CN102833759B - Cognitive radio spectrum allocation method enabling OFDM (orthogonal frequency division multiplexing) master user to realize maximum revenue - Google Patents

Abstract

The invention discloses a cognitive radio spectrum allocation method enabling an OFDM (orthogonal frequency division multiplexing) master user to realize maximum revenue and belongs to the field related to cognitive radio spectrum resource management in wireless communication. The method is characterized in that the master user serving as a manager of spectrum is used for information interaction with a secondary user, taking specificity of the OFDM multi-carrier technology into consideration, inter-carrier interferences and requirements on error rate of users are both taken into consideration during establishment of a target function, and rational spectrum price and access regulations are designed to standardize the system operational process. By the cognitive radio spectrum allocation method enabling the OFDM master user to realize the maximum revenue, real-time selection of using spectrum of the secondary user in spectrum market of multiple secondary users can be realized by the master user, so that revenue of the master user can be increased.

Description

Make the maximized cognitive radio spectrum allocation method of OFDM revenue of primary user

Technical field

The present invention relates to and a kind ofly make the maximized cognitive radio spectrum allocation method of revenue of primary user based on OFDM, belong to cognitive radio frequency spectrum resource management association area in radio communication.

Technical background

Cognitive radio (Cognitive Radio) is a kind of spectrum-sharing techniques of intelligence, it can rely on the support of artificial intelligence, the extraneous wireless communications environment of perception, according to certain study and decision making algorithm, real-time adaptive ground changes system operational parameters (as through-put power, carrier frequency, modulation system), allow the spectrum reuse carrying out multidimensional in time, space and frequency in theory, thus greatly the restriction of reduction frequency spectrum and bandwidth, to the constraint of wireless technology development, is following the most popular wireless technology by prophesy.OFDM, as one of hot technology realizing high speed radio transmission, just proposes beginning because its flexibility and adaptivity are considered to the optimal selection of cognitive radio transmission technology in cognitive radio concept.

Due to idling-resource limited frequency spectrum, competition is needed to use frequency spectrum resource between secondary user, and the priority of different secondary users, qos requirement may be different, the main purpose of spectrum allocation may is exactly the spectrum requirement different according to time user, the fair frequency spectrum resource effectively distributing some, makes systematic function improve or approaches optimum state.The cognitive radio frequency spectrum method of salary distribution of primary and secondary user cooperative effectively can be avoided interference by a small amount of information interchange between user, ensures data transmission quality.

In real-time frequency spectrum trade market, primary user has the right to obtain income from the process of selling oneself idle frequency spectrum, and by formulating corresponding frequency spectrum " lease " rule, makes interests reach maximization in long-term frequency spectrum use procedure.Based on the cognitive radio system of primary user's independent behaviour, to drive as starting point with user's individual interests, with the research of primary user under different condition and time user behavior for foundation, ensure that user's income is coordinated not conflict mutually with rationally perfect cognitive radio frequency spectrum access laws and decrees, and give full play to the independence of user behavior decision-making, primary user needs the optimizing decision making rapidly spectrum allocation may over each slot, namely at wireless environment, frequency spectrum policy, under the constraints of interference-limited, the limited user adapted with idle frequency spectrum quantity is selected from multiple potential secondary user, its access is allowed to use idle frequency spectrum.In cognitive radio system, because the signal of different sub carrier is from different users, is difficult to ensure the consistent or exact quadrature of the parameter of signal, therefore needs mutually to disturb between suppressed carrier.Secondly, also will consider that time user's bit error rate requirement and channel fading factor are on the impact of system, and comprehensive each factor is improved to model, maximum throughput can be reached while guarantee time user does not cause interference to neighboring user.

Ceaselessly fruit machine (Restless Bandits) is the theory studying a series of random sequential decision problem.Refer to series of discrete or continuous print moment point makes a policy, system will obtain the return taking decision-making relevant with status and institute, and the state of influential system residing for next decision-making moment point, and affects following finally benefited with this.It can be solved by the priority index value (priority-index) of each project, and index value can be relaxed (Linear Programming Relaxation) by linear programming and original duplicate key exploratory method (Primal-dual Index Heuristic) is tried to achieve.Priority index value can be stored in concordance list (index table) in off-line (off-line) state computation, thus can reduce online (on-line) operand significantly and implement complexity.List of references P.Whittle, " Restless bandits:Activity allocation in a changing world " .A Celebration of Applied Probability.1988 (25): 287-298, with D.Berstimas and J.Ni ~ no-Mora. " Restless bandits; linear programming relaxations; and a primal dual index heuristic " .Operations Research.2000,48 (1): 80-90.

Summary of the invention

Main purpose of the present invention is in the angle of primary user, considers that multiple users compete the situation of limited spectrum resources, turns to target, build the real-time cognitive radio frequency spectrum distribution system based on OFDM so that revenue of primary user is maximum.To solve in real time spectrum market repeatedly the decision problem under user situation, and improve revenue of primary user by the behaviour decision making optimizing primary user.

The cognitive radio networks scene that the present invention should adapt to is shown in Fig. 1.

System operation schematic diagram in technical solution of the present invention is shown in Fig. 2.

