CN105188125A - Power distribution method for integrally optimizing energy efficiency and spectrum efficiency of wireless network - Google Patents

Abstract

The invention discloses a power distribution method for integrally optimizing energy efficiency and spectrum efficiency of a wireless network. The power distribution method comprises the following steps: defining a cost function for integrally optimizing the energy efficiency and the spectrum efficiency, and performing iterative solution of the cost function so as to obtain a compromise of the energy efficiency and the spectrum efficiency. In each iteration, the user rate is approximated by searching the lower limit of the user rate; the optimal total transmission power of a base station is updated through a gradient method; and finally, the user power is distributed by utilizing a Lagrange duality method. When the user power is converged, iteration is over; and the optimal user power distribution is obtained. The invention provides the effective and rapid power distribution method capable of integrally optimizing the energy efficiency and the spectrum efficiency; the complexity due to a directly calculating method can be avoided; and simultaneously, operators can realize a compromise of the energy efficiency and the spectrum efficiency by adjusting weighting coefficients and flexibly configuring the proportion between the spectrum efficiency and the energy efficiency.

Description

The power distribution method of wireless network efficiency spectrum effect combined optimization

Technical field

The present invention relates to efficiency spectrum effect combined optimization optimization problem in mobile communication system, belong to the networking technology area in radio communication.

Background technology

Come into vogue along with the portable mobile terminal such as smart mobile phone and panel computer and popularize gradually, wireless data service amount rapid growth.But the raising speed of present spectrum efficiency of communication system does not catch up with the speed of demand data growth, and this problem is particularly severe in the hot spot region that flow of the people is intensive.

With orthogonal frequency division multiplexi (OFDM, and multi-input/output antenna technology (MIMO OrthogonalFrequencyDivisionMultiplexing), Multiple-InputMultiple-Output) technology is that the transmittability of LTE (LongTermEvolution) communication system to communication network of core is had higher requirement, therefore for the mobile communication system that frequency spectrum resource is very limited, need to promote the message transmission rate in unit transmission bandwidth, namely spectrum effect (SE, SpectralEfficiency) of system is improved.

Along with people are to the concern of energy resource consumption and environmental problem, the energy consumption problem of mobile communications network also more and more causes the concern of numerous researchers.Therefore, how to reduce the energy that mobile communications network consumes, improve the efficiency (EE, EnergyEfficiency) of mobile communication system, become a study hotspot of current mobile communications research field.In recent years, in communication system, power save transmission is more and more paid close attention to, and green communications are inexorable trends of Communication Development, and thus the efficiency of system also becomes one of important performance indexes weighing communication network.But efficiency and spectrum imitate the not always consistent change of these two kinds of Measure Indexes, sometimes or even mutually opposition, one of them index that covets will cause the decline rapidly of another index.So, the balance between two kinds of indexs and compromise, thus realize combined optimization and become a hot issue.

Summary of the invention

The object of this invention is to provide the power distribution method of a kind of LTE system efficiency spectrum effect combined optimization.By obtaining optimal user power division to distributing to the method that each user power is optimized after first optimizing total transmitting power, realize the rational combined optimization of efficiency spectrum effect.

Technical scheme of the present invention is: efficiency and spectrum effect are not increase always simultaneously or reduce, but the increase of spectrum effect can bring the sharply decline of efficiency in certain interval, therefore can not obtain both maximums simultaneously, but efficiency should be carried out and compose the compromise of effect, to realize the combined optimization that efficiency is imitated with spectrum.

The cost function of definition efficiency and spectrum effect combined optimization:

${\mathrm{\η}}_{SE-EE}=(1-\mathrm{\β})\frac{{\mathrm{\η}}_{SE}}{{\mathrm{\η}}_{SE}^{norm}}+\mathrm{\β}\frac{{\mathrm{\η}}_{EE}}{{\mathrm{\η}}_{EE}^{norm}}---\[1\]$

Wherein η _sEand η _eErepresent spectrum effect and the efficiency of system respectively, β represents weight coefficient, with for spectrum effect and efficiency are turned to same order, β represents the relative weighting of efficiency. β value is decided in its sole discretion according to network operation situation by operator.

The combined optimization problem of efficiency and spectrum effect is:

maxη _SE-EE

$\begin{array}{cc}subjectto& \underset{n=1}{\overset{N}{\Σ}}{p}_n\≤P_{\max }\end{array}---\[2\]$

Wherein p _nrepresent the power that nth user distributes, P _maxrepresent the maximum transmission power that operator allows.Therefore, the combined optimization problem of efficiency and spectrum effect is under the condition of given power constraint, is optimized transmitting power and the distribution between each user thereof, thus realizes the combined optimization of efficiency and spectrum effect.

