CN110677875A - Wireless network load balancing method facing edge computing environment - Google Patents

Wireless network load balancing method facing edge computing environment Download PDF

Info

Publication number
CN110677875A
CN110677875A CN201910926563.6A CN201910926563A CN110677875A CN 110677875 A CN110677875 A CN 110677875A CN 201910926563 A CN201910926563 A CN 201910926563A CN 110677875 A CN110677875 A CN 110677875A
Authority
CN
China
Prior art keywords
sta
matrix
load
service
threshold
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910926563.6A
Other languages
Chinese (zh)
Other versions
CN110677875B (en
Inventor
胡劲松
温泉
罗海林
孙严智
蒋丽琼
李朝广
崔晨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yunnan Power Grid Co Ltd
Original Assignee
Yunnan Power Grid Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yunnan Power Grid Co Ltd filed Critical Yunnan Power Grid Co Ltd
Priority to CN201910926563.6A priority Critical patent/CN110677875B/en
Publication of CN110677875A publication Critical patent/CN110677875A/en
Application granted granted Critical
Publication of CN110677875B publication Critical patent/CN110677875B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/0231—Traffic management, e.g. flow control or congestion control based on communication conditions
    • H04W28/0236—Traffic management, e.g. flow control or congestion control based on communication conditions radio quality, e.g. interference, losses or delay
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention relates to a wireless network load balancing method facing to an edge computing environment, and belongs to the technical field of communication. The invention considers that different service terminals have different requirements on QoS, the conversation requires low time delay, the interaction requires small packet loss rate, the stream requires high jitter, and the background requires high packet loss rate. Aiming at the situation, the method judges whether the AP is in an overload state according to the condition that whether the AP network condition meets the service requirement, and selects the optimal AP for the unsatisfied STA to associate through a TOPSIS multi-attribute judgment algorithm, so that the AP load is reduced, the load balance is realized, and the utilization rate of the network bandwidth is improved.

