CN103997743A - Effective-capacity-based resource allocation method in cognitive radio system - Google Patents

Abstract

The invention provides an effective-capacity-based resource allocation method in a cognitive radio system. The method includes the following steps: a cognitive user in the cognitive radio system perceives a channel and a transmission scheme of the cognitive user is determined according to the perception result; according to the transmission scheme, an effective capacity in the cognitive radio system is obtained and under a condition that an average power is limited, the effective capacity is optimized; and through application of a gradient descent method, a dual problem of an optimization problem is solved so that an optimal power allocation strategy is obtained. Compared with a traditional equipower scheme, the effective-capacity-based resource allocation method enables the performance of the system to be improved and the QoS guaranteeing capability of the system to be improved; and under a multi-channel scenery, the method is capable of providing reliable QoS guarantee to the cognitive user and capable of improving the frequency spectrum utilization rate of the system.

Description

Resource allocation methods based on available capacity in a kind of cognitive radio system

Technical field

The invention belongs to the Resource Allocation Formula design of cognitive radio system in wireless communication field, relate to the resource allocation methods based on available capacity in a kind of cognitive radio system.

Background technology

Due to the rise of various new business, as phone video conference, IN service etc., frequency spectrum resource becomes a kind of narrow resources.Although the link adaptation techniques, the multi-antenna technology etc. that occurred in recent years can improve the availability of frequency spectrum, to study and show, some authorize the reasonable low-frequency range availability of frequency spectrum of frequency range, especially signal propagation characteristics extremely low.In order to improve the availability of frequency spectrum, can effectively utilize idle mandate frequency range in zones of different and different period, cognitive radio receives increasing concern.The basic thought of cognitive radio is authorized user not to be produced under the prerequisite of harmful interference, and the cognitive user insertion authority frequency range of selecting a good opportunity is carried out transfer of data.Cognitive radio is a kind of smart frequency spectrum technology of sharing, is considered to solve the effective way that frequency spectrum resource is in short supply, improve the availability of frequency spectrum.

In wireless communication system, effective guarantee user's service quality (QoS) has very important meaning.But because making radio communication, the impact such as fading characteristic and burst noise of wireless channel cannot, for user provides absolute QoS as wire communication, can only provide the QoS in statistical significance.In cognitive radio system, because the transmitting power of cognitive user is to depend on whether authorized user takies band transmissions data, so just increase the difficulty that ensures the QoS of cognitive user.In existing cognitive radio system, ensureing the technology of the QoS of cognitive user, such as frequency spectrum decision-making, spectral transition etc., is all whether the QoS whether communication with cognitive user is qualitatively interrupted to weigh cognitive user is protected.Spectral transition can be divided into passive type spectral transition and active spectral transition: passive type spectral transition refers in the time that authorized user need to communicate, cognitive user just must exit this mandate frequency range, re-start frequency spectrum perception, be linked into other idle frequency ranges, passive type spectral transition can cause the temporary transient interruption of cognitive user communication, causes QoS index to worsen; Active spectral transition refer to cognitive user communication in constantly carry out frequency spectrum perception, once authorized user occurs, just transfer to immediately in other idle frequency ranges and continue to communicate by letter, thereby realize seamless branches, although this method has ensured the QoS of cognitive user, signaling and network overhead are increased greatly.

The scheme that a kind of statistics QoS that determines user in quantitative analysis wireless fading channel communication system ensures has been proposed in prior art, it utilizes the concept of available capacity, by concrete QoS index, for example time delay rate of violation, study user and add up the problem that QoS ensures, and the problem of the service ability of the time delay business of user to any qos requirement in wireless communication system.

Resource allocation algorithm in existing wireless communication system is not mostly considered time delay qos requirement or is only considered the situation of very strict and very loose qos requirement.Therefore the resource allocation algorithm that, research can provide the different QoS that require to ensure for wireless communication system is very significant.

Summary of the invention

The object of the present invention is to provide the resource allocation methods based on available capacity in a kind of cognitive radio system.

For achieving the above object, the present invention has adopted following technical scheme.

The first step, the cognitive user in cognitive radio system is carried out perception to channel, has judged whether that authorized user is using channel, if detected authorized user use channel, cognitive user not insertion authority channel communicate; If detect, with no authorized user uses channel, and cognitive user can insertion authority frequency range communicate;

Second step, determine the transmission plan of cognitive user according to the sensing results of the first step: using channel if sensing results is authorized user, for fear of cognitive user, authorized user is produced to harmful interference, now cognitive user is not carried out transfer of data, and now the transmitting power of cognitive user and transmission rate are 0; Do not use channel if sensing results is authorized user, cognitive user access frequency range, carries out transfer of data with larger transmitting power and transmission rate;

The 3rd step, obtain the available capacity in cognitive radio system according to described transmission plan, and under the limited condition of average power, described available capacity is optimized, by using gradient descent method to solve the dual problem of former optimization problem, obtain optimum power distribution strategies.