The technical scheme of the maximized cognitive radio spectrum allocation method of OFDM revenue of primary user is made to be realize by following flow process on PC:

Step (1), at least one potential secondary user base station sends the use solicited message of idle frequency spectrum to a primary user base station, wherein at least comprise: respective buffer data size, average data arrival rate λ, geographical position and bit error rate requirement, described user base station and primary user base station are hereinafter referred to as secondary user and primary user;

Step (2), the frequency spectrum that primary user collects time user described in each according to the following steps uses request:

Step (2.1), at time slot t, primary user for state sets up secondary User Status set of a t time slot, is called for short system mode vector, use with the data buffer storage amount of time user described in each under time slot t represent:

S _gexpression state is the secondary user of g, state S _gt () represents,

S ₀t () represents that buffer data size is 0, increase progressively successively,

S _gt () represents that desired data buffer memory has exceeded the storage capacity of this user data cache amount, in the urgent need to frequency band transmission data to guarantee to run well, just preparing to turn to other primary user's network,

G+1 is state sum,

Step (2.2), correspond to the different conditions of time user, primary user arranges every Mb data traffic price and State Price, and along with state increases progressively, State Price by the equal increments of setting, thus sets up following T time slot total revenue R:

$R=\underset{t=0}{\overset{T-1}{\mathrm{\Σ}}}(R_v^a\left(t\right)+R_s^a\left(t\right)),$

for time slot t unit data flow price, being set point, is time-varying,

for time slot t State Price, being set point, is time-varying,

The behavioral unity of above-mentioned two kinds of prices that secondary user occurs is expressed as a _nt ()=1 represents that the idle frequency spectrum that time user takies primary user at time slot t carries out transfer of data, a _nt ()=0 represents that time user does not take the idle frequency spectrum of primary user,

Step (2.3), the frequency spectrum leasing agreement Spectrum Lease Protocol that the above-mentioned information in step (2.2) is formed a short-term by primary user is called for short SLP, is broadcast to all described potential users in its coverage;

Step (3), primary user carries out frequency spectrum leasing preparation according to the following steps:

Step (3.1), if described time potential user does not accept the frequency spectrum leasing agreement SLP that primary user is arranged in step (2.3), then turns to the network at other primary user place, if accept, then sends to described primary user and accepts message,

Step (3.2), primary user is according to accepting the secondary number of users N of described frequency spectrum leasing agreement and primary user M the idle frequency band when time slot t, and calculate the preferential index value table from M idle frequency band selected in N number of user, M≤N, step is as follows:

Step (3.2.1), the state transition probability matrix when behavior that primary user calculates time user n is according to the following steps a

wherein:

P _eEfor current time slots t user n buffer data size is continuously the probability of 0,

P _bKfor current time slots t user n turn to other primary user's network because buffer data size exceedes cache threshold after because buffer data size reduces get back to again the probability of current primary user's network, be the estimated value of setting,

P _oTfor current time slots t user n turns to the probability after other primary user's network still not in current primary user's network, be the estimated value of setting,

p _OT+p _BK=1，

Secondary user n, when not accessing idle frequency spectrum a (t)=0, forwards the probability of state g ' to from state g be expressed as:

$P_{{\mathrm{gg}}^{\′}}^0=P\{({\mathrm{\ϵ}}_{g^{\′}-1}-\left(\frac{{\mathrm{\ϵ}}_g+{\mathrm{\ϵ}}_{g-1}}{2}\right))\≤X\≤({\mathrm{\ϵ}}_{g^{\′}}-\left(\frac{{\mathrm{\ϵ}}_g+{\mathrm{\ϵ}}_{g-1}}{2}\right))\},$

Secondary user n, when accessing idle frequency spectrum a (t)=1, forwards the probability of state g ' to from state g be expressed as:

$P_{{\mathrm{gg}}^{\′}}^1=P\{({\mathrm{\ϵ}}_{g^{\′}-1}-\left(\frac{{\mathrm{\ϵ}}_g+{\mathrm{\ϵ}}_{g-1}}{2}\right)+v_n)\≤X\≤({\mathrm{\ϵ}}_{g^{\′}}-\left(\frac{{\mathrm{\ϵ}}_g+{\mathrm{\ϵ}}_{g-1}}{2}\right)+v_n)\},$

ε _gstate S _gt the upper limit of () is set point,

g=1,2,…,g,…G,g′=1,2,…,g′,…G,g≠g′，

X is a positive integer k, at k=0, and 1,2 ..., value between K, K is a limited positive integer, is set point,

λ is time user's average data arrival rate,

V _nthe transmission rate that time user can reach,

$v_n=\mathrm{\Δf}\·{\log }_2(1+{\mathrm{\τ}}_n\·\frac{h_n{W}_n}{{\mathrm{\σ}}^2+\mathrm{\Σ}{\stackrel{\·}{I}}_{i^{\′}{j}^{\′}}}),$ Wherein:

i′=1,2,…,i′…,I,I=M,j′=1,2,…,j′…,J,J=M,i′≠j′，

The sequence number of the subcarrier of two users that i ', j ' be respectively are selected from described N number of user, data equal the sequence number of two idle frequency bands used,

Δ f is adjacent two subcarrier i ', and the frequency interval of j ' is set point,

τ _nbe a constant, with the error rate BER of secondary user n _nrelevant, for additive white Gaussian noise channel, τ _nwith BER _nbetween meet:

${\mathrm{\τ}}_n=\frac{1.5}{-\ln \left({5\mathrm{BER}}_n\right)},$

σ ²for the one-sided power spectrum density of additive white Gaussian noise, be given value,

H _nfor the channel gain of secondary user n, be given value,

W _ntotal transmitted power of each user n,

for secondary user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ', be expressed as

