CN104994593A - Rapid adaptive control method based on channel state run - Google Patents

Abstract

The invention discloses a rapid adaptive control method based on channel state run. In combination with a pPCA (p Persistent Control Algorithm), the method comprises the following specific steps: initializing a system variable; judging whether or not mod(t, RW) is equal to 0; counting the quantity of current time slot transmission nodes; and performing rapid algorithm processing. In combination with a PBCA (pseudo-Bayesian Control Algorithm), the method comprises the following specific steps: transmitting a data packet; performing rapid adaptive algorithm processing; performing normal pseudo-Bayesian algorithm processing; transmitting a request packet; counting the quantity of idle time slots in one RW; and calculating Pidle. The rapid adaptive control method has the beneficial effects that the channel utilization ratio of a burst network environment in which the node quantity changes drastically can be increased remarkably. The conflict among the nodes can be relieved when communication nodes in a channel increase dramatically, so that the probability of successful node transmission is increased. Likewise, when the quantity of the communication data nodes in the system decreases suddenly, the transmitting probability of each node can be adjusted rapidly, so that the idleness rate of the channel is lowered, and the throughput of the whole system is increased.

Description

A kind of quick self-adapted control method based on the channel status distance of swimming

Technical field

The present invention relates to Ultra-Wide Band wireless network and channel allocation technique field, particularly a kind of quick self-adapted control method based on the channel status distance of swimming.

Background technology

CDMA slotted ALOHA needs to use control algolithm to obtain stable throughput with guarantee system, and the key of control algolithm is the nodes in accurate estimating system.When the nodes in system, in certain moment, drastic change occurs, there is the longer problem of regulating time in traditional control algolithm.

CDMA slotted ALOHA and improved protocol thereof are widely used in satellite communication, under the environment that the time delays such as GSM digital cellular network and tag recognition, cognition wireless network, In-vehicle networking and underwater sensing network are longer due to its simplicity.But CDMA slotted ALOHA is unstable in essence, for solving the unsteadiness of CDMA slotted ALOHA.Prior art proposes and adopt different retransmission probability P in difference overall input load situation _rthe mode of probability obtain the stability of system.At the new bag production rate of control, allow under reject rate and transmission channel exist the multiple environment such as error probability to a certain degree, limited number of time retransmission mechanism is on the impact of the stability of a system.Continuous state and the internode collaboration of foundation system analyze systematic function, and are incorporated in channel competition by theory of games, and have studied channel utilization and the throughput of slotted ALOHA system with this.(English is pseudo-Bayesian Control Algorithms to pseudo-Bayes's control algolithm that prior art proposes, hereinafter referred to as PBCA algorithm) and p adhere to that (English is Adaptive p Persistent Control Algorithm to control algolithm, hereinafter referred to as pPCA algorithm) be all by the accurate estimation to practical communication node in system, and the data sending probability adjusting each node is to reach the object of adjustment System stability.

When system node number occurs sharply to change, will there is a large amount of idle or conflict time slot in channel, cause channel utilization very low.Research shows, when estimating that the ratio of nodes and actual node number is between 0.5 ~ 2, throughput of system is greater than 60% of theoretical maximum throughput.Therefore, if certain control algolithm can be adopted to estimate, the ratio of nodes people having the same aspiration and interest system actual node number is whole between 0.5 ~ 2, and then adopts intense adjustment algorithm, must improve the entire throughput of system.

The system of finite element network node, adopts CDMA slotted ALOHA mode sharing wireless channel.Time shaft is divided into the identical time slot of size, the time of slot length needed for transmission Frame, and each node can only just allow when time slot starts to send data.When two and above node send data simultaneously, generation is conflicted.Clash ground node and retransmit the packet producing conflict at subsequent timeslot.Therefore, channel comprises three kinds of states at any time slot: idle, success and conflict, uses that { 0,1, c} represents respectively.

Suppose at certain time slot moment t ₀system actual node number is N, and the estimation nodes of system is M, and each node carries out transfer of data to obtain maximum throughput with probability P=1/M.Then the probability of channel idle, Successful transmissions and conflict is respectively:

$\left\{\begin{array}{c}{P}_{idle}=\left(\begin{array}{c}N\\ 0\end{array}\right){(1-P_0)}^N={\left(1-\frac{1}{M}\right)}^N\≈e^{-\frac{N}{M}}=e^{-\mathrm{\β}}\\ P_{succ}=\left(\begin{array}{c}N\\ 1\end{array}\right)P_0{(1-P_0)}^{N-1}=\frac{N}{M}{\left(1-\frac{1}{M}\right)}^{N-1}\≈{\mathrm{\βe}}^{-\mathrm{\β}}\\ P_{coll}=1-P_{idle}-P_{succ}=1-(1+\mathrm{\β})e^{-\mathrm{\β}}\end{array}\right.---\left(1\right)$

Wherein, β=N/M is actual node number and the ratio estimating nodes.During β=0.5, P _idle≈ 0.61, P _coll≈ 0.09; When β=2, P _idle≈ 0.14, P _coll≈ 0.60.β is less, channel idle probability P _idlelarger, channel clashes probability P _collless; Otherwise β is larger, letter P _idlemore size, P _colllarger.

