CN102098678B - Dynamic channel assignment (DCA) method for large-scale mobile cellular communications system - Google Patents

Abstract

The invention relates to a dynamic channel assignment (DCA) method for a large-scale mobile cellular communications system, belonging to the technical field of mobile communication. In the method, a large-dimension cellular net is divided into a plurality of small-dimension cellular subnets (DCSs), and each cellular subnet is provided with an independent cellular subnet control center; and meanwhile, each cellular subnet is provided with a real-time interference channel table. Compared with the traditional DCA method, the DCA method provided by the invention has the advantages that the huge DCA calculated amount of the whole large-dimensional cellular net is shared by each cellular subnet, thus, within unit time, times for each cellular subnet to operate a DCA program can be greatly reduced; and the control center of each cellular subnet can not be overloaded. Meanwhile, because the channel reassignment times can be greatly reduced, the signalling cost of the whole cellular net can be also greatly lowered, the communication link delay is reduced, and the real-time requirement of mobile user communication is guaranteed.

Description

A kind of cellular mass mobile communication system dynamic channel assignment method

Technical field:

The present invention relates to a kind of cellular mass mobile communication system dynamic channel assignment method, belong to mobile communication technology field.

Background technology:

The dynamic channel allocation (DCA) of cell mobile communication systems (claiming again Cellular Networks), refer between each cellular cell, according to the real-time communication load in each cellular cell (number of users and traffic rate), distribute the required number of channel dynamically to each cellular cell, thereby adapt to the time dependent communication load in each cellular cell, improve QoS of customer, and improve the availability of frequency spectrum of whole Cellular Networks.

Domestic and international existing DCA method, has all only considered Cellular Networks on a small scale, and the cellular cell number of considering does not exceed at most 64, and actual Cellular Networks is all large-scale, and cellular cell number is thousands of.So, want to make DCA technology really by theory, to move towards engineering practice, just must solve the DCA problem of cellular mass net (LCN).If existing DCA technology is applied directly to actual cellular mass net, can face two difficult problems.

First, because the cellular cell number of cellular mass net is thousands of, any one cellular cell has new calling to arrive, and all can trigger the DCA operation an of the whole network, if the number of cellular cell is enough large, can think all the time so, have continual new calling to arrive each cellular cell, this also just means, the network control center will carry out DCA program continuously, thereby the network control center be can't bear the heavy load, even because responding in time, make whole network paralysis.

Secondly, DCA operation each time, the channel that all may trigger a network-wide basis heavily distributes.Channel heavily distributes and refers to: the mobile subscriber who is much communicating by letter, be forced to be switched on other channels, and even there will be certain two mobile subscriber's channel exchange.Many times, only have through channel and heavily distribute, just likely for new calling found an available idle channel.If the scale of Cellular Networks is excessive, each cellular cell has new calling to arrive continuously, trigger continual DCA operation, also will cause continual channel heavily to distribute, can make so a lot of mobile subscribers that communicating by letter, constantly from a channel, be switched to one other channel, this channel too frequently heavily distributes, not only need a large amount of network signaling overheads, but also can bring the delay of wireless communication link, thereby had a strong impact on the requirement of real-time of mobile communication.Hence one can see that, and existing DCA method can not be applicable to actual cellular mass net.

Summary of the invention:

The object of the invention is to: a kind of load that not only can alleviate existing mobile communications network control centre is provided, and can the fine cellular mass mobile communication system dynamic channel assignment method that meets mobile communication requirement of real-time.

The present invention realizes above-mentioned purpose by the following technical solutions:

A kind of cellular mass mobile communication system dynamic channel assignment method, is characterized in that: it comprises the following steps:

A), cellular mass net is divided into multiple small-scale honeycomb subnets (DCS), the size of each honeycomb subnet can arrange flexibly according to the performance of DCA, and for each honeycomb subnet, an independently honeycomb subnet control centre is set;

B), for each honeycomb subnet, set up a real-time interference channel table, this interference channel table, is to set up according to the channel allocation information of all adjacent cell subnets of each honeycomb subnet;

C), according to the channel demands of each cellular cell in the interference channel table of each honeycomb subnet, honeycomb subnet, and the channel allocation constraints between each cellular cell in each honeycomb subnet, for a cost function of each honeycomb subnet structure, and adopt noise chaotic neural network (NCNN) method, make the value minimum of this cost function; Cost function is expressed as follows: formula (1):

$E=\frac{A_e}{2}\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j\≠i}{\mathrm{\Σ}}{v}_{\mathrm{xi}}{v}_{\mathrm{xj}}{f}_{\mathrm{CSC}}(i,j)+\frac{B_e}{2}\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{y\≠x}{\underset{y\∈\mathrm{Near}}{\mathrm{\Σ}}}\underset{j\≠i}{\mathrm{\Σ}}{v}_{\mathrm{xi}}{v}_{\mathrm{yj}}{f}_{\mathrm{ACC}}(i,j)$

$+\frac{C_e}{2}\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{y\≠x}{\mathrm{\Σ}}{v}_{\mathrm{xi}}{v}_{\mathrm{yi}}{f}_{\mathrm{CCC}}(x,y)+\frac{D_e}{2}\underset{x=1}{\overset{m}{\mathrm{\Σ}}}{(\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{v}_{\mathrm{xi}}-R_x)}^2$

$+F_e\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{v}_{\mathrm{xi}}{T}_{\mathrm{xi}}+G_e\underset{x=1}{\overset{m}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{iv}}_{\mathrm{xi}};$

D), in any one honeycomb subnet, if there is new calling to arrive, need to increase the number of channel, all adjacent cell subnets are inquired about by this honeycomb subnet control centre, if now there is certain adjacent cell subnet carrying out DCA program, wait for, until all adjacent cell subnets are not all carried out DCA program, scan the channel assignment table of all adjacent cell subnets, and generate the interference channel table of this honeycomb subnet;

E), according to the channel demands number of each honeycomb subnet, and the interference channel table of each honeycomb subnet, utilize NCNN method, the value of the 6th of the cost function of each honeycomb subnet is minimized, make the value of the 1st, 2,3,4,5 of cost functions all equal zero simultaneously.