Key step comprises:

1) step one: find user rate lower bound, user rate is similar to

In LTE system, user can receive the interference from other community similar frequency bands user, and user power assignment problem becomes non-convex NP (Non-deterministicPolynomial) difficult problem, can only be solved by violence.In order to reduce the complexity of problem, by speed lower bound approximated user speed, thus simplify the relation of power between user rate and other users.By logarithmic approximation, the speed lower bound of nth user in cell i can be obtained:

$r_n^i\≥a_n^i\ln \left({\mathrm{SINR}}_n^i\right)+b_n^i---\[3\]$

Wherein,

$a_n^i=\frac{\stackrel{\‾}{{\mathrm{SINR}}_n^i}}{1+\stackrel{\‾}{{\mathrm{SINR}}_n^i}}---\[4\]$

$b_n^i=\ln (1+\stackrel{\‾}{{\mathrm{SINR}}_n^i})-a_n^i\ln \left(\stackrel{\‾}{{\mathrm{SINR}}_n^i}\right)$

When time, user rate and the middle lower bound approximately equal obtained of formula [3].Constantly adjust user power by the method for iteration to distribute, wherein for the Signal to Interference plus Noise Ratio (SINR, SignaltoInterferenceplusNoiseRatio) of nth user in cell i, for nth user's last iteration in cell i obtains the Signal to Interference plus Noise Ratio of user after optimal power allocation, during first iterative for nth user in cell i is according to the Signal to Interference plus Noise Ratio of user after initial power distribution.When user power distribute iteration to when restraining, user's Signal to Interference plus Noise Ratio approximately equal that last iteration and last iteration obtain, user rate also with the lower bound approximately equal in formula [3].

2) step 2: upgrade base station transmitting power by gradient method

for the power of nth user in cell i, for nth user in cell i receives the channel impulse response from community m, by conversion in cell i, the speed of nth user can be expressed as:

$r_n^i=a_n^i\[{\mathrm{lng}}_n^i+{\tilde{p}}_n^i-\ln (\underset{m=1,m\≠i}{\overset{M}{\Σ}}{g}_n^m{e}^{{\tilde{p}}_n^m}+N_0)\]---\[5\]$

Wherein,

$g_n^i=\frac{h_n^{ii}}{N_0+\underset{m=1,m\≠i}{\overset{M}{\Σ}}{h}_n^{mi}{p}_n^m}---\[6\]$

From formula [5], user rate be power index and logarithm value, therefore first increase the quasiconcave function subtracted afterwards.When total transmitting power one timing in base station, Modulating Power distributes the maximum system speed of lower acquisition is R ^*(P _t).To R ^*(P _t) differentiate:

$\frac{{\mathrm{dR}}^{*}\left(P_T\right)}{{\mathrm{dP}}_T}=\underset{p}{\max }\frac{1}{p_n^i}\frac{{\mathrm{dr}}_n^i}{d{\tilde{p}}_n^i}---\[7\]$

From formula [7], R ^*(P _t) be first increase the function subtracted afterwards to transmitting power.Use P _tk () represents the transmitting power obtained in kth time iteration, t represents renewal step-length, and concrete value is determined according to network condition by operator.Therefore base station best transmit power can be obtained by gradient method:

$P_T\left(k\right)=P_T(k-1)+\frac{{\mathrm{dR}}^{*}\left(P_T\right)}{{\mathrm{dP}}_T}t---\[8\]$

3) step 3: with Lagrange duality method distributing user power.

After obtaining the total transmitting power in base station by step 2, each user power solves by Lagrange duality method.Lagrange's equation now can be write as:

$L(p,\mathrm{\μ})=\underset{i=1}{\overset{M}{\Σ}}\underset{n=1}{\overset{N}{\Σ}}\left(a_n^i\ln \right(1+{\mathrm{SINR}}_n^i)+b_n^i)-\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\μ}}_i({\left(\underset{n=1}{\overset{N}{\Σ}}{p}_n^i-P_T\right)}^2-\mathrm{\ϵ})---\[9\]$

Wherein, μ=(μ ₁, μ ₂, μ ₃..., μ _n) be Lagrange multiplier, M is the number of system small area, and N is the quantity of user in each community, simplifies mark P _treplace P _tk () represents the kth time total transmitting power in iteration base station.ε is enough little threshold value, and value is decided in its sole discretion according to network operation situation by operator.