${\stackrel{\·}{I}}_{i^{\′}{j}^{\′}}(d_{i^{\′}{j}^{\′}},W_{i^{\′}})={\left|h_{i^{\′}{j}^{\′}}\right|}^2{W}_{i^{\′}}{T}_s{\∫}_{d_{i^{\′}{j}^{\′}}-\mathrm{\Δf}/2}^{d_{i^{\′}{j}^{\′}}+\mathrm{\Δf}/2}{\left(\frac{\sin \mathrm{\π}{f}_{i^{\′}}{T}_s}{\mathrm{\π}{f}_{i^{\′}}{T}_s}\right)}^2{\mathrm{df}}_{i^{\′}},$ Wherein:

H _{i ' j '}the channel gain of subcarrier i ' on subcarrier j ',

W _{i '}the secondary user n on subcarrier i ' _itotal transmitted power, W _{i '}=W _n,

T _sbeing OFDM symbol duration, is given value

D _{i ' j '}being the frequency distance between two subcarrier i ', j ', is given value

F _{i '}time user n _ithe frequency of place subcarrier i ',

Step (3.2.2), determine to make the maximized optimization object function of revenue of primary user by following formula:

Meet. ${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}{j}^{\&prime;}})<{\stackrel{\&CenterDot;}{I}}_{\mathrm{spec}}$

${\mathrm{BER}}_{n_j}<{\mathrm{BER}}_{n_j\mathrm{spec}}$ , wherein:

$\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{a}_{{\mathrm{ni}}^{\&prime;}}\left(t\right)\&le;1$

i′=1,2,…,i′…M,j′=1,2,…,j′…M,i′≠j′

Discount factor Beta β represents, guarantees the bounded of total discount income and restrains, and β is value between 0< β <1,

I ', j ' represent idle sub-carrier frequency, be given value,

be T × N matrix, row is t, represents time slot, and row are behavior a of time user _n(t),

time user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ' higher limit, be set point,

for secondary user n _jthe upper limit threshold of maximum bit error rate is set point,

A _{ni '}t () represents the secondary user n of t time slot on subcarrier i ' _ibehavior, represent on subcarrier i ', an a can only be had at most at time slot t _{ni '}any active ues of (t)=1,

Step (3.2.3), solves primary user maximum return R with ceaselessly fruit machine Restless Bandits ^*, according to state transition probability matrix primary user's idle frequency band number M, secondary user number N, discount factor Beta=β, secondary User Status number StateNumber=G+1, Alpha=zeros (N, G+1) for complete zero N × (G+1) rank matrix, lower same, R0=zeros (N, G+1), R1=ones (N, G+1) for complete 1 N × (G+1) rank matrix, P0=zeros (G+1, G+1, N), P1=P0, utilize index value function f unction [Delta]=calc_index (M, N, Beta, StateNumber, Alpha, R0, R1, P0, P1) the index value Delta in order to state use order of the idle frequency spectrum corresponding to each user is calculated, find out the secondary user corresponding to minimum index value, using this minimum index value as preferential index value, the user priority that preferential index value is corresponding uses idle frequency band, construct the enquiry form of each user index value Delta, and all stored in the database of primary user,

Step (4), primary user distributes frequency spectrum according to the following steps:

State vector { the s that step (4.1) primary user is current according to each user ₁, s ₂..., s _nenquiry form, index value is sorted from small to large, and uses corresponding idle frequency band information successively to described N number of user transmission successively, comprise access slot,

Step (4.2), if obtain allowing a _nt ()=1, accesses frequency band next time slot time user and transmits data, the normal operation of primary user base station monitoring whole system,

Step (4.3), at the end of time slot, carries out as follows,

Step (4.3.1), secondary user adds up the buffer memory of oneself and the information of previous time slot transmission data, if still there are data to need transmission and data volume ε _cachedo not exceed threshold value, then cache information is sent to primary user, 0< ε _cache≤ ε _spec,

Step (4.3.2), primary user is vectorial according to the system mode of all N number of the users of the information updating of secondary user,

Step (4.4), repeat (4.1) step to (4.3) step, until primary user does not have idle frequency spectrum to hire out, primary user sends and stops notify and regain the right to use of frequency spectrum, calculates rent that each user should pay and sends the information of rent confirmation.

Advantage of the present invention is the income of primary user in cognitive radio system to be divided into state income and flow return two parts, and secondary user pays rent according to the actual service condition of oneself.The selection course modeling of in real-time system user is become Restless Bandits model, the income of primary user can be made to improve on the one hand, in system operation, the calculating of index value can run through multiple time slot on the other hand, primary user only needs to table look-up when each time slot decision-making, improves operational efficiency.

Accompanying drawing explanation

Fig. 1, the maximized cognitive radio spectrum allocation method system figure of the revenue of primary user based on OFDM.

Fig. 2, the maximized cognitive radio spectrum allocation method schematic diagram of the revenue of primary user based on OFDM.

Fig. 3, the model algorithm schematic diagram in embodiment during disturbance restriction, in figure restless Bandits algorithm values, greedy algorithm value, random algorithm financial value.

Fig. 4, the asynchronous model algorithm schematic diagram of available band quantity in embodiment, in figure restless Bandits algorithm values, greedy algorithm value, random algorithm financial value.

Fig. 5, model algorithm schematic diagram when time number of users is available band quantity twice in embodiment, in figure restless Bandits algorithm values, greedy algorithm value, random algorithm financial value.