To P _idleboth members gets natural logrithm, can obtain real network nodes N,

${\mathrm{lnP}}_{idle}=-\frac{N}{M}\⇒N=-{\mathrm{MlnP}}_{idle}---\left(2\right)$

Wherein, ln x is to the most natural logrithm of x.According to (2) formula, the space time slot probability of (defining this time period is probability updating window RW:Renew Windows) channel in statistics a period of time, in conjunction with the system node number M of supposition, can draw the actual node number N of system.In next RW, each network node adopts new Probability p=1/N to carry out transfer of data again.By the further abbreviation of (2) formula, obtain next data sending probability P' of more each node in new window,

$\begin{array}{c}\frac{1}{N}=-\frac{1}{M}\frac{1}{{\mathrm{lnP}}_{idle}}\\ P^{\′}=-P_0/{\mathrm{lnP}}_{idle}=P_0/\ln \left(1/P_{idle}\right)\end{array}---\left(3\right)$

According to (3) formula conclusion, design pPCA algorithm control flow is as follows:

I () is at a certain initial time t ₀, initialization nodes assumed value n ₀, calculate p=1/n ₀, setting upgrades window value RW, defines a statistics timeslot number variable S, and is initialized as 0;

(ii) judge whether S equals RW, if so, then S=0 forwarding to (iv);

(iii) S adds 1, generates equally distributed random number x, judges x≤p, if so, then send data, turns to (ii);

(iv) add up the free timeslot number of a RW, calculate P _idleif, P _idle=0 or P _idle=1, p remains unchanged, otherwise p=-p/ln P _idle, forward to (iii).

In this algorithm design, more new window RW is a very important parameter.RW settings are too little, and probable error increases, and degree of regulation declines; RW too senior general causes adjustment process to delay.

Adopt PBCA algorithm slotted ALOHA system then to carry out the adjustment of next time slot sending probability according to the state of current time slots, algorithm realization step is as follows:

I () is at time slot v, assuming that system node number is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) need of next time slot send the nodes N of data _v+1estimate with following formula:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\λ},N_v+\mathrm{\λ}-1),& idleorsuccess\\ N_v+\mathrm{\λ}+{(e-2)}^{-1},& collision\end{array}\right.---\left(4\right)$

Wherein, λ is new bag arrival rate, and e is natural logrithm.

(iii) each node of next time slot is with 1/N _v+1probability send packet.

The defect of PBCA control algolithm is:

Owing to calculating this method in each adjustment, the estimation of nodes is determined according to the channel status of previous time slot, and its basic assumption is when channel clashes, and the average nodal number participating in conflict is 2.39.But in the data network system taking burst flow as main business; whether stable nodes change in system is; frequent meeting occurred sharply to change in certain moment; now PBCA algorithm needs experience just can reach stable for a long time; the throughput of system is also very low, and the utilance of system channel is also very low.

Introduce the channel status distance of swimming of CDMA slotted ALOHA below,

In the slotted ALOHA system of definition 1:N node, channel occurs that the length of certain state is called the state distance of swimming of channel continuously, and company commander is that x certain channel status sequence claims the channel status distance of swimming to be x.

Dichotomy model is used channel status to be divided into free time (or conflict) state E and busy (non conflicting) state and the probability setting E to occur is as p.With t (t=1,2 ..., RW) and represent timeslot number, represent the channel distance of swimming state in more new window from t time slot with s (t), assuming that each non-slotted channel state is independent, then s (t) is one dimension Markov chain.According to the definition of the distance of swimming, distance of swimming k >=1, in order to describe the transfer process of channel status, increasing state 0, obtaining the distance of swimming state transitions of the channel status E of slotted ALOHA system

The state transition probability of s (t) is:

$\left\{\begin{array}{cc}{P}_{i0}=1-p,& i=0,1,...,RW\\ P_{i,i+1}=p,& i=0,1,...,RW-1\end{array}\right.---\left(5\right)$

The difference equation of the channel status distance of swimming

For research slotted ALOHA system is at the run distribution of different node and transmission probability lower channel state, first with sample value element number for H (H >=2), each sample value is reference model according to the random experiment X being uniformly distributed (P=1/H), analyzes the run distribution situation of a certain state E (being assumed to X=1).In n independent experiment, population sample number is H ⁿ, occur that the number of samples of the event E that the distance of swimming is R can describe with the arrangement shown in Figure 14 and Figure 15.Wherein " " represents the value (not comprising event E) of mobile element, and "○" represents that the distance of swimming is the sequence of R, and " ▓ " represents all elements.

Make ξ=n-k, represent that in all sample spaces, the distance of swimming is k number of samples with G (ξ).As can be seen from Figure 14, G (ξ) value is:

(i) as k=n, G (0)=1;

(ii) as k=n-1, only have a mobile element, its value does not comprise event E, and the distance of swimming is that the sequence of k has two possible positions, therefore G (1)=2 (H-1);

(iii) as k=n-2, G (2)=3H ²-4H+1;

(iv) as k < n-2, the sample number G (ξ) of the distance of swimming can carry out decomposition decomposition by model as shown in Figure 15.As can be seen from Figure 15, G (ξ+2) can be analyzed to 2 G (ξ+1) and deducts and arrange containing 2 free elements (namely value can be the arbitrary value in glossary of symbols) and 1 G (ξ), namely

Solve the difference equation of (6) formula, and substitute into initial value G (1), G (2),

Replace variable ξ with k, represent that in n test, the distance of swimming is the possible occurrence number of k with RL (k), then

$RL\left(k\right)=\left\{\begin{array}{cc}1,& k=n\\ (\frac{H^2-1}{H^2}+\frac{{(H-1)}^2}{H^2}(n-k))H^{n-k},& 1\&le;k<n\end{array}\right.---\left(8\right)$

P=1/H is substituted into (7), and abbreviation obtains

$RL\left(k\right)=\{\begin{array}{cc}1,& k=0\\ \left(1-p\right)\left(2+\left(1-p\right)\left(n-k-1\right)\right)p^{k-n},& 1\&le;k<n\end{array}---\left(9\right)$

At n independent experiment, each test can value H kind, total sample number z (n)=H ⁿ=p ^-n, then in average each sample length to be the distance of swimming number of k be:

$g\left(k\right)=\left\{\begin{array}{cc}{p}^n,& k=n\\ \left(1-p\right)\left(2p+\left(1-p\right)\left(n-k-1\right)\right)p^k,& 1\&le;k<n\end{array}\right.---\left(10\right)$

In addition, if replaced by the ξ n-k in (6) formula, two ends are same divided by H ⁿ, to be then the difference equation of distance of swimming number g (k) frequency of the length k distance of swimming (claim g (k) to be certain state) of k be average each sample length

g(k)-2pg(k+1)+p ²g(k+2)＝0 (11)

Solve (11) formula and substitute into initial condition, can draw (10).