Description

Wireless network load balancing method facing edge computing environment
Technical Field
The invention belongs to the technical field of communication, and particularly relates to a wireless network load balancing method for an edge computing environment.
Background
In recent years, with the popularization of portable terminals and the development of mobile internet systems, wifi (wireless fidelity) network access parties are becoming more and more widespread. Wifi networks cover many places such as office space, school, transformer substation nowadays, compares wired network, and wifi network use volume increases year by year.
In the conventional IEEE801.11 standard, the access of an STA in the MAC layer only depends on RSSI (received signal strength) as a decision criterion, and an AP with the maximum RSSI around the STA is selected for access.
After the wireless device accesses the ap (access point), the network performance may change due to the operation of the service. Some APs may be overloaded or underloaded, so that the AP load is unbalanced, and the channel utilization of some APs is too low or too high, thereby greatly reducing the operation efficiency of the whole network. When the AP is overloaded, the average service response time becomes longer, the link is congested, the service quality cannot be guaranteed, and other APs are in an idle state, which causes waste of wireless network resources. Meanwhile, the AP has different load capacities according to different service types, and the background services have low requirements on time delay and bandwidth, so that one AP can bear a plurality of background services, the session services have high requirements on time delay and bandwidth, and only a small amount of session services can be borne by one AP, so that how to define the AP load condition according to service attributes to realize AP load balancing and ensure normal operation of services is a new solution.
In order to solve the development situation of the prior art, the existing papers and patents are searched, compared and analyzed, and the following technical information with high relevance to the invention is screened out:
the technical scheme 1: patent No. CN201520048000.9 'WIFI network system with load balancing function', relating to a method with AP load balancing function. The method is mainly completed by two steps: first, the STA sends a connection request to the APs in the load balancing group. Secondly, the AP with the least number of wireless access users in the AC control group quickly establishes network connection with the STA.
The scheme comprises an AP controller, an access point device and a wireless terminal. The AP controller and the access point devices are in communication connection to form a Wi-Fi wireless network, the STA sends a connection request to the AP in the load balancing group, and the AP with the least number of wireless access users in the AC control group quickly establishes network connection with the STA. The method has the disadvantages that the load balance condition of the wireless network is judged only by calculating the number of the STA associated with the AP, and the different requirements of different types of STAs on the network quality are ignored.
Technical scheme 2, patent number CN201810694908.5 network load balancing algorithm applied to Wi-Fi roaming area relates to a load balancing algorithm of Wi-Fi roaming area. The method is mainly completed by two steps: firstly, presetting a Wifi roaming network at least provided with more than two Wifi access points; each Wifi access point has the same ssid (serviceset identifier) access identifier and the same authentication setting; secondly, the method comprises the following steps: the method comprises the steps that a Wifi client searches a Wifi access point in a Wifi roaming network, and Wifi access equipment in the Wifi roaming network judges the saturation condition of network bandwidth; if the network bandwidth is in a non-saturated condition, performing network connection between the Wifi client and the Wifi access point; and if the network bandwidth is in a saturated condition, controlling the wireless virtual interface of the Wifi access equipment to work in a client mode (STA) and scanning surrounding Wifi access points.
According to the scheme, the Wi-Fi roaming network at least provided with more than two Wi-Fi access points is preset, the same SSID access mark and the same authentication setting are carried out on each Wi-Fi access point, and after an external Wi-Fi client is connected with any one access point in the Wi-Fi network, the external Wi-Fi client can access other Wi-Fi access points without newly verifying any information, so that the STA switching time in load balancing is shortened, and the bandwidth utilization rate of the network is improved. The method has the disadvantages of lack of STA authentication and low security.
Technical scheme 3, patent number 'a multi-AP combined control method for user balanced access in a WiFi network', relates to a load balancing method for multi-AP combined control. The method mainly comprises the following steps: firstly, authentication is established between neighbor APs periodically through a three-way handshake mechanism; secondly, when the STA sends an association request to the AP, the AP obtains the times of refusing the STA by the AP and the number and load conditions of the accessed STAs under the neighbor AP, calculates a weight value according to the three conditions and a weight value formula, and compares the weight value of the STA with that of the neighbor AP to determine whether the STA is allowed to access.
According to the scheme, mutual authentication is completed between APs through a three-way handshake mechanism, a local neighbor AP information table is formed, the table is updated along with periodic transmission of handshake information, and when an STA sends an association request to the AP, the AP determines whether to allow new STA to access or not according to the number of STAs accessed by the AP, the load condition and the times of refusing the STA by the AP in the AP and neighbor AP information tables. The method only considers the information of the AP and ignores the service information in the STA.
Therefore, how to overcome the deficiencies of the prior art is a problem to be solved in the field of communication technology.
Disclosure of Invention
In a wireless local area network with the background of the power internet of things, massive and heterogeneous terminal devices exist, and different terminal services have different requirements on communication bandwidth, network delay, coverage, safety and the like. Aiming at the scene, the invention aims to solve the defects of the prior art, and researches a wireless network load balancing method facing to an edge computing environment.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a wireless network load balancing method facing to edge computing environment includes the following steps:
step (1), defining threshold vector o, ojRepresenting STA pairsThe threshold requirement of the index j; wherein j is 1,2, and 4 respectively represents four threshold values of bandwidth, time delay, jitter and packet loss rate; o represents a gate line matrix formed by threshold vectors of all STAs associated with the AP, and the element O in OijO representing the ith STA associated with the APjA value;