Embodiment

Step (1), potential secondary user, when producing spectrum requirement, sends the use request of idle frequency spectrum, comprises the information of their buffer data size, average data arrival rate λ, geographical position, bit error rate requirement in information to primary user;

Step (2), the frequency spectrum that primary user collects described in each uses request, carries out as follows,

Step (2.1), primary user determines the criteria for classifying of state, and the state of system is made up of the state of potential secondary user, can by how many decisions of its data buffer storage amount in the state of time slot t user n, the state set of secondary user, is called for short system mode vector available represent, G represents when to have data in the data buffer area of secondary user the status number that can divide, G+1 is the sum of state.State S ₀t () represents that buffer data size is state S _g(t) represent required buffer memory more than Liao Ci user data cache district ability to bear ( threshold value for secondary user n data buffer area), in the urgent need to frequency band transmission data to ensure that it runs well, now this user has turned to the network of other primary user.According to actual conditions, secondary user can send cache information at each time slot and primary user, s alternately _nthe state S of time user n at time slot t _n(t):

$s_n=\left\{\begin{array}{cc}{S}_0\left(t\right)& {\mathrm{\&epsiv;}}^n=0\\ S_1\left(t\right)& 0<{\mathrm{\&epsiv;}}^n<{\mathrm{\&epsiv;}}_1^n\\ .& \\ .& \\ .& \\ S_g\left(t\right)& {\mathrm{\&epsiv;}}_{g-1}^n<{\mathrm{\&epsiv;}}^n<{\mathrm{\&epsiv;}}_g^n\\ .& \\ .& \\ .& \\ S_G\left(t\right)& {\mathrm{\&epsiv;}}^n>{\mathrm{\&epsiv;}}_{\mathrm{spec}}^n\end{array}\right.$

Step (2.2), primary user arranges corresponding rent, and establish system benefit function, price of spectrum is made up of two parts: data traffic price and State Price, and the total revenue R of T time slot is expressed as:

$R=\underset{t=0}{\overset{T-1}{\mathrm{\&Sigma;}}}(R_v^a\left(t\right)+R_s^a\left(t\right)),$

Secondary user n can be expressed as in the behavior of time slot t a _nt ()=1 represents that time user n enlivens at time slot t, take idle frequency spectrum and carry out transfer of data, a _nt ()=0 represents that time user does not take idle frequency spectrum, that t primary user is to all user behavior a in network _nt flow rent that () collects, the price of single Subscriber Unit data traffic is that t primary user is to all user behavior a in network _nt state rent that () collects, the set of single User Status price correspond to the state set of time user

Step (2.3), the frequency spectrum leasing agreement Spectrum Lease Protocol of above-mentioned information i.e. short-term is called for short SLP by primary user, is broadcast to all described potential users in its coverage;

Step (3), completes the preparation before frequency spectrum leasing according to the following procedure,

Step (3.1), if described time potential user does not accept the access rules that primary user is arranged, then turns to the network of other primary users, otherwise receives information to described primary user's transmission,

Step (3.2), primary user is according to accepting the secondary number of users N of described frequency spectrum leasing agreement and primary user M the idle frequency band when time slot t, and calculate the preferential index value table from M idle frequency band selected in N number of user, M≤N, step is as follows:

Step (3.2.1), asks state transition probability matrix, and the state transition probability matrix when behavior of secondary user n is a is defined as $P_n^a={\left[p_{{\mathrm{gg}}^{\&prime;}}\left(t\right)\right]}_{(G+1)\&times;(G+1)},$ Wherein

can be written as:

element depend on the behavior act a of time user _n(t), average data arrival rate λ, the transmission rate v that can reach during access frequency band _n, according to Poisson probability allocation matrix k=0,1,2 ..., X is the data volume that current time slots arrives buffer area.In OFDM transmission system, the transmission rate that secondary user n can reach represents with following formula,

$v_n=\mathrm{\&Delta;f}\&CenterDot;{\log }_2(1+{\mathrm{\&tau;}}_n\&CenterDot;\frac{h_n{W}_n}{{\mathrm{\&sigma;}}^2+\mathrm{\&Sigma;}{\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}}),$

Primary user estimates that time user's transmission rate calculates transition probability matrix, h _nthe channel gain of time user n, W _ntotal transmitted power of each user n, σ ²for the one-sided power spectrum density of additive white Gaussian noise, representative time user n _ithe interference that causes of other subcarrier of subcarrier pair, τ _nbe a constant, with the error rate BER of secondary user n _nrelevant, then additive Gaussian white noise channel (AWGN) is expressed as:

${\mathrm{\&tau;}}_n=\frac{1.5}{-\ln \left({5\mathrm{BER}}_n\right)},$

Be difficult in OFDM transmission system ensure the orthogonal of different user signal, need to consider interference effect, secondary user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ' be expressed as:

${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}})={\left|h_{i^{\&prime;}{j}^{\&prime;}}\right|}^2{W}_{i^{\&prime;}}{T}_s{\&Integral;}_{d_{i^{\&prime;}{j}^{\&prime;}}-\mathrm{\&Delta;f}/2}^{d_{i^{\&prime;}{j}^{\&prime;}}+\mathrm{\&Delta;f}/2}{\left(\frac{\sin \mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}{\mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}\right)}^2{\mathrm{df}}_{i^{\&prime;}},$

Wherein d _{i ' j '}the frequency distance between two subcarrier i ', j ', W _{i '}total transmitted power of the secondary user n on subcarrier i ', T _soFDM symbol duration, h _{i ' j '}be the channel gain of subcarrier i ' on subcarrier j ', Δ f is the frequency interval of adjacent sub-carrier, f _{i '}the frequency of time user n place subcarrier i ',

Step (3.2.2), determine the optimization object function that Restless Bandits model can solve:

Meet. ${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}{j}^{\&prime;}})<{\stackrel{\&CenterDot;}{I}}_{\mathrm{spec}}$

${\mathrm{BER}}_{n_j}<{\mathrm{BER}}_{n_j\mathrm{spec}}$

$\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{a}_{{\mathrm{ni}}^{\&prime;}}\left(t\right)\&le;1$

i′=1,2,…,i′…M,j′=1,2,…,j′…M,i′≠j′

N represents time user, i ', j ' represent free sub-band, and discount factor β guarantees the bounded of total discount income and restrains, and β is value between 0< β <1, represent primary user's all strategy sets taked when each time slot is adjudicated, wherein a kind of strategy be T × N rank matrix, wherein the capable n of t is classified as a _nt (), represents the behavior of time user n at time slot t, represent time user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ' higher limit must be less than represent time user n _jmaximum bit error rate can not exceed upper limit threshold represent can only there be at most an any active ues at time slot t on subcarrier i ',

Step (3.2.3), solves primary user maximum return R with ceaselessly fruit machine Restless Bandits ^*, according to state transition probability matrix primary user's idle frequency band number M, secondary user number N, discount factor Beta=β, secondary User Status number StateNumber=G+1, Alpha=zeros (N, G+1) for complete zero N × (G+1) rank matrix, lower same, R0=zeros (N, G+1), R1=ones (N, G+1) for complete 1 N × (G+1) rank matrix, P0=zeros (G+1, G+1, N), P1=P0, utilize index value function f unction [Delta]=calc_index (M, N, Beta, StateNumber, Alpha, R0, R1, P0, P1) the index value Delta in order to state use order of the idle frequency spectrum corresponding to each user is calculated, find out the secondary user corresponding to minimum index value, using this minimum index value as preferential index value, the user priority that preferential index value is corresponding uses idle frequency band, construct the enquiry form of each user index value Delta, and all stored in the database of primary user, step (4), primary user distributes frequency spectrum according to the following steps:

Step (4.1), the state vector { s that primary user is current according to each user ₁, s ₂..., s _nenquiry form, index value is sorted from small to large, and uses corresponding idle frequency band information successively to described N number of user transmission successively, comprise access slot,

Step (4.2), if obtain allowing a _nt ()=1, accesses frequency band next time slot time user and transmits data, the normal operation of primary user base station monitoring whole system,

Step (4.3), at the end of time slot, carries out as follows,

Step (4.3.1), secondary user adds up the buffer memory of oneself and the information of previous time slot transmission data, if still there are data to need transmission and data volume ε _cachedo not exceed threshold value, then cache information is sent to primary user, 0< ε _cache≤ ε _spec,

Step (4.3.2), primary user is vectorial according to the system mode of all N number of the users of the information updating of secondary user,

Step (4.4), repeat (4.1) step to (4.3) step, until primary user does not have idle frequency spectrum to hire out, primary user sends and stops notify and regain the right to use of frequency spectrum, calculates rent that each user should pay and sends the information of rent confirmation.

The embodiment of the maximized cognitive radio spectrum allocation method of revenue of primary user based on OFDM is described with example by reference to the accompanying drawings.

The first step, certainty annuity framework, the requirement of primary user and time user.

Fig. 1 is that the base station of primary user and time user is undertaken alternately by Common Control Channel, and the coverage of secondary user network relates to the network of multiple primary user based on the maximized cognitive radio spectrum allocation method of revenue of primary user of OFDM applicable system figure.

Suppose that the arrival rate of data to be sent meets Poisson distribution, the average data arrival rate of secondary user n is λ (n), represent that the arrival of time data volume of user in each time slot process meets the Poisson distribution that Mean Speed is λ (n), the symbol time Ts=4 μ s of OFDM, network subcarrier width Δ f=0.3125MHz.This external system white Gaussian noise variances sigma ², average channel fading gains h _n, secondary user to the bit error rate requirement of system, to the interference-limited I of intercarrier in system _specbe given value.System is current 6 idle frequency bands, has multiple users to apply for using frequency spectrum.

Second step, primary user determines the criteria for classifying and the price of spectrum of state, and obtains time user's agreement.

In this example, the state of time user is determined by data buffer storage amount.Secondary User Status is divided into according to the number of data buffer storage amount at time slot t 5 kinds, wherein S ₀t () represents that data volume is 0, secondary user does not need frequency spectrum; S ₁t () is to S ₃t the data volume of () increases gradually; S ₄t time user time () meets because can not obtain frequency spectrum and has turned to other primary user's network.State Price S is set ₁t the rent of () is 2, the every high one-level of state adds 2; Flow price is every Mb price is 2.

Decision-making is broadcast to time user by primary user, and whether secondary user gets involved this network according to the situation decision-making of oneself, is provided with 10 users and agrees to that access network uses frequency spectrum.

3rd step, primary user sets up Restless Bandits model according to Given information, the concordance list after calculating lease process starts, and the calculating of index can be carried out with the mutual of primary and secondary user simultaneously.

First calculate the transition probability matrix of time user, solve time user emission power higher limit according to channel conditions, interference requirement,

${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}})={\left|h_{i^{\&prime;}{j}^{\&prime;}}\right|}^2{W}_{i^{\&prime;}}{T}_s{\&Integral;}_{d_{i^{\&prime;}{j}^{\&prime;}}-\mathrm{\&Delta;f}/2}^{d_{i^{\&prime;}{j}^{\&prime;}}+\mathrm{\&Delta;f}/2}{\left(\frac{\sin \mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}{\mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}\right)}^2{\mathrm{df}}_{i^{\&prime;}},$

Bit error rate requirement formulae discovery each user according to secondary user uses the maximum data rate that can reach during frequency spectrum.

$v_n=\mathrm{\&Delta;f}\&CenterDot;{\log }_2(1+{\mathrm{\&tau;}}_n\&CenterDot;\frac{h_n{W}_n}{{\mathrm{\&sigma;}}^2+\mathrm{\&Sigma;}{\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}})$

Then calculate can be written as:

Wherein p _eEexpression is continuously the probability of 0 in current time slots time user cache data volume, can obtain relational expression p _bKrepresent time user because data cached exceeding after threshold value turns to other primary user's networks reduces because of data volume the probability getting back to again current primary user's network, p simultaneously _oTrepresent that time user turns at the probability of current time slots still not in current primary user's network after other primary user's networks, p _oT+ p _bX=1.Suppose state I _ithe upper bound be ε _i, the probability that secondary user is transformed into state g ' when not accessing frequency spectrum (a (t)=0) from state g can be expressed as:

$P_{{\mathrm{gg}}^{\&prime;}}^0=P\{({\mathrm{\&epsiv;}}_{g^{\&prime;}-1}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right))\&le;X\&le;({\mathrm{\&epsiv;}}_{g^{\&prime;}}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right))\},$

If a (t)=1, so

$P_{{\mathrm{gg}}^{\&prime;}}^1=P\{({\mathrm{\&epsiv;}}_{g^{\&prime;}-1}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right)+v_n)\&le;X\&le;({\mathrm{\&epsiv;}}_{g^{\&prime;}}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right)+v_n)\},$

Wherein Poisson probability allocation matrix $P(X=k)=e^{-\mathrm{\&lambda;}}\frac{{\mathrm{\&lambda;}}^k}{k!},k=\mathrm{0,1,2},...$

Next, establish the optimization object function that Restless Bandits model can solve, discount factor gets Beta=β=0.9,

Meet. ${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}{j}^{\&prime;}})<{\stackrel{\&CenterDot;}{I}}_{\mathrm{spec}}$

${\mathrm{BER}}_{n_j}<{\mathrm{BER}}_{n_j\mathrm{spec}}$

$\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{a}_{{\mathrm{ni}}^{\&prime;}}\left(t\right)\&le;1$

i′=1,2,…,i′…M,j′=1,2,…,j′…M,i′≠j′

N represents time user, i ', j ' represent free sub-band.M in time slot t in N number of user is active, and primary user obtains the rental income of any active ues then the state of secondary user according to state transition probability matrix is markoff process change.Other user pays rent and is its state transition probability matrix is income is given a discount in time, and wireless frequency spectrum assignment problem is converted into determines optimal policy make the rear income R of average folding ^*maximize.Index value function is utilized to calculate the index value of the idle frequency spectrum corresponding to each user, find out the secondary user corresponding to minimum index value, using this minimum index value as preferential index value, the user priority that preferential index value is corresponding uses idle frequency band, structure enquiry form, and all stored in the database of primary user;

4th step, starts real-time frequency spectrum leasing process.

State vector { the s that primary user is current according to each user ₁, s ₂..., s _nenquiry form, index value is sorted from small to large, and use corresponding idle frequency band information successively to described N number of user transmission successively, comprise access slot, the secondary user obtaining allowing transmits data in next time slot access frequency band, now primary user monitors the normal operation of whole system, can continue computation index table simultaneously; At the end of time slot, secondary user adds up the data volume that oneself buffer memory and previous time slot send, and needs transmission and data volume does not exceed threshold value still there being data, reports the data volume of buffer memory and transmission to primary user; Primary user is according to the information update system state vector of secondary user.

Said process repeats until primary user wishes to regain frequency spectrum or do not have secondary user to wish to continue to use frequency spectrum, then process of leasing stops, and secondary user pays rent according to agreement.

After the algorithm proposed the present invention is tested, acquired results can be monitored channel value environment more accurately.Model algorithm schematic diagram when Fig. 3 is disturbance restriction, Fig. 4 is the asynchronous model algorithm schematic diagram of available band quantity, the model algorithm schematic diagram that Fig. 5 is time number of users when being available band quantity twice.The present invention is compared with random method with greedy herein.

The present invention's the Realization of Simulation on PC uses Matlab language to programme.MATLAB is a kind of senior matrix language, comprises control statement, function, data structure, input and output and object based programming feature, is the set comprising a large amount of computational algorithm.It has the mathematical operation function will used in more than 600 engineering, can realize the various computing functions needed for user easily.The natural language of the maximized cognitive radio spectrum allocation method of the revenue of primary user based on OFDM embodiment program is as follows:

Start

The number arranging time user is 10, and sequence number 1 to 10, average data arrival rate is 1.2Mbps, and bit error rate requirement is 10 ^-7to 10 ^-3between random number;

The idle frequency spectrum number arranging primary user is 6, and data buffer storage is interval with 1.5M, arranges the relevant parameter of OFDM transmission and channel;

According in the 3rd step formula circulation solves the maximum transmission power of time user;

According to the v in the 3rd step _nthe data rate that equations time user is maximum;

The gain matrix corresponding with state and action is set, comprise flow price and State Price two-part and;

With between number of states, data arrival rate, state area, the secondary user maximum data rate that can reach, according in the 3rd step the state transition probability matrix of computational methods cycle calculations each user under different behaviour decision makings;

According to state transition probability matrix primary user's idle frequency band number M, secondary user number N, discount factor Beta=β, the N that secondary User Status number StateNumber=G+1, Alpha=zeros (N, G+1) are complete zero × (G+1) rank matrix, lower same, R0=zeros (N, G+1), the N that R1=ones (N, G+1) is complete 1 × (G+1) rank matrix, P0=zeros (G+1, G+1, N), P1=P0, utilizes index value computing function and function f unction [Delta]=calc_index (M, N, Beta, StateNumber, Alpha, R0, R1, P0, P1) calculate { s ₁, s ₂..., s ₁₀the all possible concordance list combining correspondence;

