CN101583146A - Adaptive resource allocation method for high-speed mobile 3G-HSDPA communication system - Google Patents

Abstract

The invention relates to an adaptive resource allocation method for a high-speed mobile 3G-HSDPA communication system. The system comprises a NodeB side and a UE side which mutually transmit communication signals. The method comprises the following steps that: the UE side reports a signal-to-noise ratio to the NodeB side at a certain time interval T; b, the NodeB side builds a prediction model according to the signal-to-noise ratio reported by the UE at certain time interval; c, the model is used to perform linear prediction to acquire a signal-to-noise ratio value in an actual data transmission process after the scheduling of the NodeB side; d, multi-user resource scheduling is performed according to the predicted signal-to-noise ratio value; e, according to the predicted signal-to-noise ratio value, data corresponding to a scheduled user are subjected to adaptive modulation and are transmitted to the corresponding UE side; and f, the NodeB side adjusts the parameters of the prediction model at the time interval T according to the signal-to-noise ratio reported by the UE side. The invention reduces system error rate on the basis of ensuring system throughput.

Description

High-speed mobile 3G-HSDPA communication system adaptive resource allocation method

Technical field

The invention belongs to communication technical field, relate in particular to a kind of high-speed mobile 3G-HSDPA communication system adaptive resource allocation method.

Background technology

Traditional HSDPA system is based on UE end (the UE end is user side), packet scheduling and Adaptive Modulation and Coding (AMC) are carried out in measurement according to the signal to noise ratio of pilot tone, yet UE end reports NodeB end (the NodeB end refers to the base station) and NodeB end with signal to noise ratio to carry out packet scheduling and all needs spended time, and promptly the pilot measurement of UE end and NodeB end carry out having certain delay between the transmission of data behind the packet scheduling.When the UE end is in the high speed motion (100 kilometers/hour to 250 kilometers/hour), rapid fading takes place at this section in time of delay in channel, and lifting along with UE end motion speed, the speed that rapid fading changes also can strengthen, make between signal to noise ratio that UE end reports and the signal to noise ratio in the actual data transfer process of NodeB end scheduling back and exist than mistake, thereby influence the packet scheduling and the Adaptive Modulation and Coding of HSDPA system, cause error rate of system to promote, throughput descends.

As shown in Figure 1, postpone owing to exist between the time of the time of UE measurement signal to noise ratio and transfer of data, the SNR1 that UE reports ..., the SNR1 ' of SNR4 and actual transmission channel ..., SNR4 ' is unequal, thereby has caused the error of system.

Summary of the invention

At the deficiencies in the prior art, a kind of high-speed mobile 3G-HSDPA communication system adaptive resource allocation method that improves throughput of system, reduces error rate of system that the present invention taked.

For achieving the above object, technical scheme of the present invention is: a kind of high-speed mobile 3G-HSDPA communication system adaptive resource allocation method, this system comprises the NodeB end and the UE end of mutual transmission signal of communication, and it may further comprise the steps: the a.UE end reports its snr value to hold to NodeB by the certain hour interval T; B.NodeB end is set up linear prediction model according to the UE signal to noise ratio that reports of being separated by; C. carry out linear prediction according to model, obtain the snr value in the actual data transfer process of NodeB end scheduling back; D.NodeB carries out Mulitiple user resource scheduling according to the snr value that linear prediction obtains; E. according to the snr value of predicting the data of scheduled user's correspondence are carried out Adaptive Modulation at last and send to corresponding UE end; F.NodeB holds per signal to noise ratio adjustment prediction model parameters that blanking time, T reported according to the UE end of receiving.

In step a and b, above-mentioned time interval T is 0.1ms.

In step c, the smnr data that the linear prediction method by AR model least mean-square error obtains after to interpolation is predicted.

The linear prediction method of AR model least mean-square error is as follows:

Suppose that y (n) is the SNR sequence that UE T blanking time reports NodeB, n=0,1 ..., N-1, the exponent number of AR model are IP, k _pBe reflection coefficient, e _p ^f(n), e _p ^b(n) be respectively the forward and backward predicated error,

$e_p^f\left(n\right)=e_{p-1}^f\left(n\right)+k_p{e}_{p-1}^b(n-1),$ $e_p^b\left(n\right)=e_{p-1}^b(n-1)+k_p^{*}{e}_{p-1}^f\left(n\right)$