The described first step specifically comprises the following steps:

Consider a cognitive radio system, this system contains a cognitive user, M sub-channels, M>=2, channel model is rayleigh fading channel and is piece decline, M sub-channels independently carries out channel-aware, do not interfere with each other, frame length in cognitive radio system is T, the long-time cognitive user of front N of each frame is carried out channel-aware, according to how channel-aware is modeled as Hypothesis Testing Problem by graceful Pearson criterion, and according to the prior probability of the authorized CU of channel, obtain 2 in conjunction with false alarm probability and detection probability ^2Mplant the probability expression of the state that may occur.

In described second step, the sensing results that cognitive user obtains according to the first step is selected corresponding transmission plan, if being authorized user, sensing results do not use channel, cognitive user access frequency range, meanwhile, determine corresponding power distribution strategies function and instantaneous transmission speed, and carry out transfer of data in remaining T-N is long-time, frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware.

Available capacity in described the 3rd step is expressed as:

$P\left(s_j\right)=\underset{m=1}{\overset{M}{\mathrm{\Π}}}P\left(s_{j,m}\right)=\underset{m\∈{\mathrm{\Φ}}_j}{\mathrm{\Π}}{\mathrm{\ρ}}_m\underset{m^{\′}\∈{\mathrm{\Φ}}_j^C}{\mathrm{\Π}}(1-{\mathrm{\ρ}}_{m^{\′}})$

$P\left(s_i^d\right|s_j)=\underset{m=1}{\overset{M}{\mathrm{\Π}}}P\left(s_{i,m}^d\right|s_j)$

Wherein, θ represents QoS index; s _jthe virtual condition of expression system, j=1,2 ..., 2 ^m, P (s _j) represent that the virtual condition of system is s _jtime probability; ρ _mrepresent the probability of the authorized CU of m sub-channels; ρ _m'represent the probability of the authorized CU of m' sub-channels; Φ _jthe virtual condition that system is worked as in expression is s _jtime, the set of the subchannel of authorized CU; the virtual condition that system is worked as in expression is s _jtime, the not set of the subchannel of authorized CU; represent system senses result, i=1,2 ..., 2 ^m, represent the sensing results of m sub-channels, the virtual condition that system is worked as in expression is s _j, sensing results is time conditional probability; z _mrepresent the channel gain of m sub-channels; represent the power distribution strategies function of m sub-channels; represent the instantaneous transmission speed of m sub-channels, the frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware, and M represents number of subchannels.

In the time that number of subchannels is 2 (when M=2), the optimization method of available capacity comprises the following steps:

Optimization problem is modeled as to the available capacity that maximizes system under the limited condition of system average power, is specifically expressed as:

P1＝(1-ρ _A)(1-P _f)(ρ _BP _d+(1-ρ _B)P _f)；P2＝(1-ρ _A)ρ _B(1-P _f)(1-P _d)；

P3＝ρ _A(1-ρ _B)(1-P _d)(1-P _f)；P4＝(1-ρ _A)(1-ρ _B)(1-P _f)(1-P _f)；

P5＝(1-ρ _B)(1-P _f)((1-ρ _A)P _f+ρ _AP _d)；

$P6=P_f({\mathrm{\ρ}}_A(1-{\mathrm{\ρ}}_B)+(1-{\mathrm{\ρ}}_A){\mathrm{\ρ}}_B)+(1-{\mathrm{\ρ}}_A)(1-{\mathrm{\ρ}}_B)P_f^2+{\mathrm{\ρ}}_A{\mathrm{\ρ}}_B$

Wherein, θ represents QoS index, z _aand z _brepresent respectively the channel gain of subchannel A and subchannel B; with represent respectively the power distribution strategies function of subchannel A and subchannel B, i=1,2,3,4; with represent respectively the instantaneous transmission speed of subchannel A and subchannel B; ρ _aand ρ _brepresent respectively the probability of the authorized CU of subchannel A and subchannel B; P _frepresent false alarm probability; P _drepresent detection probability, the frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware;

Because the logarithmic function in optimization aim function in formula (13) is monotonically increasing, so former optimization problem can abbreviation be:

The dual problem of the optimization problem that formula (14) represents is expressed as:

$\begin{array}{cc}\max \mathrm{imize}& D\left(\mathrm{\λ}\right)\\ s.t.& \mathrm{\λ}\≥0\end{array}---\left(15\right)$

Wherein λ represents Lagrange multiplier, and the expression of Lagrange duality function is:

Utilize gradient descent method to solve dual problem and comprise following concrete steps:

Step 1: initialization setting:

Make iterations t=1, and the initial value D of dual function is set ⁽⁰⁾and the initial value λ of Lagrange multiplier ⁽⁰⁾;

Step 2: upgrade optimized variable:

$\frac{\∂L}{{\∂\mathrm{\μ}}_A(z_A,s_i^d)}=0\⇒{\mathrm{\μ}}_A^{\left(t\right)}\left({\mathrm{\λ}}^{(t-1)}\right);\frac{\∂L}{{\∂\mathrm{\μ}}_B(z_B,s_i^d)}=0\⇒{\mathrm{\μ}}_B^{\left(t\right)}\left({\mathrm{\λ}}^{(t-1)}\right)---\left(17\right)$

Step 3: upgrade gradient factor Δ f ^(t), dual function D ^(t)and Lagrange multiplier λ ^(t):

${\mathrm{\Δf}}^{\left(t\right)}=\frac{\∂L({\mathrm{\μ}}_A^{\left(t\right)},{\mathrm{\μ}}_B^{\left(t\right)},\mathrm{\λ})}{\∂\mathrm{\λ}}{|}_{\mathrm{\λ}={\mathrm{\λ}}^{(t-1)}};D^{\left(t\right)}=L({\mathrm{\μ}}_A^{\left(t\right)},{\mathrm{\μ}}_B^{\left(t\right)},\mathrm{\λ}){|}_{\mathrm{\λ}={\mathrm{\λ}}^{(t-1)}};{\mathrm{\λ}}^{\left(t\right)}={\mathrm{\λ}}^{(t-1)}+\mathrm{step}*{\mathrm{\Δf}}^{\left(t\right)}---\left(18\right)$

Wherein step is iteration step length;

Step 4: judge whether iteration finishes:

ξ＝|D ^(t)-D ^(t-1)|；t＝t+1 (19)

If ξ is >10 ^-4, meet iterated conditional, repeat step 1 to step 3, until iteration finishes, obtain optimum power distribution strategies function.

Beneficial effect of the present invention is: the present invention is by being applied to the theory of available capacity in cognitive radio system, and its system model is extended under multi channel sight, derivation draws the available capacity expression of multichannel cognitive radio system, for research ensures that cognitive user meets the ability that different QoS requires effective and feasible method is provided in cognitive radio system, the available capacity scheme of the optimization multichannel system that the present invention proposes, compared with carrying out with traditional constant power scheme, the performance of system is got a promotion, and the supportability of the QoS of system is improved, under multichannel sight, can either ensure for cognitive user provides reliable QoS, can improve again the availability of frequency spectrum of system.

Brief description of the drawings

Fig. 1 is cognitive radio system frame structure schematic diagram, and wherein, Sensing represents channel-aware, and Data Transmission represents transfer of data;

Fig. 2 is available capacity theoretical value (theoretical performance) and simulation value (simulated performance) comparison diagram; Fig. 3 is the power allocation scheme of optimization and the available capacity comparison diagram of constant power allocative decision that the present invention proposes;

Fig. 4 is time delay rate of violation theoretical value and simulation value comparison diagram; Fig. 5 is the power allocation scheme of optimization and the time delay rate of violation comparison diagram of constant power allocative decision that the present invention proposes;

Fig. 6 is the impacts of different detection times on time delay rate of violation;

Fig. 7 is the impact of various information source speed on time delay rate of violation.

Embodiment

Below in conjunction with drawings and Examples, the present invention is elaborated.

The power allocation scheme of the optimization based on available capacity that the present invention proposes is as follows:

Consider a cognitive radio system, this system contains a cognitive user, and two sub-channels are respectively subchannel A and subchannel B, and channel model is rayleigh fading channel and is piece decline.As shown in Figure 1, wherein T represents the duration length of a frame to the frame structure of cognitive radio system, i.e. frame length, and N represents that cognitive user in each frame carries out the duration length of channel-aware, the long-time cognitive user of front N of each frame is carried out channel-aware.

The first step: two sub-channels independently carry out channel-aware, do not interfere with each other, the long-time cognitive user of front N of each frame is carried out channel-aware, judges whether busy channel of authorized user, according to how channel-aware is modeled as Hypothesis Testing Problem by graceful Pearson criterion, and obtain 2 ^2Mthe probability expression of the state that kind (M=2) may occur;

Cognitive user is carried out channel-aware, judges whether busy channel of authorized user, generally adopts the method for energy measuring, and the graceful Pearson criterion according to how can be modeled as channel-aware the problem of a hypothesis testing.Taking single channel as example, channel width is W, x[i] and y[i] the instantaneous input and output signal of a channel represented respectively; H[i] represent the fading coefficients of channel, be separate and obey the multiple Gaussian Profile of zero-mean circulation, the channel fading coefficient of every frame is independent variation; N[i] represent the independent identically distributed obedience zero-mean circulation additive white Gaussian noise of Gaussian Profile again in channel, variance is n _p[i] represents the interference of authorized user to cognitive user, obeys the multiple Gaussian Profile of zero-mean circulation, and variance is the input/output relation of channel can be expressed as so:

Hypothesis Testing Problem can be expressed as:

$\begin{array}{c}{H}_0:y\left[i\right]=n\left[i\right]\\ H_1:y\left[i\right]=n\left[i\right]+n_p\left[i\right]\end{array}---\left(2\right)$

The graceful Pearson criterion according to how, test statistics is expressed as:

$\mathrm{\&Lambda;}=\frac{1}{\mathrm{NW}}\underset{i=1}{\overset{\mathrm{NW}}{\mathrm{\&Sigma;}}}{\left|y\right[i\left]\right|}^2\underset{H_0}{\overset{H_1}{\underset{<}{>}}}{\mathrm{\&lambda;}}_0---\left(3\right)$

Wherein λ ₀represent energy measuring thresholding.

Detecting performance can represent by two very important parameters, i.e. false alarm probability P _fwith detection probability P _d:

$\left\{\begin{array}{c}{P}_f=P(\mathrm{\&Lambda;}>{\mathrm{\&lambda;}}_0|H_0)=1-\mathrm{\&gamma;}(\frac{\mathrm{NW}{\mathrm{\&lambda;}}_0}{{\mathrm{\&sigma;}}_n^2},\mathrm{NW})/\mathrm{\&Gamma;}\left(\mathrm{NW}\right)\\ P_d=P(\mathrm{\&Lambda;}>{\mathrm{\&lambda;}}_0|H_1)=1-\mathrm{\&gamma;}(\frac{\mathrm{NW}{\mathrm{\&lambda;}}_0}{{\mathrm{\&sigma;}}_n^2+{\mathrm{\&sigma;}}_p^2},\mathrm{NW})/\mathrm{\&Gamma;}\left(\mathrm{NW}\right)\end{array}---\left(4\right)\right.$

Wherein (x, a) represents incomplete gamma functions to γ, and Γ (a) represents gamma function.

For single channel, the virtual condition of channel has two kinds of possible situations, and channel is idle or busy; And cognitive user also has two kinds of possible situations to the sensing results of channel, perceive channel for idle or busy, for cognitive radio system, always have four kinds of sights that may occur, if ρ represents the probability of the authorized CU of channel, in conjunction with false alarm probability and detection probability, the probability expression of four kinds of states is as shown in table 1.

Table 1 single channel Simple Expressions

Consider the practicality of system, mainly consider multi channel situation, taking two channels as example, each channel independently carries out channel-aware and transfer of data, does not interfere with each other.In cognitive radio system, the virtual condition of channel has four kinds of possible situations, and the sensing results of cognitive user also has four kinds of possible situations, for whole system, altogether having 16 kinds of possible sights occurs, with above-mentioned single-channel state probability analysis classes seemingly, also can obtain the probability expressions of 16 kinds of states under two channel situation, illustrate, the virtual condition of channel is that channel A free time and channel B are busy, and cognitive user perception is correct, this kind of shape probability of state is (1-ρ _a) ρ _b(1-P _f) P _d, wherein ρ _aand ρ _bthe virtual condition that represents respectively channel A and channel B is the probability of busy (being the authorized CU of channel).In like manner can obtain the probability expression of other 15 kinds of states.

Second step: the sensing results that cognitive user obtains according to first step, select corresponding transmission plan, determine corresponding power distribution strategies function with and instantaneous transmission speed r _aand r _b, in remaining T-N is long-time, carry out transfer of data;

The sensing results that cognitive user obtains according to the first step, selects corresponding transmission plan to carry out transfer of data.If the sensing results in the first step is channel idle, cognitive user selects larger transmitting power and transmission rate to communicate; If the sensing results in the first step is that channel is busy, the interference for fear of cognitive user to authorized user, cognitive user does not communicate, and transmitting power and transmission rate is now 0.Four kinds of different sensing results are represented as follows:

According to different sensing results, carry out corresponding power division.If sensing results is be busy owing to perceiving channel B, according to power division principle, cognitive user, in channel B transmission data, is not therefore only carried out power division to channel A, and power distribution strategies function is if sensing results is cognitive user is carried out transfer of data at channel A and channel B simultaneously, and power distribution strategies function is with if sensing results is because two sub-channels all detect as busy condition, cognitive user does not communicate at any subchannel, and transmitting power and transmission rate under this sensing results are 0; If sensing results is cognitive user selective channel B communicates, and power distribution strategies function is after determining corresponding power distribution strategies function, can obtain the corresponding instantaneous transmission speed of two sub-channels, represent as follows respectively:

$\left\{\begin{array}{c}{r}_A=W{\log }_2(1+{\mathrm{\&mu;}}_A(z_A,s_i^d)z_A\frac{\stackrel{\&OverBar;}{P}}{W{\mathrm{\&sigma;}}_n^2})\\ r_B=W{\log }_2(1+{\mathrm{\&mu;}}_B(z_B,s_i^d)z_B\frac{\stackrel{\&OverBar;}{P}}{W{\mathrm{\&sigma;}}_n^2})\end{array}\right.---\left(5\right)$

Wherein z _a[i]=| h _a[i] | ², z _b[i]=| h _b[i] | ², the average power of expression system, represent four kinds of testing results one of them, i=1,2,3,4.

