CN112804728B - Access control method for mass terminals of power internet of things based on context learning - Google Patents

Access control method for mass terminals of power internet of things based on context learning Download PDF

Info

Publication number
CN112804728B
CN112804728B CN202110005745.7A CN202110005745A CN112804728B CN 112804728 B CN112804728 B CN 112804728B CN 202110005745 A CN202110005745 A CN 202110005745A CN 112804728 B CN112804728 B CN 112804728B
Authority
CN
China
Prior art keywords
terminal
access
base station
time slot
expressed
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
CN202110005745.7A
Other languages
Chinese (zh)
Other versions
CN112804728A (en
Inventor
周振宇
贾泽晗
廖海君
赵雄文
张磊
张素香
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.)
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Shandong Electric Power Co Ltd
North China Electric Power University
Original Assignee
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Shandong Electric Power Co Ltd
North China Electric Power University
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 State Grid Corp of China SGCC, Electric Power Research Institute of State Grid Shandong Electric Power Co Ltd, North China Electric Power University filed Critical State Grid Corp of China SGCC
Priority to CN202110005745.7A priority Critical patent/CN112804728B/en
Publication of CN112804728A publication Critical patent/CN112804728A/en
Application granted granted Critical
Publication of CN112804728B publication Critical patent/CN112804728B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W48/00—Access restriction; Network selection; Access point selection
    • H04W48/02—Access restriction performed under specific conditions
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y10/00—Economic sectors
    • G16Y10/35—Utilities, e.g. electricity, gas or water
    • 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 Security & Cryptography (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Computing Systems (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention discloses a power Internet of things mass terminal access control method based on context learning, which is applied to power Internet of things mass terminal access control.