defining a service vector f, fjA value representing a STA current index j associated with the AP; f represents a service matrix formed by service vectors of all STAs associated with the AP, wherein the element F in the FijF representing the ith STA associated with the APjA value;
time delay f2＝tunpack，tunpackIndicating queuing delay;
let the packet i' exit the queue at time t1The dequeuing time of packet i' +1 is t2The time of dequeuing packet i' +2 is t3Then dither f3＝(t3-t1)-(t2-t1)；
Packet loss rate f4＝pTMAX(ii) a Wherein TMAX is the maximum number of retransmissions,
Figure BDA0002219107190000031
TxRetry represents the number of retransmitted packets, TxFail represents the number of failed packets, TxSucc represents the number of successfully transmitted packets;
defining a benefit vector b ═ 1, -1, -1;
after the STA accesses the AP, firstly, self service information, namely a threshold vector O is sent to the AP, and after the AP receives the self service information, a threshold matrix O is constructed;
step (2), resetting an AP timer to start timing, collecting service information of the associated STA by the AP, and calculating the currently used bandwidth, time delay, jitter and packet loss rate of each STA to obtain a service matrix F;
step (3), calculating a constraint matrix U according to the service matrix F and the threshold matrix O, wherein U is F-O; then, calculating a weighted constraint matrix Y, wherein Y is U multiplied by b; judging the AP load condition according to the weighted constraint matrix Y; if all elements in the matrix Y are positive values, all network thresholds of all STAs are met, the STA is in a normal operation state, and the STA returns to the step after waiting for the next periodStep (2); if the matrix Y has negative values of elements, i.e. Y has elementsij<0, judging that the j index of the STA i is not met, and judging the AP overload state at the moment;
and (4) selecting the STA which is not satisfied, switching to other APs, waiting for the next period, and returning to the step (2).
Further, preferably, in step (4), a specific method for selecting the first STA which is not satisfied to switch to another AP is:
step a, counting the APs with the RSSI value around the STA higher than a set threshold value, and sending the BSSID to the currently connected AP; assuming that the number of the AP with the RSSI value around the STA higher than the set threshold value is u, APkDenotes the kth AP therein; kth AP load State vector Sk，SkMiddle element SkjRepresenting APkThe status of index j of (d);
wherein S iskMiddle element
Figure BDA0002219107190000041
load _ max represents the maximum bandwidth capacity of the AP;
Skmiddle element Sk2＝min{Fi2}；
SkMiddle element Sk3＝min{Fi3}；
SkMiddle element Sk4＝min{Fi4}；
Then broadcasting self-load state vector S between APskObtaining a load state matrix S;
b, combining the currently connected AP with the load state matrix S, and selecting the optimal AP to be pushed to the STA by using a TOPSIS multi-attribute decision algorithm;
and step c, the STA disconnects from the current AP and associates the pushed optimal AP.
Further, it is preferable that the specific method of step b is:
s1, load state matrix S normalization: to pairEach element S in the load state matrix SkjTransforming to obtain normalized matrix d with its elementskjThe calculation method is as follows:
Figure BDA0002219107190000043
s2, comparing the indexes pairwise according to the STA business preference, and quantizing the importance degree by using 1-9 to obtain a decision matrix Z;
Figure BDA0002219107190000051
then, carrying out normalization processing on the decision matrix Z to obtain a matrix q; element q in qhjThe calculation method is as follows:
Figure BDA0002219107190000052
Zhjan element that is Z;
then, the matrix q is summed by rows to obtain a vector G, the elements of which are
Figure BDA0002219107190000053
Then normalization processing is carried out to obtain the weight vector omega, the elements in omega
Figure BDA0002219107190000054
S3, multiplying each row element of the standardized matrix d by the weight vector omega correspondingly to obtain a standardized weighted decision matrix E; element E of EkjThe calculation method is as follows:
Ekj＝dkj×ωh,k＝1,2,...,u,h＝1,2,...,4；
s4, determining the worst solution V+And ideal solution V-：
Figure BDA0002219107190000055
Wherein,the candidate scheme refers to the AP corresponding to each row of the matrix E, and the ideal scheme is an ideal solution V+Corresponding AP, the worst scheme is worst solution V-A corresponding AP; r+Representing the distance, R, of the candidate solution to the ideal solution-Representing the distance of the candidate solution to the worst solution;
Figure BDA0002219107190000056
s6, calculating an optimal scheme:
selection of RkThe largest AP is the optimal AP.
According to the invention, the STA business terminal actively sends the requirement of the STA business terminal on the network service quality after being connected with the AP, and the AP judges whether the STA business terminal is in an overload state according to the network condition, so that the accuracy and flexibility of load judgment are improved. In the power internet of things, the terminals with different QoS requirements can be finally met. And then, selecting the optimal AP for association by using a TOPSIS multi-attribute decision algorithm according to the network preference of the terminal.
Compared with the prior art, the invention has the beneficial effects that:
the existing popular self-adaptive switching load balancing method judges the load condition of the AP by judging the MAC sending time delay of the AP, and the AP is overloaded when the time delay exceeds a preset threshold value. According to the traditional RSSI-based load balancing method, the optimal AP is selected for access by the number of the STAs associated with the AP according to the acquired signal strength of the surrounding APs. The method considers that different service terminals have different QoS requirements, the conversation requires low time delay, the interaction requires small packet loss rate, the stream requires high jitter, and the background requires high packet loss rate. Aiming at the situation, the invention judges whether the AP is in an overload state according to the condition that whether the AP network condition meets the service requirement, and selects the optimal AP for the unsatisfied STA for association through a TOPSIS multi-attribute judgment algorithm, thereby reducing the AP load, realizing load balance and improving the utilization rate of the network bandwidth.
Compared with the existing load balancing method such as a Mac transmission delay judgment method and an STA access quantity method, the method considers the QoS requirement of the terminal so as to judge whether the AP is in a load state, and the judgment accuracy is higher. By optimizing the STA selection problem, the AP load balance is realized.
Drawings
FIG. 1 is a flow chart of a method for wireless network load balancing in an edge-oriented computing environment;
FIG. 2 shows experimental results of an example of application.
Detailed Description
The present invention will be described in further detail with reference to examples.
