111.

[19]Liu H, Gao H, Long W, et al. A novel scheme for downlinkopportunistic interference alignment[C]// IEEE International Conference onTelecommunications. 2014:231-235.

[20]Benaya A M, Elsabrouty M. Two-stage opportunistic interferencealignment for downlink MU-MIMO cellular systems[C]// IEEE InternationalConference on Telecommunications. 2017:1-5.

[21]Yang H J, Shin W Y, Bang C J, et al. Opportunistic DownlinkInterference Alignment for Multi-Cell MIMO Networks[J]. IEEE Transactions onWireless Communications, 2017, 16(3):1533-1548.

[22] thank you, luhua soldiers and Zhao girl, a fair and efficient opportunity interference alignment algorithm based on a polling mechanism, a communication bulletin, 2017, 38 (10), pp.1-9.

[23]Sharif M, Hassibi B. Delay Considerations for OpportunisticScheduling in Broadcast Fading Channels[J]. IEEE Transactions on WirelessCommunications, 2007, 6(9):3353-3363.

[24]Kulkarni S S, Rosenberg C. Opportunistic scheduling policies forwireless systems with short term fairness constraints[C]// IEEE GlobalTelecommunications Conference( GLOBECOM). 2003:533-537.

[25]Sharif M, Hassibi B. On the capacity of MIMO broadcast channelswith partial side information[J]. IEEE Trans. information theory, 2005, 51(2):506-522.

[26]Ferrante M, Tagliavini A. On the coupon-collector's problem withseveral parallel collections[J]. 2016. [Online]. Avaliable: https://arxiv.org/abs/1609.04174

[27]Norris J R. Markov chains[M]. DBLP, 1998.

[28]Kemeny J G, Snell J L. Finite markovchains[M]. Van Nostrand,1976.

The interference alignment technology needs to meet more rigorous conditions when being applied^[1-4]For example, all nodes in the system need to know Global Channel State Information (GCSI), and need large-scale time domain or frequency domain spreading, or need a large number of iterative algorithms, which makes it difficult to implement in practical applications. To avoid these problems, more easily implemented opportunistic interference alignment algorithms are proposedThe method comprises the steps of (orthogonal Interference Alignment, OIA), utilizing a multi-user diversity technology, combining user scheduling and spatial domain Interference Alignment, selecting communication users according to the principle of minimization of Interference Leakage (LIF) by only utilizing Local Channel State Information (LCSI) and without channel expansion, realizing Interference Alignment and elimination, and obtaining the optimal degree of freedom of the system.

In a traditional OIA system, a base station selects communication users on the principle of minimizing interference leakage, the selection of the communication users can be considered as random (namely, the communication users of the base station are selected passively and randomly), the system cannot designate a certain user to carry out communication at a certain time, so that the user cannot be ensured to carry out access and communication within a certain time period, the communication delay of the user may exceed the maximum tolerance delay of the user, and the user experience is reduced. In addition, it is likely that such a random mechanism will result in some users being selected as communication users often for a period of time, and some users always having no opportunity to communicate. Namely, a certain fairness problem exists in a selection mechanism of a communication user in a traditional opportunistic interference alignment algorithm.

Disclosure of Invention

The technical problem to be solved by the invention is to provide a wireless communication user optimal scheduling method based on opportunistic interference alignment, and by using the method, not only the time fairness of users is greatly improved, but also the system capacity is obviously improved.

In order to solve the technical problem, the wireless communication user optimal scheduling method based on opportunistic interference alignment comprises the following steps:

1) for a TDD communication system containing G cells and K users in each cell, S users are selected from each cell to carry out communication in the same time slot and the same frequency each time;

2) all base stations transmit pilot signals, each user in the system estimates a downlink channel corresponding to each base station to obtain local channel state information, an uplink channel of each user is obtained by utilizing channel reciprocity, then each user carries out singular value decomposition on an expected channel from each user to the base station of the cell, and the maximum singular value is fed back to the base station of the cell;

3) each base station determines W active users according to the communication requirements of users in the cell, firstly, whether users which are in urgent need of communication exist in the cell is determined, if not, the base station takes the system capacity maximization as a target, and W users with the best channel quality (the largest singular value) are selected from the users which do not communicate as the active users; if yes, the base station directly takes the active users as the base station, and then selects W-1 users with the optimal channel quality from users without communication as the active users;

4) the active user takes the right singular vector corresponding to the maximum singular value as a precoding vector of the active user, and then feeds back the right singular vector to the base station through equivalent interference channels of other base stations;

5) the base station shares an equivalent channel fed back by the active user through a backbone network, designs own useful signal space and broadcasts the useful signal space to all users;

6) all users calculate the interference leakage of the users and feed the interference leakage back to the base station;

7) the base station determines S-W users with the minimum interference leakage as passive users according to the feedback information;

8) s communication users feed back equivalent channels to the base station of the cell;

9) the base station designs an intra-cell interference elimination vector according to equivalent channels fed back by S communication users;

10) and the user performs data transmission.