In formula (1), each symbol definition is: m represents a cellular cell sum in honeycomb subnet, n represents an available channel sum in honeycomb subnet, x and y represent cellular cell numberings different in a honeycomb subnet, and i and j represent available channel numberings different in a honeycomb subnet.In addition, using | x-y| represents the space length of interior two the cellular cell x of a honeycomb subnet and y; | i-j| represents that two different channels i and j are at the interval of frequency domain position, the namely distance of channel.V _xirepresent the neuron output of NCNN, if channel i has distributed to cellular cell x, v _xi=1, otherwise v _xi=0, v _xj, v _yi, v _yjdefinition and v _xidefinition similar, only footnote difference just; Six constant A _e, B _e, C _e, D _e, F _eand G _ebe corresponding every punishment parameter, they can, according to the constringency performance of NCNN, regulate respectively.

The 1st of formula (1)

it is same constraint based on sites (CSC:Co-Site Constraint), in a honeycomb subnet, CSC requires all channels that are assigned in same cellular cell, all will keep a minimum channel spacing, otherwise can produce serious inter-carrier interference; The present invention represents the interval of channel (or frequency) with L, with constraint based on sites function definition, be: formula (2):

$f_{\mathrm{CSC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<L\\ 0,|i-j|\&GreaterEqual;L\end{array}\right.$ When the interval of two channel i and j is less than L, constraint function equals 1, otherwise equals zero.If CSC is met, the value of the 1st will equal zero.Product term v _xiv _xjrepresent that (j is not equal to i) and has been distributed to cellular cell x simultaneously, and if only if constraint function f for channel i and j _cSC(i, j)=0 o'clock, the value of the 1st just can equal zero.

In formula (1) the 2nd

shi Lin road constraint (ACC), ACC requires adjacent cells can not distribute adjacent frequency, otherwise can mutually produce serious inter-carrier interference.Facing constraint function is defined as: formula (3):

$f_{\mathrm{ACC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<2\\ 0,|i-j|\&GreaterEqual;2\end{array}\right.$ When the interval of two adjacent channels is less than 2, ACC constraint function equals 1, otherwise equals 0.If ACC constraints is met, the value of the 2nd will equal zero.Product term v _xiv _yjrepresent that (j is not equal to and i) is assigned to respectively cellular cell x and y (y is not equal to x, and cellular cell y belongs to the adjacent cells of cellular cell x) simultaneously, and if only if constraint function f for channel i and j _aCC(i, j)=0 o'clock, the value of the 2nd just can equal zero.In the summation symbol of the 2nd, symbol Near represents the adjacent cells set of cellulor district x.

In formula (1) the 3rd be that people having a common goal retrains (CCC), CCC requires any two cellular cells that are assigned with same frequency, must keep at a certain distance away, otherwise can produce co-channel interference on locus.People having a common goal's constraint (CCC) function definition is: formula (4):

$f_{\mathrm{CCC}}(x,y)=\left\{\begin{array}{c}1,|x-y|<D_{\mathrm{reuse}}\\ 0,|x-y|\&GreaterEqual;D_{\mathrm{reuse}}\end{array}\right.$

Wherein D _reuserepresent channel reuse distance, if the distance of cellular cell x and y is less than reuse distance, functional value equals 1, otherwise equals 0.If CCC constraints is met, the value of the 3rd will equal zero.Product term v _xiv _yirepresent channel j to have been distributed to cellular cell x and cellular cell y (y is not equal to x), and if only if constraint function f simultaneously _cCC(x, y)=0 o'clock, the value of the 3rd just can equal zero.

In formula (1) the 4th

the channel demands bound term of cellular cell, the number of channel of and if only if each cellular cell of distributing to, while just equaling the required number of channel in each cellular cell, the 4th just can equal 0.Here use R _xrepresent the channel demands number of cellular cell x, the channel demands of all cellular cells in whole honeycomb subnet, by a channel demands, vectorial R represents, the channel demands of cellular cell x is counted R _x, be x the element of channel demands vector R.

In formula (1) the 5th

the interference channel bound term of honeycomb subnet, T _xixi the element of a honeycomb subnet interference channel table T, if channel i is the interference channel of forbidding to cellular cell x, T _xi=1, otherwise T _xi=0.Product term v _xit _xirepresent, if v _xi=0 or T _xi=0, product v _xit _xi=0, that is to say, if channel i is forbidden channel for cellular cell x, to only have v _xi=0 just can make the value of the 5th equal 0.Only have the interference channel constraints that has all met this honeycomb subnet when all cellular cells, the value of the 5th just can equal 0.

In formula (1) the 6th

be channel shortening item, in the situation that meeting each cellular cell channel demand, if can adopt the number of channel still less, can improve the availability of frequency spectrum of each honeycomb subnet and whole cellular mass net.A feasible way, is exactly the channel of the low numbering of priority allocation, and the channel of high numbering only, in the situation that low numbering channel is not enough, is just distributed gradually.Product term iv _xirepresent channel i and neuron output v _xiproduct, if it is lower to distribute to the numbering of channel i of cellular cell x, product term iv _xivalue just less, the number of channel that is assigned to all cellular cells is also just fewer, thereby the availability of frequency spectrum is also just higher.Different from other several is, last value can not equal 0, but one be greater than 0 value, and this value is unpredictable, this is because last value, its size not only will be subject to the impact of a honeycomb subnet cellular cell sum, but also will be subject to current the distributed number of channel, channel number and punishment parameter G _eimpact, and current the distributed number of channel and channel number are all uncertain, so last value is also just unpredictable.If the value of the 1st, 2,3,4,5 of formula (1) all equals zero, the 6th is one and is greater than zero value, even and NCNN rerun down, the value of the 6th also can not reduce again, thinks that NCNN has found an optimal solution.

