CN103957559A - Heterogeneous network access selection method based on user experience - Google Patents

Abstract

The invention provides a heterogeneous network access selection method based on user experience. The major steps comprise: performing QoE data collection, i.e., the wireless resource entity of each candidate network measuring the parameter of a local network and reporting the parameter to a joint resource manager, and a user terminal determining a QoE preference attribute for reporting to the joint resource manager; accordingly performing QoE data processing, i.e., the joint resource manager calculating a utility value which a user can obtain from each attribute of each network by use of a utility normalized function, determining the weight of each attribute by use of an absolute rating method and a hierarchy analytical method, and obtaining the satisfaction of the user for the candidate networks through performing weight summing on each attribute; and finally, making a QoE decision, i.e., the joint resource manager, through ordering the obtained satisfaction about the candidate networks by the user, feeding a processing result to the RRM entity of a network with greatest user satisfaction, and the RRM entity executing wireless resource distribution to realize user access.

Description

A kind of heterogeneous network access selection method of experiencing based on user

Technical field

The present invention has utilized the isomerism of heterogeneous networks and the diversity of user's request, adopts multiattribute Technology of Judgment, is applied to, for the user in heterogeneous network region selects the network that Quality of experience is higher, belong to mobile communication technology field.

Background technology

Next generation network is the heterogeneous network merging, there is dynamic adaptivity and interactivity, by multiple different radio access technologies (Radio Access Technoligy, RAT) with there is the Radio Access Network (Radio Access Network, RAN) that different QoS ensures and cover.The target of next generation wireless communication development is to allow user utilize whenever and wherever possible to have the network of heterogeneous characteristic to carry out high-quality radio communication, reach best user and experience (Quality of Experience, and network insertion system of selection and strategy one of the factor of most critical of realizing this goal beyond doubt QoE).

QoE has reflected the subjective acceptable degree of terminal use to business or network.After the lot of documents that research heterogeneous network is selected, can find: current heterogeneous network Access Algorithm mostly is multiple attribute decision making (MADM) algorithm (MADM, Multiple Attribute Decision Making), but these network selecting methods do not have the abundant angle from user, consider user's demand, or the measure adopting when considering each Attribute Significance is too subjective, can not be in the preference that objectively reflects user.Along with the isomerism of on-premise network and the diversity of user's request, original network selecting method can not meet the demand of user experience quality gradually, how from user perspective, for the user in heterogeneous network overlapping region selects the network that satisfaction is high, the access that simultaneously guarantees user does not bring inefficacy and extra-heavy result to existing network service and offered load, the integrated scheduling of realization to network, this becomes a major issue of current heterogeneous network access selection method.

Summary of the invention

Technical problem: the object of the invention is to provide a kind of access selection method of experiencing based on user under heterogeneous network environment, the method can improve user satisfaction, load reduce service blocking rate between equalizing network.

Technical scheme: the present invention is a kind of heterogeneous network access selection method of experiencing based on user, and it includes three processes: QoE Data Collection, QoE data processing and QoE decision-making.

1) QoE Data Collection: RRM(Radio Resource Managemet, the RRM of each network) this network parameter of entity measuring, as business arrival rate, QoS information etc.; Terminal is collected the preference information of user's side; Then RRM entity and terminal report joint resource management device by the data of collection;

2) QoE data processing: the normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network, adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute, then by each attribute weight summation is obtained to the satisfaction of user to candidate network;

3) make QoE decision-making: joint resource management device is by sorting in the available satisfaction of candidate network to user, then result is fed back to the RRM entity of network with the greatest satisfiction by user, by RRM entity, carry out allocation of radio resources, realize user's access.

Beneficial effect: the present invention proposes the heterogeneous network access selection method of experiencing based on user.Advantage is from user perspective, for difference preference's attribute, normalized value of utility computational methods have been provided, and calculate in user satisfaction process and combine and used absolute scale separating method and analytic hierarchy process (AHP) in weighted sum, improved the objectivity that user selects.The method has been taken into account offered load state and user preference with respect to other method for network access, has improved practicality that network selects and user's satisfaction, has reduced load between service blocking rate balanced network.

Accompanying drawing explanation

Fig. 1 is heterogeneous wireless network configuration diagram of the present invention.

Fig. 2 is method flow diagram of the present invention.

Embodiment

Heterogeneous network access selection method based on user satisfaction of the present invention, from user perspective, is provided the individual demand of self by user.This inventive method comprises following 3 processes and is undertaken by described order:

1) collect QoE data: this network parameter of RRM entity measuring of each network, as business arrival rate, available real time rate etc.; Terminal is collected the preference information of user's side; Then RRM entity and terminal report joint resource management device by the data of collection;

2) process QoE data: the normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network, adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute, then by each attribute weight summation is obtained to the satisfaction of user to candidate network;

3) make QoE decision-making: joint resource management device is by sorting in the available satisfaction of candidate network to user, then result is fed back to the RRM entity of network with the greatest satisfiction by user, by RRM entity, carry out allocation of radio resources, realize user's access.