In the difference equation about certain state-event distance of swimming adopting combined method to obtain, each event number is integer and according to being uniformly distributed appearance, i.e. Event element value H >=2, corresponding event Probability p=1/H≤0.5.But in slotted ALOHA system, although channel status is free time, Successful transmissions and conflict 3 kinds of situations, its distribution is not equiprobable.

Adopt " dichotomy " model that probability space is divided into [0, p] and [p, 1] two parts, in MATLAB, use Monte Carlo method to simulate this random experiment, then adding up the distance of swimming in each sample is the frequency g'(k of k), and obtain conclusion with (11) formula and compare, and the difference of both definition is δ (k), that is:

δ(k)＝|g'(k)-g(k)| (12)

Simulated environmental parameters is set to: sample size n sample range=100, number of samples (number of repetition in simulation) m=100000.Obtain correlated results as shown in table 1.

The distance of swimming statistical value that table 1 dichotomy generates compares with theoretical value

The numerical result display of table 1, when sample size is very large, the notional result that analog result is closely obtained by (10), and along with the error delta (k) of both increases of distance of swimming k more and more less and be tending towards 0, show g (k) and g'(k) meet same difference equation (11).

Can draw thus, if the probability that certain event E occurs is q, its exclusive events the probability occurred is 1-q.In n independent experiment, in average each sample, length is that the distance of swimming number of k meets difference equation

g(k)-2qg(k+1)+q ²g(k+2)＝0 (13)

The probability distribution of the channel status distance of swimming

The Radio Network System of application CDMA slotted ALOHA agreement shared channel, each non-slotted channel has three kinds of states: free time, Successful transmissions and conflict." dichotomy " principle is adopted to be idle condition E and busy state by slotted-ALOHA channelv state demarcation success status E and non-functional state or conflict situation E and non conflicting state in description afterwards, be referred to as state E and the mutual exclusion state of channel using the channel status of each time slot in more new window as a sample, the distance of swimming of certain state represents in sample must have this state E to occur.

Definition 2: in each sample, state E length is probability density function f (k) that the distance of swimming of k expects the ratio of the number of times expectation number of times frequency total with the various distance of swimming of this state to be length be the k state distance of swimming, that is:

$f\left(k\right)=g\left(k\right)/\underset{x=1}{\overset{n}{\&Sigma;}}g\left(x\right)---\left(14\right)$

And

$\begin{array}{c}S=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}g\left(i\right)=p^n+2\left(1-p\right)\underset{i=1}{\overset{n-1}{\mathrm{\&Sigma;}}}{p}^i{\left(1-p\right)}^2\underset{i=1}{\overset{n-1}{\mathrm{\&Sigma;}}}\left(n-i-1\right)p^i\\ =p^n+2\left(1-p\right)\underset{i=1}{\overset{n-1}{\mathrm{\&Sigma;}}}{p}^i+\left(n-1\right){\left(1-p\right)}^2\underset{i=1}{\overset{n-1}{\mathrm{\&Sigma;}}}\left(n-i-1\right)p^i-{\left(1-p\right)}^2\underset{i=1}{\overset{n-1}{\mathrm{\&Sigma;}}}{\mathrm{ip}}^i\\ =p\&lsqb;1+\left(1-p\right)\left(n-1\right)\&rsqb;\end{array}---\left(15\right)$

Substitute into (14),

$f\left(k\right)=\left\{\begin{array}{cc}\frac{p^{n-1}}{1+\left(1-p\right)\left(n-1\right)},& k=n\\ \frac{\left(1-p\right)\&lsqb;2+\left(1-p\right)\left(n-k-1\right)\&rsqb;p^{k-1}}{1+\left(1-p\right)\left(n-1\right)},& 1\&le;k<n\end{array}\right.---\left(16\right)$

As n → ∞,

$f\left(k\right)=\{\begin{array}{cc}\frac{1}{n}{p}^{n-1},& k=0\\ \left(1-p\right)p^{k-1},& 1\&le;k<n\end{array}---\left(17\right)$

Formula (17) shows when n is larger, and the length of certain state of channel is that the run distribution of k can be approximately geometry distribution variables.In the sample that the transfer of data of n time slot is configured to, represent the distance of swimming of certain state E of channel with variable R um, the probability distribution that length is not less than the distance of swimming of k is

$\begin{array}{c}F\left(k\right)=\mathrm{Pr}\left\{Rum>=k\right\}\\ F\left(k\right)=f\left(n\right)+\underset{t=k}{\overset{n-1}{\mathrm{\&Sigma;}}}f\left(t\right)\\ =\frac{p^{n-1}}{1+\left(1-p\right)\left(n-1\right)}+\underset{t=k}{\overset{n-1}{\mathrm{\&Sigma;}}}\frac{\left(1-p\right)\&lsqb;2+\left(1-p\right)\left(n-k-1\right)\&rsqb;p^{k-1}}{1+\left(1-p\right)\left(n-1\right)}\\ =\frac{1}{1+\left(1-p\right)\left(n-1\right)}\&lsqb;p^{n-1}+\underset{t=k}{\overset{n-1}{\mathrm{\&Sigma;}}}\left(1-p\right)\&lsqb;2+\left(1-p\right)\left(n-k-1\right)\&rsqb;p^{k-1}\&rsqb;\\ =\frac{1+\left(1-p\right)\left(n-k\right)}{1+\left(1-p\right)\left(n-1\right)}{p}^{k-1}\end{array}---\left(18\right)$

As n → ∞

F(k)＝P[Rum≥k]＝p ^k-1(19)

The recurrence period of the channel status k distance of swimming

G ^(k)describing carrying out the length that n data transfer obtains is in the channel status sample of n, and length is the average occurrence number of the distance of swimming of k, and formula (11) shows that the larger g (k) of n is larger.

Definition 3: when channel status probability is p, if carry out the individual transmission of the individual time slot of T (p, k), just at least may occur that 1 length is not less than the distance of swimming of the channel state of k, the distance of swimming then claiming T (p, k) to be this channel status is not less than the recurrence period of k.

In the channel status sample space formed by the transmitting procedure of n time slot, the number of times that the length that average each sample comprises is not less than the k state distance of swimming is:

$\begin{array}{c}Q\left(k\right)=g\left(n\right)+\underset{t=k}{\overset{n-1}{\mathrm{\&Sigma;}}}g\left(t\right)=p^n+\left(1-p\right)\&lsqb;2+\left(1-p\right)\left(n-k-1\right)\&rsqb;p^k\\ =\&lsqb;1+\left(1-p\right)\left(n-k\right)\&rsqb;p^k\end{array}---\left(20\right)$

According to the definition of T (p, k), its value is sample length n divided by the average time going out length in average each sample and be not less than the channel status distance of swimming of k, then

$T(p,k)=\frac{n}{Q\left(k\right)}=\frac{n}{\&lsqb;1+(1-p)(n-k)\&rsqb;p^k}---\left(21\right)$

When sample length n equals recurrence period T (p, the k) of the channel status distance of swimming, then occur in sample that length is not less than number of times Q (k)=1 of the channel status distance of swimming of k, i.e. [1+ (1-p) (n-k)] p ^k=1, Xiang Yingyou:

$T(p,k)=k+\frac{p^{-k}-1}{1-p}---\left(22\right)$

Summary of the invention

The present invention is directed to the defect of prior art, provide a kind of computational methods based on the adaptive control of the channel status distance of swimming, effectively can solve above-mentioned prior art Problems existing.

In order to realize above goal of the invention, the technical scheme that the present invention takes is as follows:

Based on a quick self-adapted control method for the channel status distance of swimming, it is characterized in that: in conjunction with pPCA algorithm with in conjunction with PBCA algorithm realization;

The described concrete steps in conjunction with pPCA algorithm are:

I () is located at initial time t ₀before, systems stabilisation nodes is n ₀, each node is with Probability p=1/n ₀send data; First carry out system variable initialization, comprise and upgrade window value RW, simulation time timeslot number t; At t ₀after moment, system actual node number is n; Nodes M is estimated, idle and conflict distance of swimming variable SS in current more new window ₀, SS ₁;

(ii) determine whether that mod (t, RW)=0 forwards to (v), otherwise, forward to (iii);

(iii) time slot counter adds 1 (t=t+1), the number S of statistics current time slots transmission node ₀and determine channel status S ₁;

(iv) fast algorithm process is carried out

if(S1＝＝0)then

{

SS1＝0；SS0＝SS0+1；

if(SS0>7)then M＝M/2；p＝1/M；goto(iii)

else if(S2＝＝0)then

{

SS0＝0；SS1＝SS1+1；

if(SS1>7)then M＝2M；p＝1/M；goto(iii)

}

Else

goto(ii)

As preferably, the described concrete steps in conjunction with PBCA algorithm are:

I () is at time slot v, assuming that the nodes in system is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) quick self-adapted algorithm process: if when systems axiol-ogy is the idle condition of 7 to the distance of swimming, by N _v+1=N _v/ 2; If when systems axiol-ogy is the conflict situation of 7 to the distance of swimming, by N _v+1=2N _v; Jump to (iv);

(iii) normal pseudo-bayesian algorithm process: the nodes following formula of the need transmission data group of next time slot is estimated:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\&lambda;},N_v+\mathrm{\&lambda;}-1),& idleorsuccess\\ N_v+\mathrm{\&lambda;}+{(e-2)}^{-1},& collision\end{array}\right.$

Wherein, λ is new bag arrival rate;

(iv) each node of next time slot is with 1/N _v+1probability send request grouping;

V the free timeslot number in () statistics RW, calculates P _idle, calculate next more new window interior nodes sending probability p=-p/ln P _idle, forward to (iii).

As preferably, the concrete steps of described pPCA algorithm are:

I () is located at initial time t ₀, initialization nodes assumed value n ₀, calculate p=1/n ₀, setting upgrades window value RW, defines a statistics timeslot number variable S, and is initialized as 0;

(ii) judge whether S equals RW, if so, then S=0 forwarding to (iv), otherwise, forward to (iii);

(iii) S adds 1, generates equally distributed random number x, judges x≤p, if so, then send data, turns to (ii);

(iv) add up the free timeslot number of a RW, calculate P _idleif, P _idle=0 or P _idle=1, p remains unchanged, otherwise p=-p/ln P _idle, forward to (iii).

As preferably, the concrete steps of described PBCA algorithm are:

I () is at time slot v, assuming that system node number is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) need of next time slot send the nodes N of data _v+1estimate with following formula:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\&lambda;},N_v+\mathrm{\&lambda;}-1),& idleorsuccess\\ N_v+\mathrm{\&lambda;}+{(e-2)}^{-1},& collision\end{array}\right.$

Wherein, λ is new bag arrival rate, and e is natural logrithm;

(iii) each node of next time slot is with 1/N _v+1probability send packet.