It will be appreciated by those skilled in the art that the following examples are illustrative of the invention only and should not be taken as limiting the scope of the invention. The examples do not specify particular techniques or conditions, and are performed according to the techniques or conditions described in the literature in the art or according to the product specifications. The materials or equipment used are not indicated by manufacturers, and all are conventional products available by purchase.
The load balancing strategy based on the service attributes mainly comprises two steps, wherein the first step is to judge the AP load condition, calculate the QoS condition of the STA associated with the current AP, such as time delay, bandwidth, jitter, packet loss rate and other factors, and judge whether the network state of the AP meets the STA service requirement, and the second step is to select the optimal AP through a TOPSIS algorithm when the network state of the AP does not meet the STA service requirement, and associate the network quality with the AP if the network quality does not meet the STA requirement.
An AP load judgment method.
Wireless communication type services are generally classified into four types, a conversational type, a streaming type, an interactive type, and a background type, according to contents. The four traffic types QoS requirements are as follows.
(1) The conversation type service comprises voice conversation, video conversation and the like, and is mainly characterized in that the end-to-end time delay is small, and the service volume is symmetrical up and down or almost symmetrical. For the session service, the QoS index to be examined is mainly transmission delay and delay jitter. The human senses are less sensitive to packet loss and packet mishap rates, allowing some short speech pauses and picture mosaicism.
(2) The stream service includes video on demand, FTP download, etc. Although real-time, a cache is usually provided locally to maintain service continuity for a certain time, and this type of service is not sensitive to the delay parameter, but has high requirements on jitter and service bandwidth.
(3) The interactive service refers to a service for performing online data interaction between a user and a remote device, including Web browsing, database retrieval, network games, and the like, and is characterized by high request response requirements, generally longer time delay than a session service, but shorter time delay than a streaming service, high requirements for packet loss rate, and generally not allowing packet loss.
(4) Background services include automatic background E-mail receiving and sending, SMS, file receiving and database downloading, which have no high requirement on transmission time, but have high requirement on packet loss rate, generally zero packet loss rate. The requirements on time delay and jitter are low, and a best-effort mode is adopted for forwarding.
The following AP load determination model is defined:
(1) defining a threshold vector o, ojIndicating the threshold requirement of the STA for the index j. J is 1,2, and 4 respectively indicate four threshold values of bandwidth, time delay, jitter, and packet loss rate. O represents a gate line matrix formed by threshold vectors of all the STAs related to the AP; element O in OijO representing the ith STA associated with the APjA value;
(2) defining a service vector f, fjIndicating the value of the STA current index j associated with the AP. F represents a service matrix formed by service vectors of all STAs associated with the AP, wherein the element F in the FijF representing the ith STA associated with the APjA value;
(a) bandwidth f1. And counting the number of data frames transmitted in unit time.
(b) Time delay f2。
f2＝trec+tque+tunpack+tpacket+tsend
trec，tque，tunpack，tpacket，tsendRespectively representing receiving, queuing, unpacking and analyzing, packaging and sending time delay. Generally, the receiving and sending, unpacking and analyzing, and the packet delay of the Mac frame are fixed values, and the network delay is mainly consumed in the queuing process, which means that the queuing time is considered:
f2＝tunpack
the queuing delay can be obtained by subtracting the queue time value from the packet dequeue time. Queuing delay can well represent network congestion conditions.
(c) Jitter f3And means the difference between RTTs of two adjacent packets. The measurement of delay jitter may be obtained by the time interval of the 802.11Mac layer dequeue. Suppose packet i' is dequeued at time t1The dequeuing time of packet i' +1 is t2The time of dequeuing packet i' +2 is t3Jitter is
f3＝(t3-t1)-(t2-t1)
(d) Packet loss rate f4It can be calculated by counting the number of failed packets, the number of retransmitted packets and the number of successfully transmitted packets. Where TMAX is the maximum number of retransmissions.
Figure BDA0002219107190000081
f4＝pTMAX；
TxRetry represents the number of retransmitted packets, TxFail represents the number of failed packets, TxSucc represents the number of successfully transmitted packets;
(3) defining a benefit vector b as (1, -1, -1, -1), wherein the bandwidth is gain type, the larger the bandwidth is, the better the benefit value is 1, and the delay, jitter and packet loss rate are cost type, the smaller the benefit value is, the better the benefit value is-1.
And judging the load state according to a threshold matrix O and the network performance F, calculating a constraint matrix U as F-O, and multiplying each row in U by a corresponding column b to obtain a weighted constraint matrix Y.
Y＝U×b
When all the elements of Y are positive values, it can be known that all the network thresholds of all the STAs are satisfied and are in a normal operation state. And when negative values are present, i.e. Y is the element Y in Yij<0, the j index of the STA i is not satisfied, and the AP overload state is judged at the moment.
Before the STA is connected with the AP for the first time, the RSSI signal receiving intensity of the surrounding APs is firstly scanned and acquired, and the AP with the RSSI larger than the set threshold value is selected for connection. The AP load state judgment process comprises the following steps:
(1) after the STA accesses the AP, firstly, self service information, namely a threshold vector O is sent to the AP, and after the AP receives the self service information, a threshold matrix O is constructed;
(2) resetting an AP timer to start timing, collecting service information of associated STAs by the AP, and calculating the currently used bandwidth, time delay, jitter and packet loss rate of each STA to obtain a service matrix F;
(3) and calculating a weighted constraint matrix according to the service matrix F and the threshold matrix O to judge the AP load condition. If all elements in the matrix Y are positive values, the matrix Y is in a normal state, and the step (2) is returned to after the next period; if the element in the matrix Y is a negative value, overload is caused;
(4) selecting the first STA which is not satisfied to be switched to other APs, waiting for the next period, and returning to the step (2);
the AP load determination flowchart is shown in fig. 1.
An AP selection method.