In step 2), all users not communicating perform singular value decomposition on the useful channel from the user to the base station of the cell, that is:

wherein

Arranging channels from large to small on a main diagonal for a diagonal matrix

Singular value of

。

In the step 3), taking the maximum value of the singular values

Reflecting the quality of the user channel.

In the step 5), each selected communication user sends the equivalent interference channel to other base stations

Feeding back to the base station of the cell; all base stations share all equivalent interference channels through a backbone network to obtain all equivalent interference channels of own cells; then, the base station BS _g By designing the receiving matrix U _g Completely eliminate interference from other cell "active users"; since each cell has been selectedWOne 'active user', so that base stationgReceiving to from a celliThe interference of all active users in the system can be expressed as:

to eliminate these interferences, the receiving matrix of each base station is designed as follows:

in order to prevent the useful signal of each cell from being interfered by 'active users' of other cells, the useful signal of each cell hasSThe useful signal space for receiving the signal transmitted by the user in the local cell is defined by the following formula _g Is provided withSAnd (3) maintaining solution, namely:

therefore, the number of receiving antennas of the base station needs to satisfy:

after determining its own useful signal space, each base station broadcasts it to all users.

In the present invention, the matrix and the vector are represented by upper and lower bold type, respectively, a^T、A ^H 、pinv(A) Andnull(A) denotes the transpose, conjugate transpose, pseudo-inverse of matrix A, respectively, and the orthonormal basis of the null space of column vector space of A, I and E denote the identity matrix and the full "1" matrix, respectively, A^1*And A^*1Respectively representing the first row and the first column of matrix a,sum(A) representing the summation of all elements in matrix a. Representing collections by handwriting, e.g. representing collections by

To do so by

A collection is represented. Express geta，bThe smaller of (1) represents takingaAnd 0.

The algorithm of the invention has the main ideas that: for a TDD communication system with G cells and K users in each cell, S users are selected from each cell to communicate in the same time slot and the same frequency, firstly, W users with the best current channel state are obtained through singular value decomposition as active users, and then, a useful signal space matrix U of each base station is designed on the principle of completely eliminating active user interference from other cells _g ,Then, the rest is selected based on the principle of minimizing the interference to other cellsS-WA 'passive user', and finally designing a decoding vector f of each user_g,s. Simulation shows that the improved algorithm of the invention not only greatly improves the time fairness, but also obviously improves the system capacity.

Drawings

Fig. 1 is a system model diagram of a multi-cell cellular communication system according to an embodiment;

fig. 2 is a schematic diagram illustrating interference alignment in a state where an algorithm of the present invention selects two communication users in three cells and each cell;

FIG. 3 is a flow chart of the algorithm of the present invention;

FIG. 4 is

Comparing the theoretical value with the simulated value;

FIG. 5 is a drawing showing

Compare plots as a function of K;

fig. 6 is a comparison graph of total system capacity at G =3, Nt =2, Nr =4, S =2, and K is 20 and 100 respectively for the two algorithms of the present invention and the prior art;

FIG. 7 is a graph of the log of total interference leakage as a function of K;

fig. 8 is a diagram of the variation of capacity with the signal-to-noise ratio in downlink communication.

