CN102457338B - Method for selecting multi-user sensing channel in cognitive sensor network - Google Patents

Abstract

In cognitive radio, frequency spectrum sensing is an important premise and a core link for realizing the frequency spectrum dynamic access. The invention aims to provide a method for selecting a multi-user sensing channel in a cognitive sensor network, which comprises the following steps of: on an energy detection method, sensing normalized throughput acquired by a primary user channel from a maximized secondary user; using a detection time optimal scheme to budget the maximum throughput of all the secondary users and the selected channels of the secondary users; constructing a maximum throughput matrix; using a bidiagraph to express the channel selection problem under a multi-user multi-channel environment; converting a maximization and minimization problem into a unified minimization solution problem; and under the limit of target detection probability, adopting an improved Hungarian algorithm for the optimal selection of the sensing channel. By the method disclosed by the invention, the minimized throughput of the whole secondary user network can be greatly improved.

Description

The system of selection of multi-user sensing channel in cognitive sensor network

Technical field

The present invention relates to a kind ofly for channel perception selection scheme under cognitive radio networks multi-user and multi-channel scene, belong to communication technical field.

Background technology

In recent years, along with the broad development of radio communication service, increasingly deficient phenomenon has appearred in frequency spectrum resource, and its main cause is the irrational utilization of wireless frequency spectrum access technology, and existing wireless communication system all adopts inefficient fixed channel assignment strategy.Cognitive radio is considered to address this problem one of the most effective new technology, and its allows cognitive user to share this frequency range under the prerequisite that does not affect validated user proper communication in frequency band.

The CRN(CognitiveRadioNetwork cognitive radio networks) in; CRN to wireless environment intermediate frequency spectrum on every side take situation accurately, fast detecting is to improve the availability of frequency spectrum to protect the primary user to avoid the important prerequisite and the core link that disturb simultaneously, so frequency spectrum perception is a vital technology in cognitive radio.

Physical layer detection method in cognitive radio (CognitiveRadio, CR) comprises matched filter detection method, energy measuring method and feature detection method etc.But, in the practical application scene, due to a variety of causes, the CR detection technique is simply by virtue of simple physical layer detection algorithm, for improving the availability of frequency spectrum and protecting the primary user that certain limitation is arranged.So be necessary detected parameters is chosen and optimized by the CR testing mechanism based on the MAC layer, improve the validity of physical layer detection algorithm, find fast the frequency spectrum hole, improve detection performance and the efficiency of CR user to frequency spectrum, realize the primary user is disturbed to the availability of frequency spectrum that maximizes time user under minimum.

In addition, the performance of frequency spectrum perception not only relies on detection method, but also depends on the perception behavior of how dispatching.The purpose of frequency spectrum perception scheduling is in order to help CR user primary user's signal effectively to be detected and to identify the usable spectrum chance, mainly comprising the selection of detection time and channel perception.Research intermediate frequency spectrum perception in recent years scheduling all is based on cycle perception frame structure, in each frame, comprise one detection time a piece and data transmitting time piece.Be very important parameter in the frequency spectrum perception algorithm detection time, and its value directly affects the detection probability in frequency spectrum hole and the collision probability between the primary and secondary user.Different detection-phase is optimized to have different optimization formula detection time: search for the new idle channel stage band is outer, optimize from the average search time that minimizes outband channel; Periodic spectral detection-phase in chance takies the band of frequency spectrum, be optimized from maximizing obtainable average throughput.Can also, from the maximum channel utilance, adopt the numerical optimization algorithm to carrying out optimization detection time.Under the primary user is subject to the situation of adequately protecting, can be from making time user's the effective throughput maximum angle optimum detection time, and the application distribution frequency spectrum sensing method can make performance further promote.Further, under the multi-user and multi-channel environment, inferior user independently selective channel carries out frequency spectrum perception, and this need to study new channel perception allocative decision and make the throughput-maximized of whole user network.

Summary of the invention

Technical problem: a kind of method that provides channel perception optimization to select under the multi-user and multi-channel environment is provided.The method, based on the energy measuring method, from maximizing obtainable Normalized throughput of primary user's channel of user awareness one time, under the target detection probability restriction, adopts improved Hungary Algorithm to carry out the optimal selection of channel perception.Adopt the method can greatly improve the Normalized throughput of whole user network under cognitive radio multi-user and multi-channel scene.