The 3rd step: derive and obtain the expression formula of multichannel cognitive radio system available capacity, be based upon the optimization problem that maximizes available capacity under the limited condition of average power, and it is solved and obtains optimum power distribution strategies function;

After determining that cognitive user is for the corresponding transmission plan of different sensing results, can obtain the expression of the available capacity of cognitive radio system, thereby can optimization problem be carried out modeling and be solved, finally obtain optimum power distribution strategies.The expression formula of available capacity is as follows:

Wherein θ represents QoS index, represent accumulated time service process.In the theory of available capacity, can weigh the performance that system QoS ensures by time delay rate of violation, instantaneous time delay exceedes the probability of time delay threshold value, and its expression formula is as follows:

$P(D\left(t\right)\&GreaterEqual;D_{\max })\&ap;e^{-\mathrm{\&theta;\&mu;}{D}_{\max }}---\left(7\right)$

Wherein D (t) represents instantaneous time delay, D _maxrepresent time delay threshold value, μ represents information source speed.Can obtain the expression formula of the available capacity under two channel situation in cognitive radio system according to (6) formula.First try to achieve the expression formula of four kinds of exponential parts under different actual channel state, as follows:

1) channel A idle channel B is busy

$e^{-\mathrm{\&theta;S}\left(t\right)}=(1-{\mathrm{\&rho;}}_A){\mathrm{\&rho;}}_B[(1-P_f)P_d{e}^{-\mathrm{\&theta;}(T-N)r_A\left({\mathrm{\&mu;}}_A(z_A,s_1^d)\right)}+(1-P_f)(1-P_d)e^{-\mathrm{\&theta;}(T-N)r_A\left({\mathrm{\&mu;}}_A(z_A,s_2^d)\right)}+P_f{P}_d+P_f(1-P_d)]---\left(8\right)$

2) the channel A idle channel B free time

$e^{-\mathrm{\&theta;S}\left(t\right)}=(1-{\mathrm{\&rho;}}_A)(1-{\mathrm{\&rho;}}_B)\left[\begin{array}{c}(1-P_f)P_f{e}^{-\mathrm{\&theta;}(T-N)r_A\left({\mathrm{\&mu;}}_A(z_A,s_1^d)\right)}+(1-P_f)(1-P_f)e^{-\mathrm{\&theta;}(T-N)(r_A\left({\mathrm{\&mu;}}_A(z_A,s_2^d)\right)+r_B\left({\mathrm{\&mu;}}_B(z_B,s_2^d)\right))}\\ +P_f{P}_f+P_f(1-P_f)e^{-\mathrm{\&theta;}(T-N)r_B\left({\mathrm{\&mu;}}_B(z_B,s_4^d)\right)}\end{array}\right]---\left(9\right)$

3) the busy channel B free time of channel A

$e^{-\mathrm{\&theta;S}\left(t\right)}={\mathrm{\&rho;}}_A(1-{\mathrm{\&rho;}}_B)[(1-P_d)P_f+(1-P_d)(1-P_f)e^{-\mathrm{\&theta;}(T-N)r_B\left({\mathrm{\&mu;}}_B(z_B,s_2^d)\right)}+P_d{P}_f+P_d(1-P_f)e^{-\mathrm{\&theta;}(T-N)r_B\left({\mathrm{\&mu;}}_B(z_B,s_4^d)\right)}]---\left(10\right)$

4) the busy channel B of channel A is busy

e ^-θS(t)＝ρ _Aρ _B[(1-P _d)P _d+(1-P _d)(1-P _d)+P _dP _d+P _d(1-P _d)] (11)

The expression that can obtain the available capacity under two channel situation of cognitive radio system through be simplified by integration is:

P1＝(1-ρ _A)(1-P _f)(ρ _BP _d+(1-ρ _B)P _f)；P2＝(1-ρ _A)ρ _B(1-P _f)(1-P _d)；

P3＝ρ _A(1-ρ _B)(1-P _d)(1-P _f)；P4＝(1-ρ _A)(1-ρ _B)(1-P _f)(1-P _f)；

P5＝(1-ρ _B)(1-P _f)((1-ρ _A)P _f+ρ _AP _d)；

$P6=P_f({\mathrm{\&rho;}}_A(1-{\mathrm{\&rho;}}_B)+(1-{\mathrm{\&rho;}}_A){\mathrm{\&rho;}}_B)+(1-{\mathrm{\&rho;}}_A)(1-{\mathrm{\&rho;}}_B)P_f^2+{\mathrm{\&rho;}}_A{\mathrm{\&rho;}}_B$