Detailed Description

1. System model

As shown in FIG. 1, this embodiment studies a TDD multi-cell cellular communication system, in which the number of cells interfering with each other is G (the firstgThe base stations in a cell are denoted as BS _g ) The base stations are connected with each other through a wired network. In each cell there isKA user to be communicated, wherein at a certain moment, the user is selected in each cellSA user communicates (gIn a cellsOne communication user being denoted as MS _{g s,}) Each communication user transmits only one data stream. The user side hasN _tA transmitting antenna at the base stationN _rA receiving antenna, andN _t≤N _rthe present invention considersN _r<GSIn the case of (when)N _r≥GSIn time, the base station can directly eliminate all interference through a zero forcing algorithm and obtain signals sent by users in the cell). For simplicity, the uplink is taken as an example in this embodiment, but the algorithm of the present invention is also applicableThe system model is suitable for downlink communication, and is shown in fig. 1, in which a solid line represents a desired signal of a user in a cell, and a dotted line represents an interference signal of another cell. Suppose a subscriber MS _{g s,}To the base station BS _i Is represented by

，MS _{g s,}The data symbols transmitted are

It sends a precoding vector of v _{g s,}Average transmit power per user.

The invention considers a block fading channel model, i.e. the channel coefficient is invariant in a transmission block, each transmission block comprises scheduling time and data transmission time, the change of the channel coefficient between the continuous transmission blocks is independent, the channel coefficient obeys independent complex Gaussian distribution, i.e. the channel coefficient is subjected to independent complex Gaussian distribution

. Base station BS _g The received signal may be expressed as:

wherein the content of the first and second substances,

represents BS _g The additive white Gaussian noise has the elements of zero mean and zero variance

。

Base station BS _g First using a receiving matrix U _g Inter-cell user interference from other cells is cancelled. Namely:

(2)

then the base station decodes the vector f by designing _{g s,}To eliminate the interference between users in the cell, that is:

subscriber MS _{g s,}Can be expressed as

The degrees of freedom of the system can be expressed as

In all scheduling mechanisms, the time fairness of the polling mechanism is optimal, which can ensure the time difference of communication once after the user communicates, and the communication time difference of each user is equal. In the conventional OIA algorithm, the selection of the communication users is random, the time for obtaining the next communication opportunity after the user communicates once has a very large uncertainty, the time difference between two communications may be very small or very large, and the system cannot ensure the maximum time delay of the user communicating once again. Therefore, in the present invention, we measure the fairness of the system by "maximum delay", i.e. the number of transport blocks required for all users in the system to communicate at least onceTIn order to be a measure of the standard,Tthe smaller the communication delay of the user is, the better the fairness of the system is, otherwise,Tthe larger the system the worse the fairness.

2. Opportunistic interference alignment algorithm based on capacity improvement and instant access

2.1 Algorithm scheme design

The invention provides an OIA algorithm based on the real-time communication requirement of a system and considering the expected channels of part of users, aiming at the problem that the traditional OIA only uses the minimization of the interference of the users to other users as the only standard for selecting communication users and does not completely consider the channel condition of the communication users in the system and randomly select the communication users, and the algorithm of the invention is described in detail below.

The fundamental reason why the maximum delay of users in the conventional OIA algorithm may be large is that all communication users selected each time are random, and the system cannot autonomously determine any communication user, thereby failing to control the maximum delay. Based on this, in the algorithm of the present invention, we use the base station cooperation to divide the communication users selected by each cell into two categories, one category is the communication users actively selected by the system according to the real-time communication demand or capacity improvement in each cell and is called "active users", and the other category is the communication users selected by the principle of minimizing the interference to other cells and is called "passive users". The overall algorithm scheme is as follows:

I. selecting "active users"

The active user selection principle firstly considers whether there is a user in the cell which is in urgent need of communication, if so, the user can be directly arranged as a communication user of the time block, and the cell without special communication requirement selects users which have not communicated according to the principle of improving the system capacity performance of the cellW（W<S) The user with the best channel quality is the active user.

Each user estimates a downlink channel according to a pilot signal sent by the base station, and can obtain an uplink channel of the user by utilizing channel reciprocity in the TDD system. In order to maximize system capacity and time fairness, all users which do not communicate perform singular value decomposition on useful channels from the users to the base station of the cell, namely:

wherein

As diagonal matrices, master pairsArranging channels from large to small on a corner line

Singular value of

Maximum of singular values since only one data stream is transmitted per user

Can reflect the quality of the user channel, and each user which has not communicated can determine the user with the best channel quality

Feedback to the base station BS of the cell _i Base stationiComparing all feedback information of the local cell to select the largest feedback informationWThe users communicate, and the cell is determinediIn (1)WAn "active user". The base station broadcasts the information selected by the 'active users', and in order to achieve the best transmission performance, the 'active users' select the right singular vector corresponding to the maximum singular value, namely

As its precoding vector

Therefore, the communication quality of the active user can be guaranteed.