After the AP load condition is obtained, an overload signal is sent to the STA of which the service requirement is not met, and the STA is requested to be switched to other light-load APs, namely an AP selection problem is solved. The APs are communicated with each other, load information is exchanged, and the optimal AP is selected for association through a multi-attribute decision algorithm by combining the RSSI information of the APs around the STA.
The following AP selection model is defined:
counting APs with peripheral RSSI values higher than a set threshold value by the STA with the QoS not satisfied, and setting BSSI of the APsD, sending the data to the currently connected AP to obtain a load state matrix S, wherein the load state matrix S is formed by combining load state vectors, and assuming that the number of the APs meeting the requirement is u, the APskIndicating the k-th AP therein. The specific description is as follows:
(1) defining APkLoad state vector Sk，SkjRepresenting APkThe status of the index j. By APkThe service quality matrix F can be calculated to obtain the current APkThe state vector of (2). The AP broadcasts its own load status vector every period T. Element F in FijF representing the ith STA associated with the APjThe value is obtained.
(a) For bandwidth Sk1Definition of
Figure BDA0002219107190000091
(b) For time delay Sk2Definition of Sk2＝min{Fi2}
(c) For jitter Sk3Definition of Sk3＝min{Fi3}。
(d) For packet loss rate Sk4Definition of Sk4＝min{Fi4}。
(2) After the STA receives the overload signal, the RSSI information of the peripheral APs is obtained by scanning the peripheral APs, and the self load state vector S is broadcasted among the APskObtaining a load state matrix S, wherein S is expressed as an AP load state matrix meeting STA requirements:
Figure BDA0002219107190000101
Skjrepresenting APkJ index case of (1).
(3) Defining a weight vector ω ═ ω1,ω2,ω3ω4And selecting the optimal AP for association through a multi-attribute decision algorithm.
The optimal AP algorithm flow is selected as follows:
① STA that QoS is not satisfied counts AP whose peripheral RSSI value is higher than the set threshold, and sends its BSSID to the AP connected currently to obtain the load state matrix S.
② AP combines the load state matrix, and selects the optimal AP to push to the STA by using TOPSIS multi-attribute decision algorithm.
③ the STA disconnects from the current AP and associates with the push AP.
TOPSIS is a classical multi-attribute decision algorithm based on the principle that a candidate solution is closest to the ideal solution and farthest to the worst solution. The ideal scheme is composed of the optimal attribute values of all candidate APs, the worst scheme is composed of the worst attribute values of all candidate APs, and the distance is calculated by using Euclidean distance.
1) Assuming that there are u APs meeting STA service requirements, a load state matrix S is formed by the state vectors of the candidate APs.
Figure BDA0002219107190000102
2) The load state matrix S is normalized. Performing the following operation on each element in the load state matrix S to obtain a standardized matrix d;
Figure BDA0002219107190000103
3) the feature vector method defines attribute weights.
a) According to the STA business preference, pairwise comparison is carried out on the indexes, the importance degree is quantized by using 1-9, wherein 1 is equal and important, 9 is very important, 2-8 is between 1 and 9, and the importance degree is sequentially increased to obtain a decision matrix Z. For example z12The matrix may be set by experience of the operation and maintenance personnel, and the present invention is not limited in particular.
Figure BDA0002219107190000111
b) And (5) carrying out normalization processing on the decision matrix. For each element in Z, calculating
Figure BDA0002219107190000112
A matrix q is obtained.
c) The summation is row-wise to obtain a vector. For each row calculation
Figure BDA0002219107190000113
d) Normalization treatment:the resulting ω is the weight vector.
4) And correspondingly multiplying element weight vectors of each row of the standardized matrix to calculate a standardized weighted decision matrix E.
Ekj＝dkj×ωh,k＝1,2,...,u,h＝1,2,...,4
5) The worst solution V + and the ideal solution V-are determined. Because the bandwidth is an indicator of the benefit, the maximum value in all schemes is ideally solved, and the minimum value is solved in the worst way. And jitter, time delay and packet loss rate are forming indexes, the minimum value in all schemes is ideally solved, and the maximum value is worst solved.
Figure BDA0002219107190000115
Figure BDA0002219107190000116
6) And (3) distance calculation: calculating Euclidean distance from each candidate scheme to an ideal scheme and a worst scheme, wherein the candidate scheme refers to the AP corresponding to each row of the matrix E, and the ideal scheme is an ideal solution V+Corresponding AP, the worst scheme is worst solution V-A corresponding AP; r+Representing the distance, R, of the candidate solution to the ideal solution-Representing the distance of the candidate solution to the worst solution.
Figure BDA0002219107190000117
Figure BDA0002219107190000118
7) And calculating an optimal scheme.
Figure BDA0002219107190000121
Selection of RkThe largest AP connects.
Examples of the applications
And carrying out simulation experiments by using a simulation tool based on the strategy. In the experiment, m APs (m is 3) are set, the maximum bandwidth capacity is 30Mbps, 40Mbps and 50Mbps respectively, the jitter and the delay of the APs are set to comply with poisson distribution in the experiment, and the packet loss rate is set to be 0 by default. 12 STAs are set, wherein, the STA in the background class, the STA in the interactive class, the STA in the stream class and the STA in the conversation class are respectively three. The QoS threshold for each service is set as follows:
the test results are described using the load ratio standard deviation Lrsd representing the load ratio standard deviation representing m APs, where load _ max represents the maximum bandwidth capacity of the AP, load _ curr represents the current AP bandwidth usage,
Figure BDA0002219107190000123
expressing the load ratio of the AP, the formula is as follows:
Figure BDA0002219107190000124
Figure BDA0002219107190000125
the experimental results are shown in fig. 2, with the abscissa representing the time progression in seconds, with one STA associated to the AP every other time unit.
(1) And in the period of 0-5, the load ratio standard deviation is in an ascending state, and in the period, only a plurality of STAs are associated with the AP, the AP possibly is in an idling state, and the load ratio standard deviation gradually ascends.
(2) And in the period of 5-12, the STA associates the optimal AP according to the factors such as the residual bandwidth, the time delay, the jitter and the like of the AP, so that the requirement of self service operation is met, and the load of the AP with heavier load is reduced.
(3) The peak appears at 13 because the session STA is suddenly associated to the AP, the bandwidth requirement of the session service is high, the AP load balancing state is broken, and then the AP gradually converges to 0.1 in load ratio standard deviation through the load balancing strategy.
Most of the existing technical methods only judge the AP load condition by acquiring the AP state and neglect the service operation condition of the STA.
The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (3)