As preferably, the prerequisite of described design pPCA algorithm control flow is:

If time slot moment t ₀system actual node number is N, and the estimation nodes of system is M, and each node carries out transfer of data to obtain maximum throughput with probability P=1/M, then the probability of channel idle, Successful transmissions and conflict as shown in Equation 1:

$\left\{\begin{array}{c}{P}_{idle}=\left(\begin{array}{c}N\\ 0\end{array}\right){(1-P_0)}^N={\left(1-\frac{1}{M}\right)}^N\&ap;e^{-\frac{N}{M}}=e^{-\mathrm{\&beta;}}\\ P_{succ}=\left(\begin{array}{c}N\\ 1\end{array}\right)P_0{(1-P_0)}^{N-1}=\frac{N}{M}{\left(1-\frac{1}{M}\right)}^{N-1}\&ap;{\mathrm{\&beta;e}}^{-\mathrm{\&beta;}}\\ P_{coll}=1-P_{idle}-P_{succ}=1-(1+\mathrm{\&beta;})e^{-\mathrm{\&beta;}}\end{array}\right.---\left(1\right).$

As preferably, according to formula 1

To P _idleboth members gets natural logrithm, can obtain real network nodes N, can derive formula 2 to be:

${\mathrm{lnP}}_{idle}=-\frac{N}{M}\&DoubleRightArrow;N=-{\mathrm{MlnP}}_{idle}.$

As preferably, be that according to formula 2, the space time slot probability of channel in statistics a period of time, combines the system node number M supposed, draws the actual node number N of system to the most natural logrithm of x according to lnx in formula 2; In next RW, each network node adopts new Probability p=1/N to carry out transfer of data again; By formula 2 abbreviation, obtain next data sending probability P' of more each node in new window, can derive formula 3 is:

$\begin{array}{c}\frac{1}{N}=-\frac{1}{M}\frac{1}{{\mathrm{lnP}}_{ilde}}\\ P^{\&prime;}=-P_0/{\mathrm{lnP}}_{ilde}=P_0/\ln \left(1/P_{ilde}\right)\end{array}.$

As preferably, the conclusion of described formula 3 is the foundation of design pPCA algorithm.

Compared with prior art the invention has the advantages that: quick self-adapted control algolithm (FA) is according to the channel status detected, utilize Run Theory, when Channel Detection is to 7 consecutive collision time slots, immediately the estimation nodes of system is adjusted to 1/2 of previous time slot, when Channel Detection is to 7 continuous free timeslots, the estimation nodes of system is adjusted to 2 times of previous time slot.And revise the transmission data probability of each node immediately.This algorithm can significantly improve the channel utilization of nodes nonbursty network environment jumpy.Alleviate the conflict of node point when the communication node in channel sharply increases, improve the probability that node successfully transmits, equally, when the communication data nodes in system reduces suddenly, can the sending probability of each node of rapid adjustment, reduce the idleness of channel, improve the throughput of whole system.

Accompanying drawing explanation

Fig. 1 is the graph of a relation of embodiment of the present invention T (k) and Probability p;

Fig. 2 is embodiment of the present invention M=50 when becoming 1,2,10, FA-pPCA and pPCA adjustment process throughput schematic diagram;

Fig. 3 is embodiment of the present invention M=100 when becoming 1,2,10, FA-pPCA and pPCA adjustment process throughput schematic diagram;

Fig. 4 is embodiment of the present invention M=5 when becoming 20,50,100, FA-pPCA and pPCA adjustment process throughput schematic diagram;

Fig. 5 is embodiment of the present invention M=10 when becoming 20,50,100, FA-pPCA and pPCA adjustment process throughput schematic diagram;

Fig. 6 is embodiment of the present invention M=50 when becoming 1,2,10, FA-PBCA and PBCA throughput performance schematic diagram;

Fig. 7 is embodiment of the present invention M=100 when becoming 1,2,10, FA-PBCA and PBCA throughput performance schematic diagram;

Fig. 8 is embodiment of the present invention M=5 when becoming 20,50,100, FA-PBCA and PBCA throughput performance schematic diagram;

Fig. 9 is embodiment of the present invention M=10 when becoming 20,50,100, FA-PBCA and PBCA throughput performance schematic diagram;

Figure 10 is that embodiment of the present invention M=50 becomes other values constantly, and four kinds of control algolithms are at 100 time slot average throughput schematic diagrames;

Figure 11 is that embodiment of the present invention M=100 becomes other values constantly, and four kinds of control algolithms are at 100 time slot average throughput schematic diagrames;

Figure 12 is that embodiment of the present invention M=5 becomes other values constantly, and four kinds of control algolithms are at 100 time slot average throughput schematic diagrames;

Figure 13 is that embodiment of the present invention M=10 becomes other values constantly, and four kinds of control algolithms are at 100 time slot average throughput schematic diagrames;

Figure 14 is the distance of swimming of the present invention is k number of samples arrangement mode schematic diagram;

Figure 15 is sample number G (ξ) the arrangement mode schematic diagram of the distance of swimming of the present invention.

Embodiment

For making object of the present invention, technical scheme and advantage clearly understand, to develop simultaneously embodiment referring to accompanying drawing, the present invention is described in further details.

Adopt p to adhere to the slotted ALOHA system of control algolithm, through the transmission of RW time slot, count this more probability of free timeslot in new window.Then, each network node calculate according to (3) formula in next more new window probability P '=P ₀/ ln (1/P _idle) carry out data transmission.As can be seen from (3) formula:

(i) when 1, ln (1/P _idle) → 0, P' increases and is tending towards 1, but works as P _idlewhen=1, (3) formula is nonsensical.In fact, P is worked as _idlewhen=1, channel idle in whole RW, node does not carry out the transmission of data, and from (1) formula, N is very little, and M is very large.Namely be a very little value at some time etching system node from a very large value transition.

(ii) when 0, ln (1/P _idle) → ∞, P' reduces and is tending towards 0; But, work as P _idlewhen=0, (3) formula is nonsensical equally.And P _idlewhen=0, namely each time slot is carrying out transfer of data, and from formula (1), N is very large, and M is very little.Namely be a very large value at some time etching system node from a very little value transition.

Quick self-adapted algorithm principle:

Channel distribution is that (probability is p) and busy state to idle condition E by theorem 1: in slotted ALOHA system at one after the transmission of n (n > 7) individual time slot, if detect, channel clear length is the distance of swimming sequence of 7, then think p>=exp (-0.5) ≈ 0.61 in the one-sided confidential interval of 0.95, and the actual node number of system is less than and estimates 1/2 of nodes.

Prove: the distance of swimming representing channel clear E by variable R, P [R < k] >=1-α is the 1-α confidential interval of R, should have P [R >=k]≤α mutually.Again because of

P[R＜k]＝1-P[R≥k]＝1-F(k) (23)

From (17) formula, P [R < k] >=1-α be met, namely require that F (k) meets:

F(k)＝P[R≥k]＜1-0.95＝0.05 (24)

When n is larger, p=exp (-0.5) ≈ 0.61 is substituted into (19) formula, the inequality of formula (23) is transformed to:

Easily known by (19), F (k) is p ∈ [0,1] monotonically increasing function, formula (25) shows, when p < exp (-0.5), detect that the less F (k) of probability F (k) < 0.05, p of the distance of swimming of length k >=7 of channel clear E is also less.Otherwise, when p >=exp (-0.5), probability F (k) >=0.05 of the distance of swimming of length k >=7 of event E detected, and the larger F (k) of p is also larger.Therefore 0.95 confidential interval, when the distance of swimming of k=7 being detected, Probability p >=exp (-0.5) ≈ 0.61 that event E occurs can be thought.

By (1) formula, p=P _idleduring=exp (-N/M)>=exp (-0.5),

$N/M\&le;0.5\&DoubleRightArrow;N\&le;\frac{M}{2}---\left(26\right)$

Theorem 1 must be demonstrate,proved.

Adhere in control algolithm at the p of CDMA slotted ALOHA, the choose reasonable upgrading window value RW is significant.By theorem 1, when detecting the channel clear distance of swimming of length k=7, namely thinking that system node number is less than 1/2 of actual node number, then carrying out index replacement.From (9) formula, larger for the timeslot number n adding up transmitting procedure, generation length is that the idle condition distance of swimming number of k is more.On the other hand, when n is too small, even if channel clear Probability p > 0.61 is very large, still cannot obtain the idle condition distance of swimming that length is 7.Thus cannot immediate mode adjustment be carried out.

Formula (22) shows, k determines situation, and the Probability p that channel clear E occurs is larger, and length is that distance of swimming return period T (k) of k is less; When p determines, k is less, and T (k) is less.MATLAB computational tool is used to obtain the relation of the Probability p that T (k) occurs with channel clear E as shown in Figure 1.As seen from Figure 1, as p=0.7, approximately every 40 tests will obtain the distance of swimming sequence that a length is not less than 7, i.e. T (0.7,7) ≈ 40.About 30 tests can obtain and obtain the distance of swimming sequence that a length is not less than 6, T (0.7,6) ≈ 30.And as p=0.61, T (0.61,7) ≈ 80, T (0.61,6) ≈ 50.Channel clear probability of happening is larger, and length is that the distance of swimming of k is shorter for recurrence period.

When channel clear probability of happening p=exp (-N/M)=0.7, by (1) formula, N/M ≈ 0.37, namely estimates that nodes is about 3 times of actual node number.In application FA algorithmic system, the distance of swimming occurring at most being more than or equal to for 1 time 7 in the window of a renewal must be ensured.And when estimating that node is actual 3 times, also can only carry out a rapid adjustment, therefore, more the value of new window should meet as far as possible

$T(0.7,6)<RW\&le;T(0.7,7)\&DoubleRightArrow;30<RW\&le;40---\left(27\right)$

In pPCA, renewal window value is set to 32.

Quick self-adapted algorithm design:

Last joint theorem shows, if be consecutively detected 7 idle conditions (length is the channel clear distance of swimming of 7) at more new window RW, the one-sided confidential interval 0.95 thinks channel idle probability P _idle>=0.61, system actual node number is less than for current window estimates 0.5 of node.Estimation nodes M is updated to 1/2 original (i.e. M=M/2), and node uses new probability P=1/M to carry out transfer of data simultaneously.In like manner, channel status is divided into conflict and non conflicting two states, by (1) formula, when the actual node number N of system is 2 times that estimate nodes, the probability P of channel confliction _coll≈ 0.6.And β=N/M is larger, P _colllarger.When systems axiol-ogy length is the conflict situation distance of swimming of 7, the confidential interval one-sided 0.95 also can think that actual node number N is greater than 2 times that estimate nodes M, and the sending probability of control algolithm and adjustable nodal is original 1/2.

Row index adjustment process of going forward side by side based on the channel status distance of swimming detected is called quick self-adapted adjustment algorithm (FA:Fast Adaptive).Adhere to that quick self-adapted p that algorithm is set up adheres to that control algolithm (FA-pPCA) performs flow process in conjunction with p as follows:

I () is located at certain initial time t ₀before, systems stabilisation nodes is n ₀, each node is with Probability p=1/n ₀send data; First carry out system variable initialization, comprise and upgrade window value RW, simulation time timeslot number t; At t ₀after moment, system actual node number is n; Nodes M is estimated, idle and conflict distance of swimming variable SS in current more new window ₀, SS ₁;

(ii) mod (t, RW)=0 is judged? forward to (v);

(iii) time slot counter adds 1 (t=t+1), the number S of statistics current time slots transmission node ₀and determine channel status S ₁;

(iv) fast algorithm process is carried out

if(S1＝＝0)then

{

SS1＝0；SS0＝SS0+1；

if(SS0>7)then M＝M/2；p＝1/M；goto(iii)

else if(S2＝＝0)then

{

SS0＝0；SS1＝SS1+1；

if(SS1>7)then M＝2M；p＝1/M；goto(iii)

}

Else

goto(ii)

V the free timeslot number in () statistics RW, calculates P _idle, calculate next more new window interior nodes sending probability p=-p/ln P _idle, forward to (iii).

In like manner, quick self-adapted pseudo-Bayes's control algolithm (FA-PBCA) algorithm realization step is as follows:

I () is at time slot v, assuming that the nodes in system is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) quick self-adapted algorithm process: if when systems axiol-ogy is the idle condition of 7 to the distance of swimming, by N _v+1=N _v/ 2; If when systems axiol-ogy is the conflict situation of 7 to the distance of swimming, by N _v+1=2N _v; Jump to (iv);

(iii) normal pseudo-bayesian algorithm process: the nodes following formula of the need transmission data group of next time slot is estimated:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\&lambda;},N_v+\mathrm{\&lambda;}-1),& idleorsuccess\\ N_v+\mathrm{\&lambda;}+{(e-2)}^{-1},& collision\end{array}\right.---\left(28\right)$

Wherein, λ is new bag arrival rate

(iv) each node of next time slot is with 1/N _v+1probability send request grouping.

Algorithm simulating and checking:

Throughput of system is the important index of evaluating network performance, and in the MAC protocol based on competition, high throughput also means low time delay.Utilize under MATLAB instrument and FA-pPCA is emulated in the throughput of stability control events and regulating time.Simulated environment arranges as follows: initial time t ₀=0, nodes is changed to n by m, and simulation time is 100 time slots.

As shown in Figure 2,3, be at m for comparatively large (50,100) value drastic change for comparatively trifle is counted in (2,5,10) situation, in pPCA algorithm and FA-pPCA adjustment process system throughput and reach stable throughput required time comparison diagram.As can be seen from Fig. 2,3, as β=m/n > 2, and the value of m and n is when differing greatly, FA-pPCA once adjusts with regard to needs substantially through 7 time slots, as 1 < m/n < 2, the adjustment process of FA-pPCA is identical with pPCA, namely often just once adjusts through CW (CW=32) time slot.Simulation result shows, and pPCA algorithm adjustment process has obvious jumping characteristic, and FA-pPCA is in course of adjustment throughput then relative smooth.β larger FA-pPCA governing speed is more obvious.

A system having a stable state of counting compared with trifle, at moment t ₀when the drastic change of system node number is bigger numerical, FA-pPCA and pPCA stability control events and throughput thereof are as shown in Figure 3,4.Estimation nodes m can adjust in [0.5,1] scope of system actual node number n by simulation result display FA-pPCA algorithm fast.In adjustment process, the throughput of FA-pPCA is higher than the throughput of pPCA.Due to the impact by window value CW, when nodes is very large, in pPCA control system, channel confliction aggravation, causes throughput of system very low.Generally need could obtain stable approximation theory maximum throughput through 4 adjustment window (about 128 time slots) systems.And through rapid adjustment several times, FA-pPCA control algolithm only need can estimate that nodes be tied to [0.5 of system actual node number n, 1] in scope, and then the maximum system throughput that just substantially can reach close to theory is regulated through a widow time CW.Notice that the ratio beta=m/n as the estimation nodes m of system and the actual node number n of system is moderate, a certain section of slot range FA-pPCA throughput a little less than pPCA throughput, this is because FA-pPCA is after rapid adjustment, its time slot counter will delay 7k (k is rapid adjustment number of times) individual time slot, therefore, 7k time slot has also just been delayed in the adjustment of relative pPCA, when β value is moderate, idle probability statistics are relatively more accurate, and after adjustment, system can reach substantially close to theoretical maximum stable throughput.

Fig. 6,7 is under smaller value (2,5,10) environment for system start node number is higher value (m=50,100) drastic change, the average throughput of each time in system fading margin process.Fig. 6,7 Emulating display, when initial nodes is larger, and change deutomerite and count less, PBCA throughput less, the time reached needed for stable maximum throughput is longer.And throughput of system can be adjusted to stable maximum throughput by FA-PBCA fast.Fig. 8,9 is the less (m=5 of start node number, 10) system FA-PBCA and PBCA algorithm adjusting function comparison diagram when drastic change is higher value (20,50,100), can find out that FA-PBCA can significantly improve the adjusting function of system from simulation result, accelerate governing speed.Comparison diagram 6,7 and Fig. 8,9, number of network node is greater than network node from the regulating time that larger situation becomes less situation and becomes higher value from smaller value.

For the system throughput performance under different node situation of change of test FA-PBCA, PBCA, FA-pPCA and pPCA, emulate at MATLAB, simulation time is 100 time slots, calculates the average throughput of system under various scene.

As can be seen from Figure 10,11,12,13, estimate nodes m and actual node number n difference larger, comparatively pPCA throughput performance is higher for FA-pPCA, and it is high that FA-PBCA compares PBCA throughput performance.Under start node number comparatively large scene, pPCA average throughput performance is higher than PBCA throughput performance, and under the less scene of start node number, PBCA average throughput performance is then higher than pPCA throughput performance.As β ∈ [0.5,1] (β=m/n), FA-pPCA is basically identical compared with the throughput performance of pPCA, and FA-PBCA and PBCA throughput is also substantially identical.

Those of ordinary skill in the art will appreciate that, embodiment described here is to help reader understanding's implementation method of the present invention, should be understood to that protection scope of the present invention is not limited to so special statement and embodiment.Those of ordinary skill in the art can make various other various concrete distortion and combination of not departing from essence of the present invention according to these technology enlightenment disclosed by the invention, and these distortion and combination are still in protection scope of the present invention.

Claims (8)

1. based on a quick self-adapted control method for the channel status distance of swimming, it is characterized in that: in conjunction with pPCA algorithm with in conjunction with PBCA algorithm realization;

The described concrete steps in conjunction with pPCA algorithm are:

I () is located at initial time t ₀before, systems stabilisation nodes is n ₀, each node is with Probability p=1/n ₀send data; First carry out system variable initialization, comprise and upgrade window value RW, simulation time timeslot number t; At t ₀after moment, system actual node number is _n; Nodes M is estimated, idle and conflict distance of swimming variable SS in current more new window ₀, SS ₁;

(ii) determine whether that mod (t, RW)=0 forwards to (v), otherwise, forward to (iii);

(iii) time slot counter adds 1 (t=t+1), the number S of statistics current time slots transmission node ₀and determine channel status S ₁;

(iv) fast algorithm process is carried out

if(S1＝＝0)then

{

SS1＝0；SS0＝SS0+1；

if(SS0>7)then M＝M/2；p＝1/M；goto(iii)

else if(S2＝＝0)then

{

SS0＝0；SS1＝SS1+1；

if(SS1>7)then M＝2M；p＝1/M；goto(iii)

}

Else

goto(ii)。

2. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 1, is characterized in that: the described concrete steps in conjunction with PBCA algorithm are:

I () is at time slot v, assuming that the nodes in system is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) quick self-adapted algorithm process: if when systems axiol-ogy is the idle condition of 7 to the distance of swimming, by N _v+1=N _v/ 2; If when systems axiol-ogy is the conflict situation of 7 to the distance of swimming, by N _v+1=2N _v; Jump to (iv);

(iii) normal pseudo-bayesian algorithm process: the nodes following formula of the need transmission data group of next time slot is estimated:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\&lambda;},N_v+\mathrm{\&lambda;}-1),& idleorsuccess\\ N_v+\mathrm{\&lambda;}+{(e-2)}^{-1},& collision\end{array}\right.$

Wherein, λ is new bag arrival rate;

(iv) each node of next time slot is with 1/N _v+1probability send request grouping;

V the free timeslot number in () statistics RW, calculates P _idle, calculate next more new window interior nodes sending probability p=-p/lnP _idle, forward to (iii).

3. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 1, is characterized in that: the concrete steps of described pPCA algorithm are:

I () is located at initial time t ₀, initialization nodes assumed value n ₀, calculate p=1/n ₀, setting upgrades window value RW, defines a statistics timeslot number variable S, and is initialized as 0;

(ii) judge whether S equals RW, if so, then S=0 forwarding to (iv), otherwise, forward to (iii);

(iii) S adds 1, generates (0 ~ 1) equally distributed random number x, judges x≤p, if so, then send data, turn to (ii);

(iv) add up the free timeslot number of a RW, calculate P _idleif, P _idle=0 or P _idle=1, p remains unchanged, otherwise p=-p/lnP _idle, forward to (iii).

4., according to a kind of quick self-adapted control method based on the channel status distance of swimming of claim 1 or 2, it is characterized in that: the concrete steps of described PBCA algorithm are:

I () is at time slot v, assuming that system node number is N _v, each node is with probability q _r(N _v)=min{1,1/N _vsend packet;

(ii) need of next time slot send the nodes N of data _v+1estimate with following formula:

$N_{v+1}=\left\{\begin{array}{cc}max(\mathrm{\&lambda;},N_v+\mathrm{\&lambda;}-1),& idleorsuccess\\ N_v+\mathrm{\&lambda;}+{(e-2)}^{-1},& collision\end{array}\right.$

Wherein, λ is new bag arrival rate, and e is natural logrithm;

(iii) each node of next time slot is with 1/N _v+1probability send packet.

5. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 3, is characterized in that: the prerequisite of described design pPCA algorithm control flow is:

If time slot moment t ₀system actual node number is N, and the estimation nodes of system is M, and each node carries out transfer of data to obtain maximum throughput with probability P=1/M, then the probability of channel idle, Successful transmissions and conflict as shown in Equation 1:

$\left\{\begin{array}{c}{P}_{idle}=\left(\begin{array}{c}N\\ 0\end{array}\right){\left(1-P_0\right)}^N={\left(1-\frac{1}{M}\right)}^N\&ap;e^{-\frac{N}{M}}=e^{-\mathrm{\&beta;}}\\ P_{succ}=\left(\begin{array}{c}N\\ 1\end{array}\right)P_0{\left(1-P_0\right)}^{N-1}=\frac{N}{M}{\left(1-\frac{1}{M}\right)}^{N-1}\&ap;{\mathrm{\&beta;e}}^{-\mathrm{\&beta;}}\\ P_{cill}=1-P_{idle}-P_{succ}=1-(1+\mathrm{\&beta;})e^{-\mathrm{\&beta;}}\end{array}\right..$

6. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 5, is characterized in that: according to formula 1

To P _idleboth members gets natural logrithm, can obtain real network nodes N, can derive formula 2 to be:

$ln{P}_{idle}=-\frac{N}{M}\&DoubleRightArrow;N=-Mln{P}_{idle}.$

7. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 6, it is characterized in that: be to the most natural logrithm of x according to lnx in formula 2, according to formula 2, the space time slot probability of channel in statistics a period of time, in conjunction with the system node number M of supposition, draw the actual node number N of system; In next RW, each network node adopts new Probability p=1/N to carry out transfer of data again; By formula 2 abbreviation, obtain next data sending probability P' of more each node in new window, can derive formula 3 is:

$\frac{1}{N}=-\frac{1}{M}\frac{1}{ln{P}_{idle}}.$

P'＝-P ₀/lnP _idle＝P ₀/ln(1/P _idle)

8. a kind of quick self-adapted control method based on the channel status distance of swimming according to claim 7, is characterized in that: the conclusion of described formula 3 is the foundation of design pPCA algorithm.