II, designing a useful signal space matrix U of each base station _g

After determining the active users, each selected communication user will use its equivalent interference channel to other base station to design the useful signal space of the base station

And feeding back to the base station of the cell. All base stations share all equivalent interference channels through a backbone network to obtain all equivalent interference channels of own cellsAnd (4) disturbing the channel. Then, the base station BS _g By designing the receiving matrix U _g The interference from other cell "active users" is completely cancelled. Since each cell has been selectedWOne 'active user', so that base stationgReceiving to from a celliThe interference of all active users in the system can be expressed as:

to eliminate these interferences, the receiving matrix of each base station is designed as follows:

in order to prevent the useful signal of each cell from being interfered by 'active users' of other cells, the useful signal of each cell hasSThe useful signal space for receiving the signal transmitted by the user in the local cell is defined by the following formula _g Is provided withSAnd (3) maintaining solution, namely:

therefore, the number of receiving antennas of the base station needs to satisfy:

after determining its own useful signal space, each base station broadcasts it to all users.

III, selecting passive users in each cell "

Since the number of communication users per cell isSIn which has selectedWAn "active user", and a reception matrix U for each cell _g Can completely eliminate signals from other cellsW(G-1) "active user" interference. The rest is selected according to the principle of minimizing the interference to other cellsS-WA "passive user", i.e. v _g,k So that it satisfies:

interference channel for the user to other cells

And the useful space thereof

The matrix of products is Z _g,k Namely:

singular value decomposition is carried out on the obtained data, namely:

to minimize interference leakage, a precoding vector v is taken _{g k,}Is a matrix Z _g,k Minimum singular value of

Corresponding right singular vector

. The square of the smallest singular value is the interference leakage of the user

，MS _{g k,}Feeding back the data to the base station of the local cell, and selecting the smallest data from the data by the base stationS-WThe user is used as a passive user, and then the base station uses the passive userS-WSelected information of a "passive user" is broadcast. When the number of users per cellKWhen the interference is larger, the interference of the 'passive user' selected by the principle of minimum interference leakage to other cells in the system can be almost completely eliminated.

IV, decoding vector f of each user_g,sDesign (2) of

By designing in II, III, the selectedWAn 'active user' andS-Wthe interference of a passive user to other cells is almost zero, and the cell selectsSInterference between individual users still exists and needs to be per cellSDesign of decoding vector F for individual communication usersg,sTo eliminate IUI.

Take cell 1 as an example, whichSThe communication user needs to make its equivalent channel vector

Feeding back to the base station of the cell, and designing a decoding vector f of the base station _{g s,}Comprises the following steps:

the decoding vectors for other cells can be calculated in the same manner. FIG. 2 shows the algorithm of the present invention selecting two communication users per cell, one active user and one passive user (i.e., the active user and the passive user) in three cellsG=3,S=2,WInterference alignment diagram of = 1) (only BS is shown in the figure)₂Alignment diagram of interference at (a).

2.2 Algorithm steps of the invention

As can be seen from fig. 3, according to the time sequence and the processing sequence of the base station and the user side, the detailed steps of the algorithm of the present invention are summarized as follows:

1) all base stations transmit pilot signals;

2) each user in the system estimates the downlink channel corresponding to each base station to obtain the local channel state information, and can obtain the uplink channel of the user by utilizing the channel reciprocity, and then each user carries out singular value decomposition on the expected channel from the user to the base station of the cell and feeds back the maximum singular value to the base station;

3) each base station determines active users according to the communication requirements of the users in the cell, firstly, whether the users in urgent need of communication exist in the cell is determined, and if not, the base station determines the active usersThe base station selects the user with the best channel quality from the users not communicating with the base station with the aim of maximizing the system capacityWThe individual user is taken as an active user; if so, the base station directly takes the active users as the active users and then selects the users with the best channel quality from the users without communicationW-1 user as active user;

4) the active user takes the right singular vector corresponding to the maximum singular value as a precoding vector of the active user, and then feeds back the right singular vector to the base station through equivalent interference channels of other base stations;

5) the base station shares an equivalent channel fed back by the active user through a backbone network, designs own useful signal space and broadcasts the useful signal space to all users;

6) all users calculate the interference leakage of the users and feed the interference leakage back to the base station;

7) the base station determines according to the feedback informationS-WThe user with the minimum interference leakage is a passive user;

8)Sthe communication user feeds back the equivalent channel to the base station of the local cell;

9) the base station is based onSDesigning an intra-cell interference elimination vector by an equivalent channel fed back by each communication user;

10) and the user performs data transmission.