1.A wireless network load balancing method facing to edge computing environment is characterized by comprising the following steps:
step (1), defining threshold vector o, ojRepresenting the threshold requirement of the STA on the index j; wherein j is 1,2, and 4 respectively represents four threshold values of bandwidth, time delay, jitter and packet loss rate; o represents a threshold matrix formed by threshold vectors of all STAs associated with the AP, and an element O in the OijO representing the ith STA associated with the APjA value;
defining a service vector f, fjA value representing a STA current index j associated with the AP; f represents a service matrix formed by service vectors of all STAs associated with the AP, wherein the element F in the FijIndicating association with an APF of the ith STA ofjA value;
time delay f2＝tunpack，tunpackIndicating queuing delay;
let the packet i' exit the queue at time t1The dequeuing time of packet i' +1 is t2The time of dequeuing packet i' +2 is t3Then dither f3＝(t3-t1)-(t2-t1)；
Packet loss rate f4＝pTMAX(ii) a Wherein TMAX is the maximum number of retransmissions,
Figure FDA0002219107180000011
TxRetry represents the number of retransmitted packets, TxFail represents the number of failed packets, TxSucc represents the number of successfully transmitted packets;
defining a benefit vector b ═ 1, -1, -1;
after the STA accesses the AP, firstly, self service information, namely a threshold vector O is sent to the AP, and after the AP receives the self service information, a threshold matrix O is constructed;
step (2), resetting an AP timer to start timing, collecting service information of the associated STA by the AP, and calculating the currently used bandwidth, time delay, jitter and packet loss rate of each STA to obtain a service matrix F;
step (3), calculating a constraint matrix U according to the service matrix F and the threshold matrix O, wherein U is F-O; then, calculating a weighted constraint matrix Y, wherein Y is U multiplied by b; judging the AP load condition according to the weighted constraint matrix Y; if all elements in the matrix Y are positive values, all network thresholds of all STAs are met, the STA is in a normal operation state, and the STA returns to the step (2) after waiting for the next period; if the matrix Y has negative values of elements, i.e. Y has elementsij<0, judging that the j index of the STA i is not met, and judging the AP overload state at the moment;
and (4) selecting the STA which is not satisfied, switching to other APs, waiting for the next period, and returning to the step (2).
2. The method for balancing load of a wireless network facing an edge computing environment according to claim 1, wherein in the step (4), the specific method for selecting the STA which is not satisfied to switch to the other AP is:
step a, counting the APs with the RSSI value around the STA higher than a set threshold value, and sending the BSSID to the currently connected AP; assuming that the number of the AP with the RSSI value around the STA higher than the set threshold value is u, APkDenotes the kth AP therein; kth AP load State vector Sk，SkMiddle element SkjRepresenting APkThe status of index j of (d);
wherein S iskMiddle element
Figure FDA0002219107180000021
load _ max represents the maximum bandwidth capacity of the AP;
Skmiddle element Sk2＝min{Fi2}；
SkMiddle element Sk3＝min{Fi3}；
SkMiddle element Sk4＝min{Fi4}；
Then broadcasting self-load state vector S between APskObtaining a load state matrix S;
Figure FDA0002219107180000022
b, combining the currently connected AP with the load state matrix S, and selecting the optimal AP to be pushed to the STA by using a TOPSIS multi-attribute decision algorithm;
and step c, the STA disconnects from the current AP and associates the pushed optimal AP.
3. The method for balancing load of the wireless network facing the edge computing environment according to the claim 3, wherein the specific method of the step b is as follows:
s1, load state matrix S normalization: for each element S in the load state matrix SkjTransforming to obtain normalized matrix d with its elementskjThe calculation method is as follows:
s2, comparing the indexes pairwise according to the STA business preference, and quantizing the importance degree by using 1-9 to obtain a decision matrix Z;
then, carrying out normalization processing on the decision matrix Z to obtain a matrix q; element q in qhjThe calculation method is as follows:
Figure FDA0002219107180000032
zhjan element that is Z;
then, the matrix q is summed by rows to obtain a vector G, the elements of which are
Then normalization processing is carried out to obtain the weight vector omega, the elements in omega
Figure FDA0002219107180000034
S3, multiplying each row element of the standardized matrix d by the weight vector omega correspondingly to obtain a standardized weighted decision matrix E; element E of EkjThe calculation method is as follows:
Ekj＝dkj×ωh,k＝1,2,...,u,h＝1,2,...,4；
s4, determining the worst solution V+And ideal solution V-：
Figure FDA0002219107180000035
S5, distance calculation: calculating Euclidean distances from each candidate scheme to an ideal scheme and a worst scheme, wherein the candidate scheme refers to the AP corresponding to each row of the matrix E, and the ideal scheme is an ideal solution V+Corresponding AP, the worst scheme is worst solution V-A corresponding AP; r+Representing the distance, R, of the candidate solution to the ideal solution-Representing the distance of the candidate solution to the worst solution;
Figure FDA0002219107180000036
s6, calculating an optimal scheme:
Figure FDA0002219107180000037
selection of RkThe largest AP is the optimal AP.
CN201910926563.6A 2019-09-27 2019-09-27 Wireless network load balancing method facing edge computing environment Active CN110677875B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910926563.6A CN110677875B (en) 2019-09-27 2019-09-27 Wireless network load balancing method facing edge computing environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910926563.6A CN110677875B (en) 2019-09-27 2019-09-27 Wireless network load balancing method facing edge computing environment

