Summary of the invention
Goal of the invention: it is an object of the invention to provide a kind of multiple-user network system of selection in heterogeneous network converged environment, in conjunction with fair multi-subscriber dispatching technology, take into account user's request and network performance, based on the Optimum Matching asking for weighting bipartite graph, dynamic coordinate heterogeneous network resource, ensure service quality and the access probability of user, ensure Network Load Balance and efficent use of resources simultaneously.
Technical scheme: the network selecting method being based on Weighted optimal bipartite graph matching that the present invention proposes, in conjunction with multi-subscriber dispatching technology, consider the many factors such as heterogeneous network environment, alternative network parameter, QoS, take into account user's request and network performance, dynamic coordinate heterogeneous network resource, network select permeability in isomerization network fusion system is modeled as bipartite model, and realizes, according to Weighted optimal matching process, the network selection that matching degree is maximum. This network selecting method includes herein below:
(1) multi-user's equity dispatching that network relates in selecting and the Optimum Matching based on weighting bipartite graph all carry out in the high in the clouds processing center of isomerization network fusion system. Current service quality and business demand, network side collection network performance parameter are collected in user side, collect result processing center all beyond the clouds and are applied to multi-user's equity dispatching and weighting bipartite graph matching network selection procedures as input parameter.
(2) if number of users is more than alternative network quantity, network to carry out multi-user's equity dispatching before selecting. QoS of customer requires, as schedule constraints, user to be scheduling the tax power of priority, and user's access demand that priority is high is the most urgent. The one group of user selecting the highest access demand enters network selection procedures.
(3) the network select permeability in heterogeneous network converged environment is modeled as bipartite model, and realizes, according to Weighted optimal matching process, the network selection that matching degree is maximum.
(4) matching degree calculates: consider the many factors such as heterogeneous network environment, alternative network parameter, QoS, the parameter vector of user and network both sides is carried out Similarity Measure, respective parameter is had skewed popularity weights by user and network, incorporating parametric weights and similarity vector, calculate the matching degree of every pair of user's request and network performance.
(5), when overloaded network or blocked user occur, the user in network of problem is re-started network and selects, the real-time performance resources balance of optional matching degree suboptimum.
Beneficial effect: owing to heterogeneous network converged environment has variability, the parameter that each network selects also can dynamically change, and therefore parameter and weights will carry out with newly before the method performs. The network selecting method of present invention default network environment in the process of implementation and relevant parameter remain unchanged method the term of execution. The present invention takes into account user's request and network performance, dynamic coordinate heterogeneous network resource, is effectively improved service quality and the access probability of user, ensures Network Load Balance and efficent use of resources simultaneously.
Detailed description of the invention
Below in conjunction with accompanying drawing, the invention will be further described.
As shown in Figure 1, in heterogeneous network converged scene, each access network can provide multiple different application service, and network selects to take into account performance parameter and the demand of user and network two aspect, and processing center carries out integrated treatment beyond the clouds. Consider heterogeneous network environment, alternative network parameter, QoS, dynamic coordinate heterogeneous network resource.
The present invention provides a kind of network selecting method based on Weighted optimal bipartite graph matching, in conjunction with multi-subscriber dispatching technology in isomery UNE, consider heterogeneous network environment, alternative network parameter, QoS and take into account user's request and network performance, dynamic coordinate heterogeneous network resource, network select permeability in isomerization network fusion system is modeled as bipartite model, and realizes, according to Weighted optimal matching process, the network selection that matching degree is maximum. Network selecting method in the present invention mainly comprises the steps that
(1) multi-user's equity dispatching that network relates in selecting and the Optimum Matching based on weighting bipartite graph all carry out in the high in the clouds processing center of isomerization network fusion system.Current service quality and business demand, network side collection network performance parameter are collected in user side, collect result processing center all beyond the clouds and are applied to multi-user's equity dispatching and weighting bipartite graph matching network selection procedures as input parameter.
(2) if number of users is more than alternative network quantity, network to carry out multi-user's equity dispatching before selecting. QoS of customer requires, as schedule constraints, user to be scheduling the tax power of priority, and user's access demand that priority is high is the most urgent. The one group of user selecting the highest access demand enters network selection procedures.