By reference to the accompanying drawings, the present invention program is designed and does further concrete analysis and describe.

Fig. 1 is heterogeneous wireless network configuration diagram of the present invention.Heterogeneous network is comprised of RAN1 and RAN2, and the RRM entity of each network is responsible for this network parameter of strategy, and measurement parameter is reported to joint resource management device.RAN1 can be 3G Cellular Networks or LTE network, and RAN2 can be WLAN or WiMAX network.User in RAN1 or RAN2 overlay area can only select its overlay network separately, and in RAN1 and the overlapping covered user of RAN2, can select RAN1 or RAN2, the problem that we will solve is in network overlapped overlay area, for user selects network with the greatest satisfiction by user, accesses.

When user will start new business or switch, first determine the available network collection of user's periphery, the network in this set meets the basic QoS demand that user is using business; Then trigger the network insertion system of selection of experiencing based on user, idiographic flow is as Fig. 2.

1) QoE Data Collection

QoE Data Collection comprises two processes below: this network parameter of RRM entity measuring of each network also reports joint resource management device; Terminal is collected user's request property set and is reported joint resource management device.

(1) this network parameter of RRM entity measuring of each network report joint resource management device

Suppose that the number of users having existed in the RRM entity measuring current network of candidate network i has U _iindividual, the shared bandwidth of business of k user's use is B _k, the bandwidth that the current customer service that will access need to take is B _j, the available network total bandwidth of candidate network i is C _i.Then RRM entity reports joint resource management device by metrical information, so joint resource management device end can to obtain the load factor of network i as follows:

$L_i=\left\{\begin{array}{c}0,C_i-\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k<B_j\\ 1-\frac{\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k+B_j}{C_i},C_i-\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k>B_j\end{array}\right.---\left(1\right)$

In formula, expression takes into account existing user the remaining bandwidth of rear candidate network i.

Meanwhile, RRM entity sends to the information of radio resource manager to have the QoS information of current candidate network i, as time delay, bandwidth and price etc.According to load factor defined above, if residue of network organization bandwidth is not enough to support user institute use business, illustrate that current network is saturated, so the offered load factor is made as to 0, now refuse this user's access; If remaining bandwidth is enough to support user institute use business, and after user adds network, occupied bandwidth is less for network, illustrate that current network belongs to light-load state, what the offered load factor is arranged is larger, after user satisfaction by weighted calculation also larger, this is conducive to load between equalizing network.

(2) terminal is collected user's request property set and is reported joint resource management device

The business that user is used by user the satisfaction of network or business and user's preference determine, every kind of business has its association attributes requirement and associated user's perceived quality.Consider that each user there are differences the preference attribute of used business, the numerical value unit between different attribute and value are also not quite similar, and cannot directly compare, and the inventive method takes following method to be normalized to facilitate comparison to different attributes.

Note R ^cand={ r ₁, r ₂, r ₃... r _mbe the set of m candidate's access network, Q={q ₁, q ₂..., q _nbe that user requires set, P={p to the n of current use business subjective attribute _i1, p _i2..., p _inthe value of n the attribute that can provide in real time for i candidate network.

When describing user to the demand of certain attribute, from user friendliness and user habit, to consider, the inventive method specifically represents user's attribute demand by a tlv triple, specific as follows: req _j={ th _j, g _j, r _j.Wherein, th _jthe ideal value of j the attribute of setting for user; r _jrepresent the th that user sets self _jdefinite degree of parameter, value interval (0,1] in, r _j=1 represents the ideal value th that user sets self _jvery definite, receptible j the property value of user is to a great extent close to th _j.G _jrepresent the attention degree of user to this attribute.First the inventive method is used absolute scale separating method to characterize the attention degree of user to this attribute.The different attention degrees of different mark representative of consumer to attribute, as shown in table 1.

The significance level that the different marks of table 1 are corresponding

2) QoE data processing

QoE data handling procedure comprises, the normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network; Adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute; By each attribute weight summation is obtained to the satisfaction of user to candidate network.

(1) normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network

This process mainly provides unified normalization computational methods to the preference attribute of user's appointment.The numerical value that normalization obtains is called value of utility, represents the potential income that this attribute can bring for user.

In order to provide unified normalization computational methods, according to the physical significance of attribute, the inventive method is divided into two large classes by attribute: positive attribute and passive attribute.For positive attribute, the property value that network can provide is larger, and the satisfaction that user obtains is higher, such as fail safe, bandwidth etc.; For passive attribute, the property value that network can provide is less, and the satisfaction that user obtains is higher, such as expense, delay etc.

The inventive method is thought tlv triple req _j={ th _j, g _j, r _jin th _jfor the expectation ideal value of user to this attribute, and non-threshold.Such as, for the ideal value th of passive attribute _jif the property value that candidate network provides is less than th _j, the inventive method thinks that j the property value that this candidate network provides meets user's request completely; If the property value that candidate network provides is greater than th _j, the inventive method also not exclusively excludes candidate's scope by this network, but the property value providing in candidate network departs from th _jwhen not far, user still can accept this network.It should be noted that ideal value th _jcan change along with the variation of user's confidence index.

According to the above description, the value of utility of dissimilar attribute can calculate by following formula:

The value of utility computational methods of positive attribute:

$u_{i^{\mathrm{positive}}}\left(j\right)=\left\{\begin{array}{c}1,p_{\mathrm{ij}}>{\mathrm{th}}_j(2-r_j)\\ p^{M*(1-\frac{p_{\mathrm{ij}}}{{\mathrm{th}}_j})},p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j),0<p<1\end{array}\right.---\left(2\right)$

The value of utility computational methods of passive attribute:

$u_{\mathrm{inegative}}\left(j\right)=\left\{\begin{array}{c}1,p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j)\\ p^{M*(1-\frac{{\mathrm{th}}_j}{p_{\mathrm{ij}}})},p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j),0<p<1\end{array}\right.---\left(3\right)$

Wherein, p _ij/ th _jand th _j/ p _ijreflected the property value p that candidate network can provide _ijideal value th with user's proposition _jbetween departure degree.M is amplification factor.P is the constant of a span between (0,1), and its concrete numerical value can maybe need to adjust according to different application scenarioss.

(2) adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute

User, provide the attention grade { g to each attribute ₁, g ₂..., g _nafter, the inventive method combine adopt a kind of convenience and effectively analytic hierarchy process (AHP) the subjective weight of judgement index carried out to reassignment, omitted user to the comparing between two of each attribute, concrete grammar is as shown in table 2.

Table 2 analytic hierarchy process (AHP) scheme

If attribute i is d to the step analysis assignment of attribute j _ij, according to method shown in table 2, carry out assignment, can obtain assignment matrix and be:

$D={\left(d_{\mathrm{ij}}\right)}_{n\&times;n}=\left[\begin{array}{cccccc}{d}_{11}& d_{12}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{1n}\\ d_{21}& d_{22}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{2n}\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ d_{n1}& d_{n2}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{\mathrm{nn}}\end{array}\right]---\left(4\right)$

Calculate the excellent ordinal number O of step analysis preference of each attribute _j:

$O_j=\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_{\mathrm{jk}}(j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,n)---\left(5\right)$

And get:

$w_j=O_j/\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{O}_k(j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,n)---\left(6\right)$

The subjective preferences weight vectors that can obtain candidate network all properties is:

W＝(w ₁,w ₂,…,w _n)(7)

Combine and use absolute scale separating method and analytic hierarchy process (AHP) to make to have guaranteed the objectivity of each attribute weights when facilitating user to select, improved accuracy and the practicality of decision-making.

(3) by each attribute weight summation is obtained to the satisfaction of user to candidate network

Obtain the value of utility vector U of each attribute _i=(u _i1, u _i2..., u _in) and subjective preferences weight vectors W=(w ₁, w ₂..., w _n) after, the total satisfaction that can try to achieve each attribute that user provides this candidate network according to weighted mean method is:

$S_i=\underset{j}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}{w}_i---\left(8\right)$

3) make QoE decision-making

Joint resource management device, by user is sorted in the available satisfaction of candidate network, then feeds back to result the RRM entity of network with the greatest satisfiction by user, by RRM entity, carries out allocation of radio resources, realizes user's access.

Claims (3)

1. a heterogeneous network access selection method of experiencing based on user, is characterized in that comprising following process:

Step 1: collect QoE data: this network parameter of RRM entity measuring of each network, terminal is collected the preference information of user's side, and then RRM entity and terminal report collected data to joint resource management device;

Step 2: process QoE data: the normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network, adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute, then by each attribute weight summation is obtained to the satisfaction of user to candidate network;

Step 3: make QoE decision-making: joint resource management device is by sorting in the available satisfaction of candidate network to user, then result is fed back to the RRM entity of network with the greatest satisfiction by user, by RRM entity, carry out allocation of radio resources, realize user's access.