The average power of predicated error is ${\mathrm{\ρ}}_p=\frac{1}{2}({\mathrm{\ρ}}_{p,e}+{\mathrm{\ρ}}_{p,b}),$ Wherein ${\mathrm{\ρ}}_{p,e}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^f\left(n\right)\right|}^2,$ ${\mathrm{\ρ}}_{p,b}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^b\left(n\right)\right|}^2,$

Utilize $\frac{\∂{\mathrm{\ρ}}_p}{\∂k_p}=0,$ Try to achieve the reflection coefficient k that makes predicated error average power minimum _p,

$k_p=\frac{-2\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{e}_{p-1}^f\left(n\right)e_{p-1}^{f*}(n-1)}{\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}({\left|e_{p-1}^f\left(n\right)\right|}^2+{\left|e_{p-1}^b(n-1)\right|}^2)}$

Initialization $e_0^f\left(n\right)=x\left(n\right),$ $e_0^b\left(n\right)=x\left(n\right),$ This moment p=0,

P=p+1 substitution k _pThe derivation formula obtains k ₁, and then substitution e _p ^f(n), e _p ^b(n) the derivation formula obtains e ₁ ^f(n), e ₁ ^b(n), afterwards according to a _{P, p}=k _p, $a_{p,i}=a_{p-1,i}+k_p{a}_{p-1,p-i}^{*}$ i＝1，2，…，p-1

By iteration repeatedly,, then can obtain whole coefficient a of IP rank AR model up to p=IP _{P, i}, i=1,2 ..., IP,

With the substitution of AR model coefficient $z\left(n\right)=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{a}_{p,i}y(n-i)$ The signal to noise ratio sequence z (n) that obtains predicting.

In step e, when the error rate of snr value requires 10 ^-2During rank, the signal to noise ratio sequential value is carried out the Adaptive Modulation of QPSK, 16QAM.

Requiring in the snr value error rate is 10 ^-2During the order of magnitude, the Adaptive Modulation process is: when SNR≤25dB, adopt the QPSK modulation system; When snr value SNR＞25dB, adopt the 16QAM modulation system.

Relative prior art, beneficial effect of the present invention is:

According to simulation result invent as can be seen proposed in high-speed mobile enhancement mode 3G-HSDPA communication system, utilize the channel linearity Predicting Technique, can make the scheduling of NodeB end and Adaptive Modulation predicted value according to signal to noise ratio in the actual data transfer process more accurately, thereby on the basis that guarantees throughput of system, reduced the error rate of system, effectively improve high speed UE end, i.e. the service quality of mobile client.In addition since invention suggest plans to changing lessly with traditional 3G-HSDPA communication system, and linear prediction calculates all to be held by NodeB and finishes, and can guarantee finishing of complicated algorithm, has higher feasibility and application value.

Description of drawings

Fig. 1 is the comparison diagram that prior art UE end reports the signal to noise ratio in signal to noise ratio and the NodeB end scheduling back actual data transfer process;

Fig. 2 is the handling process of signal to noise ratio of the present invention;

Fig. 3 is in the Rayleigh channel, and the UE translational speed is respectively 100 kilometers/hour, and 150 kilometers/hour and 250 kilometers/hour, carrier frequency is 2GHz, when fd (Doppler frequency shift)=185Hz, 278Hz, 500Hz, and signal envelope autocorrelation curve in time;

Fig. 4 is that NodeB utilizes the AR model to carry out the process of linear prediction according to the SNR sequential value in the invention system;

Fig. 5 is that the UE translational speed is 150 kilometers/hour in high-speed mobile enhancement mode 3G-HSDPA communication system, uses the comparison diagram between the actual signal to noise ratio of signal to noise ratio and system after interpolation of the present invention and the linear prediction;

Fig. 6 is that QPSK under the Rayleigh channel, 16QAM modulation system are 10 in error rate requirement ^-2Bit error rate performance curve during the order of magnitude;

Fig. 7 is to be 10 in the error rate ^-2The order of magnitude, UE translational speed are 150 kilometers/hour, the present invention and traditional HSDPA system under the identical situation of throughput, the comparison curves of error rate of system.

Embodiment

The invention provides a kind of high-speed mobile 3G-HSDPA communication system adaptive resource allocation method, this system comprises the NodeB end and the UE end of mutual transmission signal of communication, and as shown in Figure 2, it may further comprise the steps:

The a.UE end reports its snr value to hold to NodeB by the certain hour interval T;

The b.NodeB end is set up linear prediction model according to the signal to noise ratio (snr) that reports of being separated by;

C. carry out linear prediction according to model, obtain the snr value in the actual data transfer process of NodeB end scheduling back;

D. the snr value that obtains according to linear prediction again carries out Mulitiple user resource scheduling;

E. according to the snr value of predicting the data of scheduled user's correspondence are carried out Adaptive Modulation at last and send to corresponding UE end;

F.NodeB holds per signal to noise ratio adjustment prediction model parameters that blanking time, T reported according to the UE end of receiving.