3. Algorithm performance analysis of the invention

3.1 minimum transport Block number analysis of the inventive Algorithm

For analyzing and comparing the time fairness, the number of transmission blocks needed by all users in the system to communicate at least once is usedTIs a measure of^[24]，TSmaller numbers indicate better system fairness, and conversely,Tthe larger the system, the worse the fairness, and obviously in this system configuration,

. When the system configuration is fixed, all users communicate the required transport block number at least onceTIs randomly varied and for analysis, we compute the mathematical expectation and variance of the required transport block number.

At one comprisesGEach cell isCellKIn a cellular system for individual users, each transport block is selected from each cellSThe users communicate, and the expectation of the number of transmission blocks required by all users in the system to communicate at least once is that

(wherein the matrix F₁A base matrix consisting of instantaneous states of combinations of all possible communicated user numbers per transport block before all users communicate at least once), variance

。

And (3) proving that:Geach cell of a cellKEach user selects W users with the best channel quality from the users which are not communicated in each cell as active users, because of the channel coefficient

Each element of (a) is an independent and equally distributed random variable, so the probability of selecting each user from among the users who have not communicated is the same. Similarly, due to channel coefficients

Each element of (a) is an independent and equally distributed random variable, unitary matrix

Independently randomly generated by each base station, thus the matrix

Each element in (a) is also an independent and identically distributed random variable, and the interference leakage is

The probability of selecting any user from any cell at any one time is also the same.

The minimum transport block number is used as the measure of the fairness of the system,and applies Markov chain correlation theory^[26-28]Its expectation and variance are calculated. We have the following

Indicating the number of users that have communicated in each cell after t time blocks have elapsed, the system can be used

Viewed as a discrete-time Markov chain whose state space is all possible states before all users communicate at least once, represented as a set

The number of state spaces is

(24)

Wherein

Definition of

Suppose that

And is and

then its state transition probability is:

(25)

wherein

The transition probability between each state is expressed by matrix, so that the state transition probability matrix of the whole system can be obtained (for convenience of expression, the state sequence numbers are used for representing each state, such as sequence number 0 and sequence number

Respectively represent initial states

And final state

Then probability of

Indicating slave status

One step transition to state

Then the whole system is likely to appear

The states):

in particular, since all users have not communicated until the first moment in time, the communication is not allowed to proceed

And only one absorption state in all states

The other states are transient states.

Defining random variables

Then its condition is expected

I.e. the expected value of the number of transport blocks required for all users in the system to communicate at least once

. Defining vectors

，

Then

Should be the least non-negative solution of the following system of linear equations^[27]：

It is not easy to solve the system of equations directly, and we assume that the matrix Q₁Is a state transition probability matrix P₁Sub-matrices corresponding to all transient states (i.e. pre-matrices)

Rows and columns

A matrix of columns), a markov chain may be defined

Of the basic matrix^[28]Is composed of

And is

. By using [28]Can be obtained

(10)

Therein obtained

I.e. the number of transport blocks required for communicating all users in the system at least onceT，

Namely, it is

。

The following calculationTVariance of (2)

Defining a conditional variance vector

(11)

Then conditional variance

(12)

I.e. the variance found, is obtained from [28]

(13)

Then

Obtain the syndrome.

Table 1 shows the minimum transport block number comparison between the present invention algorithm and the conventional OIA algorithm in different configurations, whenGWhen =3, takeS=2，W= 1; when in useGWhen =4, takeS=3，WAnd (2). It can be seen from the table that the number of transport blocks required by the algorithm of the present invention under various configurations is significantly reduced compared with the conventional algorithm, which indicates that the present inventionThe inventive algorithm has higher time fairness.

Table 1 comparison of minimum transport block number in different configurations of the algorithm of the present invention and the conventional algorithm

3.2 Algorithm analysis of System configurations in different time Per cell

In an actual cellular system, the configuration of each cell may be different, and the algorithm and fairness analysis method of the present invention are also applicable to the case of asymmetric configuration, and we will introduce the OIA algorithm and fairness analysis in the case of asymmetric configuration. Suppose to shareGCell, cellgIs provided withK _g A subscriber to be communicated, its base station BS _g The number of antennas isN _rgThe number of transmitting antennas of all users isN _t(since each user only sends one data stream, the number of antennas at the user side will not cause algorithm change), and each transport block is transmitted from the cellgIn selectionS _g The individual users communicating, includingW _g The number of active users is increased, and the number of active users is increased,S _g -W _g a passive user.