According to Lagrange duality method, problem [9] can be decomposed into lagrange duality problem with Lagrange duality function:

$d\left(\mathrm{\μ}\right)=\underset{p}{\max }L(p,\mathrm{\μ})---\[10\]$

By conversion formula [6] obtains kth time iteration optimal user power division by differentiate:

$e^{{\tilde{p}}_n^{i*}}=\frac{-c+\sqrt{c^2+4{\mathrm{\μ}}_i(a_n^i+2{\mathrm{\μ}}_i{P}_T)}}{4{\mathrm{\μ}}_i}---\[11\]$

Wherein,

$c=-\underset{m=1,m\≠i}{\overset{M}{\Σ}}{a}_n^m\left(\frac{h_n^{im}}{N_0+I_n^m}\right)-2{\mathrm{\μ}}_i(\underset{j=1,j\≠n}{\overset{N}{\Σ}}{p}_n^i-P_T)---\[12\]$

for the interference signal intensity that nth user in the m of community receives.

4) step 4: iteration judges

Step 3 judges power division after obtaining kth time iteration optimal power allocation whether restrain.When power division does not restrain, return step one and carry out next iteration; When power division restrains, iteration terminates, and power division is now optimum user power and distributes.

Technical scheme of the present invention has following beneficial effect: the present invention establishes the combined optimization problem of spectrum effect and efficiency, gives a kind of power distribution method that can realize spectrum effect and efficiency combined optimization.

Tool of the present invention has the following advantages:

1. give a kind of power distribution method effectively and rapidly that can realize spectrum effect and efficiency combined optimization, avoid the complexity that violence method for solving brings.

2. operator can pass through to regulate weight coefficient, the proportion between flexible configuration spectrum effect with efficiency, thus realizes efficiency and compose the compromise of effect.

Embodiment

Below the execution mode of the present invention's power division is in the wireless network described further.

(1) Base station initialization user power is distributed, and operator can select to distribute according to actual conditions.Through type [3], formula [4] obtain user rate lower bound.

(2) base station is according to the selected weight coefficient of operator and user rate lower bound, and calculation cost function (formula [1]), about the derivative of total transmitting power, upgrades total transmitting power by gradient French [8].

(3) base station carries out power division according to formula [11] to user.

(4) if power distribution result and last time power distribution result squared norm be greater than threshold value, then through type [3], formula [4] upgrade user rate lower bound returning (2); Otherwise namely complete spectrum effect, efficiency optimal power allocation when trading off.Threshold value is selected according to concrete network condition by operator.

Claims (2)

1. a power distribution method for wireless network efficiency spectrum effect combined optimization, described method is carried out efficiency and is composed the compromise of effect, to realize efficiency and to compose the combined optimization of imitating, is specially:

The cost function of definition efficiency and spectrum effect combined optimization:

${\mathrm{\η}}_{SE-EE}=(1-\mathrm{\β})\frac{{\mathrm{\η}}_{SE}}{{\mathrm{\η}}_{SE}^{norm}}+\mathrm{\β}\frac{{\mathrm{\η}}_{EE}}{{\mathrm{\η}}_{EE}^{norm}}---\[1\]$

Wherein η _sEand η _eErepresent spectrum effect and the efficiency of system respectively, β represents weight coefficient, with for spectrum effect and efficiency are turned to same order, β represents the relative weighting of efficiency. β value is decided in its sole discretion according to network operation situation by operator;

The combined optimization problem of efficiency and spectrum effect is:

maxη _SE-EE

$\begin{array}{cc}subjectto& \underset{n=1}{\overset{N}{\Σ}}{p}_n\≤P_{\max }\end{array}---\[2\]$

Wherein p _nrepresent the power that nth user distributes, P _maxrepresent the maximum transmission power that operator allows.

2. the power distribution method of wireless network efficiency spectrum effect combined optimization according to claim 1, its key step comprises:

1) step one: find user rate lower bound, user rate is similar to

By speed lower bound approximated user speed, by logarithmic approximation, the speed lower bound of nth user in cell i can be obtained:

$r_n^i\≥a_n^i\ln \left({\mathrm{SINR}}_n^i\right)+b_n^i---\[3\]$

Wherein,

$a_n^i=\frac{\stackrel{\‾}{{\mathrm{SINR}}_n^i}}{1+\stackrel{\‾}{{\mathrm{SINR}}_n^i}}---\[4\]$

$b_n^i=\ln (1+\stackrel{\‾}{{\mathrm{SINR}}_n^i})-a_n^i\ln \left(\stackrel{\‾}{{\mathrm{SINR}}_n^i}\right)$