The invention has the advantages that:

First, the DCA amount of calculation that whole cellular mass net is huge, by each honeycomb subnet, shared, because the scale of each honeycomb subnet is very little, so all cellular cells in each honeycomb subnet, within the unit interval, the new arrival number of times of calling out can be not too many, thereby in the unit interval, the number of times of each honeycomb subnet operation DCA program can greatly reduce, and therefore the control centre of each honeycomb subnet can not transship.Meanwhile, the number of times heavily distributing because of channel greatly reduces, and the signaling consumption of whole Cellular Networks also can reduce greatly, and the delay of communication link also can reduce, thereby has guaranteed the requirement of real-time of mobile subscriber's communication.

Secondly, although DCA program is independently to carry out in each honeycomb subnet, but each honeycomb subnet is before operation DCA program, capital is first inquired about its adjacent cell subnet and whether is being carried out DCA program, if all adjacent cell subnets are not now all carried out DCA program, scans the channel assignment table of all adjacent cell subnets, and calculate the interference channel table of this honeycomb subnet, and then operation DCA program, so between adjacent cell subnet, can not produce interference.In addition, because the scale of each honeycomb subnet is all very little, so that adjacent cell subnet needs to carry out the collision probability of DCA program is very little simultaneously, even if clash, needed wait time is also very little.

Accompanying drawing explanation

Fig. 1 is the cellular mass net example (being divided into the honeycomb subnet of 9 49 cellular cells) of 449 cellular cells;

Fig. 2 is the interference channel table (considering the interference of 8 adjacent cell subnets) of honeycomb subnet E;

Fig. 3 is the allocated channel table (considering CSC, ACC, CCC and frequency compression) of honeycomb subnet E;

Fig. 4 is the allocated channel table (consider CSC, ACC, CCC, but without frequency compression) of honeycomb subnet E;

Fig. 5 is the allocated channel table (only considering CCC and frequency compression) of honeycomb subnet E;

Fig. 6 be honeycomb subnet E DCA Performance Ratio.

Embodiment:

This cellular mass mobile communication system dynamic channel assignment method, it comprises the following steps:

Cellular mass net is divided into multiple small-scale honeycomb subnets (DCS), and the size of each honeycomb subnet can arrange flexibly according to the performance of DCA, and is that each honeycomb subnet arranges individual independently honeycomb subnet control centre;

For each honeycomb subnet is set up a real-time interference channel table, this interference channel table, is to set up according to the channel allocation information of all adjacent cell subnets of each honeycomb subnet;

According to the channel demands of each cellular cell in the interference channel table of each honeycomb subnet, honeycomb subnet, and the channel allocation constraints between each cellular cell in each honeycomb subnet, for a cost function of each honeycomb subnet structure, and adopt noise chaotic neural network (NCNN) method, make the value minimum of this cost function; Cost function is expressed as follows: formula (1):

$E=\frac{A_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{v}_{\mathrm{xj}}{f}_{\mathrm{CSC}}(i,j)+\frac{B_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{y\&NotEqual;x}{\underset{y\&Element;\mathrm{Near}}{\mathrm{\&Sigma;}}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{v}_{\mathrm{yj}}{f}_{\mathrm{ACC}}(i,j)$

$+\frac{C_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{y\&NotEqual;x}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{v}_{\mathrm{yi}}{f}_{\mathrm{CCC}}(x,y)+\frac{D_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}{(\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xi}}-R_x)}^2$

$+F_e\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xi}}{T}_{\mathrm{xi}}+G_e\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{iv}}_{\mathrm{xi}};$

In formula (1), each symbol definition is: m represents a cellular cell sum in honeycomb subnet, n represents an available channel sum in honeycomb subnet, x and y represent cellular cell numberings different in a honeycomb subnet, and i and j represent available channel numberings different in a honeycomb subnet.In addition, using | x-y| represents the space length of interior two the cellular cell x of a honeycomb subnet and y; | i-j| represents that two different channels i and j are at the interval of frequency domain position, the namely distance of channel.V _xirepresent the neuron output of NCNN, if channel i has distributed to cellular cell x, v _xi=1, otherwise v _xi=0, v _xj, v _yi, v _yjdefinition and v _xidefinition similar, only footnote difference just; Six constant A _e, B _e, C _e, D _e, F _eand G _ebe corresponding every punishment parameter, they can, according to the constringency performance of NCNN, regulate respectively.

The 1st of formula (1) it is same constraint based on sites (CSC:Co-Site Constraint), in a honeycomb subnet, CSC requires all channels that are assigned in same cellular cell, all will keep a minimum channel spacing, otherwise can produce serious inter-carrier interference; The present invention represents the interval of channel (or frequency) with L, with constraint based on sites function definition, be: formula (2):

$f_{\mathrm{CSC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<L\\ 0,|i-j|\&GreaterEqual;L\end{array}\right.$ When the interval of two channel i and j is less than L, constraint function equals 1, otherwise equals zero.If CSC is met, the value of the 1st will equal zero.Product term v _xiv _xjrepresent that (j is not equal to i) and has been distributed to cellular cell x simultaneously, and if only if constraint function f for channel i and j _cSC(i, j)=0 o'clock, the value of the 1st just can equal zero.

In formula (1) the 2nd

shi Lin road constraint (ACC), ACC requires adjacent cells can not distribute adjacent frequency, otherwise can mutually produce serious inter-carrier interference.Facing constraint function is defined as: formula (3):

$f_{\mathrm{ACC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<2\\ 0,|i-j|\&GreaterEqual;2\end{array}\right.$

When the interval of two adjacent channels is less than 2, ACC constraint function equals 1, otherwise equals 0.If ACC constraints is met, the value of the 2nd will equal zero.Product term v _xiv _yjrepresent that (j is not equal to and i) is assigned to respectively cellular cell x and y (y is not equal to x, and cellular cell y belongs to the adjacent cells of cellular cell x) simultaneously, and if only if constraint function f for channel i and j _aCC(i, j)=0 o'clock, the value of the 2nd just can equal zero.In the summation symbol of the 2nd, symbol Near represents the adjacent cells set of cellulor district x.