The algorithm of the invention can also realize the opportunity interference alignment under the asymmetric configuration, the flow of the algorithm is consistent with the prior art, but the following conditions are required to be met:

wherein, the formula (a) is the requirement of the number of antennas discussed in the present invention, that is, the number of antennas can not make the user directly use the zero forcing algorithm to obtain the desired signal; (b) The formula being such as to ensure that each base station has a sufficient spatial dimension to accommodate the cellS _g (ii) a signal and interference cancellation for active users in other cells, ((iii))c) This is to ensure that the system allows each user to communicate at least once within a certain time.

From the above (a), (b), (c) and (c)b）、（c) The number of the antennas at the user end can be seen by three inequalitiesN _rg Number of communication subscribers per selectionS _g And number of active users per selection

Can be dynamically changed under the condition that the three constraints are met, and the following three cases discuss the BS _g Number of antennasN _rg Increased fairness of the system:

(1) if it is notS _g AndW _g all of which are unchanged, the number of users required for other cells can be reduced under the condition of obtaining the same degree of freedom

Is required to^[11]The number of users per cell is reduced,Tthe fairness becomes smaller, so that the fairness of the whole system is better;

(2) if it is maintainedS _g Does not change, increases

Then let BS _g More dimension space is aligned with interference signals, thereby reducing the number of users needed by other cells

In addition, the requirements of

The users in the cell can also communicate once in shorter time, so the fairness of the system is better in the case;

(3) if it is maintained

Does not change, increasesS _g The dimensional space for aligning the interfering signals is reduced,the number of users required for other cells increases with the same degree of freedom

Requirement (exponential growth)^[11]) Due to the fact thatS _g Linearly increasing and thus system fairness deteriorates.

From the above analysis, it can be seen that it is desirable to increase the number of antennas of a base stationNrg and number of active users selectedWg, thereby reducing the requirement of the number of users, improving the fairness of the whole system, which is certainly the optimization problem of the whole system, and in order to make the fairness of the whole system reach the optimum, the configuration of each cell is preferably closer.

Time fairnessTThe calculation of (a) is also consistent with the steps described above, and only the differences are described here. Definition of

Without loss of generality, we assume

The same as we do

Represents passing through

After a time slot, the number of users already communicated in each cell is different

Representing a state, the number of state spaces of the system is:

wherein

Similarly, we define

Suppose that

And is

Then its state transition probability is:

wherein

The state transition probability matrix P can be derived from the state transition probabilities as above₂In particular, the amount of the acid present,

and only one absorption state in all states

The other states are transient states. Hypothesis matrix Q₂Is a state transition probability matrix P₂Submatrices corresponding to all transient states, and defining Markov chains as in section 3.1

Of the basic matrix

Then calculate out

And

。

4. simulation results and analysis

This section utilizes Matlab software for simulation and is in accordance with the document [11]]And document [17 ]]Document [18]]Reference [21 ]]And comparing the data and the data to verify the performance of the algorithm and the theoretical analysis result of the invention. In the simulation, the channels are assumed to be rayleigh block fading channels, each component obeys complex gaussian distribution with mean value 0 and variance 1, the noise in the system obeys mean value 0 and variance 1σ ²Complex gaussian distribution of =1, each user at maximum powerPSignals are sent, simulation results are taken from 10⁵Mean value of sub-realizations.

In order to verify the performance and theoretical analysis of the algorithm of the present invention, the algorithm of the present invention and the literature are mainly used in the simulation [11]]SVD-OIA Algorithm and literature [17 ]]The medium R-ICAP-OIA algorithm is compared, and compared performances comprise time fairness, interference leakage and system capacity. In order to ensure the fairness of comparison, the weight coefficient in the R _ ICAP _ OIA algorithm is usedαAnd path loss coefficientβAre all set to 1. Since the algorithm of the present invention is also applicable to downlink channels, we will also finally find the inventive downlink algorithm and document [21 ]]The medium ODIA algorithm and the SE-ODIA algorithm are compared.