Publications (2)

Publication Number Publication Date
CN110677875A true CN110677875A (en) 2020-01-10
CN110677875B CN110677875B (en) 2022-09-09

Family

ID=69079659

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910926563.6A Active CN110677875B (en) 2019-09-27 2019-09-27 Wireless network load balancing method facing edge computing environment

Country Status (1)

Country Link
CN (1) CN110677875B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111510958A (en) * 2020-04-27 2020-08-07 广州天源迪科信息技术有限公司 Message access load balancing method and system
CN111629046A (en) * 2020-05-22 2020-09-04 中国联合网络通信集团有限公司 Edge computing cooperation method, edge computing equipment and terminal
CN112911668A (en) * 2021-01-28 2021-06-04 浙江大华技术股份有限公司 Data transmission method, data transmission device, storage medium and electronic device
CN115567536A (en) * 2022-09-22 2023-01-03 中国联合网络通信集团有限公司 Computing power network scheduling method and device, electronic equipment and storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2734138A1 (en) * 2011-03-02 2012-09-02 3Inova Networks Inc. Traffic management in distributed wireless networks
US20140293795A1 (en) * 2013-03-27 2014-10-02 Fujitsu Limited Qoe optimization in wireless networks
CN104602301A (en) * 2014-12-30 2015-05-06 奇点新源国际技术开发(北京)有限公司 Wireless network load balancing method and device
WO2016165585A1 (en) * 2015-04-15 2016-10-20 电信科学技术研究院 Load balancing method and equipment
WO2018082594A1 (en) * 2016-11-07 2018-05-11 中兴通讯股份有限公司 Method and device for accessing satellite communications network