When time, user rate and the middle lower bound approximately equal obtained of formula [3]; Constantly adjust user power by the method for iteration to distribute, wherein for the Signal to Interference plus Noise Ratio SINR of nth user in cell i, for nth user's last iteration in cell i obtains the Signal to Interference plus Noise Ratio of user after optimal power allocation, during first iterative for nth user in cell i is according to the Signal to Interference plus Noise Ratio of user after initial power distribution; When user power distribute iteration to when restraining, user's Signal to Interference plus Noise Ratio approximately equal that last iteration and last iteration obtain, user rate also with the lower bound approximately equal in formula [3];

2) step 2: upgrade base station transmitting power by gradient method

for the power of nth user in cell i, for nth user in cell i receives the channel impulse response from community m, by conversion in cell i, the speed of nth user can be expressed as:

$r_n^i=a_n^i\[{\mathrm{lng}}_n^i+{\tilde{p}}_n^i-\ln (\underset{m=1,m\≠i}{\overset{M}{\Σ}}{g}_n^m{e}^{{\tilde{p}}_n^m}+N_0)\]---\[5\]$

Wherein,

$g_n^i=\frac{h_n^{ii}}{N_0+\underset{m=1,m\≠i}{\overset{M}{\Σ}}{h}_n^{mi}{p}_n^m}---\[6\]$

From formula [5], user rate be power index and logarithm value, therefore first increase the quasiconcave function subtracted afterwards; When total transmitting power one timing in base station, Modulating Power distributes the maximum system speed of lower acquisition is R ^*(P _t); To R ^*(P _t) differentiate:

$\frac{{\mathrm{dR}}^{*}\left(p_T\right)}{{\mathrm{dP}}_T}=\underset{p}{\max }\frac{1}{p_n^i}\frac{{\mathrm{dr}}_n^i}{d{\tilde{p}}_n^i}---\[7\]$

R ^*(P _t) be first increase the function subtracted afterwards to transmitting power; Use P _tk () represents the transmitting power obtained in kth time iteration, t represents renewal step-length, and concrete value is determined according to network condition by operator; Therefore base station best transmit power can be obtained by gradient method:

$P_T\left(k\right)=P_T(k-1)+\frac{{\mathrm{dR}}^{*}\left(P_T\right)}{{\mathrm{dP}}_T}t---\[8\]$

3) step 3: with Lagrange duality method distributing user power

After obtaining the total transmitting power in base station by step 2, each user power solves by Lagrange duality method, and Lagrange's equation now can be write as:

$L(p,\mathrm{\μ})=\underset{i=1}{\overset{M}{\Σ}}\underset{n=1}{\overset{N}{\Σ}}\left(a_n^i\ln \right(1+{\mathrm{SINR}}_n^i)+b_n^i)-\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\μ}}_i({\left(\underset{n=1}{\overset{N}{\Σ}}{p}_n^i-P_T\right)}^2-\mathrm{\ϵ})---\[9\]$

Wherein, μ=(μ ₁, μ ₂, μ ₃..., μ _n) be Lagrange multiplier, M is the number of system small area, and N is the quantity of user in each community, simplifies mark P _treplace P _tk () represents the kth time total transmitting power in iteration base station, ε is enough little threshold value, and value is decided in its sole discretion according to network operation situation by operator;

According to Lagrange duality method, problem [9] can be decomposed into lagrange duality problem with Lagrange duality function:

$d\left(\mathrm{\μ}\right)=\underset{p}{\max }L(p,\mathrm{\μ})---\[10\]$

By conversion formula [6] obtains kth time iteration optimal user power division by differentiate:

$e^{{\tilde{p}}_n^{i*}}=\frac{-c+\sqrt{c^2+4{\mathrm{\μ}}_i(a_n^i+2{\mathrm{\μ}}_i{P}_T)}}{4{\mathrm{\μ}}_i}---\[11\]$

Wherein,

$c=-\underset{m=1,m\≠i}{\overset{M}{\Σ}}{a}_n^m\left(\frac{h_n^{im}}{N_0+I_n^m}\right)-2{\mathrm{\μ}}_i(\underset{j=1,j\≠n}{\overset{N}{\Σ}}{p}_n^i-P_T)---\[12\]$

for the interference signal intensity that nth user in the m of community receives;

4) step 4: iteration judges

Step 3 judges power division after obtaining kth time iteration optimal power allocation whether restrain; When power division does not restrain, return step one and carry out next iteration; When power division restrains, iteration terminates, and power division is now optimum user power and distributes.