In order to verify the correctness of the theoretical analysis of the time fairness in the text, the algorithm of the invention, the SVD-OIA algorithm in the document [11] and the R-ICAP-OIA algorithm in the document [18] are simulated in the case that three cells and 20 users to be communicated are selected in each cell, and two communication users are selected each time (one active user is selected in the algorithm of the invention).

FIG. 4 is

Comparison of theoretical and simulated values of (1), (2)G=3,S=2,N _t=2,N _r=4,W= 1); from FIG. 4, it can be seen whether the algorithm of the present invention or the document [11]][18]The respective theoretical calculation values of the algorithms are basically completely coincided with the simulation result of the actual model, and the time fairness introduced in the text is explainedThe calculation method of (2) is correct. In addition, it can be seen from the figure that the time fairness of the SVD-OIA algorithm and the R-ICAP-OIA algorithm is basically consistent, because the communication users in the two algorithms are randomly selected with equal probability, and only the selection mechanism is inconsistent.

FIG. 5 is a drawing showing

Followed byKVariation comparison graph (G= 3), in particular a comparison of several algorithms in terms of time fairness, it can also be observed from the figure that the algorithm of the invention is comparable to the document [11]]、[18]、[21]The time fairness of the algorithm is greatly improved, and the gap is larger and larger as the number of users per cell is increased, because the system selects users which have not communicated in each time block in the algorithmWThe user with better channel communicates, and other algorithms randomly select from all users in each time blockSThe more users communicate, the greater the uncertainty that arises from this random selection. In addition, we can observe two obvious results from the figure, all the algorithms areSRatio of =3SLess time blocks are needed when = 2; in the algorithm of the present invention, the number of users is selected at each timeSWhen the phase difference is equal to each other,Wthe larger the number of time blocks required, the fewer.

FIG. 6 shows three algorithmsG=3，N _t=2，N _r=4，S=2，KA comparison of total system capacity at 20 and 100 hours, respectively. As can be seen from the figure, the algorithm of the invention has better performance than the SVD-OIA algorithm and the R _ ICAP _ OIA algorithm, and is characterized in thatKCapacity ratio SVD-OIA algorithm at =20KCapacity is better when =100, close to the R _ ICAP _ OIA algorithmKCapacity at = 100. The capacity boosting part of the system is derived from a useful signal space U _g As shown in fig. 2, at the base station 2, from the active users MS in the cell 1_1,1Interference signal of

And active user MS in cell 3_3,1Interference signal of

Interference signal space U fully aligned to base station 2₂Thus, the inter-cell interference from the active users is completely eliminated, and only the interference of the passive users is received; in addition, the base station BS₂The user with the best channel quality is selected as the active user from the users which do not communicate, and the capacity of the cell is enhanced.

FIG. 7 is a graph of total interference leakageKA variable logarithmic graph (SNR =0 dB), specifically, the total interference leakage of the system is determined according to the number of users per cell under the two conditions of three cells and four cells by using three algorithmsKThe change in increase is compared to the figure. The horizontal and vertical coordinates in the simulation diagram are all processed logarithmically, and the SNR =0dB is setGWhen the ratio is not less than =3,S=2，N _t=2，N _r= 4; when in useGWhen the number of the carbon atoms is not less than 4,S=3，N _t=3，N _rand (6). As can be seen from the figure, the number of users to be communicated per cell is variedKThe interference leakage of the three algorithms is continuously reduced, and the number of cells is increasedGInterference leakage at =3 is significantly lower thanG=4 and the rate of decline is faster because of whenGIf the number of communication users is larger and the number of interference terms received by each user is larger than 4, a larger user base is required to reduce interference leakage. And the number of users communicatingSIn both cases, the inventive algorithm showed less interference leakage than the SVD-OIA algorithm and the R _ ICAP _ OIA algorithm, verifying the simulation results of fig. 6.

FIG. 8 is a diagram of the variation of capacity with signal-to-noise ratio in downlink communication (C:)G=3,K=20,S=2,W=1,N _t=4,N _r= 2); the algorithm of the present invention is also suitable for downlink channels, and we also simulate the capacity performance of the downlink channels and compare with the document [21 ]]Comparing the two algorithms of the middle ODIA algorithm and the SE-ODIA algorithm, and selecting in simulationG=3，K=20，S=2，W=1, number of transmission antennas of base stationN _t=4, number of receiving antennas of userN _rAnd (2). As can be seen from the simulation plot, the present invention calculates after SNR is greater than 4dBThe method has larger capacity improvement compared with other two algorithms, and the reason is that the sending beam forming matrix at each base station in the algorithm is completely orthogonal with the equivalent interference channel vectors of all other cell active users, and all interference signals from the base station to the other cell active users can be completely eliminated, so that the algorithm can ensure to obtain the interference signals when being applied to downlink channels no matter how many users to be communicated exist in the systemGWThis degree of freedom, as can also be seen in fig. 8, the capacity of the algorithm of the present invention rises linearly.

