CN106973413B - Self-adaptive QoS control method for wireless sensor network - Google Patents

Self-adaptive QoS control method for wireless sensor network Download PDF

Info

Publication number
CN106973413B
CN106973413B CN201710193889.3A CN201710193889A CN106973413B CN 106973413 B CN106973413 B CN 106973413B CN 201710193889 A CN201710193889 A CN 201710193889A CN 106973413 B CN106973413 B CN 106973413B
Authority
CN
China
Prior art keywords
network
qos
application
scheduling
bandwidth
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.)
Active
Application number
CN201710193889.3A
Other languages
Chinese (zh)
Other versions
CN106973413A (en
Inventor
余成波
田桐
孙梦娜
余奕佳
李彩虹
罗培根
杨亚
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.)
Chongqing University of Technology
Original Assignee
Chongqing University of Technology
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 Chongqing University of Technology filed Critical Chongqing University of Technology
Priority to CN201710193889.3A priority Critical patent/CN106973413B/en
Publication of CN106973413A publication Critical patent/CN106973413A/en
Application granted granted Critical
Publication of CN106973413B publication Critical patent/CN106973413B/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
    • H04W28/16—Central resource management; Negotiation of resources or communication parameters, e.g. negotiating bandwidth or QoS [Quality of Service]
    • H04W28/24—Negotiating SLA [Service Level Agreement]; Negotiating QoS [Quality of Service]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W84/00—Network topologies
    • H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
    • 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)
  • Quality & Reliability (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention discloses a self-adaptive QoS control method facing a wireless sensor network, which is characterized in that an RBF neural network health degree evaluation model is constructed based on four parameters of time delay, packet loss rate, throughput and energy consumption, the overall performance of the network is evaluated, and decision support is provided for network control and adjustment; obtaining the current state and the change trend of the network according to the health degree evaluation result, and dynamically modifying the QoS grade of an application user through a dynamic QoS negotiation mechanism based on the network situation; the packet scheduling strategy for dynamically adjusting the QoS priority is realized by the base station and the subscriber station together, the bandwidth is dynamically allocated to the service flows with different QoS priorities, the end-to-end time delay of the service flows can be reduced, and the utilization rate of bandwidth resources is improved.

Description

Self-adaptive QoS control method for wireless sensor network
Technical Field
The invention belongs to the technical field of wireless sensor network data communication, and particularly relates to a self-adaptive QoS control method for a wireless sensor network.
Background
With the rapid development of computer, communication and network technologies, Wireless sensor networks (Wireless sensor networks) have come into play. At present, a wireless sensor network is widely applied to the fields of intelligent traffic, environment monitoring, military safety, intelligent home furnishing, modern agriculture, medical health, aerospace and the like. It is believed that future wireless sensor network technologies will find wider application.
The wireless sensor network meets the requirements of high reliability, high availability and safety of the wireless sensor network, but the core problems of the research of the wireless sensor network system are resource limitation and energy expenditure. According to different application environment requirements, on the premise that network system resources are generally limited, various resources of the wireless sensor network are optimally configured through a series of service quality control technologies, and service quality in the aspects of data perception, transmission, processing, storage and the like is improved and promoted to become a great hotspot of the wireless sensor network technology. Therefore, it is very necessary to design a reasonable and efficient adaptive QoS control method for the user QoS requirement of the wireless sensor network.
Disclosure of Invention
The invention aims to provide a self-adaptive QoS control method for a wireless sensor network, which can provide better fairness and effectively reduce time delay.
The invention relates to a self-adaptive QoS control method facing a wireless sensor network, which comprises the following steps:
step 1, QoS health degree evaluation based on RBF neural network;
step2, network situation perception based on health degree;
step 3, QoS negotiation based on network situation;
and step 4, dynamically adjusting the QoS priority.
In the step 1, the method for evaluating the QoS health degree based on the RBF neural network comprises the following steps:
step 1a, selecting an evaluation variable:
selecting four network QoS indexes of network delay, packet loss rate, throughput and energy consumption as evaluation variables of the QoS health degree of the wireless sensor network;
step 1b, determining parameters of the basis functions:
1b-1, hidden layer initialization, given an initial class center Cj(1) J is 1,2, and k is the number of hidden layer nodes;
1b-2, in the r-th iteration, sample set { xiThe classification method is as follows: for all j, h | | | x 1,2, … k, j ≠ hi-ch(r)||<||xi-cj(r) |, then xi∈sh(r) wherein xiI sample representing a sample set, ch(r) class center when the number of hidden layer nodes is h, cj(r) represents the class center when the number of hidden layer nodes is j;
1b-3, reacting s obtained in step 1b-2h(r) a new class center of ch(r +1), minimizing the value of the metric, let:
Figure BDA0001256889020000021
minimum, (h ═ 1, 2.., k), then
Figure BDA0001256889020000022
Wherein N ishIs s ishNumber of samples in (r), HhIs the value of the metric;
1b-4, for all h ═ 1,2, … k, if ch(r+1)＝ch(r), terminating, otherwise, returning to the step 1 b-2;
step 1c, determination of RBF network weight:
according to the result of learning training, the network QoS health can be given by the following formula:
Figure BDA0001256889020000023
wherein the content of the first and second substances,
Figure BDA0001256889020000024
delta is an optional parameter, determines the width of the basis function, k is the number of nodes of the hidden layer, cjFor class centers when the number of hidden layer nodes is j, | | x-cjI denotes the norm of the vector, Gj(x) Representing the response of the jth basis function to the input vector, GjFor all Gj(x) Formed matrix, y being all yiFormed vector, ωjtThen the weights between the hidden layer and the output layer are represented, and x represents any n-dimensional input vector.
In the step2, the network situation awareness based on the health degree specifically includes:
judging whether the current network QoS health degree is smaller than a threshold Th of the network QoS health degree, if not, judging the network health, if so, judging whether the network QoS health degree at the last moment of the network is smaller than the threshold Th of the network QoS health degree, if not, indicating the network health, and if so, indicating the network is in sub-health.
In step 3, the policy of QoS negotiation includes the following elements:
step 3a, obtaining an application QoS index and a corresponding weight value after the application proposed by a user passes QoS mapping; extracting the QoS indexes, and inquiring whether the indexes in the network are consistent with the indexes in the network resources; if the network does not have the index, the index is fed back to the user to prompt that the application cannot be met; if yes, performing step 3 b;
step 3b, according to the situation of the network residual resources, negotiating the application judged in the step 3 a;
the network maintains the resource residual situation and the network situation evaluation at any time, when the application judged in the step 3a arrives, the network residual resource and the application QoS index are compared, whether the network resource can meet the application requirement is judged, and the support situation of the network QoS is decided; when the network resource can not be satisfied, the following processing is carried out:
3b-1, feeding back to the user, coordinating and reducing the application priority level so as to enable the network resources to support;
3b-2, terminating the application with lower priority in the network, and reserving resources for the application with high priority coming newly;
3b-3, rejecting the application and feeding back that the user network cannot meet the application.
In step 3b, the negotiation modes for the application are divided into two types: a QoS static negotiation mode and a QoS dynamic negotiation mode;
QoS static negotiation mode: before starting application program, user submits QoS requirement of application to QNM, QNM submits it to resource manager to apply resource, resource manager reserves resource according to resource condition of current system, informs resource scheduler to set corresponding scheduling parameter;
QoS dynamic negotiation mode: the user starts the application first, and dynamically obtains the QoS requirement according to the actual condition of the QoS provided by the system, the QNM submits the QoS requirement to the resource manager to negotiate resource allocation, and the resource manager informs the resource scheduler to modify the corresponding scheduling parameter according to the allocation result.
In step 4, the packet scheduling policy for dynamically adjusting the QoS priority includes base station side scheduling and user side scheduling;
the base station end scheduling comprises:
4a, distributing the broadband to different service flows:
the bandwidth allocated to rtPS (real-time Polling Services), nrtPS (non-real-time Polling Services) and BE (Best Effort) traffic streams can BE calculated by the following equation:
BWi＝BW_PB×fi
BW_PB＝BWall-BWUGS-BWr-BWBE
wherein: BW (Bandwidth)iIs the total time slot allocated to the ith service flow, BW _ PB is the transmission time slot allocated to the rtPS, nrtPS and BE service flow data, BWallIs the total slot, BW, of transmission of an uplink subframeUGSIs a transmission slot, BW, allocated to UG traffic in an uplink subframerPolling time slot, BW, allocated to rtPS, nrtPS servicesBEIs a bandwidth request slot allocated to the BE service, fiIs the bandwidth calculation factor of the ith service flow;
4b, allocating bandwidth to different users:
4b-1, initializing the bandwidth allocated to each user;
4b-2, calculating the residual bandwidth;
4b-3, distributing the residual bandwidth to the residual users;
4b-4, repeatedly executing the step 4b-2 and the step 4b-3 for the rest users;
4b-5, completing the allocation of all the bandwidths, ending, and if the rtPS service bandwidth of a user is not met, continuing to allocate the rtPS service bandwidth in the next frame;
the user side scheduling comprises the following steps:
designing different scheduling algorithms for different service flows, and scheduling data among different queues in the same service flow; the rtPS service flow has high real-time requirement, and an EDF algorithm is adopted; when nrtPS service flow bandwidth scheduling is carried out, in order to ensure scheduling fairness, a WFQ algorithm is adopted; in order to reduce the complexity of BE best effort service flow scheduling and ensure the scheduling fairness among queues, a packet priority scheduling algorithm with the minimum packet length is adopted.
The invention has the beneficial effects that:
(1) firstly, analyzing network situation of a wireless sensor network QoS health degree evaluation result; then, according to the situation of the network situation, the QoS requirement of the application is negotiated so as to support the application; the applications are distinguished according to QoS types and enter a dynamic QoS priority grouping queue, so that the applications with special QoS requirements can be smoothly realized. Therefore, the self-adaptive QoS control technology of the wireless sensor network can self-adjust according to the real-time change of the network, and plays the roles of optimizing the network, saving resources and meeting the application requirements.
(2) The invention designs a QoS health degree evaluation method based on an RBF neural network, and enhances the QoS control of a wireless sensor network.
(3) The invention adds QoS negotiation strategy based on the original static QoS priority, judges the support condition of QoS demand through the network situation result, and negotiates with the user to modify QoS grade, thus forming dynamic QoS priority strategy.
(4) The invention designs a packet scheduling algorithm for dynamically adjusting QoS priority, which is realized by a base station and a subscriber station together, allocates bandwidth for service flows with different priorities together, can provide better fairness for a system, reduces end-to-end time delay of the service flows, and improves the utilization rate of the bandwidth.
Drawings
FIG. 1 is a general structure diagram of an adaptive QoS control method;
FIG. 2 is a QoS health assessment model based on RBF neural network;
FIG. 3 is a diagram of a RBF neural network topology;
FIG. 4 is a flow chart of network situational awareness based on health;
FIG. 5 is a flow chart of QoS negotiation based on network situation;
FIG. 6 is a flow chart of a dynamic QoS negotiation management algorithm;
fig. 7 is a packet queue based on QoS dynamic priority.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
as shown in fig. 1, the adaptive QoS control method for a wireless sensor network mainly solves the problem that in a dynamically changing wireless sensor network environment, the weights of various QoS parameters are difficult to adaptively adjust along with the environmental change. The method comprises the following steps: an RBF neural network health degree evaluation model constructed based on four parameters of time delay, packet loss rate, throughput and energy consumption is used for evaluating the overall performance of the network and providing decision support for network control and adjustment; obtaining the current state and the change trend of the network according to the health degree evaluation result, and dynamically modifying the QoS grade of an application user through a dynamic QoS negotiation mechanism based on the network situation; the packet scheduling strategy for dynamically adjusting the QoS priority is realized by the base station and the subscriber station together, the bandwidth is dynamically allocated to the service flows with different QoS priorities, the end-to-end time delay of the service flows can be reduced, and the utilization rate of bandwidth resources is improved. The invention optimizes the end-to-end performance of the network, ensures the service quality of users and improves the utilization rate of bandwidth resources by constructing a network health degree evaluation model, dynamic QoS negotiation and a grouping scheduling strategy for adjusting QoS priority under the network environment adaptive to dynamic change.
1. QoS health degree assessment based on RBF neural network
The network QoS health degree is the evaluation of the network operation state, and the final purpose is to ensure that the network integrally operates well and to ensure that the user requirements are met. The network QoS health degree is related to QoS information, network evaluation can be carried out on the network QoS health degree through QoS indexes, measurement, analysis and calculation are carried out through a series of meaningful parameters reflecting the current and historical states of the wireless sensor network, the actual and evaluated results are used for guiding configuration and management of the network through evaluation of the RBF neural network, and the intelligence and the adaptability of the wireless sensor network are reflected. Fig. 2 is a QoS health assessment model.
Step 1a, selection of evaluation variables
QoS index of wireless sensor network: the relation among indexes such as delay, bandwidth, packet loss rate, throughput and energy consumption is complex, and meanwhile, excessive resources are consumed to a certain extent by acquiring and processing each index, so that the network load is increased. Therefore, in consideration of the characteristics of uncertainty and variability of network application-oriented network, four network QoS indexes of network delay, packet loss rate, throughput and energy consumption are selected as the QoS health degree evaluation variables of the wireless sensor network.
Network average delay: suppose that there are n paths P from all end nodes of the network to the system manager1,P2…,PnAt intervals of time T (T)<T) statistics on path PiThe set of delays from the end node to the system manager is di1,di2…,diIT/T, path PiAverage time delay in time T interval
Figure BDA0001256889020000061
Can be expressed as:
Figure BDA0001256889020000062
then, carrying out weighted average on the time delays on m paths in the network to obtain the network average time delay TD in a time interval T:
Figure BDA0001256889020000063
furthermore, it is possible to provide a liquid crystal display device,
Figure BDA0001256889020000064
packet loss rate: within a specified time in the network, in all data packets, correctThe ratio of received data to total data packets, during a time interval T, assuming Pi_getNumber of data received for the system manager, Pi_sendIf the number of data sent by the network terminal node is less than the threshold, the packet loss rate DL of the network in the time interval T is:
Figure BDA0001256889020000065
throughput: the sum of all the loads in a given time in the network. During time interval T, TP is assumed to be the total data load received by the system manager, denoted by TP.
Network energy consumption rate: the network energy consumes energy over a period of time as a ratio of the total energy. In time interval T, the consumed energy is NwasterThe total energy is initially fixed by NallThat means, the network energy consumption rate EW in time T is:
Figure BDA0001256889020000066
the four indexes cannot be directly used in an evaluation model of the RBF neural network, and data needs to be normalized.
1) And (3) processing time delay: and processing data according to the condition that the current time delay occupies the whole time delay range.
Figure BDA0001256889020000067
Wherein, TDValue represents the value of the average time delay of the network, TDmin represents the minimum value of the whole time delay range, and TDmax represents the maximum value of the whole time delay range.
2) And (3) processing the packet loss rate: and processing the data according to the condition that the packet loss rate is prolonged in the whole packet loss rate range.
Figure BDA0001256889020000068
The DLValue represents a value of the average network packet loss rate, DLmin represents a minimum value of the whole packet loss rate range, and DLmax represents a maximum value of the whole packet loss rate range.
3) Treatment of energy consumption: and processing data according to the proportion of the residual energy to the total energy sum.
Figure BDA0001256889020000071
Wherein, EWValue represents the value of the average lost energy consumption of the network, EW represents the residual energy, and EWtotal represents the total energy.
4) And (3) processing the flow: and processing data according to the condition that the flow accounts for the whole data flow range.
Figure BDA0001256889020000072
Wherein, TPValue represents the average network flow, TPmin represents the minimum value of the whole flow range, and TPmax represents the maximum value of the whole flow range.
And (4) carrying out the four steps of processing on the network evaluation variables, and taking the processing result as final data of health degree evaluation.
Design of RBF neural network
The RBF neural network is a three-layer feedforward network comprising an input layer, an implicit layer and an output layer. And after the input layer node acquires the input vector, transmitting the input vector to the hidden layer. The structure is shown in fig. 3.
The input layer node obtains the input vector and transmits the input vector to the hidden layer. The hidden layer nodes of the RBF neural network are composed of radial basis functions, the radial basis functions adopt Gaussian functions, and the output layer is linear functions. The key problem in establishing the RBF neural network is that the center of a radial basis function is determined according to a given training sample, and the closer an input vector is to the center of the basis function, the larger the response of a hidden layer node is.
Step 1b, determination of parameters of basis functions
Determining the Gauss function of each hidden layer node according to the input vectorHeart value CjThe method comprises the following specific steps:
suppose that:
the ith input vector of the input layer of the RBF neural network, namely xi＝[TD′i,DL′i,EW′i,TP′i]T
Wherein: TD'i＝TDValue，DL′i＝DLValue，EW′i＝EWValue，TP′i＝TPValue
Step 1b-1, hidden layer initialization, initial class center Cj(1),j＝1,2,...,k。
Step 1b-2, in the r-th iteration, sample set { xiThe classification method is as follows: for all j, h | | | x 1,2, … k, j ≠ hi-ch(r)||<||xi-cj(r) |, then xi∈sh(r) wherein xiI sample representing a sample set, ch(r) class center when the number of hidden layer nodes is h, cj(r) represents the class center for the number j of hidden layer nodes.
Step 1b-3, reacting s obtained in step 1b-2h(r) a new class center of ch(r +1), we minimize the value of the metric below. Order:
Figure BDA0001256889020000073
minimum, (h ═ 1, 2.., k), then
Figure BDA0001256889020000074
Wherein N ishIs s ishNumber of samples in (r), HhExpressed as a value of a metric. .
Step 1b-4, for all h ═ 1,2, … k, if ch(r+1)＝ch(r), terminate otherwise return to Step 2.
Step 1c, determination of RBF network weight:
suppose that: the output response of the jth node of the hidden layer is:
Figure BDA0001256889020000081
where, δ is an optional parameter, determining the width of the basis function, k is the number of nodes of the hidden layer, | x-cjI represents a vector (x-c)j) Norm of (G)j(x) Representing the response of the jth basis function to the input vector.
Substituting the result of RBF network learning training into a formula:
Figure BDA0001256889020000082
wherein, according to the least square principle, the following is obtained:
Figure BDA0001256889020000083
Gjfor all Gj(x) A matrix is formed, y is all yiThe resulting vector is the desired output vector. New data outcomes, i.e., network QoS health, can be predicted.
2. Network situation awareness based on health degree
The network situation refers to the current state and the change trend of the whole network formed by the operating conditions of various network devices, network behaviors, user behaviors and other factors. The situation emphasizes the environment, the dynamics and the relationship among the entities, and is a state trend, an integral and macroscopic concept.
Fig. 4 is a flow chart of network situation analysis. The threshold Th of the network QoS health degree represents a minimum allowable QoS value for guaranteeing that the network is in a healthy state. The value of Th is used as a health threshold of the network QoS, and is generally set according to empirical data, and can be set artificially under special conditions. Setting of the health threshold Th of QoS has a significant influence on network evaluation, and requires careful treatment. The significance of analyzing the network situation by selecting the QoS health degree of the network at the current moment and the QoS health degree of the network at the last moment is to eliminate the influence of accidental error interference on the network. Whether the evaluation result of the network QoS health degree at the last moment is accurate or not can also influence the perception of the network situation.
Judging whether the current network QoS health degree is smaller than a threshold Th of the network QoS health degree, if not, judging the network health, if so, judging whether the network QoS health degree at the last moment of the network is smaller than the threshold Th of the network QoS health degree, if not, indicating the network health, if so, considering that the error of the network QoS health degree evaluation or the fluctuation of the network QoS health degree exists at the moment, and at the moment, judging that the network is in a sub-health state.
3. Network situation based QoS negotiation
The QoS negotiation of the wireless sensor network has the effect of more favorably processing and controlling network resources and better giving QoS support to users. Firstly, judging the support condition of the QoS requirement of the application according to the network situation result. When the network can support, the network gives corresponding resources according to the application QoS requirement. When the network can not support, the network negotiates with the user through a feedback mechanism, and reduces the application QoS grade or the application QoS requirement, thereby effectively adjusting the network resource, guaranteeing the high-grade QoS requirement as much as possible, and providing a satisfactory end-to-end QoS service for the user.
Based on the above considerations, the QoS negotiation algorithm of the wireless sensor network is as follows, and the flow chart is shown in fig. 5:
and 3a, obtaining an application QoS mapping result, and firstly, carrying out relevant judgment on the result.
And after the application proposed by the user passes the QoS mapping, obtaining an application QoS index and a corresponding weight. The QoS metrics are extracted and it is queried whether the metrics match the metrics in the network resources in the network. If the network does not have the index, the index is fed back to the user, and the application cannot be met. If so, the next step is performed.
And step 3 b: according to the situation of the network residual resources, the network negotiates the application passing through the step 1.
And (3) the network maintains the resource residual situation and the network situation evaluation at any time, when the application judged in the step (1) arrives, the network residual resource and the application QoS index are compared, whether the network resource can meet the application requirement is judged, and the support situation of the network QoS is decided.
When the network resource is satisfied, the network gives corresponding QoS service support to the application; when the network resources cannot be satisfied, the following processing can be performed:
and 3b-1, feeding back to the user, and coordinating to lower the application priority level so as to enable the network resources to support.
3b-2, terminating applications in the network at a lower priority, reserving resources for newly arriving applications with a high priority.
3b-2, rejecting the application and feeding back that the user network cannot meet the application.
The negotiation in step 3b is actually operated, and the QoS negotiation mode is as follows: fig. 6 is a flow chart of a QoS negotiation management algorithm.
QNM (QoS negotiation manager) is a link for connecting users and network system to carry out QoS negotiation, and the QoS negotiation mode designed by the invention is divided into two types: static and dynamic.
1) QoS static negotiation mode: before starting an application program, a user submits a QoS requirement of the application to a QNM, the QNM submits the QoS requirement of the application to a resource manager to apply for resources, the resource manager determines whether the requirement can be met according to the resource condition of a current system, if the requirement can be met, the resources are reserved for the application, a resource scheduler is informed to set corresponding scheduling parameters and return confirmation information to the QNM, the QNM immediately informs the user and starts the application, and the resource scheduler schedules the resources according to the preset parameters to meet the requirement of the application; otherwise, the resource manager returns rejection information and QoS information that the system can provide, QNM returns rejection information to the user immediately and displays the QoS suggested by the system on the negotiation interface, the user can modify the QoS requirement and re-negotiate with the system; it is also possible to wait for sufficient resources in the system before performing QoS negotiation with the system.
2) QoS dynamic negotiation mode: the user starts the application first, then inputs the QoS requirement dynamically according to the actual condition of the QoS provided by the system, the QNM submits the QoS requirement to the resource manager to negotiate resource allocation, and the resource manager informs the resource scheduler to modify the corresponding scheduling parameter according to the allocation result.
4. Packet scheduling with dynamic adjustment of QoS priority
Scheduling at a base station end:
4a, allocating bandwidth to different traffic flows
In the wireless sensor network adaptive QoS control technology, an application queuing strategy needing to be processed is as follows: and classifying according to the QoS requirements, wherein the classification comprises time delay, reliability, load and the like, and the applications of different classes are respectively put into corresponding caches according to the receiving time sequence of the applications to wait for the next processing.
The network manager provides end-to-end QoS for different types of traffic flows, of which there are mainly four different types: an active authorization service (UGS) for supporting a periodically generated trial-and-error service stream of fixed bit packets; real-time Polling Services (rtPS) for supporting real-time traffic of periodic long packet data; non-real-time polling services (non-real-time polling services) for supporting non-real-time variable bit rate service flows of non-periodic and variable-length packets; best effort service be (best effort) for supporting best effort type traffic flows without reliability guarantee. Fig. 7 is a diagram of a packet queue based on QoS requirements.
The bandwidth allocated to rtPS, nrtPS and BE traffic streams can BE calculated by:
BWi＝BW_PB×fi(14)
BW_PB＝BWall-BWUGS-BWr-BWBE(15)
wherein, BWiIs the total time slot allocated to the ith traffic stream, BW _ PB is the transmission time slot allocated to the rtPS, nrtPS and BE traffic stream data, BWallIs the total slot, BW, of transmission of an uplink subframeUGSIs a transmission slot, BW, allocated to UG traffic in an uplink subframerPolling time slot, BW, assigned to rtPS and nrtPS servicesBEIs a bandwidth request slot allocated to the BE service, fiIs the bandwidth calculation factor for the ith service flow. It can BE concluded that the BE traffic gets at least a part of the bandwidth after the high priority traffic increases, which avoids the starvation of the low priority traffic to some extent.
4b, allocating bandwidth to different users
And distributing the bandwidth obtained by the service flow to different users through a maximum-minimum fair sharing (MMFS) scheduling algorithm, wherein the algorithm can fully utilize the bandwidth and fairly distribute the bandwidth among the users. The specific steps of the algorithm are as follows:
suppose that: with SpsMIndicates the bandwidth obtained by user M, for Sps1,Sps2,Sps3L SpsKThat is, the user: 1,2,3LK; satisfies the following conditions: (S)ps1≤Sps2≤Sps3L≤SpsK) There are bandwidth requests for rtPS:
CrtPSis the total bandwidth of the rtPS traffic flow in the uplink.
The algorithm comprises the following steps:
4b-1, initializing the bandwidth allocated to each user.
4b-2, calculating the residual bandwidth: crtPS/K+(CrtPS/K-Sps1)/(K-1)。
4b-3, allocating the remaining bandwidth to the remaining users.
4b-4, and repeatedly executing the step 4b-2 and the step 4b-3 for the rest users.
4b-5, completing the allocation of all the bandwidths, ending the allocation, and if the rtPS service bandwidth of the user is not met, continuing to allocate the rtPS service bandwidth in the next frame.
And user side scheduling:
different scheduling algorithms are designed for different service flows, and data scheduling is carried out among different queues in the same service flow. The rtPS service flow has high real-time requirement, and an EDF (early dead time First) algorithm is adopted. When nrtPS traffic bandwidth scheduling, in order to ensure fairness of scheduling, a WFQ (Weighted fair queuing) algorithm is used. In order to reduce the complexity of BE best effort service flow scheduling and ensure the scheduling fairness among queues, a packet priority scheduling algorithm with the minimum packet length is adopted
(1) EDF algorithm
The real-time polling service is sensitive to time delay, so that the earliest time limit priority algorithm (EDF) for ensuring time delay is applied to ensure the real-time performance of rtPS service flow. EDF algorithm priority scheduling deadtimeThe earliest packet in time. Wherein, TdeadlineIndicates the deadline of the packet, TarriveIndicating the arrival time, T, of packets in the queuemaxdelayIndicating the maximum delay, T, allowed for a packetcAnd representing the current time of the system, calculating the time of delta T, and arranging the packets in the queue according to the sequence of delta T from small to large.
Tdeadline＝Tarrive+Tmaxdelay(16)
ΔT＝Tdeadline-Tc(17)
(2) Weighted fair queuing algorithm (WFQ) algorithm
The non-real-time polling service queue adopts a weighted fair queue algorithm (WFQ) algorithm taking rate as weight so as to ensure the fairness of nrtPS service flow scheduling. In the WFQ algorithm, the arrival rate of the packet in each queue is calculated to serve as the weight in the WFQ algorithm, a larger weight is distributed to the link with good link state, better fairness is obtained among the queues, and more bandwidth resources are distributed to the link with good link state.
(3) Packet priority scheduling algorithm with minimum packet length
The best effort service queue adopts a method of carrying out priority scheduling on the packet with the minimum packet length, which can avoid the unfairness to the short packet service flow service caused by the long service time of the long packet in the BE to a certain extent.

Claims (2)

1. A self-adaptive QoS control method facing a wireless sensor network is characterized by comprising the following steps:
step 1, QoS health degree evaluation based on RBF neural network, specifically:
step 1a, selecting an evaluation variable:
selecting four network QoS indexes of network delay, packet loss rate, throughput and energy consumption as evaluation variables of the QoS health degree of the wireless sensor network;
step 1b, determining parameters of the basis functions:
1b-1, hidden layer initialization, given an initial class center Cj(1) J 1,2, k, k is an implicit layer segmentThe number of points;
1b-2, in the r-th iteration, sample set { xiThe classification method is as follows: for all j, h | | x |, 1,2i-ch(r)||＜||xi-cj(r) |, then xi∈sh(r) wherein xiI sample representing a sample set, ch(r) class center when the number of hidden layer nodes is h, cj(r) represents the class center when the number of hidden layer nodes is j;
1b-3, reacting s obtained in step 1b-2h(r) a new class center of ch(r +1), minimizing the value of the metric, let:
Figure FDA0002280731880000011
minimum, (h ═ 1, 2.., k), then
Figure FDA0002280731880000012
Wherein N ishIs s ishNumber of samples in (r), HhIs the value of the metric;
1b-4, for all h 1,2h(r+1)＝ch(r), terminating, otherwise, returning to the step 1 b-2;
step 1c, determination of RBF network weight:
according to the result of learning training, the network QoS health can be given by the following formula:
Figure FDA0002280731880000013
wherein the content of the first and second substances,
Figure FDA0002280731880000014
delta is an optional parameter, determines the width of the basis function, k is the number of nodes of the hidden layer, cjFor class centers when the number of hidden layer nodes is j, | | x-cjI denotes the norm of the vector, Gj(x) Representing the response of the jth basis function to the input vector, GjFor all Gj(x) Formed matrix, y being all yiForm aVector of (a), ωjtThen representing the weight between the hidden layer and the output layer, and x represents any n-dimensional input vector;
step2, network situation perception based on the health degree is specifically carried out;
the network situation awareness based on the health degree specifically comprises the following steps:
judging whether the current network QoS health degree is smaller than a threshold Th of the network QoS health degree, if not, judging the network health, if so, judging whether the network QoS health degree at the last moment of the network is smaller than the threshold Th of the network QoS health degree, if not, indicating the network health, and if so, indicating the network is in sub-health;
and step 3, QoS negotiation based on network situation, wherein the strategy of QoS negotiation comprises the following elements:
step 3a, obtaining an application QoS index and a corresponding weight value after the application proposed by a user passes QoS mapping; extracting the QoS indexes, and inquiring whether the indexes in the network are consistent with the indexes in the network resources; if the network does not have the index, the index is fed back to the user to prompt that the application cannot be met; if yes, performing step 3 b;
step 3b, according to the situation of the network residual resources, negotiating the application judged in the step 3 a;
the network maintains the resource residual situation and the network situation evaluation at any time, when the application judged in the step 3a arrives, the network residual resource and the application QoS index are compared, whether the network resource can meet the application requirement is judged, and the support situation of the network QoS is decided; when the network resource can not be satisfied, the following processing is carried out:
3b-1, feeding back to the user, coordinating and reducing the application priority level so as to enable the network resources to support;
3b-2, terminating the application with lower priority in the network, and reserving resources for the application with high priority coming newly;
3b-3, rejecting the application and feeding back that the user network cannot meet the application;
step 4, dynamically adjusting the QoS priority, specifically:
the grouping scheduling strategy for dynamically adjusting the QoS priority comprises base station end scheduling and user end scheduling;
the base station end scheduling comprises:
4a, distributing the broadband to different service flows:
the bandwidth allocated to rtPS, nrtPS and BE traffic streams can BE calculated by:
BWi＝BW_PB×fi
BW_PB＝BWall-BWUGS-BWr-BWBE
wherein: BW (Bandwidth)iIs the total time slot allocated to the ith service flow, BW _ PB is the transmission time slot allocated to the rtPS, nrtPS and BE service flow data, BWallIs the total slot, BW, of transmission of an uplink subframeUGSIs a transmission slot, BW, allocated to UG traffic in an uplink subframerPolling time slot, BW, allocated to rtPS, nrtPS servicesBEIs a bandwidth request slot allocated to the BE service, fiIs the bandwidth calculation factor of the ith service flow;
4b, allocating bandwidth to different users:
4b-1, initializing the bandwidth allocated to each user;
4b-2, calculating the residual bandwidth;
4b-3, distributing the residual bandwidth to the residual users;
4b-4, repeatedly executing the step 4b-2 and the step 4b-3 for the rest users;
4b-5, completing the allocation of all the bandwidths, ending, and if the rtPS service bandwidth of a user is not met, continuing to allocate the rtPS service bandwidth in the next frame;
the user side scheduling comprises the following steps:
designing different scheduling algorithms for different service flows, and scheduling data among different queues in the same service flow; the rtPS service flow has high real-time requirement, and an EDF algorithm is adopted; when nrtPS service flow bandwidth scheduling is carried out, in order to ensure scheduling fairness, a WFQ algorithm is adopted; in order to reduce the complexity of BE best effort service flow scheduling and ensure the scheduling fairness among queues, a packet priority scheduling algorithm with the minimum packet length is adopted.
2. The adaptive QoS control method for wireless sensor networks according to claim 1, wherein: in step 3b, the negotiation modes for the application are divided into two types: a QoS static negotiation mode and a QoS dynamic negotiation mode;
QoS static negotiation mode: before starting application program, user submits QoS requirement of application to QNM, QNM submits it to resource manager to apply resource, resource manager reserves resource according to resource condition of current system, informs resource scheduler to set corresponding scheduling parameter;
QoS dynamic negotiation mode: the user starts the application first, and dynamically obtains the QoS requirement according to the actual condition of the QoS provided by the system, the QNM submits the QoS requirement to the resource manager to negotiate resource allocation, and the resource manager informs the resource scheduler to modify the corresponding scheduling parameter according to the allocation result.
CN201710193889.3A 2017-03-28 2017-03-28 Self-adaptive QoS control method for wireless sensor network Active CN106973413B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710193889.3A CN106973413B (en) 2017-03-28 2017-03-28 Self-adaptive QoS control method for wireless sensor network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710193889.3A CN106973413B (en) 2017-03-28 2017-03-28 Self-adaptive QoS control method for wireless sensor network

Publications (2)

Publication Number Publication Date
CN106973413A CN106973413A (en) 2017-07-21
CN106973413B true CN106973413B (en) 2020-04-28

Family

ID=59335869

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710193889.3A Active CN106973413B (en) 2017-03-28 2017-03-28 Self-adaptive QoS control method for wireless sensor network

Country Status (1)

Country Link
CN (1) CN106973413B (en)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10805222B2 (en) * 2017-05-01 2020-10-13 General Electric Company Resilient network configuration for time sensitive traffic
CN108259367B (en) * 2018-01-11 2022-02-22 重庆邮电大学 Service-aware flow strategy customization method based on software defined network
US11425735B2 (en) * 2018-04-05 2022-08-23 Qualcomm Incorporated Scheduling channel state information (CSI) processes in association with periodic traffic in a CoMP network
CN110519178B (en) * 2019-08-07 2023-05-12 京信网络系统股份有限公司 Low-delay data processing method, device and system
CN112888076B (en) * 2019-11-29 2023-10-24 华为技术有限公司 Scheduling method and device
CN112996125B (en) * 2019-12-13 2023-04-07 华为技术有限公司 Scheduling method and device
CN112291495B (en) * 2020-10-16 2021-08-03 厦门大学 Wireless video low-delay anti-interference transmission method based on reinforcement learning
CN112153702B (en) * 2020-11-05 2021-08-13 广州竞远安全技术股份有限公司 Local area network bandwidth resource allocation method, storage device and equipment
CN113038538B (en) * 2021-03-01 2024-02-20 许昌学院 Optimization distribution method and device for WSNs communication data transmission bandwidth of intelligent power distribution network
CN116016278A (en) * 2022-12-22 2023-04-25 四川九州电子科技股份有限公司 Dynamic adjustment method for EDCA parameters
CN117313004B (en) * 2023-11-29 2024-03-12 南京邮电大学 QoS flow classification method based on deep learning in Internet of things

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1645860A (en) * 2005-01-20 2005-07-27 中国科学院计算技术研究所 Terminal up dispatching method for voice service in wide band wireless cut-in network
CN1925365A (en) * 2006-09-25 2007-03-07 北京北方烽火科技有限公司 System and method for bandwidth distributing and dispatching of wireless broad band network
CN101222441A (en) * 2008-01-29 2008-07-16 北京北方烽火科技有限公司 WiMAX ascending scheduling and band width distribution method
CN101262651A (en) * 2008-04-18 2008-09-10 中国科学院计算技术研究所 Base station downlink data dispatching method and system in broadband wireless MAN
CN101790239A (en) * 2010-02-23 2010-07-28 中国电信股份有限公司 Packet dispatching method and forward service dispatcher
CN101827027A (en) * 2009-12-25 2010-09-08 中国科学院声学研究所 Interlayer coordination-based home network QoS guarantee method
CN104038957A (en) * 2014-06-13 2014-09-10 杭州大光明通信系统集成有限公司 4G base station operation maintenance information analysis process method based on integration structure

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1645860A (en) * 2005-01-20 2005-07-27 中国科学院计算技术研究所 Terminal up dispatching method for voice service in wide band wireless cut-in network
CN1925365A (en) * 2006-09-25 2007-03-07 北京北方烽火科技有限公司 System and method for bandwidth distributing and dispatching of wireless broad band network
CN101222441A (en) * 2008-01-29 2008-07-16 北京北方烽火科技有限公司 WiMAX ascending scheduling and band width distribution method
CN101262651A (en) * 2008-04-18 2008-09-10 中国科学院计算技术研究所 Base station downlink data dispatching method and system in broadband wireless MAN
CN101827027A (en) * 2009-12-25 2010-09-08 中国科学院声学研究所 Interlayer coordination-based home network QoS guarantee method
CN101790239A (en) * 2010-02-23 2010-07-28 中国电信股份有限公司 Packet dispatching method and forward service dispatcher
CN104038957A (en) * 2014-06-13 2014-09-10 杭州大光明通信系统集成有限公司 4G base station operation maintenance information analysis process method based on integration structure

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
IP网络的QoS参数预测和优化方案生成方法;王征;《CNKI》;20130101;正文第3章 *
WiMAX系统的分组调度算法研究;陈东;《万方》;20080829;正文第4章 *
认知网络QoS动态自适应监控机制研究;章鹏程;《万方》;20120806;摘要、正文3-5章 *

Also Published As

Publication number Publication date
CN106973413A (en) 2017-07-21

Similar Documents

Publication Publication Date Title
CN106973413B (en) Self-adaptive QoS control method for wireless sensor network
Sun et al. Autonomous resource slicing for virtualized vehicular networks with D2D communications based on deep reinforcement learning
CN108965024B (en) Virtual network function scheduling method based on prediction for 5G network slice
CN112039965B (en) Multitask unloading method and system in time-sensitive network
Ruan et al. Machine learning-based bandwidth prediction for low-latency H2M applications
CN111211830B (en) Satellite uplink bandwidth resource allocation method based on Markov prediction
KR100233091B1 (en) Atm traffic control apparatus and method
Alsahag et al. Fair uplink bandwidth allocation and latency guarantee for mobile WiMAX using fuzzy adaptive deficit round robin
Richart et al. Slicing with guaranteed quality of service in wifi networks
Ang et al. A performance analysis on packet scheduling schemes based on an exponential rule for real-time traffic in LTE
Zhou et al. Joint multi-objective optimization for radio access network slicing using multi-agent deep reinforcement learning
Chen et al. Learning-based proactive resource allocation for delay-sensitive packet transmission
Mardani et al. Robust resource allocation scheme under channel uncertainties for LTE-A systems
Fountoulakis et al. Dynamic power control for time-critical networking with heterogeneous traffic
Kahlon An embedded fuzzy expert system for adaptive WFQ scheduling of IEEE 802.16 networks
US20020168940A1 (en) Predictive fair polling mechanism in a wireless access scheme
US7801152B2 (en) Method and system for scheduling utilization of resources, related communication network and computer program product
CN116074260A (en) Service slice scheduling method in power network
Samanta et al. On low latency uplink scheduling for cellular haptic communication to support tactile internet
Rezaee et al. A fuzzy algorithm for adaptive multilevel queue management with QoS feedback
Kumar et al. Neural network based Scheduling Algorithm for WiMAX with improved QoS constraints
Al-Kashoash et al. Optimization-based hybrid congestion alleviation
Ambika et al. An effective queuing architecture for elastic and inelastic traffic with different dropping precedence in MANET
Pan et al. Optimizing Internet of Things Fog Computing: Through Lyapunov-Based Long Short-Term Memory Particle Swarm Optimization Algorithm for Energy Consumption Optimization
Farzaneh IRCTGP: a novel intra-node resource control protocol based on traffic priority and evolutionary games theory for wireless sensor networks

Legal Events

Date Code Title Description
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