Disclosure of Invention
In order to effectively improve the transmission rate of the whole system, the invention provides a communication resource allocation method based on non-orthogonal multiple access under a multi-beam satellite system, which comprises the steps of constructing an objective function model for multi-beam satellite communication resource allocation, obtaining an optimal resource allocation scheme by solving the model, executing a grouping algorithm on users to obtain a user grouping set in the process of solving the objective function model, and then executing power allocation on each group of users; the objective function model of the multi-beam satellite communication resource allocation is represented as:
wherein G is
tRepresenting a set of user groupings;
the transmission power of the ith user under the nth beam of the time slot t is represented; t is the total time slot number; n is the total beam number;
representing the signal-to-interference-and-noise ratio of a strong user receiving end;
representing the signal-to-interference-and-noise ratio of a receiving end of a weak user;
representing user i served by beam n at time slot t,
indicating that the user belongs to the t-th slot,
indicating that the user does not belong to the t-th time slot, wherein the value of i indicates the type of the user, and indicates a strong user when i is 1 and indicates a weak user when i is 2; p represents the total transmitting power of the transmitting end;
represents the transmission rate of user i served by beam n at time slot t; r
minIndicating the minimum transmission rate.
Further, in the process of solving the objective function, a grouping algorithm is executed for multiple users to obtain a user grouping set, and the method specifically comprises the following steps:
step 101, initializing parameters including total user number K
totalNumber of beams N, user set
Number of time slot groups T, number of users K in single time slot
tSet of
slot groups 2 × N
User grouping set
Step 102, judging whether the number of the time slots is less than or equal to the number of the time slot groups, if so, executing step 103, otherwise, executing step 104;
103, initializing each user group according to a user grouping selection standard, and distributing first users of each group;
step 104, judging whether the user set is an empty set, if not, executing step 105, otherwise, executing step 106;
105, according to the channel correlation, putting the selected proper user into a corresponding time slot group, and if the length of the time slot group is more than or equal to the number of users contained in a single time slot, reallocating a new time slot;
step 106, judging whether the number of the time slots is less than or equal to the number of the time slot groups, if so, executing step 107, otherwise, executing step 111;
step 107, judging whether the number of users is a user in the corresponding time slot, if so, executing step 108, otherwise, executing step 110;
108, clustering the selected proper users according to the channel gain difference, and determining strong and weak users according to the channel gain of the single user;
step 109, updating the current user grouping set and the time slot group set, changing the number of users, and executing step 107;
step 110, adding 1 to the number of time slots, and executing step 106;
and step 111, outputting the final user grouping set.
Further, in step 103, the criterion for initializing the user group to select the first user is to take the user with the largest channel vector norm, and the channel vector norm calculation includes:
in step 105, selecting a user with a user standard of maximum channel vector norm and a user in each time slot group to respectively calculate a channel correlation coefficient, and placing the user in the time slot group with the minimum channel correlation coefficient, wherein the channel gain calculation comprises:
in step 108, the criterion for selecting suitable users to form clusters is to calculate channel gain difference between two users in each time slot group, two users with large channel gain difference form a cluster, and the calculation of channel gain difference includes:
d(i,j)=||hi|-|hj||;
wherein,
represents the channel gain for user k;
representing the channel gain of user i in t time slot; h is
iWhich represents the channel gain of the user i,
further, power allocation is performed on each group of users after user grouping, and when power allocation is performed, the objective function model can be equivalent to T power allocation subproblems, which are expressed as:
further, in order to solve the T power allocation subproblems, the numerator denominator after the numerator of the subproblem is transformed, and the power allocation subproblem is reconstructed based on the properties of the exponential function and the logarithmic function, which is expressed as:
wherein m is
n、y
n、u
n、v
nA numerator denominator corresponding to the nth wave beam after the power distribution subproblem is divided
Carrying out transformation to obtain a relaxation variable; y is
nFrom noise power
Determination, expressed as
m is x in the relaxation variables of all beams
nIs expressed as m ═ m
1,...,m
N]
T(ii) a u is the relaxation variable in all beams
nIs expressed as u ═ u
1,...,u
N]
T(ii) a v is the v in the relaxation variables in all beams
nIs expressed as v ═ v [ v ]
1,...,v
N]
T;
The transmission power of a strong user under the nth wave beam of the t time slot is represented;
represents the channel gain of a strong user served by beam n at time slot t; w is a
nRepresenting the nth column of the ZF precoding matrix generated by all users under the t time slot;
the transmission power of the weak user under the nth wave beam of the t time slot is represented;
representing the channel gain of the weak user served by beam n at time slot t.
Further, inter-beam interference is eliminated by zero-forcing precoding matrix, so that the satellite transmits signal x through beam nnExpressed as:
wherein, w
nRepresents the nth column of the zero-forcing precoding matrix generated by all users in the t slot,
to representThe transmission power of the kth user in the nth beam and the transmission data are expressed as
And satisfy
Further, the strong user eliminates the intra-cluster interference according to the serial interference elimination technology, so that the signal-to-interference-and-noise ratio of the receiving end of the strong user
Expressed as:
signal-to-interference-and-noise ratio of weak user receiving end
Expressed as:
wherein,
representing the noise power.
Further, the numerator denominator after the power distribution subproblem is divided
Expressed as:
further, the constraint C8 is transformed into a convex constraint, that is, the constraint C8 is transformed into:
wherein,
is a point of linearization. -
Compared with the prior art, the invention can ensure the fairness among users by adopting the user grouping algorithm, reduce the time slot group number, increase the service frequency of a single time slot and also reduce the resource waste of wave beams. The power distribution algorithm effectively improves the transmission rate of the whole system according to the grouped condition on the premise of meeting the minimum service quality of each user.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
In this embodiment, a process of constructing an objective function model for multi-beam satellite communication resource allocation is specifically described.
The invention is suitable for a high-flux multi-beam satellite communication scene, and the system mechanism of the invention is shown in figure 2. In the forward link of the system, a single broadband multibeam satellite is KtotalEach user provides service and adopts a full frequency multiplexing mode to improve the frequency spectrum efficiency.
Suppose K
totalThe users are evenly distributed in N beams generated by the satellite, and each beam is averagely served
And (4) users. To serve multiple users simultaneously, multi-beam satellites employ both Time Division Multiple Access (TDMA) and non-orthogonal multiple access techniques. The system provides service for a plurality of users simultaneously in a frame unit through a TDMA technology, each frame comprises T time slots, and each beam under each time slot selects one user cluster for data transmission through the NOMA technology. The number of user clusters served by the NOMA technology can be more than or equal to two, and it is assumed herein that there are two users in each user cluster under the same time slot per beam, and the strong and weak users are divided according to the magnitude of the channel gain of the users.
At the t-th time slot, from K
totalSelecting K from individual users
t(K
t2N) users are served. All time slots for guaranteeing user service fairnessEach user has one and only one chance to be served under all beams, defining a set of timeslots T ═ 1,2
User group at each time slot is G
tTherefore, there are
Because the high-flux multi-beam satellite communication system mostly adopts a Ku/Ka frequency band, when a satellite channel model is established, the influence of the path propagation loss, the multi-beam antenna gain and the rainfall attenuation on the system performance needs to be considered. Then the channel gain of the satellite and any user k in the ground is expressed as:
wherein h iskGain of the channel of the satellite and any user k on the ground; bmaxRepresents the free space loss, consisting of path propagation loss and noise, expressed as:
where λ is the carrier wavelength, d0Denotes the distance from the satellite to the beam center of the beam corresponding to the ground user terminal, d denotes the distance from the user to the beam center, k is the Bolmatz constant, TRIs the receiver noise temperature, BWIs the noise bandwidth.
Among the various atmospheric effects, the influence of rain attenuation is the most severe, and the corresponding rain attenuation vector Φ is expressed as:
wherein the rain attenuation gain xi
dB(in dB) obeys a lognormal distribution, i.e.
Phi is a phase vector that follows a uniform distribution over 0,2 pi).
The gain of the multi-beam antenna mainly depends on the antenna beam gain of the satellite and the user position, and the antenna beam gain g of the user kkIs expressed as gk={gk,1,gk,2,...,gk,n,...,gk,NH, the beam gain g of user k under antenna nk,nExpressed as:
wherein u is 2.07123sin (theta)k,n)/sin(θk,3dB),J1,J3First order first class and third order first class bezier functions, respectively; thetak,3dBRepresents the 3dB angle of the beam of the user k, and the angle between the kth user and the nth beam center is thetak,nRepresents; gmaxIndicating the maximum satellite antenna gain.
At the t-th slot, assume that K is selected in totaltUser group G formed by individual userstIs provided with
Wherein,
indicating strong and weak users served by beam n at time slot t simultaneously. At this time slot, receive the signal
Is represented as follows:
wherein, y
n,1Indicating a strong user's received signal in the n-th beam, y
n,2Indicating the received signal of the weak user in the nth beam,
is satellite and user group G under t time slot
tThe channel matrix in between is used to determine,
is the transmission signal of the satellite in the t time slot,
means mean 0 and variance
White additive gaussian noise.
To mitigate the inter-beam interference caused by the channel matrix, the inter-beam interference is removed by zero-forcing (ZF) precoding matrix, so that the satellite transmission signal xnCan be expressed as:
wherein,
represents the nth column of the ZF precoding matrix generated by all users at t slot,
indicating the transmission power of the kth user on the nth beam for data transmission
Represents and satisfies
Because the precoding matrix is generated according to the channel gain vector of the strong user, the interference between the beams of the strong user can be completely eliminated. Meanwhile, the strong user can eliminate the intra-cluster interference according to a Successive Interference Cancellation (SIC) technique. Therefore, the signal received by the strong user at the receiving end under the time slot is:
while weak users cannot eliminate inter-beam interference and intra-cluster interference, the signals received at the receiving end are:
therefore, the signal to interference plus noise ratio (SINR) at the receiving end of the strong and weak users is:
the user transmission rate of the nth wave beam under t time slot can be obtained by Shannon formula
Comprises the following steps:
then the user group G under t time slot
tTotal transmission rate of
Is shown as
Then the total transmission rate R of the system for the satellite through TDMA servicetotalIs shown as
Wherein maximizing the system transmission rate maximizes the user group transmission rate per time slot, which is selected by K at a single time slot
tDetermined by the individual user. And K selected at each time slot
tThe users are all different. The power allocation may be made to the users in each time slot to maximize the system transmission rate. Therefore, the resource distribution problem of the multi-beam satellite communication system can be solved by optimizing the user group G
tInter-group user power
An optimization objective function for the following resource allocation problem is established.
Where C1 indicates the number of users per time slot and is only 2N, C2 indicates that each user can be divided into only one time slot, C3 indicates that the power obtained by the user per time slot is not negative, C4 indicates the total power constraint per time slot, C5 indicates the minimum quality of service constraint per user,
indicating that the user belongs to the t-th slot,
indicating that the user does not belong toThe t-th time slot.
Example 2
The objective function model for multi-beam satellite communication resource allocation proposed in embodiment 1 is a non-convex problem, and is difficult to directly solve. For this purpose, it is proposed to perform a grouping algorithm for multiple users first, and then perform a power allocation algorithm for users in each slot to obtain a sub-optimal solution. The present embodiment proposes a packet algorithm for multiple users.
Clustering based NOMA user grouping. The flow chart of the designed user grouping method is shown in fig. 3, and comprises the following steps:
step 1, constructing a multi-beam satellite communication system, and initializing system parameters including total user number K
totalNumber of beams N, user set
Number of time slot groups T, number of users K in single time slot
tSet of
slot groups 2 × N
User grouping set
Step 2, judging whether the number of the time slots is less than or equal to the number of the time slot groups, if so, executing step 3, otherwise, executing step 4;
step 3, circularly calculating the user set
The vector norm of each user channel is selected, and the user u with the maximum norm value is selected
kI.e. by
As G 'per group'
tAnd G 'of a first user, and'
t=G′
t∪{u
k},
Until T is equal to T, the time slot group initialization is finished;
step 4, judging a user set
Whether the set is an empty set or not, if not, executing the
step 5, otherwise, executing the
step 6;
step 5, collecting the sets
User u with the largest channel vector norm in (1)
kAnd according to a channel correlation coefficient calculation formula:
calculate user u separatelykAnd each slot group G'tObtaining 1 channel correlation coefficient matrix C' from the channel correlation coefficient of the inner user, and connecting the user ukAnd putting the time slots into a time slot group corresponding to the minimum element in the matrix. If the length of the time slot group is more than or equal to the number of users contained in a single time slot, reallocating a new time slot;
step 6, judging whether the number of the time slots is less than or equal to the number of the time slot groups, if so, executing step 7, otherwise, executing step 11;
step 7, judging whether the number of users is a user in the corresponding time slot, if so, executing step 8, otherwise, executing step 10;
step 8, according to a calculation formula of the channel gain difference:
calculate slot group G'tUser uiAnd user uj(j∈{i+1,...,Kt) } of the channel gain difference di,j}→DiIn the set DiTake the maximum value, and take user uiAnd the userujDividing into the same cluster, defining strong and weak users according to the channel gain of single user, KtIs KtotalWrite between;
step 9, updating the current user grouping set Gt=Gt∪{ui,ujAnd slot group set G't=G′t\{ui,ujExecuting step 7;
step 10, adding 1 to the number of time slots, and executing step 6;
step 11, outputting the final user grouping set Gt。
Example 2
The present embodiment proposes a method for performing power allocation to each group of users by performing calculation after grouping the users. In this embodiment, the iterative power allocation algorithm for maximizing the system transmission rate specifically includes, as shown in fig. 4:
after user grouping is completed, power distribution is performed based on system transmission rate maximization, and an objective function can be equivalently solved for T power distribution subproblems, namely:
although the constraint conditions are convex conditions, the objective function is a multivariable coupling and non-convex optimization problem and cannot be directly solved. Firstly, carrying out the following variable replacement on the numerator denominator after the passing score of the objective function:
using the properties of exponential and logarithmic functions, the objective function can be written as
Wherein m is
n,u
n,v
nIs the relaxation variable, and yn is the noise power
Constant of determination, i.e.
Thus by relaxing the variable m ═ m
1,...,m
N]
T,u=[u
1,...,u
N]
T,v=[v
1,...,v
N]
TThe objective function can be reconstructed as:
since C8 is a non-convex constraint, it needs to be transformed into a convex constraint to solve the objective function using a convex optimization problem. Therefore, for C8, a first order Taylor expansion approximation is used.
By initialisation
Obtaining an initial feasible point through a first-order Taylor expansion
The initial value of the transmitting power of the strong user under the beam n of the t time slot is obtained, and when the difference value of the transmission rates obtained by two adjacent iterations is smaller than a set threshold value or reaches the maximum valueAnd (5) iteration times, ending iteration and returning to the maximum transmission rate. The constraint C8 may be approximated at each iteration.
Wherein, each wave beam is iterated respectively, and is obtained through N times of iterations
Is shown as
Is a point of linearization. Through the above transformation, the non-convex constraint C8 is transformed into a convex constraint, by which both the objective function and the constraint have been transformed into a convex function. The objective function can be re-expressed as:
the problem can be solved using standard mathematical optimization tools such as CVX.
Example 4
In this example, simulation experiments were performed based on the theories provided in examples 1 to 3.
In this embodiment, the multi-beam satellite communication system includes a single multi-beam GEO satellite equipped with N-16 beams, and the coverage areas of the multiple beams form the coverage area of the entire satellite. And each beam simultaneously serves k-2 users in a single timeslot using power domain NOMA techniques.
The invention provides an algorithm and the existing multi-antenna downlink orthogonal grouping algorithm (MADAC) scheme to carry out simulation comparison verification experiment: the MADOC algorithm is a trade-off between fairness in user service and system transmission rate, so that the grouping threshold epsilon is 0.3 and 0.35 respectively. The satellite carrier frequency is 20GHz and,the user link bandwidth B is 500MHz, the user antenna gain is 41.7dBi, the ground user receiving quality factor G/T is 17.68dB/K, the Bolmann constant K is 1.38 multiplied by 10 (-23) J/K, and the passing K B is
WT
RNormalizing the noise power and the noise variance
The total transmission rate of the system is defined as follows:
fig. 5 compares the total transmission rate performance of the system under different satellite transmission powers according to the algorithm proposed by the present invention with MADOC algorithm-0.3 and MADOC algorithm-0.35. Suppose total number of users K total320, the total time slot number T is 10. As can be seen from fig. 5, as the satellite transmission power increases, the total transmission rate of the system also increases, wherein the invention proposes that the total transmission rate of the algorithm system is kept highest. The invention not only maximizes the number of service users in the time slot group and improves the fairness of user service, but also considers the power distribution among users and maximizes the transmission rate of the system as far as possible on the premise of ensuring the minimum quality of user service.
Fig. 6 compares the transmission rate of each time slot group between the proposed algorithm and the MADOC algorithm-epsilon 0.3 and between the proposed algorithm and the MADOC algorithm-epsilon 0.35 under a single time slot group, and three histograms in each group of 1-10 time slot groups in the picture are the MADOC algorithm-epsilon 0.35, the proposed algorithm and the MADOC algorithm-epsilon 0.3 sequentially from left to right. Because the invention provides the algorithm and adds the power domain NOMA technology to increase the number of users served by a single beam in the same time slot, the transmission rate calculation of the MADOC algorithm in the single time slot group is the sum and average of the adjacent time slot groups. As can be seen from fig. 6, the transmission rate of the algorithm proposed by the present invention is always higher than the MADOC algorithm under any timeslot group, no matter whether the packet threshold of the MADOC algorithm is 0.3 or 0.35.
Fig. 7 compares the variation of the proposed algorithm with MADOC algorithm-0.3 and MADOC algorithm-0.35 in the timeslot groups for different total number of users. It can be seen from fig. 7 that as the total number of users increases, the number of timeslot groups used by each algorithm increases, but the number of timeslot groups used by the algorithm is the least as proposed by the present invention. Since the more slot groups, the lower the service frequency of a single slot group, the better the number of slot groups should be, in order to increase the service frequency of each slot group. The algorithm provided by the invention can ensure that the number of the time slot groups is minimized under the total number of users, and the number of service users of each time slot group is maximized.
The simulation result shows that the user service fairness of the invention is better, the number of user time slot groups is further reduced on the basis of the MADOC algorithm, the number of service users of a single time slot group is increased, the frequency spectrum utilization rate of the system is improved, and the transmission rate of the system is improved.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.