In formula (1) the 3rd

be that people having a common goal retrains (CCC), CCC requires any two cellular cells that are assigned with same frequency, must keep at a certain distance away, otherwise can produce co-channel interference on locus.People having a common goal's constraint (CCC) function definition is: formula (4):

$f_{\mathrm{CCC}}(x,y)=\left\{\begin{array}{c}1,|x-y|<D_{\mathrm{reuse}}\\ 0,|x-y|\&GreaterEqual;D_{\mathrm{reuse}}\end{array}\right.$

Wherein D _reiserepresent channel reuse distance, if the distance of cellular cell x and y is less than reuse distance, functional value equals 1, otherwise equals 0.If CCC constraints is met, the value of the 3rd will equal zero.Product term v _xiv _yirepresent channel i to have been distributed to cellular cell x and cellular cell y (y is not equal to x), and if only if constraint function f simultaneously _cCC(x, y)=0 o'clock, the value of the 3rd just can equal zero.

In formula (1) the 4th

the channel demands bound term of cellular cell, the number of channel of and if only if each cellular cell of distributing to, while just equaling the required number of channel in each cellular cell, the 4th just can equal 0.Here use R _xrepresent the channel demands number of cellular cell x, the channel demands of all cellular cells in whole honeycomb subnet, by a channel demands, vectorial R represents, the channel demands of cellular cell x is counted R _x, be x the element of channel demands vector R.

In formula (1) the 5th

the interference channel bound term of honeycomb subnet, T _xixi the element of a honeycomb subnet interference channel table T, if channel i is the interference channel of forbidding to cellular cell x, T _xi=1, otherwise T _xi=0.Product term v _xit _xirepresent, if v _xi=0 or T _xi=0, product v _xit _xi=0, that is to say, if channel i is forbidden channel for cellular cell x, to only have v _xi=0 just can make the value of the 5th equal 0.Only have the interference channel constraints that has all met this honeycomb subnet when all cellular cells, the value of the 5th just can equal 0.

In formula (1) the 6th

be channel shortening item, in the situation that meeting each cellular cell channel demand, if can adopt the number of channel still less, can improve the availability of frequency spectrum of each honeycomb subnet and whole cellular mass net.A feasible way, is exactly the channel of the low numbering of priority allocation, and the channel of high numbering only, in the situation that low numbering channel is not enough, is just distributed gradually.Product term iv _xirepresent channel i and neuron output v _xiproduct, if it is lower to distribute to the numbering of channel i of cellular cell x, product term iv _xivalue just less, the number of channel that is assigned to all cellular cells is also just fewer, thereby the availability of frequency spectrum is also just higher.Different from other several is, last value also can not equal 0, but one be greater than 0 value, and this value is unpredictable, this is because last value, its size not only will be subject to the impact of a honeycomb subnet cellular cell sum, but also will be subject to current the distributed number of channel, channel number and punishment parameter G _eimpact, and current the distributed number of channel and channel number are all uncertain, so last value is also just unpredictable.If the value of the 1st, 2,3,4,5 of formula (1) all equals zero, the 6th is one and is greater than zero value, even and NCNN rerun down, the value of the 6th also can not reduce again, thinks that NCNN has found an optimal solution.

In any one honeycomb subnet, if there is new calling to arrive, need to increase the number of channel, all adjacent cell subnets are inquired about by this honeycomb subnet control centre, if now there is certain adjacent cell subnet carrying out DCA program, wait for, until all adjacent cell subnets are not all carried out DCA program, scan the channel assignment table of all adjacent cell subnets, and generate the interference channel table of this honeycomb subnet;

According to the channel demands number of each honeycomb subnet, and the interference channel table of each honeycomb subnet, utilize NCNN method, the value of the 6th of the cost function of each honeycomb subnet is minimized, make the value of the 1st, 2,3,4,5 of cost functions all equal zero simultaneously.

For making object of the present invention, technical scheme and advantage more cheer and bright, below in conjunction with accompanying drawing, the present invention is described in further detail for example.

Fig. 1 is example of the present invention, is the cellular mass net (LCN) of 441 cellular cells.First, by the cellular mass net shown in Fig. 1, be divided into the honeycomb subnet of 9 49 cellular cells, these 9 honeycomb subnets are expressed as A, B, C, D, E, F, G, H, I, 49 cellular cells in each honeycomb subnet are numbered 1 to 49, and for example cellular cell G43 refers to the 43rd cellular cell in honeycomb subnet G.Whole Cellular Networks has 70 available channels, and which honeycomb subnet no matter each cellular cell belong to, and has the chance of equality to be assigned to any available channel.For the ease of distinguishing each honeycomb subnet, in Fig. 1, each cellular cell of honeycomb subnet B, D, F, H is filled to grey, all the other honeycomb subnets are not filled, and are blank.

For three constraintss (CSC, ACC, CCC) of channel allocation are described, take the cellular cell I25 of the honeycomb subnet I in Fig. 1 as example.Three round R1, R2 centered by the I25 of cellular cell, R3, run through respectively three class cellular cells, is called R1 class cellular cell, R2 class cellular cell, the R3 class cellular cell of cellular cell I25.For CSC, if the L=3 in formula (2), suppose that No. 10 channels have been assigned to cellular cell I25, according to formula (2), in order to meet CSC, 8,9,11, No. 12 channels can not have been reallocated to cellular cell I25, but numbering is less than or equal to 7, is more than or equal to 13 channel and can be assigned to I25; For ACC, if No. 10 channel has been assigned to I25, according to formula (3), 8, No. 9 channels can not be distributed to R1 class cellular cell (I31, I32, I24, I26, I18, I19), but other channels except 8, No. 9 channels, can distribute to R1 class cellular cell.For CCC, if No. 10 channel has been assigned to I25, according to formula (4), No. 10 channels can not be reallocated to R1 class cellular cell, the R2 class cellular cell of I25, but can distribute to R3 class cellular cell, and other cellular cells outside circle R3, channel reuse distance B in formula (4) _reusedefinition, just refer to the distance of I25 and R3 class cellular cell, all and distance cellular cell I25 is more than or equal to D _reusecellular cell, can distribute to and channel identical in I25 (as No. 10 channels).

Then, set up an interference channel table to each honeycomb subnet.For ease of analyzing, in Fig. 1, only, take honeycomb subnet E as considering object, because it is surrounded by 8 adjacent cell subnets (A, B, C, D, F, G, H, I), its network environment approaches a real honeycomb subnet most.For honeycomb subnet E, it has 66 to disturb cellular cell, lay respectively in 8 adjacent cell subnets, and as two 66 cellular cells that dotted line ring Loop1 and Loop2 are run through in Fig. 1, be exactly the interference cellular cell of honeycomb subnet E.So the interference channel table of honeycomb subnet E, only need to consider these 66 allocated channels that disturb cellular cells, other cellular cells in 8 adjacent cell subnets, because of and the distance of honeycomb subnet E be greater than reuse distance D _reuse, can not cause interference to honeycomb subnet E, thereby need not consider.

In order to simulate a real networking operational environment, give 66 of honeycomb subnet E some channels that disturbed cellular cell Random assignments, these channels have also avoided 66 to disturb the phase mutual interference between cellular cell (to meet CSC simultaneously, ACC, tri-constraintss of CCC), be allocated as follows: A49 (17, 28, 47), B36 (2, 10, 53, 69), B37 (14, 20, 31, 58, 64, 67), B38 (4, 11, 56, 62), B39 (19, 22, 35, 68), B40 (1, 26, 29, 49, 60, 63), B41 (16, 37, 52), B42 (2, 19, 27, 39, 48, 58), B43 (22, 25, 35, 49, 55), B44 (18, 27, 51, 60), B45 (7, 16, 24), B46 (53, 58), B47 (3, 14, 23, 33), B48 (6, 66), B49 (12, 32, 46, 60, 64), C36 (14, 17, 36, 53, 56, 62), C37 (6, 10, 47, 65), C43 (29, 34), C44 (4, 21, 45, 54, 61), F1 (7, 42, 69), F2 (14, 27, 33, 36, 63), F8 (22, 39, 51, 57, 66), F9 (9, 46, 60), F15 (37, 43, 68), F16 (7, 18, 58), F22 (49, 52, 62, 65), F23 (55), F29 (8, 38, 44, 47), F30 (1, 13, 41, 53, 59, 63), F36 (15, 26, 61, 70), F37 (4, 45, 49, 56), F43 (18, 36, 39, 42, 58), F44 (24, 30, 33, 47, 54, 60), I1 (16), H1 (11, 21, 28, 31, 55), H2 (1, 9, 15, 23, 33, 36, 42, 67), H3 (57), H4 (6, 10, 32, 61), H5 (4, 45, 54, 65), H6 (8, 11, 14, 37, 57, 60), H7 (5, 22, 25, 31, 66, 69), H8 (25, 44, 47), H9 (59, 62), H10 (34, 46, 66, 69), H11 (2, 18, 26, 36, 39, 52, 58, 63), H12 (24), H13 (17, 35), H14 (19, 29, 38, 41, 56, 59), G6 (18, 29, 32, 56, 62, 65), G7 (3, 26, 37, 40, 43, 46, 68), G13 (5, 10, 20, 23, 48, 53, 60), G14 (8, 50, 57, 63, 66), D6 (39, 50), D7 (1, 19, 32, 44, 52, 62, 65, 70), D13 (16, 48, 56), D14 (3, 26, 38, 46), D20 (7, 10, 69), D21 (14, 20, 33, 40, 43, 64), D27 (12, 22, 30, 47, 52, 60), D28 (2, 5, 18, 24, 27), D34 (15, 37, 49, 58, 65), D35 (21, 34, 39, 45, 69), D41 (23, 61), D42 (4, 19, 25, 28, 55, 64), D48 (11, 14, 35, 38, 44, 47, 59, 67, 70), D49 (6, 9, 16, 22, 49, 52).The channel number that numeral in bracket is distributed.These 66 channels that disturb cellular cell to distribute have all met CSC, ACC, the large constraints of CCC tri-, have simulated an actual networking operational environment.

According to above-mentioned 66 channel informations that disturb cellular cell to distribute, according to formula (2), (3), (4), honeycomb subnet control centre just can calculate the interference channel table of honeycomb subnet E, as shown in Figure 2, black lattice represents that this channel is interference channel to affiliated cellular cell, forbid, and blank grid represents this channel, can distribute to honeycomb subnet E.Honeycomb subnet E, when carrying out DCA program, must avoid these black lattices, namely avoids its interference channel.An interesting phenomenon is, 17th, 18,19,24,25,26,31,32,33 row are all blank grid, this be because these cellular cells in the center of honeycomb subnet E, they are irrelevant with eight adjacent cell subnets outside the reuse distance of eight adjacent cell subnets.

Based on formula (1), the equation of motion of the NCNN that can derive is as follows:

Formula (5):

$\frac{{\mathrm{du}}_{\mathrm{xi}}}{\mathrm{dt}}=-\frac{\&PartialD;E}{{\&PartialD;v}_{\mathrm{xi}}}=-A_e\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xj}}{f}_{\mathrm{CSC}}(i,j)-B_e\underset{y\&NotEqual;x}{\underset{y\&Element;\mathrm{Near}}{\mathrm{\&Sigma;}}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{yj}}{f}_{\mathrm{ACC}}(i,j)$

$-C_e\underset{y\&NotEqual;x}{\mathrm{\&Sigma;}}{v}_{\mathrm{yi}}{f}_{\mathrm{CCC}}(x,y)-D_e(\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xj}}-R_x)-F_e{T}_{\mathrm{xi}}-G_{e}i$

Wherein u _xirepresent the neuron output v with NCNN _xicorresponding input.

[0032] adopt Euler's method, obtain the NCNN discrete time model of can software emulation realizing:

Formula (6):

$u_{\mathrm{xi}}(t+1)={\mathrm{ku}}_{\mathrm{xi}}\left(t\right)-A_a\mathrm{\&alpha;}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xj}}\left(t\right)f_{\mathrm{CSC}}(i,j)-B_e\mathrm{\&alpha;}\underset{y\&NotEqual;x}{\underset{y\&Element;\mathrm{Near}}{\mathrm{\&Sigma;}}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{yj}}\left(t\right)F_{\mathrm{ACC}}(i,j)$

$-C_e\mathrm{\&alpha;}\underset{y\&NotEqual;x}{\mathrm{\&Sigma;}}{v}_{\mathrm{yi}}\left(t\right)f_{\mathrm{CCC}}(x,y)-D_e\mathrm{\&alpha;}(\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xj}}\left(t\right)-R_x)-\mathrm{\&alpha;}{F}_e{T}_{\mathrm{xi}}-\mathrm{\&alpha;}{G}_{e}i$

$-z_{\mathrm{xi}}\left(t\right)[v_{\mathrm{xi}}\left(t\right)-I_0]+n_{\mathrm{xi}}\left(t\right)$

Formula (7):

$v_{\mathrm{xi}}=\frac{1}{1+\exp (-{\mathrm{\&mu;}}_0{u}_{\mathrm{xi}})}$

Formula (8):

z _xi(t)＝(1-β ₁)z _xi(t-1)

Formula (9):

A _m[n _xi(t+1)]＝(1-β ₂)A _m[n _xi(t)]

T in formula (6) represents the time; The neuron activation functions that formula (7) is NCNN, μ ₀it is sharpness factor; Formula (8) is neuron self feed back attenuation function, z _xi(t) be the self feed back connection weight of neuron xi, wherein z _xi(t)>=0, β ₁z _xi(t) decay factor, and 0 < β ₁< 1, the z of decay _xi(t) for generation of transient chaos; Formula (9) is the noise attentuation function of NCNN, n _xi(t) be the input that is injected into neuron xi, obey equally distributed random noise, m _xi(t) actual value excursion is [A _m, A _m], A _mnoise n _xi(t) amplitude; β ₂noise amplitude A _mdecay factor, 0 < β ₂< 1.

6 punishment parameters in formula are set to: A _e=2, B _e=3, C _e=2.5, D _e=8, F _e=5, G _e=0.05; Other parameters, according to general NCNN model, are set to: k=0.95, α=0.12, β ₁=β ₂=0.06, μ ₀=0.55, z _xi(0)=0.87, I ₀=0.65, A _m=2; Maximum iteration time is set to 1000.

About 6 punishment parameter A in formula (1) _e, B _e, C _e, D _e, F _e, G _erule is set, because last of formula (1) can not converge to zero, so can first suppose G _e=0, the initial value of all the other 5 parameters is all set to 1, i.e. A _e=B _e=C _e=D _e=F _e=1, then according to the performance of NCNN, increase respectively or reduce parameter A _e, B _e, C _e, D _e, F _evalue.For example, suppose end of run of NCNN, find that the value of Section 1 is larger than other values of several, and the increase of the iteration step length of NCNN, even converge to a trivial solution (iterations exceedes 1000), increase accordingly A _evalue, otherwise, A _eremain unchanged or reduce.Adopt and use the same method parameter A _e, B _e, C _e, D _e, F _evalue set, finally parameters G again _evalue.Because last value can not converge to zero, for fear of NCNN, converge to a trivial solution, first suppose an optimal solution arbitrarily, and then investigate the cost function value of the adjacent states of this optimal solution, require the cost function value of adjacent states must be greater than the value of optimal solution.Method is to be from 1 to 0 by any one neuron output modifications of optimal solution, and the first three items of formula (1) and the Section 5 null value that all can preserve value is constant like this, but the value of Section 4 will increase D _e/ 2, and last value will reduce G _ei (i=1,2 ..., 69,70), in order to suppress the trivial solution of adjacent states, just necessarily require the recruitment of the value of Section 4, be greater than the reduction of last value, that is: formula (10):

$\frac{D_e}{2}>G_{e}i$

Because the maximum possible value of i in formula (10) is i=70, so require D _e> 140G _e., Here it is, and this example parameter is set to D _e=8, G _e=0.05 reason.

In addition, because the neuron output of NCNN is continuous, rather than discrete (binary system), how by the continuous neuron output of NCNN, being converted to discrete binary system output, this problem, is the stopping criterion for iteration problem of a NCNN in fact.The method that this example adopts is, by the matrix V that 49 row 70 are listed as, represent the neuron output of a honeycomb subnet 49 cellular cell 70 channels, in the running of NCNN, every single-step iteration finishes, all check all row of V, for example x is capable, and stopping criterion for iteration is, choose x capable in maximum R _xindividual element, by this R _xindividual element is all set to 1, and remaining element is all set to 0, if now the value of the 1st, 2,3,5 of formula (1) all equals 0, and NCNN reruns more than 10 steps, last value does not also decline again, thinks that NCNN has found an optimal solution, and iteration stops.

Based on above-mentioned parameter setting, and the interference channel table of Fig. 2, this example distributes 5 channel (R to each cellular cell of honeycomb subnet E _x=5, x=1,2 ..., 48,49), as shown in Figure 3, black lattice represents that channel allocation is to respective cells to corresponding channel assignment table, blank grid represents unallocated.

By comparison diagram 2 and Fig. 3, can find out, all interference channels in Fig. 2 have all been avoided in Fig. 3, be black lattices all in Fig. 2, same position place in Fig. 3, is all blank grid, that is to say that the grid of same position in Fig. 2 and Fig. 3 is not all black simultaneously; The channel that in Fig. 3, arbitrary cellular cell is distributed, has at least kept 3 frequency intervals (CSC); The channel that any two adjacent cells are distributed, has at least kept 2 frequency intervals (ACC); Any two have been assigned with the cellular cell of same channel, and their space length is all more than or equal to reuse distance D _reuse(CCC).

In Fig. 3, the channel of all distribution has all shunk to the channel of low numbering, and this has benefited from cost function last, and along with the increase of frequency numbering, the number of times that channel is assigned with reduces gradually.For example: the 64th and No. 69 channels be only assigned with once, the 62nd, 63,67,68, No. 70 channel did not once also distribute, in idle condition.Whole Cellular Networks total number of channel used reduces, and means that the availability of frequency spectrum has improved.

If cancel channel shortening item, last of cost function punishment parameter is set to: G _e=0, other parameters remain unchanged, and as shown in Figure 4, the channel distributing has still been avoided the interference channel in Fig. 2 to corresponding channel assignment table, and has met CSC, ACC, tri-constraintss of CCC.But the number of times that all channels are assigned with is all roughly equal, Here it is does not have the result of channel shortening.

If only consider CCC bound term, first and second punishment parameter of formula (1) is set to: A _e=0, B _e=0, all the other parameters remain unchanged, corresponding channel assignment table as shown in Figure 5, the cellular cell of all distribution same channels, its space length is all more than or equal to reuse distance D _reuse, but violate CSC and ACC constraints (CCC).For example: the 17th, 18 and No. 19 channel has been distributed to cellular cell No. 44 simultaneously, has violated CSC constraint, and No. 19, No. 20 channel has been assigned to respectively cellular cell No. 48, No. 49, has violated ACC constraint.

In order to further illustrate the DCA method of the present invention for cellular mass net, then the cellular mass net shown in Fig. 1 is carried out to system emulation, further to verify the validity of DCA algorithm proposed by the invention.CCC bound term is only considered in system emulation, for ease of performance evaluation explanation, and DCA method of the present invention, independent operating DCA program in each honeycomb subnet, referred to herein as DCA-DCS algorithm; Traditional DCA algorithm, to whole cellular mass network operation DCA program, is referred to as DCA-LCN algorithm.All cellular cells are assumed that to have the identical traffic, and every cellular cell average traffic is ρ=λ/μ (Ireland), μ=180, and emulation value is set as: 6,6.5,7,7.5,8,8.5,9,9.5,10 (Ireland), the calls λ of the corresponding average initiation per hour in every cellular cell is: 120,130,140,150,160,170,180,190,200 (number of calls/hour).DCA-DCS and DCA-LCN program are respectively moved 10 times, then get the mean value of blocking rate, obtain blocking rate performance curve as shown in Figure 6, and carry out Performance Ratio.

In Fig. 6, for convenient function comparison, the performance curve of FCA is drawn out according to the following allusion quotation Erlang's formula of warp: formula (11):

$P_b=\frac{A^N/N!}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{A}^i/i!}$

In formula (11), P _brepresent blocking rate, N represents total number of available channels, and A represents the traffic (unit: Ireland).

As can be seen from Figure 6, two DCA curves have just started almost to overlap, and along with the increase of the traffic, when the traffic reaches 7 Ireland when above, two DCA curves separate gradually.When average traffic exceedes approximately 9.4 Ireland, the blocking rate of DCA-LCN has even exceeded FCA.Curve D CA-DCS is when the traffic increases, larger with the distance of DCA-LCN, and the superiority of DCA-DCS algorithm has been described thus.Visible, DCA method provided by the invention (DCA-DCS) is better than traditional DCA method (DCA-LCN) in performance.

Claims (4)

1. a cellular mass mobile communication system dynamic channel assignment method, is characterized in that: it comprises the following steps:

A), cellular mass net is divided into multiple small-scale honeycomb subnets (DCS), the size of each honeycomb subnet arranges flexibly according to the performance of dynamic channel allocation, and for each honeycomb subnet, an independently honeycomb subnet control centre is set, be responsible for dynamic channel allocation (DCA) computing of the cell mobile communication systems of this honeycomb subnet;

B), for each honeycomb subnet, set up a real-time interference channel table, this interference channel table, is to set up according to the channel allocation information of all adjacent cell subnets of each honeycomb subnet;

C), according to the channel demands of each cellular cell in the interference channel table of each honeycomb subnet, honeycomb subnet, and the channel allocation constraints between each cellular cell in each honeycomb subnet, for a cost function of each honeycomb subnet structure, and adopt noise chaotic neural network (NCNN) method, make the value minimum of this cost function; Cost function is expressed as follows: formula (1):

$\begin{array}{c}E=\frac{A_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{v}_{\mathrm{xj}}{f}_{\mathrm{CSC}}(i,j)+\frac{B_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{\underset{y\&NotEqual;x}{y}\&Element;\mathrm{Near}}{\mathrm{\&Sigma;}}\underset{j\&NotEqual;i}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{v}_{\mathrm{yj}}{f}_{\mathrm{ACC}}(i,j)\\ +\frac{C_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{y\&NotEqual;x}{\mathrm{\&Sigma;}}{v}_{\mathrm{xi}}{y}_{\mathrm{yi}}{f}_{\mathrm{CCC}}(x,y)+\frac{D_e}{2}\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}{(\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xi}}-R_x)}^2\\ +F_e\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{v}_{\mathrm{xi}}{T}_{\mathrm{xi}}+G_e\underset{x=1}{\overset{m}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{iv}}_{\mathrm{xi}}\end{array};$

D), in any one honeycomb subnet, if there is new calling to arrive, need to increase the number of channel, all adjacent cell subnets are inquired about by this honeycomb subnet control centre, if now there is certain adjacent cell subnet carrying out DCA program, wait for, until all adjacent cell subnets are not all carried out DCA program, scan the channel assignment table of all adjacent cell subnets, and generate the interference channel table of this honeycomb subnet;

E), according to the channel demands number of each honeycomb subnet, and the interference channel table of each honeycomb subnet, utilize NCNN method, the value of the 6th of the cost function of each honeycomb subnet is minimized, make the value of the 1st, 2,3,4,5 of cost functions all equal zero simultaneously;

In formula (1), each symbol definition is: m represents a cellular cell sum in honeycomb subnet, n represents an available channel sum in honeycomb subnet, x and y represent cellular cell numberings different in a honeycomb subnet, and i and j represent available channel numberings different in a honeycomb subnet; | x-y| represents the space length of interior two the cellular cell x of a honeycomb subnet and y; | i-j| represents that two different channels i and j are at the interval of frequency domain position, v _xirepresent the neuron output of NCNN, if channel i has distributed to cellular cell x, v _xi=1, otherwise v _xi=0, v _xj, v _yi, v _yjdefinition and v _xidefinition similar, six constant A _e, B _e, C _e, D _e, F _eand G _ebe corresponding every punishment parameter, they,, according to the constringency performance of NCNN, regulate respectively;

The 1st of formula (1) it is same constraint based on sites (CSC:Co-Site Constraint), in a honeycomb subnet, CSC requires all channels that are assigned in same cellular cell, all will keep a minimum channel spacing, otherwise can produce serious inter-carrier interference; L represents the interval of channel, with constraint based on sites function definition, is: formula (2):

$f_{\mathrm{CSC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<L\\ 0,|i-j|\&GreaterEqual;L\end{array}\right.$

When the interval of two channel i and j is less than L, constraint function equals 1, otherwise equals zero, if CSC is met, the value of the 1st will equal zero, product term v _xiv _xjrepresent that channel i and j have been distributed to cellular cell x simultaneously, and if only if constraint function f _cSC(i, j)=0 o'clock, the value of the 1st just can equal zero;

In formula (1) the 2nd shi Lin road constraint (ACC), ACC requires adjacent cells can not distribute adjacent frequency, otherwise can mutually produce serious inter-carrier interference, faces constraint function and is defined as: formula (3):

$f_{\mathrm{ACC}}(i,j)=\left\{\begin{array}{c}1,|i-j|<2\\ 0,|i-j|\&GreaterEqual;2\end{array}\right]$

When the interval of two adjacent channels is less than 2, ACC constraint function equals 1, otherwise equals 0, if ACC constraints is met, the value of the 2nd will equal zero; Product term v _xiv _yjrepresent that channel i and j are assigned to respectively cellular cell x and y simultaneously, and if only if constraint function f _aCC(i _{, j})=0 o'clock, the value of the 2nd just can equal zero, and in the summation symbol of the 2nd, symbol Near represents the adjacent cells set of cellulor district x;

In formula (1) the 3rd

that people having a common goal retrains (CCC), CCC requires any two cellular cells that are assigned with same frequency, on locus, must keep at a certain distance away, otherwise can produce co-channel interference, people having a common goal's constraint (CCC) function definition is: formula (4):

$f_{\mathrm{CCC}}(x,y)=\left\{\begin{array}{c}1,|x-y|<D_{\mathrm{reuse}}\\ 0,|x-y|\&GreaterEqual;D_{\mathrm{reuse}}\end{array}\right.$

Wherein D _reuserepresent channel reuse distance, if the distance of cellular cell x and y is less than reuse distance, functional value equals 1, otherwise equals 0; If CCC constraints is met, the value of the 3rd will equal zero; Product term v _xiv _yirepresent that channel i has been distributed to cellular cell x and cellular cell y simultaneously, and if only if constraint function f _cCC(x, y)=0 o'clock, the value of the 3rd just can equal zero.

2. a kind of cellular mass mobile communication system dynamic channel assignment method according to claim 1, is characterized in that:

In formula (1) the 4th

the channel demands bound term of cellular cell, the number of channel of and if only if each cellular cell of distributing to, while just equaling the required number of channel in each cellular cell, the 4th just can equal 0; R _xrepresent the channel demands number of cellular cell x, the channel demands of all cellular cells in whole honeycomb subnet, by a channel demands, vectorial R represents, the channel demands of cellular cell x is counted R _x, be x the element of channel demands vector R.

3. a kind of cellular mass mobile communication system dynamic channel assignment method according to claim 1, is characterized in that:

In formula (1) the 5th

the interference channel bound term of honeycomb subnet, T _xixi the element of a honeycomb subnet interference channel table T, if channel i is the interference channel of forbidding to cellular cell x, T _xi=1, otherwise T _xi=0; Product term v _xit _xirepresent, if v _xi=0 or T _xi=0, product v _xit _xi=0, if channel i is forbidden channel for cellular cell x, only have v _xi=0 just can make the value of the 5th equal 0; Only have the interference channel constraints that has all met this honeycomb subnet when all cellular cells, the value of the 5th just can equal 0.

4. a kind of cellular mass mobile communication system dynamic channel assignment method according to claim 1, is characterized in that: in formula (1) the 6th be channel shortening item, in the situation that meeting each cellular cell channel demand, if can adopt the number of channel still less, can improve the availability of frequency spectrum of each honeycomb subnet and whole cellular mass net; Product term iv _xirepresent channel i and neuron output v _xiproduct, if it is lower to distribute to the numbering of channel i of cellular cell x, product term iv _xivalue just less, the number of channel that is assigned to all cellular cells is also just fewer, thereby the availability of frequency spectrum is also just higher; Different from other several is, last value also can not equal 0, but one be greater than 0 value, and this value is unpredictable, this is because last value, its size not only will be subject to the impact of a honeycomb subnet cellular cell sum, but also will be subject to current the distributed number of channel, channel number and punishment parameter G _eimpact, and current the distributed number of channel and channel number, all uncertain, if the value of the 1st, 2,3,4,5 of formula (1) all equals zero, the 6th is one and is greater than zero value, even if NCNN reruns down, last value also can not reduce again, thinks that NCNN has found an optimal solution.

Images