Therefore optimization problem is modeled as under the limited condition of system average power, maximizes the available capacity of system, is specifically expressed as:

Because the logarithmic function in optimization aim function is monotonically increasing function, so former optimization problem can abbreviation be:

Because this optimization aim function is a convex function, and restrictive condition is linear, so this optimization problem is a protruding optimization problem, meets strong duality.But owing to cannot directly by Lagrangian method, it being solved, so by the method that solves its dual problem, former optimization problem is solved.The dual problem of former problem is expressed as:

$\begin{array}{cc}\max \mathrm{imize}& D\left(\mathrm{\&lambda;}\right)\\ s.t.& \mathrm{\&lambda;}\&GreaterEqual;0\end{array}---\left(15\right)$

Wherein λ represents Lagrange multiplier, and the expression of Lagrange duality function is as follows:

Utilize gradient descent method to solve dual problem, thereby obtain the optimal solution of the optimized variable of former problem, concrete solution procedure comprises:

Step 1: initialization setting, makes iterations t=1, and the initial value D of dual function is set ⁽⁰⁾and the initial value λ of Lagrange multiplier ⁽⁰⁾.

Step 2: upgrade optimized variable:

$\frac{\&PartialD;L}{{\&PartialD;\mathrm{\&mu;}}_A(z_A,s_i^d)}=0\&DoubleRightArrow;{\mathrm{\&mu;}}_A^{\left(t\right)}\left({\mathrm{\&lambda;}}^{(t-1)}\right);\frac{\&PartialD;L}{{\&PartialD;\mathrm{\&mu;}}_B(z_B,s_i^d)}=0\&DoubleRightArrow;{\mathrm{\&mu;}}_B^{\left(t\right)}\left({\mathrm{\&lambda;}}^{(t-1)}\right)---\left(17\right)$

Step 3: upgrade the gradient factor, dual function and Lagrange multiplier:

${\mathrm{\&Delta;f}}^{\left(t\right)}=\frac{\&PartialD;L({\mathrm{\&mu;}}_A^{\left(t\right)},{\mathrm{\&mu;}}_B^{\left(t\right)},\mathrm{\&lambda;})}{\&PartialD;\mathrm{\&lambda;}}{|}_{\mathrm{\&lambda;}={\mathrm{\&lambda;}}^{(t-1)}};D^{\left(t\right)}=L({\mathrm{\&mu;}}_A^{\left(t\right)},{\mathrm{\&mu;}}_B^{\left(t\right)},\mathrm{\&lambda;}){|}_{\mathrm{\&lambda;}={\mathrm{\&lambda;}}^{(t-1)}};{\mathrm{\&lambda;}}^{\left(t\right)}={\mathrm{\&lambda;}}^{(t-1)}+\mathrm{step}*{\mathrm{\&Delta;f}}^{\left(t\right)}---\left(18\right)$

Wherein step is iteration step length.

Step 4: judge whether iteration finishes:

ξ＝|D ^(t)-D ^(t-1)|；t＝t+1 (19)

If ξ is >10 ^-4, meet iterated conditional, repeat step 1-step 3, until iteration finishes, finally obtain optimum power distribution strategies function.

Simulated effect of the present invention is as follows: considers a cognitive radio system, in system, contains a user, and two sub-channels, channel model adopts rayleigh fading channel, and the bandwidth of every sub-channels is 10kHz, detection threshold value λ ₀the duration T that is 1.4, one frames is 0.1 second, and the long-time cognitive user of front N of each frame is carried out channel-aware, and its value is 0.02 second, the probability ρ of the authorized CU of subchannel A and subchannel B _aand ρ _bthe span that is 0.1, QoS index θ is [0,1].

Theoretical value and the simulation value of the available capacity of the system model that the present invention is proposed contrast, and the theoretical curve of the available capacity of system and simulation curve are substantially identical as seen from Figure 2, and this has proved correctness and the feasibility of the system model of the present invention's proposition.Along with the increase of QoS index, index reduces the curve of available capacity as can be seen from Figure 2, describes and meets with formula (6).Due to the stricter delay requirement of larger QoS exponential representation, the less looser delay requirement of QoS exponential representation, when QoS index increases, when delay requirement is stricter, the available capacity of system reduces.In the time that QoS index is tending towards 0, system can be born the time delay of any length, and in the time that QoS index is tending towards infinity, system cannot be born any time delay.Power allocation scheme and the constant power allocative decision of the optimization that the present invention is proposed compare, contrast with constant power allocative decision (equal power adaption algorithm) as can be seen from Figure 3, adopt after the power allocation scheme (proposed power adaption algorithm) of optimizing, the available capacity (Effective Capacity) of system of the present invention has had certain lifting, the supportability under different QoS requires increases to cognitive user to make system, the ability of the real-time service of the stricter delay requirement of cognitive user service simultaneously also has a certain upgrade.