Technical scheme: a kind of method that the present invention provides channel perception optimization to select under cognitive radio multi-user and multi-channel environment.Based on the energy measuring method, at first the method maximizes obtainable Normalized throughput of primary user's channel of user awareness one time, then, under the target detection probability restriction, adopts improved Hungary Algorithm to carry out the optimal selection of channel perception.The method can greatly improve the Normalized throughput of whole user network under cognitive radio multi-user and multi-channel scene.

The method comprises the following steps:

A. time user i is detected channel j, i=1, and 2 ..., K, K is time number of users, j=1,2 ..., N, N is the number of channel, calculates time obtainable Normalized throughput of user i

Wherein T is sense cycle, and τ is detection time, P _fFor the false alarm probability required, H ₀Mean that the primary user does not exist, P (H ₀) mean that there is not primary user's probability in channel j;

B. when T is fixed value, find optimum τ, be designated as τ ^*, make R _Ij(τ) maximum, be designated as R _Ij

C.K time user calculates and detects N the obtainable maximum throughput R of channel according to received signal to noise ratio _Ij, this K * N maximum throughput forms the maximum throughput moment matrix $A_{K,N}=\left[\begin{array}{ccccc}{R}_{11}& R_{12}& R_{13}& ......& R_{1N}\\ R_{21}& R_{22}& R_{23}& ......& R_{2N}\\ ...& ...& ...& ......& ...\\ R_{i1}& R_{i2}& R_{i3}& ......& R_{\mathrm{iN}}\\ ...& ...& ...& ......& ...\\ R_{K1}& R_{K2}& R_{K3}& ......& R_{\mathrm{KN}}\end{array}\right]$

D. a shilling C is matrix A _K,NMiddle maximum, use C-R _IjReplace A _K,NThe element R of correspondence in matrix _IjThen each row element deducts the minimum value in this row element, and finally each column element deducts the minimum value in this column element;

E. use minimum straight line to cover all 0 elements in matrix, suppose to have used l bar straight line, if l<K, making m is the minimum value be not capped in element, the element not capped to each deducts m, and each elements of two straight line infalls is all added to m, repeating step e; If l=K, enter step f;

F. according to the row order, from the first row, start to carry out successively Channel Assignment, while only only having 0 element in one's own profession, the channel that 0 element is corresponding is assigned to the inferior user that this row is corresponding, once complete appointment, deletes immediately these row of one's own profession; If can not complete K row of assigning and existing in element more than 0, consider row corresponding to this row 0 element, select to contain the minimum row of 0 element, this is listed as to corresponding Channel Assignment to inferior user corresponding to this row; Repeating above-mentioned appointment process, until carried out abundant unique appointment, if still have ungratifiedly, is that remaining user assigns arbitrarily a channel.

The present invention for frequency spectrum perception scheduling be based on cycle perception frame structure.As shown in Figure 1, the frame structure of cognitive radio system periodic spectral perception, each frame comprise one detection time a piece and data transmitting time piece.Suppose that sense cycle (being frame duration) is T, be τ detection time.Cognitive radio users, every T time detecting one secondary channel, accesses in idle condition the transmission that this channel carries out the cognitive user business if channel detected; If channel detected in busy state cognitive user transport service and wait for that T-τ detects this channel again after the time not within this cycle.

In cognitive radio, the multi-user and multi-channel scene as shown in Figure 2.Suppose total K time user in time user network, total N channel and N primary user in primary user's network, each primary user j takies a channel j separately, and each interchannel is separate, does not interfere with each other.Suppose that channel is rayleigh fading channel, α _Ij(i≤K, j≤N) means that the primary user of busy channel j is to time power fading coefficient at user i place, if use P _jThe transmitting power that means the primary user of busy channel j, N ₀The power that means noise in channel, the received signal to noise ratio at time user i place

Suppose the transmitting power P of primary user j _jNecessarily, when different inferior users detects same channel, because same primary user arrives the not channel power fading coefficients α at homogeneous user place _IjNot identical, therefore for same channel, the signal to noise ratio snr that the homogeneous user does not obtain at receiving terminal _iNot identical.

When with a user, detecting different channels, different channels is to the power fading factor alpha with a user place _IjNot identical, and different primary users' transmitting power P on different channels _jAlso different, therefore the signal to noise ratio snr that inferior user i receiving terminal obtains _iNot identical yet.

So different inferior users detect different channels, the signal to noise ratio that inferior user's receiving terminal obtains is not identical.Therefore in order to meet the requirements of the detection performance, required detection time is all not different during homogeneous user awareness different channels in the energy measuring method.

If expect the Normalized throughput of whole user network maximum, will consider that each user's How to choose channel carries out perception, how to be the Normalized throughput maximum that each inferior user assignment channel makes whole user network in other words.The present invention is based on the improvement Hungary Algorithm, from the optimized detection time, for each inferior corresponding channel of user assignment carries out frequency spectrum perception, make the Normalized throughput maximum of whole user network acquisition.In perception, each time user can only select a channel to carry out perception, and each channel can only be assigned to one user and carries out perception in other words.

The present invention adopts the detection time optimization scheme to calculate in advance the obtainable maximum throughput of combination of all users and its selected channel, form the maximum throughput moment matrix, use two-dimensional plot to mean multi-user and multi-channel environment lower channel selection problem, to ask maximum Assignment Problems to convert to and ask minimum Assignment Problems, classical Hungary Algorithm is improved, eliminate the randomness of its appointment, use the minimum Assignment Problems of improved Hungarian Method, obtain optimum channel perception, make the Normalized throughput maximum of whole user network.

Beneficial effect: distribute for the channel perception under cognitive radio multi-user and multi-channel environment, the present invention is based on the energy measuring method, from maximizing obtainable Normalized throughput of primary user's channel of user awareness one time, under the target detection probability restriction, carry out the optimal selection of channel perception based on improved Hungary Algorithm, the Normalized throughput of whole user network is maximized.

The accompanying drawing explanation

Fig. 1 cognitive radio system periodic spectral perception frame structure,

Fig. 2 multi-user and multi-channel sensory perceptual system model.

Embodiment

(1) optimization of alone family single channel scenario lower detection time

When primary user j does not exist, use C ₀Mean time throughput of user i place acquisition.The signal to noise ratio at inferior user i place can be expressed as P _sThe received power that means time user, N ₀Mean noise power.C ₀=log ₂(1+SNR _i).For the given frequency band that utilizes, definition P (H ₁) mean the probability that primary user j exists, P (H ₀) the non-existent probability of expression primary user j, visible P (H ₀)+P (H ₁)=1.

During inferior user i channel perception j, if primary user j does not exist, inferior user i can not produce false-alarm, and time accessible throughput of user i is The probability that this situation occurs is (1-P _f) P (H ₀), P wherein _fMean false alarm probability.

Time obtainable throughput of user i

Can be expressed as follows:

${\tilde{R}}_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)=\frac{T-\mathrm{\&tau;}}{T}{C}_0(1-P_f)P\left(H_0\right)---\left(1\right)$

Normalization can obtain:

${\hat{R}}_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)=\frac{{\tilde{R}}_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)}{C_0}=\frac{T-\mathrm{\&tau;}}{T}(1-P_f)P\left(H_0\right)---\left(2\right)$

Obviously, for given sense cycle T, detection time, τ was longer, and available data transmitting time (T-τ) is just shorter.On the other hand, for given target detection probability , detection time is longer, and false alarm probability is lower, and inferior user i is used the chance of channel j just higher.Determine that in each frame, optimum detection time τ is protected the obtainable Normalized throughput maximization of time user i while primary user fully.Therefore, this optimization problem can be expressed as with mathematic(al) representation:

$\underset{\mathrm{\&tau;}}{\max }{\hat{R}}_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)---\left(3\right)$

$s.t.P_d(\mathrm{\&epsiv;},\mathrm{\&tau;})\&GreaterEqual;{\stackrel{-}{P}}_d---\left(4\right)$

Threshold value when wherein ε is energy measuring,

Be that target detection probability is that detection probability reaches this value primary user and just adequately protected, the maximum throughput of acquisition is designated as R _Ij.

The uses energy detection method, order

Can obtain:

${\hat{R}}_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)=P\left(H_0\right)(1-\frac{\mathrm{\&tau;}}{T})(1-Q(\mathrm{\&alpha;}+\sqrt{\mathrm{\&tau;}{f}_s}\mathrm{\&gamma;}))---\left(5\right)$

Wherein $Q\left(x\right)=\frac{1}{\sqrt{2\mathrm{\&pi;}}}\underset{x}{\overset{\&infin;}{\&Integral;}}\exp \left(\frac{{-y}^2}{2}\right)\mathrm{dy},$ $\mathrm{\&alpha;}=\sqrt{2\mathrm{\&gamma;}+1}{Q}^{-1}\left({\stackrel{-}{P}}_d\right).$ Can find out that from (5) formula time obtainable throughput of user i is the function of τ detection time.Right

Differentiate can obtain:

$\frac{{\hat{R}}_{\mathrm{ij}}^{\&prime;}\left(\mathrm{\&tau;}\right)}{P\left(H_0\right)}=\frac{\mathrm{\&gamma;}\sqrt{f_s}(1-\frac{\mathrm{\&tau;}}{T})}{2\sqrt{2\mathrm{\&pi;\&tau;}}}\exp (-\frac{{(\mathrm{\&alpha;}+\sqrt{\mathrm{\&tau;}{f}_s}\mathrm{\&gamma;})}^2}{2})-\frac{1}{T}+\frac{1}{T}Q(\mathrm{\&alpha;}+\sqrt{\mathrm{\&tau;}{f}_s}\mathrm{\&gamma;})---\left(6\right)$

Obviously, $\underset{\mathrm{\&tau;}\&RightArrow;T}{\lim }\frac{{\hat{R}}_{\mathrm{ij}}^{\&prime;}\left(\mathrm{\&tau;}\right)}{P\left(H_0\right)}<(-\frac{1}{T}+\frac{1}{T}Q\left(\mathrm{\&alpha;}\right))<0---\left(7\right)$

$\underset{\mathrm{\&tau;}\&RightArrow;0}{\lim }\frac{{\hat{R}}_{\mathrm{ij}}^{\&prime;}\left(\mathrm{\&tau;}\right)}{P\left(H_0\right)}=+\&infin;---\left(8\right)$

And (8) formula while showing that τ is very little (7)

Increase and reduce when τ approaches T.Therefore, (0, T) in the interval, exist an optimal check time to make

Maximum, maximum throughput is R _Ij.

(2) detect the optimal selection of channel

Can be influential to the throughput of whole user network when inferior user selects different channels to carry out perception, before carrying out frequency spectrum perception, channel perception is in optimized selection, can make the throughput maximum of whole user network, so the present invention has provided the detection channel optimal selection method based on improving Hungary Algorithm.

Front has proved the Normalized throughput maximum that exists an optimal check time that time user awareness channel is obtained, and is designated as R _Ij.The signal to noise ratio that detects different channels time user's receiving terminal due to different inferior users is not identical, so from the optimized detection time, K time user can precompute and detect N the obtainable maximum throughput of channel according to received signal to noise ratio, and this K * N maximum throughput forms the maximum throughput matrix A _K,N.Finally, use improved Hungary Algorithm for each inferior channel of user assignment carries out perception, make the Normalized throughput maximum of whole user network acquisition.The maximum throughput matrix A _K,NCan be expressed as:

$A_{K,N}=\left[\begin{array}{ccccc}{R}_{11}& R_{12}& R_{13}& ......& R_{1N}\\ R_{21}& R_{22}& R_{23}& ......& R_{2N}\\ ...& ...& ...& ......& ...\\ R_{i1}& R_{i2}& R_{i3}& ......& R_{\mathrm{iN}}\\ ...& ...& ...& ......& ...\\ R_{K1}& R_{K2}& R_{K3}& ......& R_{\mathrm{KN}}\end{array}\right]---\left(9\right)$

Definition of T _i,jFor channel is selected the factor, if the final selective channel j of inferior user i carries out perception, T _i,j=1, if not, T _i,j=0.The channel under the multi-user and multi-channel environment selects problem to be expressed as:

$\max \underset{i,j}{\mathrm{\&Sigma;}}{T}_{i,j}\&CenterDot;R_{\mathrm{ij}}\left(\mathrm{\&tau;}\right)---\left(10\right)$

$s.t.\underset{i}{\mathrm{\&Sigma;}}{T}_{i,j}\&le;1$ And $\underset{j}{\mathrm{\&Sigma;}}{T}_{i,j}\&le;1---\left(11\right)$

This condition of formula (11) must meet, and it shows that each user can only select a channel to carry out perception and can only distribute to the user one time for fear of channel of interference.

Channel is selected to be divided into two steps to be realized:

The first step: use the detection time optimization scheme can be easy to calculate in advance the obtainable maximum throughput R of combination of all users and its selected channel _Ij, form the maximum throughput matrix A _K,N.

Second step: use two-dimensional plot to mean multi-user and multi-channel environment lower channel selection problem: given K time user and N channel.Target of the present invention is exactly to find the distribution method that maximizes whole user network Normalized throughput.Adopt above-mentioned method for expressing, channel is selected to have become an assignment problem, can solve by Hungary Algorithm.

The basic thought of Hungary Algorithm is the row or column of revising the benefit matrix, make in each row or column and have an element that is zero at least, through after correction until have a neutral element at least in different rows, different lines, thereby obtain a complete allocative decision corresponding with these neutral elements.The theoretical foundation of the method is: in any row or column of benefit matrix, can not change optimum allocation after adding or deduct a constant.But its complexity is high, often enter endless loop, trace it to its cause be due to the application Hungary Algorithm, when also having unappropriated neutral element, choose at random neutral element in unmarked row or column as primary distribution, this assigned at random often causes entering endless loop just.In order to reduce the complexity of Hungary Algorithm, eliminate this randomness, the present invention has adopted improved Hungary Algorithm, and improved thought is: while in every row element, more than 0 element occurring, select the minimum row of 0 element to start to assign.Under the Hungary Algorithm of application enhancements solution multi-user and multi-channel environment, the step of channel perception assignment problem is as follows:

The first step: use optimization scheme detection time, K time user calculates in advance and detects N the obtainable maximum throughput of channel (at this, only considering the situation of K≤N), forms the maximum throughput matrix A _K,N.

Second step: will ask maximum Assignment Problems to convert to and ask minimum Assignment Problems, method is: making C is A _K,NMiddle maximum, use C-R _IjSubstitute all R _Ij.

The 3rd step: each row element deducts the minimum value of one's own profession; Each column element deducts the minimum value of these row.

The 4th step: use minimum straight line to cover all 0 in matrix.Suppose to have used the l bar: if 1. l<K, making m is the minimum value be not capped in element, and each does not cover element and deducts m, and m is added on the element that two straight lines intersect, and repeats the 4th step.If 2. l=K, enter the 5th step.

The 5th step: from the first row, always start downwards to assign.Only in one's own profession, only have one 0 o'clock, this channel is assigned to 0 inferior user.Delete immediately these row of one's own profession once complete appointment.If can not complete K assign and every row element in more than one 0, from row, start, select the minimum row of 0 element to start appointment.Repeat the appointment of row and column, until carried out abundant unique appointment, if still have ungratifiedly, assign arbitrarily one 0.

Improved Hungary Algorithm has been eliminated the randomness in former algorithm, not there will be endless loop, and complexity will reduce greatly than former algorithm, and performance does not reduce.

Claims (1)

1. the system of selection of a multi-user sensing channel in cognitive sensor network is characterized in that the method comprises the following steps:

A. time user i is detected channel j, i=1, and 2 ..., K, K is time number of users, j=1,2 ..., N, N is the number of channel, calculates time obtainable Normalized throughput of user i Wherein T is sense cycle, and τ is detection time, P _fFor the false alarm probability required, H ₀Mean that the primary user does not exist, P (H ₀) mean that there is not primary user's probability in channel j;

B. when T is fixed value, find optimum τ, be designated as τ ^*, make R _Ij(τ) maximum, be designated as R _Ij

C.K time user calculates and detects N the obtainable maximum throughput R of channel according to received signal to noise ratio _Ij, this K * N maximum throughput forms the maximum throughput moment matrix $A_{K,N}=\left[\begin{array}{ccccc}{R}_{11}& R_{12}& R_{13}& ......& R_{1N}\\ R_{21}& R_{22}& R_{23}& ......& R_{2N}\\ ...& ...& ...& ......& ...\\ R_{i1}& R_{i2}& R_{i3}& ......& R_{\mathrm{iN}}\\ ...& ...& ...& ......& ...\\ R_{K1}& R_{K2}& R_{K3}& ......& R_{\mathrm{KN}}\end{array}\right]$

D. a shilling C is matrix A _K,NMiddle maximum, use C-R _IjReplace A _K,NThe element R of correspondence in matrix _IjThen each row element deducts the minimum value in this row element, and finally each column element deducts the minimum value in this column element;

E. use minimum straight line to cover all 0 elements in matrix, suppose to have used l bar straight line, if l<K, making m is the minimum value be not capped in element, the element not capped to each deducts m, and each elements of two straight line infalls is all added to m, repeating step e; If l=K, enter step f;

F. according to the row order, from the first row, start to carry out successively Channel Assignment, while only only having 0 element in one's own profession, the channel that 0 element is corresponding is assigned to the inferior user that this row is corresponding, once complete appointment, deletes immediately these row of one's own profession; If can not complete K row of assigning and existing in element more than 0, consider row corresponding to this row 0 element, select to contain the minimum row of 0 element, this is listed as to corresponding Channel Assignment to inferior user corresponding to this row; Repeating above-mentioned appointment process, until carried out abundant unique appointment, if still have ungratifiedly, is that remaining user assigns arbitrarily a channel.

Images