(3) the network select permeability in heterogeneous network converged environment is modeled as bipartite model, and realizes, according to Weighted optimal matching process, the network selection that matching degree is maximum.
(4) matching degree calculates: consider the many factors such as heterogeneous network environment, alternative network parameter, QoS, the parameter vector of user and network both sides is carried out Similarity Measure, respective parameter is had skewed popularity weights by user and network, incorporating parametric weights and similarity vector, calculate the matching degree of every pair of user's request and network performance.
(5), when overloaded network or blocked user occur, the user in network of problem is re-started network and selects, the real-time performance resources balance of optional matching degree suboptimum.
As shown in Figure 2, it is assumed that total M user, N kind access network, each multi-subscriber dispatching result is one group of user of M/N packet medium priority sequence foremost. The state of user's current time includes the constraintss such as queue length, time delay, number of retransmissions, carries out composing power according to the service scenario of the current individual constraints of user and constrained, and the target of scheduling is weights finding one group of user that user is organized and maximum. Weights and the maximum access demand representing user are the most urgent, preferentially carry out network selection. Criterion is mainly from several aspects:
(1) the current time delay of user's calling is closer to the time delay thresholding of tolerance, and weights are more big, and the urgency of access demand is more high;
(2) queue length of station camp-on is more long, and weights are more big, and the urgency of access demand is more high;
(3) number of retransmissions of user's calling is closer to maximum retransmission, and weights are more big, and the urgency of access demand is more high;
Multi-subscriber dispatching detailed process can include herein below:
(1) threshold value of the constraints of each user is confirmed, such as time delay restriction, queue length, maximum retransmission.
(2) according to the present case of user's calling and average transmission rate, the state of counting user current time calling: time delay state, quene state, retransmission state etc.
(3) calculating the priority weighting of each user, according to descending order descending, be divided into M/N packet according to number of users and network number, the packet priority coming first carries out network selection.
As shown in Figure 3, the network select permeability in heterogeneous network converged environment is modeled as bipartite model, realizes the maximum network of matching degree according to Weighted optimal matching process and select. Network select permeability in isomerization network fusion system can be modeled as bipartite graph G=(C, N, E), wherein C represents the user's set asking in isomerization network fusion system to access, N represents alternative access collection of network, has m user and n alternative network, and each user is likely to exist potential link between several alternative network, E represents the set of potential link between user and alternative network, namely the limit e=in bipartite graph (c, n), e �� E, c �� C, n �� N.Each edge has weights, and the weights on limit are matching degree l in the method for the inventionM, n��
The calculating of matching degree includes herein below:
(1) requirement of service quality is expressed as vector form by user, it is considered to factor includes the network bandwidth, access fee, power consumption, call business type etc.; Performance parameter corresponding for each access network also is indicated as vector form.
(2) similarity between every couple of user and network relevant parameter is calculated, it is thus achieved that similarity vector.
(3) different performance parameter has different weights, the matching degree according to every kind of parameter every pair of link of weight computing of similarity vector.
The method of Weighted optimal bipartite graph matching is to find subgraph G '=(C, N, E '), E ' �� E so that every other subgraph coupling limit weights and less than or equal to this subgraph coupling limit weights and. This Optimum Matching is exactly the result that network selects.
As shown in Figure 4, when there is overloaded network or blocked user, the user in network of problem is re-started network and selects by available bipartite model, the real-time performance resources balance of optional matching degree suboptimum. The process of this dynamic coordinate heterogeneous network resource, it is possible to reducing packet loss and blocking rate, balanced load and the utilization of resources, this process includes herein below:
(1) if there is network over loading, the user in overloaded network can reselect network. Overloaded network is removed from the optional link of user, and the bipartite graph updated is mated again, and matching result is equivalent to user and selects the network of matching degree suboptimum to access.
(2) if there being the user blocked by current network, this user can reselect network. The bipartite graph updated, by currently occurring the network blocked from selecting removal link, is mated by user again, and matching result is equivalent to user and selects the network of matching degree suboptimum to access.