5. Conclusion

Aiming at the problems of randomness and the like of communication user selection of the traditional opportunistic interference alignment algorithm, the time fairness of the opportunistic interference alignment algorithm is firstly analyzed, the user groups which have communicated in each cell are combined into one state, and the expectation and the variance of the number of transmission blocks required by all users to communicate at least once in the system are calculated by utilizing the correlation property of a discrete time Markov chain. Based on the method, the traditional opportunistic interference alignment algorithm is improved, the opportunistic interference alignment algorithm based on capacity improvement and instant access is provided, and the cooperation among base stations is utilized to ensure that part of communication users selected in each transmission block can be determined by the system independently; when there is a user in the system which is in urgent need of communication, the system can arrange the communication preferentially, thereby improving the satisfaction degree of the user; when users which do not need to communicate urgently, the system selects users with better channel quality from the users who do not communicate to communicate by taking the improvement of capacity performance and time fairness as targets. Finally, simulation shows that compared with the traditional opportunistic interference alignment algorithm, the algorithm provided by the invention has the advantages that the time fairness, the system capacity and the like are obviously improved.

Claims (1)

1. A wireless communication user optimal scheduling method based on opportunistic interference alignment comprises the following steps:

1) for a TDD communication system containing G cells and K users in each cell, S users are selected from each cell to carry out communication in the same time slot and the same frequency each time;

2) all base stations transmit pilot signals, each user in the system estimates a downlink channel corresponding to each base station to obtain local channel state information, an uplink channel of each user is obtained by utilizing channel reciprocity, then each user carries out singular value decomposition on an expected channel from each user to the base station of the cell, and the maximum singular value is fed back to the base station of the cell;

3) each base station determines W active users according to the communication requirements of users in the cell, firstly, whether users which are in urgent need of communication exist in the cell is determined, if not, the base station takes the system capacity maximization as a target, and the W users with the largest singular value, namely the best channel quality, are selected from the users which do not communicate as the active users; if yes, the base station directly takes the active users as the base station, and then selects W-1 users with the optimal channel quality from users without communication as the active users;

4) the active user takes the right singular vector corresponding to the maximum singular value as a precoding vector of the active user, and then feeds back the right singular vector to the base station through equivalent interference channels of other base stations;

5) the base station shares an equivalent channel fed back by the active user through a backbone network, designs own useful signal space and broadcasts the useful signal space to all users;

6) all users calculate the interference leakage of the users and feed the interference leakage back to the base station;

7) the base station determines S-W users with the minimum interference leakage as passive users according to the feedback information;

8) s communication users feed back equivalent channels to the base station of the cell;

9) the base station designs an intra-cell interference elimination vector according to equivalent channels fed back by S communication users;

10) the user transmits data;

in step 2), all users not communicating perform singular value decomposition on the useful channel from the user to the base station of the cell, that is:

wherein

Arranging channels from large to small on a main diagonal for a diagonal matrix

Singular value of

；

In the step 3), taking the maximum value of the singular values

Reflecting the quality of the user channel;

in the step 5), each selected communication user sends the equivalent interference channel to other base stations

Feeding back to the base station of the cell; all base stations share all equivalent interference channels through a backbone network to obtain all equivalent interference channels of own cells; then, the base station BS _g By designing the receiving matrix U _g Completely eliminate interference from other cell "active users"; since each cell has been selectedWOne 'active user', so that base stationgReceiving to from a celliThe interference of all active users in the system can be expressed as:

to eliminate these interferences, the receiving matrix of each base station is designed as follows:

in order to prevent the useful signal of each cell from being interfered by 'active users' of other cells, the useful signal of each cell hasSThe useful signal space for receiving the signal transmitted by the user in the local cell is defined by the following formula _g Is provided withSAnd (3) maintaining solution, namely:

therefore, the number of receiving antennas of the base station needs to satisfy:

after determining its own useful signal space, each base station broadcasts it to all users.

Images