The present invention adopts the channel linearity Predicting Technique, can make the scheduling of NodeB end and Adaptive Modulation predicted value according to signal to noise ratio in the actual data transfer process more accurately, thereby on the basis that guarantees throughput of system, reduced the error rate of system, effectively improved UE end at a high speed.

In step a and b, above-mentioned time interval T is 0.1ms.

As shown in Figure 3, among the present invention, the carrier frequency of signal is 2GHz, and when fd (Doppler frequency shift)=185Hz, 278Hz, 463Hz, corresponding respectively UE end translational speed is 100 kilometers/hour, 150 kilometers/hour and 250 kilometers/hour.As time interval T, be data dispatching unit when being 2ms, the linear dependence that declines between the adjacent thread illustrates in high speed HSDPA system all less than 0.5, be that the linear dependence that declines between every frame of frame length is too little with 2ms, therefore can not directly utilize model to carry out linear prediction.Yet when thread during less than 2ms, for example when thread was 0.1ms, the UE translational speed was 100 kilometers/hour, and the linear dependence that declines between the adjacent thread is 0.9966; When the UE translational speed is 150 kilometers/hour, the linear dependence that declines between the adjacent thread is 0.9924; When the UE translational speed is 250 kilometers/hour, the linear dependence that declines between the adjacent thread is 0.9790.And thread is more little as shown in Figure 3, and the decline linear dependence is high more between the adjacent thread.In order to guarantee the feasibility of next step linear prediction, the linear dependence that declines between the adjacent thread require when the UE translational speed reach 250 kilometers/hour on be not less than 0.9 in limited time, promptly UE reports the signal to noise ratio of one-shot measurement at least every 0.1ms.

In step c, predict according to the smnr data that obtains by the linear prediction method of AR model least mean-square error.

The AR model is a kind of linear prediction, be a known N data, can release the data of N point front or back by model, if release the P point, its essence is similar to interpolation, its purpose all is in order to increase valid data, and just the AR model is by N point recursion, and interpolation is to remove the multiple spot of deriving by 2 points (or minority what time).

Among the present invention, the linear prediction method of AR model least mean-square error is specific as follows:

Suppose that y (n) is the UE SNR sequence that blanking time, T reported, n=0,1 ..., N-1, the exponent number of AR model are IP, kp is a reflection coefficient, e _p ^f(n), e _p ^b(n) be respectively the forward and backward predicated error,

$e_p^f\left(n\right)=e_{p-1}^f\left(n\right)+k_p{e}_{p-1}^b(n-1),$ $e_p^b\left(n\right)=e_{p-1}^b(n-1)+k_p^{*}{e}_{p-1}^f\left(n\right)$

The average power of predicated error is ${\mathrm{\ρ}}_p=\frac{1}{2}({\mathrm{\ρ}}_{p,e}+{\mathrm{\ρ}}_{p,b}),$ Wherein ${\mathrm{\ρ}}_{p,e}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^f\left(n\right)\right|}^2,$ ${\mathrm{\ρ}}_{p,b}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^b\left(n\right)\right|}^2,$

Utilize $\frac{\∂{\mathrm{\ρ}}_p}{\∂k_p}=0,$ Try to achieve the reflection coefficient k that makes predicated error average power minimum _p, ρ wherein _p, k _pAnd e _pDeng in p refer to identical parameter,

$k_p=\frac{-2\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{e}_{p-1}^f\left(n\right)e_{p-1}^{f*}(n-1)}{\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}({\left|e_{p-1}^f\left(n\right)\right|}^2+{\left|e_{p-1}^b(n-1)\right|}^2)}$

Initialization $e_0^f\left(n\right)=x\left(n\right),$ $e_0^b\left(n\right)=x\left(n\right),$ This moment p=0,

P=p+1 substitution k _pThe derivation formula obtains k ₁, and then substitution e _p ^f(n), e _p ^b(n) the derivation formula obtains e ₁ ^f(n), e ₁ ^b(n), afterwards according to a _{P, p}=k _p, $a_{p,i}=a_{p-1,i}+k_p{a}_{p-1,p-i}^{*}$ i＝1，2，…，p-1，

By iteration repeatedly,, then can obtain whole coefficient a of IP rank AR model up to p=IP _{P, i}, i=1,2 ..., IP,

With the substitution of AR model coefficient $z\left(n\right)=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{a}_{p,i}y(n-i)$ The signal to noise ratio sequence z (n) that obtains predicting.