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2734138A1 (en) * 2011-03-02 2012-09-02 3Inova Networks Inc. Traffic management in distributed wireless networks
US20140293795A1 (en) * 2013-03-27 2014-10-02 Fujitsu Limited Qoe optimization in wireless networks
CN104602301A (en) * 2014-12-30 2015-05-06 奇点新源国际技术开发(北京)有限公司 Wireless network load balancing method and device
WO2016165585A1 (en) * 2015-04-15 2016-10-20 电信科学技术研究院 Load balancing method and equipment
WO2018082594A1 (en) * 2016-11-07 2018-05-11 中兴通讯股份有限公司 Method and device for accessing satellite communications network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
FEI MA: "Distributed Load Balancing Allocation of Virtual Machine in Cloud Data Center", 《 2012 IEEE INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE AND AUTOMATION ENGINEERING》 *
孙严智: "遗传算法电力通信网关键业务备选路由配置", 《云南电力技术第45卷》 *
盛亚辛: "异构无线IP网络的无缝垂直切换机制研究", 《中国优秀硕士学位论文全文数据库》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111510958A (en) * 2020-04-27 2020-08-07 广州天源迪科信息技术有限公司 Message access load balancing method and system
CN111510958B (en) * 2020-04-27 2024-02-20 中国联合网络通信有限公司广东省分公司 Message access load balancing method and system
CN111629046A (en) * 2020-05-22 2020-09-04 中国联合网络通信集团有限公司 Edge computing cooperation method, edge computing equipment and terminal
CN112911668A (en) * 2021-01-28 2021-06-04 浙江大华技术股份有限公司 Data transmission method, data transmission device, storage medium and electronic device
CN112911668B (en) * 2021-01-28 2023-03-24 浙江大华技术股份有限公司 Data transmission method, data transmission device, storage medium and electronic device
CN115567536A (en) * 2022-09-22 2023-01-03 中国联合网络通信集团有限公司 Computing power network scheduling method and device, electronic equipment and storage medium
CN115567536B (en) * 2022-09-22 2024-04-16 中国联合网络通信集团有限公司 Computing power network scheduling method and device, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN110677875B (en) 2022-09-09

Similar Documents

Publication Publication Date Title
CN110677875B (en) Wireless network load balancing method facing edge computing environment
US8346275B2 (en) System and method for controlling access point load
RU2277300C2 (en) Method and device for monitoring and prediction of multi-thread load
CN105208624B (en) The multi-access network based on business selects system and method in a kind of heterogeneous wireless network
CN104023357B (en) The load-balancing method and access controller of Centralized Wireless LAN
CN110677886B (en) Wireless communication access network switching scheduling method facing edge computing environment
TW201717694A (en) Quality of service aware access point and device steering
CN109219071B (en) Vertical switching method based on service classification in heterogeneous wireless network
Amani et al. Programmable policies for data offloading in LTE network
CN108055677B (en) Load balancing method based on software defined wireless network
Anedda et al. Adaptive real-time multi-user access network selection algorithm for load-balancing over heterogeneous wireless networks
JP2013106326A (en) Control device, communication system, control program, and control method
CN112996057B (en) Load balancing method and system based on distributed wireless network
CN113395712B (en) High-load cell optimization method and device
JP2019169783A (en) Routing control system, routing control method, and program
WO2014187126A1 (en) Service unloading method based on different qos
CN107770816A (en) A kind of LTE load-balancing methods, call access control module and system
Mazza et al. A user-satisfaction based offloading technique for smart city applications
González et al. Dynamic access control and slice allocation algorithm for diverse traffic demand over 5G heterogeneous networks
WO2019158034A1 (en) Resource allocation method and apparatus
CN110933691A (en) Vertical switching method based on relative entropy and ideal solution for special converged network
CN110649952A (en) Antenna switching method and device, storage medium and electronic equipment
EP3289806A1 (en) Method and apparatus for using call admission control for client balancing
Yeznabad et al. Cross-layer joint optimization algorithm for adaptive video streaming in mec-enabled wireless networks
CN105516636A (en) Heterogeneous network multi-access resource distribution method based on video communication

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant