Distributed base station clustering method based on hedonic game
Technical Field
The invention relates to the field of wireless communication, in particular to a distributed base station clustering method based on a hedonic game.
Background
with the development of wireless communication technology and the advent of the information age, the demand for communication and networks has increased. In wireless communication networks, interference handling techniques directly impact the performance of the system. When the precoding technology is used for interference processing of the whole network, each node needs to know global Channel State Information (CSI) to carry out precoding design of sending and receiving, but obtaining the global CSI can cause a large amount of signaling overhead, and when the number of users is too large, occupied time-frequency resources even offset performance gain obtained through precoding.
under the condition of limited sending signal-to-noise ratio, the reachable rate is mainly influenced by strong interference, the influence of weak interference is limited, and more time-frequency resources are used for eliminating all interference, so that the interference is obviously irrevocable. By utilizing a base station clustering method, the transmitting and receiving pairs with strong interference are divided into the same cluster, and a good reachable rate can be obtained when CSI cost is low.
At present, a lot of algorithms for clustering base stations exist, geographically close base stations are usually filled into clusters with fixed sizes, and the method is greatly influenced by search sequences and has poor stability.
Disclosure of Invention
the purpose of the invention is as follows: the invention aims to provide a distributed base station clustering method based on a hedonic game, which can solve the defects in the prior art.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the following technical scheme:
The distributed base station clustering method based on the hedonic game comprises the following steps:
S1: defining a set of base stations in a network asInitially, a single base station is a cluster, i.e. piinital{1}, {2}, { K } }, initializing Πstate=Πinital;
S2:The middle base station carries out the following operations according to the lexicographic order:
S2.1: let base station k currently be in a clusterCalculating preference value of base station k to current cluster according to defined preference function, and recording as Vstate(k);
S2.2: base station k tries andperforming switching operation on the middle cluster, calculating the preference value of the base station k to each cluster according to a defined preference function, and enabling the preference value to be larger than Vstate(k) Marking the clusters as candidate clusters, and sorting the candidate clusters according to the size of the preference value;
S2.3: judging the candidate clusters according to the sequence, if the candidate clusters meet the exchange principle, allowing the exchange, and updating pistate(ii) a If not, keeping the original cluster structure;
s3: when in useif pi is greater than pi after the middle base station completes the operation in step S2stateIf the updating is carried out, the stability is not reached, and the step S2 is skipped to continue the iteration; if pistateKeeping the state unchanged, the base station reaches the individual equilibrium state to obtain the final stateII with cluster structurefinal=Πstate。
Further, the preference function is a rate loss caused by interference from base station j, calculated according to equation (1):
In the formula (1), vi(j) In order to be a function of the preference,As shown in formula (2);
In the formula (2), PiIs the transmission power, P, of base station ijIs the transmit power of the base station j,For receiving conjugate transposes of precoding vectors, vjfor the transmitted precoding vector of user j, viA precoding vector is transmitted for user i. H is the channel matrix, HjiRepresenting the channel between user j and base station i, HjjRepresenting the channel between user j and base station j,For noise power, E represents expectation.
Further, the exchange principle in step S2.3 is: at the present clusternext, base station k leaves the current clusterAdding any one of the other clustersk attempts to andany element j in (1) is exchanged, if and only ifAnd to anysatisfy the requirement ofWhen true, the swap operation executes.
Has the advantages that: the invention discloses a distributed base station clustering method based on a hedonic game, which adopts a game theory to cluster base stations, and from the viewpoint of the game theory, the base stations form stable clusters by dynamically adjusting the cluster structures where the base stations are located, which is a game process. Compared with a heuristic method, the method has great advantages in performance; compared with a clustering method based on the traversal idea, the clustering method has the advantages that the difference in performance is small, and meanwhile, the complexity is greatly reduced.
drawings
FIG. 1 is a flow chart of a clustering method in an embodiment of the present invention;
FIG. 2 is a diagram illustrating the clustering results in an embodiment of the present invention;
Fig. 3 is a graph showing the average total rate change curve under different clustering algorithms when the number of base stations is 8 in the embodiment of the present invention;
fig. 4 is a graph of the average total rate change under different clustering algorithms when the number of base stations is 16 in the embodiment of the present invention;
Fig. 5 is a graph illustrating average total rate variation for different numbers of base station users according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention will be further described with reference to the following detailed description and accompanying drawings.
The present embodiment is described with reference to base station clustering under interference alignment, where Interference Alignment (IA) is a radio transmission strategy for interference coordination proposed in recent years, and interference can be perfectly eliminated when the cluster size meets the IA feasibility. Assuming that 16 base stations are randomly distributed in a square area of 2km × 2km, users are randomly distributed on a ring at a distance of 150m from the corresponding base station. Each base station end is equipped with 3 transmitting antennas, the receiving end is equipped with 2 receiving antennas, the path loss rho is 15.3+37.6log10(distance[m]) Small scale fading follows rayleigh distribution. The noise power spectral density is-174 dbm/hz. Assuming that the transmission power of all transmitting ends is the same, i.e. Pj=P,a distributed base station clustering method based on a hedonic game is shown in figure 1 and comprises the following steps:
s1: defining a set of base stations in a network asinitially, a single base station is a cluster, i.e. piinital{1}, {2}, {16} }, initializing Πstate=Πinital;
S2:The middle base station carries out the following operations according to the lexicographic order:
s2.1: let base station k currently be in a clustercalculating preference value of base station k to current cluster according to defined preference function, and recording as Vstate(k);
S2.2: base station k tries andthe middle cluster is subjected to the switching operation,Calculating preference value of base station k for each cluster according to defined preference function, and making preference value be greater than Vstate(k) Marking the clusters as candidate clusters, and sorting the candidate clusters according to the size of the preference value;
S2.3: judging the candidate clusters according to the sequence, if the candidate clusters meet the switching principle and the cluster size limitation, allowing the switching, and updating the pistate(ii) a If not, keeping the original cluster structure; wherein the cluster size limit is determined by IA feasibility,I.e., cluster size is limited to 4;
S3: when in useIf pi is greater than pi after the middle base station completes the operation in step S2stateIf the updating is carried out, the stability is not reached, and the step S2 is skipped to continue the iteration; if pistateKeeping the base station unchanged, enabling the base station to reach an individual balanced state, and obtaining a final clustering structure IIfinal=Πstate。
wherein the preference function is a rate loss caused by interference from base station j, calculated according to equation (1):
In the formula (1), vi(j) In order to be a function of the preference,As shown in formula (2);
In the formula (2), PiIs the transmission power, P, of base station ijis the transmit power of the base station j,For receiving conjugate transposes of precoding vectors, vjSending pre-order for user jCoding vector, viA precoding vector is transmitted for user i. H is the channel matrix, Hjirepresenting the channel between user j and base station i, HjjRepresenting the channel between user j and base station j,For noise power, E represents expectation.
The exchange principle in step S2.3 is: at the present clusterNext, base station k leaves the current clusterAdding any one of the other clustersk attempts to andAny element j in (1) is exchanged, if and only ifAnd to anysatisfy the requirement ofWhen true, the swap operation executes.
As shown in fig. 3, the clustering method proposed in this embodiment approaches the optimal clustering on the average total rate curve, and the performance is greatly improved in the case of non-clustering. The optimal clustering is obtained by an exhaustion method, the calculated amount is increased exponentially along with the increase of the number of transmitting and receiving pairs, and the clustering by the exhaustion method is not practical. The clustering method proposed by the present embodiment makes a good trade-off between performance and computational complexity. As shown in fig. 4, when the number of base stations increases to 16, the clustering method proposed in this embodiment still maintains a great advantage. Fig. 5 is a curve of average total rate change with an increase in the number of base station users, and the clustering method provided in the present embodiment maintains significant advantages. In summary, in the base station clustering scenario under interference alignment, the clustering method provided by the present embodiment exhibits good performance.