By the index that calculates and state vector { s ₁, s ₂..., s ₁₀corresponding in table;

The time arranging each time slot is 1s, emulates 1000 time slots altogether, starts to emulate real time spectrum lease process, arranges initial state vector { s ₁, s ₂..., s ₁₀}={ 1,1..., 1}, perform follow procedure at each time slot:

State corresponding to current each user is found out in circulation, determines current state vector;

Table look-up according to state vector and find out sequence number corresponding to 6 times minimum users of index under this state, be placed in active set;

Secondary user 1 is produced to the random number R and of 0 to 1, judge time user 1 whether in active set, if, income adds the income of last user 1 under current state and transmission rate, judges time state of user 1 at the end of current time slots with the random number R and produced according to the state transition probability matrix of during a (t)=1 user 1; If judge not in active set, income adds 0, judges time state of user 1 at the end of current time slots with Rand according to the state transition probability matrix of during a (t)=0 user 1;

Circulate and residue time user is carried out to the operation of single step above, obtain new state vector;

Obtain the primary user's total revenue added up at the end of 1000 time slots, program performs and is repeatedly averaging total revenue.

Terminate

Claims (1)

1. make the maximized cognitive radio spectrum allocation method of OFDM revenue of primary user, it is characterized in that, the Realization of Simulation according to the following steps on PC:

Step (1), at least one potential secondary user base station sends the use solicited message of idle frequency spectrum to a primary user base station, wherein at least comprise: respective buffer data size, average data arrival rate λ, geographical position and bit error rate requirement, described user base station and primary user base station are hereinafter referred to as secondary user and primary user;

Step (2), the frequency spectrum that primary user collects time user described in each according to the following steps uses request:

Step (2.1), at time slot t, primary user for state sets up secondary User Status set of a t time slot, is called for short system mode vector, use with the data buffer storage amount of time user described in each under time slot t represent:

S _gexpression state is the secondary user of g, state S _gt () represents,

S ₀t () represents that buffer data size is 0, increase progressively successively,

S _gt () represents that desired data buffer memory has exceeded the storage capacity of this user data cache amount, in the urgent need to frequency band transmission data to guarantee to run well, just preparing to turn to other primary user's network,

G+1 is state sum,

Step (2.2), correspond to the different conditions of time user, primary user arranges every Mb data traffic price and State Price, and along with state increases progressively, State Price by the equal increments of setting, thus sets up following T time slot total revenue R:

$R=\underset{t=0}{\overset{T-1}{\mathrm{\&Sigma;}}}(R_v^a\left(t\right)+R_s^a\left(t\right)),$

for time slot t unit data flow price, being set point, is time-varying,

for time slot t State Price, being set point, is time-varying,

The behavioral unity of above-mentioned two kinds of prices that secondary user occurs is expressed as a _nt ()=1 represents that the idle frequency spectrum that time user takies primary user at time slot t carries out transfer of data, a _nt ()=0 represents that time user does not take the idle frequency spectrum of primary user,

Step (2.3), the frequency spectrum leasing agreement Spectrum Lease Protocol that the above-mentioned information in step (2.2) is formed a short-term by primary user is called for short SLP, is broadcast to all potential users in its coverage;

Step (3), primary user carries out frequency spectrum leasing preparation according to the following steps:

Step (3.1), if described time potential user does not accept the frequency spectrum leasing agreement SLP that primary user is arranged in step (2.3), then turns to the network at other primary user place, if accept, then sends to described primary user and accepts message,

Step (3.2), primary user is according to accepting the secondary number of users N of described frequency spectrum leasing agreement and primary user M the idle frequency band when time slot t, and calculate the preferential index value table from M idle frequency band selected in N number of user, M≤N, step is as follows:

Step (3.2.1), the state transition probability matrix when behavior that primary user calculates time user n is according to the following steps a

wherein:

P _eEfor current time slots t user n buffer data size is continuously the probability of 0,

P _bKfor current time slots t user n turn to other primary user's network because buffer data size exceedes cache threshold after because buffer data size reduces get back to again the probability of current primary user's network, be the estimated value of setting,

P _oTfor current time slots t user n turns to the probability after other primary user's network still not in current primary user's network, be the estimated value of setting,

p _OT+p _BK＝1，

Secondary user n, when not accessing idle frequency spectrum a (t)=0, forwards the probability of state g ' to from state g be expressed as:

$P_{{\mathrm{gg}}^{\&prime;}}^0=P\{({\mathrm{\&epsiv;}}_{g^{\&prime;}-1}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right))\&le;X\&le;({\mathrm{\&epsiv;}}_{g^{\&prime;}}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right))\},$

Secondary user n, when accessing idle frequency spectrum a (t)=1, forwards the probability of state g ' to from state g be expressed as:

$P_{{\mathrm{gg}}^{\&prime;}}^1=P\{({\mathrm{\&epsiv;}}_{g^{\&prime;}-1}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right)+v_n)\&le;X\&le;({\mathrm{\&epsiv;}}_{g^{\&prime;}}-\left(\frac{{\mathrm{\&epsiv;}}_g+{\mathrm{\&epsiv;}}_{g-1}}{2}\right)+v_n)\},$

ε _gstate S _gt the upper limit of () is set point,

g＝1,2,…,g,…G,g′＝1,2,…,g',…G,g≠g′，

X is a positive integer k, at k=0, and 1,2 ..., value between K, K is a limited positive integer, is set point, λ is time user's average data arrival rate,

V _nthe transmission rate that time user can reach,

$v_n=\mathrm{\&Delta;f}\&CenterDot;{\log }_2(1+{\mathrm{\&tau;}}_n\&CenterDot;\frac{h_n{W}_n}{{\mathrm{\&sigma;}}^2+\mathrm{\&Sigma;}{\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}}),$ Wherein:

i′＝1,2,…i′…,I,I＝M,j′＝1,2,…,j′…,J,J＝M,i′≠j′，

The sequence number of the subcarrier of two users that i ', j ' be respectively are selected from described N number of user, data equal the sequence number of two idle frequency bands used,

Δ f is adjacent two subcarrier i ', and the frequency interval of j ' is set point,

τ _nbe a constant, with the error rate BER of secondary user n _nrelevant, for additive white Gaussian noise channel, τ _nwith BER _nbetween meet:

${\mathrm{\&tau;}}_n=\frac{1.5}{-\ln \left(5{\mathrm{BER}}_n\right)},$

σ ²for the one-sided power spectrum density of additive white Gaussian noise, be given value,

H _nfor the channel gain of secondary user n, be given value,

W _ntotal transmitted power of each user n,

I _{i ' j '}for secondary user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ', I _{i ' j '}be expressed as

${\stackrel{\&CenterDot;}{I}}_{i^{\&prime;}{j}^{\&prime;}}(d_{i^{\&prime;}{j}^{\&prime;}},W_{i^{\&prime;}})={\left|h_{i^{\&prime;}{j}^{\&prime;}}\right|}^2{W}_{i^{\&prime;}}{T}_s{\&Integral;}_{d_{i^{\&prime;}{j}^{\&prime;}}-\mathrm{\&Delta;f}/2}^{d_{i^{\&prime;}{j}^{\&prime;}}+\mathrm{\&Delta;f}/2}{\left(\frac{\sin \mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}{\mathrm{\&pi;}{f}_{i^{\&prime;}}{T}_s}\right)}^{2}d{f}_{i^{\&prime;}},$ Wherein:

H _{i ' j '}the channel gain of subcarrier i ' on subcarrier j ',

W _{i '}the secondary user n on subcarrier i ' _itotal transmitted power, W _{i '}=W _n,

T _sbeing OFDM symbol duration, is given value

D _{i ' j '}being the frequency distance between two subcarrier i ', j ', is given value

F _{i '}time user n _ithe frequency of place subcarrier i ',

Step (3.2.2), determine to make the maximized optimization object function of revenue of primary user by following formula:

Meet .I _{i ' j '}(d _{i ' j '}, W _{i ' j '}) < I _spec

${\mathrm{BER}}_{n_j}<{\mathrm{BER}}_{n_j\mathrm{spec}},$ Wherein:

$\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{a}_{{\mathrm{ni}}^{\&prime;}}\left(t\right)\&le;1$

i′＝1,2,…,i′…M,j′＝1,2,…,j′…M,i′≠j′

Discount factor Beta β represents, guarantees the bounded of total discount income and restrains, β in 0 < β <, 1 value,

I ', j ' represent idle sub-carrier frequency, be given value,

be T × N matrix, row is t, represents time slot, and row are behavior a of time user _nt,

time user n _isubcarrier i ' to secondary user n _jthe interference that causes of subcarrier j ' higher limit, be set point, for secondary user n _jthe upper limit threshold of maximum bit error rate is set point,

α _{ni '}t () represents the secondary user n of t time slot on subcarrier i ' _ibehavior, represent on subcarrier i ', an a can only be had at most at time slot t _{ni '}any active ues of (t)=1,

Step (3.2.3), solves primary user maximum return R with ceaselessly fruit machine Restless Bandits ^*, according to state transition probability matrix primary user's idle frequency band number M, secondary user number N, discount factor Beta=β, secondary User Status number StateNumber=G+1, Alpha=zeros (N, G+1) for complete zero N × (G+1) rank matrix, lower same, R0=zeros (N, G+1), R1=ones (N, G+1) for complete 1 N × (G+1) rank matrix, P0=zeros (G+1, G+1, N), P1=P0, utilize index value function f unction [Delta]=calc_index (M, N, Beta, StateNumber, Alpha, R0, R1, P0, P1) the index value Delta in order to state use order of the idle frequency spectrum corresponding to each user is calculated, find out the secondary user corresponding to minimum index value, using this minimum index value as preferential index value, the user priority that preferential index value is corresponding uses idle frequency band, construct the enquiry form of each user index value Delta, and all stored in the database of primary user,

Step (4), primary user distributes frequency spectrum according to the following steps:

State vector { the s that step (4.1) primary user is current according to each user ₁, s ₂..., s _nenquiry form, index value is sorted from small to large, and uses corresponding idle frequency band information successively to described N number of user transmission successively, comprise access slot,

Step (4.2), if obtain allowing a _nt ()=1, accesses frequency band next time slot time user and transmits data, the normal operation of primary user base station monitoring whole system,

Step (4.3), at the end of time slot, carries out as follows,

Step (4.3.1), secondary user adds up the buffer memory of oneself and the information of previous time slot transmission data, if still there are data to need transmission and data volume ε _cachedo not exceed threshold value, then cache information is sent to primary user, 0 < ε _cache≤ ε _spec,

Step (4.3.2), primary user is vectorial according to the system mode of all N number of the users of the information updating of secondary user,

Step (4.4), repeat (4.1) step to (4.3) step, until primary user does not have idle frequency spectrum to hire out, primary user sends and stops notify and regain the right to use of frequency spectrum, calculates rent that each user should pay and sends the information of rent confirmation.