Theoretical value and the simulation value of the time delay rate of violation of the system model that the present invention is proposed contrast, the theoretical curve of the time delay rate of violation of system and simulation curve are basic coincideing as seen from Figure 4, have further verified the correctness of system model proposed by the invention.As can be seen from Figure 4, time delay rate of violation (Delay violation probability) reduces along with the increase of time delay threshold value is index.Power allocation scheme and the constant power allocative decision of the optimization that the present invention is proposed contrast, and can be found out by Fig. 5, adopt the power allocation scheme of optimizing can make the time delay rate of violation of system reduce, and have promoted the performance of system.Fig. 6 and Fig. 7 have compared respectively different channels detecting period and the impact of various information source speed on Time Delay of Systems rate of violation.As seen from Figure 6, along with the increase of channel-aware time, time delay rate of violation increases, and this is because detecting period increases, and can cause more packet congested at transmit queue place, therefore can cause time delay rate of violation to increase.As seen from Figure 7, along with the increase of information source speed, the time delay rate of violation of system increases, and information source speed is larger, equally also can cause the packet of storing excess in transmit queue, finally causes the time delay rate of violation of system to increase.

Can find out by above simulated effect, the power allocation scheme of optimizing in multichannel cognitive radio system that the present invention proposes gets a promotion the available capacity of system, time delay rate of violation declines to some extent, thereby the ability that makes user serve the real-time service of different QoS requirement is improved, and the ability that makes system provide reliable QoS to ensure gets a promotion.

Claims (6)

1. the resource allocation methods based on available capacity in cognitive radio system, is characterized in that: comprise the following steps:

The first step, the cognitive user in cognitive radio system is carried out perception to channel, has judged whether that authorized user is using channel;

Second step, determines the transmission plan of cognitive user: using channel if sensing results is authorized user, now cognitive user is not carried out transfer of data, and the transmitting power of cognitive user and transmission rate are 0 according to the sensing results of the first step; Do not use channel if sensing results is authorized user, cognitive user access frequency range, carries out transfer of data;

The 3rd step, obtain the available capacity in cognitive radio system according to described transmission plan, and under the limited condition of average power, described available capacity is optimized, by using the dual problem of gradient descent method solving-optimizing problem, obtain optimum power distribution strategies.

2. the resource allocation methods based on available capacity in a kind of cognitive radio system according to claim 1, is characterized in that:

Cognitive radio system contains a cognitive user, M sub-channels, M>=2, channel model is rayleigh fading channel and is piece decline, M sub-channels independently carries out channel-aware, do not interfere with each other, frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware, according to how channel-aware is modeled as Hypothesis Testing Problem by graceful Pearson criterion, and according to the prior probability of the authorized CU of channel, obtain 2 in conjunction with false alarm probability and detection probability ^2Mplant the probability expression of the state that may occur.

3. the resource allocation methods based on available capacity in a kind of cognitive radio system according to claim 1, it is characterized in that: in described second step, the sensing results that cognitive user obtains according to the first step is selected corresponding transmission plan, if being authorized user, sensing results do not use channel, cognitive user access frequency range, simultaneously, determine corresponding power distribution strategies function and instantaneous transmission speed, and carry out transfer of data in T-N is long-time, frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware.

4. the resource allocation methods based on available capacity in a kind of cognitive radio system according to claim 1, is characterized in that: the available capacity in described the 3rd step is expressed as:

$P\left(s_j\right)=\underset{m=1}{\overset{M}{\mathrm{\&Pi;}}}P\left(s_{j,m}\right)=\underset{m\&Element;{\mathrm{\&Phi;}}_j}{\mathrm{\&Pi;}}{\mathrm{\&rho;}}_m\underset{m^{\&prime;}\&Element;{\mathrm{\&Phi;}}_j^C}{\mathrm{\&Pi;}}(1-{\mathrm{\&rho;}}_{m^{\&prime;}})$

$P\left(s_i^d\right|s_j)=\underset{m=1}{\overset{M}{\mathrm{\&Pi;}}}P\left(s_{i,m}^d\right|s_j)$

Wherein, θ represents QoS index; s _jthe virtual condition of expression system, j=1,2 ..., 2 ^m, P (s _j) represent that the virtual condition of system is s _jtime probability; ρ _mrepresent the probability of the authorized CU of m sub-channels; ρ _m'represent the probability of the authorized CU of m' sub-channels; Φ _jthe virtual condition that system is worked as in expression is s _jtime, the set of the subchannel of authorized CU; the virtual condition that system is worked as in expression is s _jtime, the not set of the subchannel of authorized CU; represent system senses result, i=1,2 ..., 2 ^m, represent the sensing results of m sub-channels, the virtual condition that system is worked as in expression is s _j, sensing results is time conditional probability; z _mrepresent the channel gain of m sub-channels; represent the power distribution strategies function of m sub-channels; represent the instantaneous transmission speed of m sub-channels, the frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware, and M represents number of subchannels.