2. a kind of heterogeneous network access selection method of experiencing based on user according to claim 1, is characterized in that in described step 1, collects QoE data and comprises two processes;

(1) this network parameter of RRM entity measuring of each network report joint resource management device:

Suppose that the number of users having existed in the RRM entity measuring current network of candidate network i has U _iindividual, the shared bandwidth of business of k user's use is B _k, the bandwidth that the current customer service that will access need to take is B _j, the available network total bandwidth of candidate network i is C _i; Then RRM entity reports joint resource management device by metrical information, so joint resource management device end can to obtain the load factor of candidate network i as follows:

$L_i=\left\{\begin{array}{c}0,C_i-\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k<B_j\\ 1-\frac{\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k+B_j}{C_i},C_i-\underset{k=1}{\overset{U_i}{\mathrm{\&Sigma;}}}{B}_k>B_j\end{array}\right.---\left(1\right)$

In formula, expression takes into account existing user the remaining bandwidth of rear candidate network i;

Meanwhile, the RRM entity of candidate network i sends to the information of radio resource manager to comprise the QoS information of current network, and it can have: time delay, bandwidth and pricing information;

(2) terminal is collected user's request property set and is reported joint resource management device:

Different attributes is normalized to note R ^cand={ r ₁, r ₂, r ₃... r _mbe the set of m candidate's access network, Q={q ₁, q ₂..., q _nbe that user requires set, P={p to the n of current use business subjective attribute _i1, p _i2..., p _inthe value of n the attribute that can provide in real time for i candidate network;

The attribute demand that specifically represents user by a tlv triple, req _j={ th _j, g _j, r _j, wherein, th _jthe ideal value of j the attribute of setting for user; r _jrepresent the th that user sets self _jdefinite degree of parameter, value interval (0,1] in, r _j=1 represents the ideal value th that user sets self _jvery definite, receptible j the property value of user is to a great extent close to th _j; g _jrepresent the attention degree of user to this attribute;

Use absolute scale separating method to characterize the attention degree of user to this attribute, the different attention degrees of different mark representative of consumer to attribute, can be:

The significance level that the different marks of table 1 are corresponding

3. a kind of heterogeneous network access selection method of experiencing based on user according to claim 1, is characterized in that in step 2, processes QoE data procedures and comprises;

(1) normalization of joint resource management device is calculated each preference attribute of user at the obtainable value of utility of candidate network:

Attribute is divided into two large classes: positive attribute and passive attribute, actively attribute can be fail safe, bandwidth; Passive attribute can be expense, delay;

The value of utility of dissimilar attribute can calculate by following formula:

The value of utility computational methods of positive attribute:

$u_{i^{\mathrm{positive}}}\left(j\right)=\left\{\begin{array}{c}1,p_{\mathrm{ij}}>{\mathrm{th}}_j(2-r_j)\\ p^{M*(1-\frac{p_{\mathrm{ij}}}{{\mathrm{th}}_j})},p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j),0<p<1\end{array}\right.---\left(2\right)$

The value of utility computational methods of passive attribute:

$u_{\mathrm{inegative}}\left(j\right)=\left\{\begin{array}{c}1,p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j)\\ p^{M*(1-\frac{{\mathrm{th}}_j}{p_{\mathrm{ij}}})},p_{\mathrm{ij}}<{\mathrm{th}}_j(2-r_j),0<p<1\end{array}\right.---\left(3\right)$

Wherein, p _ij/ th _jand th _j/ p _ijreflected the property value p that candidate network can provide _ijideal value th with user's proposition _jbetween departure degree; M is amplification factor; P is the constant of a span between (0,1);

(2) adopt absolute scale separating method and analytic hierarchy process (AHP) to determine the weight of each attribute:

User, provide the attention grade { g to each attribute ₁, g ₂..., g _nafter, the subjective weight of judgement index is carried out to reassignment, concrete grammar is as shown in table 2:

Table 2 analytic hierarchy process (AHP) scheme

if attribute i is d to the step analysis assignment of attribute j _ij, according to method shown in table 1, carry out assignment, can obtain assignment matrix and be:

$D={\left(d_{\mathrm{ij}}\right)}_{n\&times;n}=\left[\begin{array}{cccccc}{d}_{11}& d_{12}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{1n}\\ d_{21}& d_{22}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{2n}\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & & & \&CenterDot;\\ d_{n1}& d_{n2}& \&CenterDot;& \&CenterDot;& \&CenterDot;& d_{\mathrm{nn}}\end{array}\right]---\left(4\right)$

Calculate the excellent ordinal number O of step analysis preference of each attribute _j:

$O_j=\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_{\mathrm{jk}}(j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,n)---\left(5\right)$

And get:

$w_j=O_j/\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{O}_k(j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,n)---\left(6\right)$

The subjective preferences weight vectors that can obtain candidate network all properties is:

W＝(w ₁,w ₂,…,w _n)(7)

(3) each attribute weight summation is obtained to the satisfaction of user to candidate network:

Obtain the value of utility vector U of each attribute _i=(u _i1, u _i2..., u _in) and subjective preferences weight vectors W=(w ₁, w ₂..., w _n) after, the total satisfaction that can try to achieve each attribute that user provides this candidate network according to weighted mean method is:

$S_i=\underset{j}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}{w}_i---\left(8\right).$