As shown in Figure 4, the SNR that invention adopts the AR model to utilize the UE end to report, predict to obtain the signal to noise ratio in the actual data transfer process behind the base station scheduling, carry out having certain problem that postpones between the transmission of data behind the packet scheduling thereby solved the pilot measurement of UE as shown in Figure 1 and NodeB.Suppose that herein the AR model is 20 rank, then last 20 data predictions of the SNR sequence that reports according to UE are to next SNR value constantly.

Adopt prediction scheme proposed by the invention emulation in high-speed mobile enhancement mode 3G-HSDPA communication system, result such as Fig. 5, the SNR of prediction can well follow the tracks of the variation of real system SNR, and error rate of system is 10 ^-2During the order of magnitude, the mean square error MSE between predicted value and the actual value＜4dB (MSE=(actual SNR-prediction SNR) ²), report the mean square error between snr value and the actual value on average to reduce 60dB than UE in the traditional 3G-HSDPA system.

In the step of front of the present invention, obtained 0.1ms and be current channel SNR value at interval, therefore can do Adaptive Modulation on this basis, thereby improved throughput of system and reliability of data transmission by linear prediction.

As shown in Figure 6, under the Rayleigh channel bit error rate performance of QPSK, 16QAM modulation system as can be known, in step e, when the error rate of snr value requires 10 ^-2During rank, can transmit QPSK, the 16QAM Adaptive Modulation of data according to the signal to noise ratio sequential value of prediction.

The bit error rate performance of emulation QPSK, 16QAM modulation system under Rayleigh channel, as Fig. 6, we can obtain table 1, and wherein table 1 is 10 for the error rate ^-2During the order of magnitude, the relation table of modulation system and channel signal to noise ratio (snr).Consider the lifting of throughput of system, we choose QPSK, 16QAM as adaptive modulation system, and obtaining requiring in the error rate is 10 ^-2The adaptive modulation scheme during order of magnitude: when SNR≤25dB, adopt the QPSK modulation system; When SNR＞25dB, adopt the 16QAM modulation system.

Signal to noise ratio (dB)	The error rate order of magnitude is 10 -3The Shi Kexuan modulation system
		SNR≤25	QPSK
SNR＞25	16QAM，QPSK

Table 1

Adopt scheme proposed by the invention in high-speed mobile enhancement mode 3G-HSDPA communication system, obtain SNR in the real data transport process based on interpolation and linear prediction method, adopting maximum S R algorithm to carry out resource allocation and Adaptive Modulation between the multi-user, is 10 in the error rate ^-2During the order of magnitude, as shown in Figure 7, under the throughput of system situation identical with traditional HSDPA system, the error rate has reduced about 68%.

In order to further describe, the present invention has enumerated a specific embodiment, and existing with 20 rank AR forecast models, it is 10 that the error rate requires ^-2The order of magnitude is the example explanation:

Step 1.NodeB receives the SNR value that UE reports every 0.1ms, and 8ms obtains 80 SNR values altogether;

Step 2 is got 80 SNR values that the first step obtains and is carried out 20 rank AR model coefficients and calculate a _{20, i}, i=1,2 ..., 20;

Step 3. get 80 SNR values that the first step obtains last 20 as y (n) (n=1,2 ..., 20) and substitution $z\left(n\right)=\underset{i=1}{\overset{20}{\mathrm{\Σ}}}{a}_{20,i}y(n-i),$ The SNR value of the 80.1ms channel that obtains predicting;

Step 4. obtains the foundation of 80.1ms channel SNR value as NodeB data packet dispatching this moment according to prediction, can adopt maximum CI algorithm or direct ratio fair algorithm to carry out the multi-user data packet scheduling;

User in step 5. scheduling obtains 80.1ms channel SNR value according to table 1 Adaptive Modulation and transmission according to its prediction, and requiring in the error rate is 10 ^-2During the order of magnitude: when SNR≤25dB, adopt the QPSK modulation system; When 25dB＜SNR, adopt the 16QAM modulation system.

Step 6.NodeB receives the SNR value of the 80.1ms that UE reports, adjusts prediction model parameters

Step 7. repeats the 2-5 step, promptly measurable 80.2ms channel SNR value, go forward side by side line data packet scheduling and Adaptive Transmission;

Step 8. repeating step 1-7 can finish the invention high-speed mobile enhancement mode 3G-HSDPA telecommunication system resources adaptive resource allocation method that proposes.

Claims (8)

1, a kind of high-speed mobile 3G-HSDPA communication system adaptive resource allocation method, this system comprises the base station and the user side of mutual transmission signal of communication, it is characterized in that may further comprise the steps:

A. user side reports its snr value to give the base station by the certain hour interval T;

B. linear prediction model is set up according to the user side signal to noise ratio that reports of being separated by in the base station;

C. carry out linear prediction according to linear prediction model, obtain the snr value in the actual data transfer process behind the base station scheduling;

D. the snr value that obtains according to linear prediction carries out Mulitiple user resource scheduling;

E. the snr value according to prediction carries out Adaptive Modulation with the data of scheduled user's correspondence and sends to corresponding user side;

2, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 1 is characterized in that: also comprise step f, base station per blanking time of T adjusts the linear prediction model parameter according to the signal to noise ratio that the user side of receiving reports.

3, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 1 is characterized in that: in step a and b, above-mentioned time interval T is 0.1ms.

4, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 1, it is characterized in that: in step c, d, the smnr data that the base station obtains is predicted by the linear prediction method of linear prediction model least mean-square error.

5, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 1, it is characterized in that: in step b, the linear prediction model method for building up is as follows:

Suppose the signal to noise ratio sequence that y (n) reports every time T for user side, n=0,1 ..., N-1, the exponent number of linear prediction model are IP, k _pBe reflection coefficient, e _p ^f(n), e _p ^b(n) be respectively the forward and backward predicated error,

$e_p^f\left(n\right)=e_{p-1}^f\left(n\right)+k_p{e}_{p-1}^b(n-1),$ $e_p^b\left(n\right)=e_{p-1}^b(n-1)+k_p^{*}{e}_{p-1}^f\left(n\right)$

The average power of predicated error is ${\mathrm{\ρ}}_p=\frac{1}{2}({\mathrm{\ρ}}_{p,e}+{\mathrm{\ρ}}_{p,b}),$ Wherein ${\mathrm{\ρ}}_{p,e}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^f\left(n\right)\right|}^2,$

${\mathrm{\ρ}}_{p,b}=\frac{1}{N-p}\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{\left|e_p^b\left(n\right)\right|}^2,$

Utilize $\frac{\∂{\mathrm{\ρ}}_p}{\∂k_p}=0,$ Try to achieve the reflection coefficient k that makes predicated error average power minimum _p,

$k_p=\frac{-2\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}{e}_{p-1}^f\left(n\right)e_{p-1}^{*}(n-1)}{\underset{n=p}{\overset{N-1}{\mathrm{\Σ}}}({\left|e_{p-1}^f\left(n\right)\right|}^2+{\left|e_{p-1}^b(n-1)\right|}^2)}$

Initialization $e_0^f\left(n\right)=x\left(n\right),$ $e_0^b\left(n\right)=x\left(n\right),$ This moment p=0,

P=p+1 substitution k _pThe derivation formula obtains k ₁, and then substitution e _p ^f(n), e _p ^b(n) the derivation formula obtains e ₁ ^f(n), e ₁ ^b(n), afterwards according to a _{P, p}=k _p, $a_{p,i}=a_{p-1,i}+k_p{a}_{p-1,p-i}^{*},$ i＝1，2，…，p-1，

By iteration repeatedly,, then can obtain whole coefficient a of IP rank linear prediction model up to p=IP _{P, i}, i=1,2 ..., IP.

6, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 5 is characterized in that: with the substitution of linear prediction model coefficient $z\left(n\right)=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{a}_{p,i}y(n-i)$ The signal to noise ratio z that obtains predicting (n).

7, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 1 is characterized in that: in step e, when the error rate of snr value requires 10 ^-2During rank, snr value is carried out the Adaptive Modulation of QPSK, 16QAM.

8, high-speed mobile 3G-HSDPA communication system adaptive resource allocation method according to claim 7 is characterized in that: requiring in the snr value error rate is 10 ^-2During the order of magnitude, the Adaptive Modulation process is: when SNR≤25dB, adopt the QPSK modulation system; When SNR＞25dB, adopt the 16QAM modulation system.

Images