5. the resource allocation methods based on available capacity in a kind of cognitive radio system according to claim 1, is characterized in that: in the time that number of subchannels is 2, the optimization method of available capacity comprises the following steps:

Optimization problem is modeled as to the available capacity that maximizes system under the limited condition of system average power, is specifically expressed as:

P1＝(1-ρ _A)(1-P _f)(ρ _BP _d+(1-ρ _B)P _f)；P2＝(1-ρ _A)ρ _B(1-P _f)(1-P _d)；

P3＝ρ _A(1-ρ _B)(1-P _d)(1-P _f)；P4＝(1-ρ _A)(1-ρ _B)(1-P _f)(1-P _f)；

P5＝(1-ρ _B)(1-P _f)((1-ρ _A)P _f+ρ _AP _d)；

$P6=P_f({\mathrm{\&rho;}}_A(1-{\mathrm{\&rho;}}_B)+(1-{\mathrm{\&rho;}}_A){\mathrm{\&rho;}}_B)+(1-{\mathrm{\&rho;}}_A)(1-{\mathrm{\&rho;}}_B)P_f^2+{\mathrm{\&rho;}}_A{\mathrm{\&rho;}}_B$

Wherein, θ represents QoS index, z _aand z _brepresent respectively the channel gain of subchannel A and subchannel B; with represent respectively the power distribution strategies function of subchannel A and subchannel B, i=1,2,3,4; with represent respectively the instantaneous transmission speed of subchannel A and subchannel B; ρ _aand ρ _brepresent respectively the probability of the authorized CU of subchannel A and subchannel B; P _frepresent false alarm probability; P _drepresent detection probability, the frame length in cognitive radio system is T, and the long-time cognitive user of front N of each frame is carried out channel-aware;

Because the logarithmic function in optimization aim function in formula (13) is monotonically increasing, so former optimization problem abbreviation is:

The dual problem of the optimization problem that formula (14) represents is expressed as:

$\begin{array}{cc}\max \mathrm{imize}& D\left(\mathrm{\&lambda;}\right)\\ s.t.& \mathrm{\&lambda;}\&GreaterEqual;0\end{array}---\left(15\right)$

Wherein λ represents Lagrange multiplier, and the expression of Lagrange duality function is:

6. the resource allocation methods based on available capacity in a kind of cognitive radio system according to claim 5, is characterized in that: utilize gradient descent method to solve dual problem and comprise following concrete steps:

Step 1: initialization setting:

Make iterations t=1, and the initial value D of dual function is set ⁽⁰⁾and the initial value λ of Lagrange multiplier ⁽⁰⁾;

Step 2: upgrade optimized variable:

$\frac{\&PartialD;L}{{\&PartialD;\mathrm{\&mu;}}_A(z_A,s_i^d)}=0\&DoubleRightArrow;{\mathrm{\&mu;}}_A^{\left(t\right)}\left({\mathrm{\&lambda;}}^{(t-1)}\right);\frac{\&PartialD;L}{{\&PartialD;\mathrm{\&mu;}}_B(z_B,s_i^d)}=0\&DoubleRightArrow;{\mathrm{\&mu;}}_B^{\left(t\right)}\left({\mathrm{\&lambda;}}^{(t-1)}\right)---\left(17\right)$

Step 3: upgrade gradient factor Δ f ^(t), dual function D ^(t)and Lagrange multiplier λ ^(t):

${\mathrm{\&Delta;f}}^{\left(t\right)}=\frac{\&PartialD;L({\mathrm{\&mu;}}_A^{\left(t\right)},{\mathrm{\&mu;}}_B^{\left(t\right)},\mathrm{\&lambda;})}{\&PartialD;\mathrm{\&lambda;}}{|}_{\mathrm{\&lambda;}={\mathrm{\&lambda;}}^{(t-1)}};D^{\left(t\right)}=L({\mathrm{\&mu;}}_A^{\left(t\right)},{\mathrm{\&mu;}}_B^{\left(t\right)},\mathrm{\&lambda;}){|}_{\mathrm{\&lambda;}={\mathrm{\&lambda;}}^{(t-1)}};{\mathrm{\&lambda;}}^{\left(t\right)}={\mathrm{\&lambda;}}^{(t-1)}+\mathrm{step}*{\mathrm{\&Delta;f}}^{\left(t\right)}---\left(18\right)$

Wherein step is iteration step length;

Step 4: judge whether iteration finishes:

ξ＝|D ^(t)-D ^(t-1)|；t＝t+1 (19)

If ξ is >10 ^-4, meet iterated conditional, repeat step 1 to step 3, until iteration finishes, obtain optimum power distribution strategies function.