Description

Access control method for mass terminals of power internet of things based on context learning
Technical Field
The invention relates to the technical field of power internet of things, in particular to a context learning-based access control method for a mass terminal of the power internet of things.
Background
The electric power internet of things is an industrial-grade internet of things for realizing the mutual connection and the man-machine interaction of all links of an electric power system, the level of accurate control and intelligent scheduling of a power grid can be improved based on the depth perception capability and the advanced information communication technology, and the transition of the traditional electric power system to the energy internet is promoted. However, due to the limitation of wireless resources and computing resources, the concurrent access of mass terminals in the power internet of things greatly increases the bearing pressure of the access network, causes the problems of network congestion, overload and the like, and seriously threatens the safe and stable operation of the power grid. Existing access control techniques can be divided into two broad categories, contention and non-contention. The contention access control technique improves the access success probability by restricting access of a part of terminals during a high congestion period. However, the contention access control technique relies on frequent signaling interaction between the base station and the terminal, and is only suitable for a scenario with low connection density, which is difficult to meet the access requirement of the explosively increased terminal. Compared with the competitive access control technology, the non-competitive access control technology has the advantages of low signaling overhead, high resource utilization rate, high bearing capacity and the like. 3GPP introduces a fast uplink authorization technology in Release14, allowing a terminal obtaining uplink permission to directly perform data transmission on a channel pre-allocated by a base station under the condition of not sending any scheduling request, and reducing the signaling overhead and the probability of access conflict. However, most of the existing studies are based on the assumption that global information is known, and have limitations in practical applications.
Therefore, at the present stage, research on access control of a mass terminal of the power internet of things is still in a starting stage, and several key challenges are not solved:
(1) firstly, the power internet of things terminal is in an active state only when data is transmitted, the rest time slots are in a dormant state, and a set of active terminals needs to be accurately predicted by using a fast uplink authorization technology.
(2) Secondly, due to the limitation of network resources and signaling overhead, the base station cannot accurately obtain all information of a large number of terminals, including terminal states, channel gains, queue backlogs and the like, and an uplink authorization strategy needs to be formulated in an information uncertain scene.
(3) Finally, different types of terminals have different access requirements, and the base station needs to meet the different access service quality requirements of a large number of terminals.
Disclosure of Invention
The purpose of the invention is: the method overcomes the defects in the prior art, dynamically adjusts the uplink authorization strategy without knowing information such as future channel state, terminal queue backlog and the like in advance, and can improve the total network energy efficiency while meeting the requirement of terminal access service quality.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
1. building a system model
(1) Access model
An access model of a multi-cell mass terminal of an electric power internet of things based on fast uplink authorization is shown in fig. 1. The whole network consists of J cells and K (K > J) terminals, each cell comprises a base station, an edge server and a plurality of terminals to be accessed, wherein the base station provides wireless access service for the terminals, and the edge server and the base station are positioned at the same position and provide calculation service. The base station set is expressed as Σ ═ s1,s2...,sJDenoted as Y ═ u }, set of terminals1,u2,...,uK}. Defining base stations sjThe set of covered terminals is represented as
Figure BDA0002883076690000021
Base station sjAnd base station sj′The set of terminals in the coverage overlap area is denoted as Nj∩j′＝Nj∩Nj′. The invention adopts a time slot model, a total time period is divided into T time slots with equal length, the length of each time slot is tau, and a total time slot set is represented as tau {1, 2. It is assumed that channel state information between a base station and a terminal is constant within one slot and varies from slot to slot. Due to the characteristics of event-driven activation or periodic activation and the like of the power internet of things terminal, the terminal is supposed to have two states, namely an active state and a dormant state, and when the terminal has a data transmission requirement, the terminal is automatically in the active state, otherwise, the terminal is in the dormant state. At the beginning of each time slot, the base station predicts the terminal set in an active state and selects the terminal to distribute uplink permission, and the authorized terminal can establish connection with the base station and perform data transmission.
(2) Task transmission model
The invention adopts a task division model, and the task is divided into a limited number of subtasks with the same size.Suppose that each slot arrives at terminal u firstkHas a number of subtasks ofk(t) each subtask size is ρ, and all data is first stored in ukIn the local task cache. When u iskAnd when the connection with the base station is established, the task data is unloaded to the edge server for calculation. Is stored in ukTask data in local cache is modeled as a queue qkQueue backlog of Qk(t) the initial queue backlog at time t +1 time slot is
Qk(t+1)＝max{Qk(t)-Uk(t)+ρAk(t),0} (1)
Wherein, Uk(t) represents ukThe amount of task data offloaded to the edge server at the t-th slot. Defining an active indicator variable as ak(t) when there is a data transmission need for the terminal, i.e. Qk(t) > 0, the terminal is in an active state and is represented as ak(t) ═ 1; otherwise, the system is in a dormant state and is denoted as ak(t) is 0, when the terminal has a data transmission requirement, i.e. Qk(t) > 0, the terminal is in an active state and is represented as ak(t) ═ 1; otherwise, the system is in a dormant state and is denoted as ak(t) is 0. Defining an authorization indicator variable as xk,j(t)，xk,j(t) '1' indicates that terminal u is in time slot tkObtaining a base station sjOtherwise xk,j(t)＝0。
Considering uplink data transmission, terminal ukAnd base station sjThe signal-to-noise ratio of the data transmission between can be expressed as
Figure BDA0002883076690000031
Wherein, PTX,kIs the transmission power, gk,j(t) is the t-th time slot ukAnd sjChannel gain between, Bk,jIs the transmission bandwidth, N0Is the noise power spectral density. Thus a transmission rate of
Figure BDA0002883076690000032
Terminal ukTransmitting to base station s in the t time slotjIs expressed as
Uk,j(t)＝xk,j(t)min{Qk(t)+ρAk(t),τRk,j(t)} (4)
Terminal ukThe throughput at the t-th slot is expressed as
Figure BDA0002883076690000033
(3) Energy efficiency model
In the t-th time slot, terminal ukOffloading task data to base station sjThe energy consumption of (1) is the product of transmission power and transmission delay, i.e.
Figure BDA0002883076690000034
ukOffloading task data to sjIs defined as throughput Uk,j(t) energy consumption Ek,jThe ratio of (t), i.e., the size of the amount of data that can be transmitted per unit energy (bits/J), is expressed as
Figure BDA0002883076690000035
(4) Access quality of service requirement model
Let Xk,TAnd T'kRespectively represent terminals ukObtaining the total time slot number authorized by the base station and the total time slot number in the active state in T time slots, then
Figure BDA0002883076690000036
Figure BDA0002883076690000041
The access quality of service requirement model can be defined as
Figure BDA0002883076690000042
Wherein eta isk∈(0,1]Represents ukAccess quality of service constraints.
2. Problem modeling
The aim of the invention is to maximize the overall energy efficiency of the network under the long-term constraints of the access quality of service requirements. Thus, the optimization objective is expressed as:
Figure BDA0002883076690000043
Figure BDA0002883076690000044
Figure BDA0002883076690000045
Figure BDA0002883076690000046
wherein, C1The number of terminals which can be authorized by the base station is M; c2Indicating that each terminal can only be authorized by one base station to carry out data transmission in each time slot; c3Long term constraints are imposed on access quality of service requirements.
3. Algorithm design
(1) Problem transformation
P1 is difficult to solve directly because short-term base station decisions are coupled with long-term optimization objectives and constraints. The access service quality requirement can be restrained for a long time by the concept of virtual queues in Lyapunov optimization3Translating into a queue stability constraint. Defining an access quality of service requirementVirtual queue of deficit Fk(t), the update formula is as follows
Figure BDA0002883076690000047
Its meaning is terminal ukDeviation between actual access performance and regulatory requirements at the t-th time slot.
According to Lyapunov theory, a vector ψ (t) ═ F is definedk(t)]And represents a Lyapunov function of
Figure BDA0002883076690000048
Lyapunov drift is defined as the expected value of L (θ (t)) that varies in two consecutive time slots, expressed as
ΔL(ψ(t))＝E[L(ψ(t+1))-L(ψ(t))|ψ(t)] (14)
Under long-term constraints of access quality of service requirements, a drift minus reward is defined to trade off minimizing drift or maximizing reward, i.e., minimizing access quality of service requirement deficit or maximizing energy efficiency, expressed as
ΔVL(ψ(t))＝ΔL(ψ(t))-VE[ζ(t)|ψ(t)] (15)
Wherein the content of the first and second substances,
Figure BDA0002883076690000051
v is a non-negative weight parameter, whose larger value tends to maximize energy efficiency, whereas it tends to minimize access quality of service requirement deficit.
Substituting equations (13) and (14) into equation (15) and reducing, the upper bound of the available drift minus reward is
Figure BDA0002883076690000052
Wherein C is a constant and does not affect the Lyapunov optimization. Thus, P1 translates to an upper bound for minimizing the Drift minus reward (or the inverse of the upper bound for maximizing the Drift minus reward), expressed as
Figure BDA0002883076690000053
θk,j(t) is the inverse of the upper bound of the Drift minus reward, i.e. the weighted sum of energy efficiency and Access Performance, expressed as
Figure BDA0002883076690000054
Wherein, VEEAnd VFFkAnd (t) weights corresponding to energy efficiency and access performance respectively.
(2) Access control algorithm based on context-aware learning
Based on the multi-arm slot Machine (MAB) theory, the base station and the terminal are respectively modeled into a player and a rocker arm, and the problem of maximization of the accumulated income of the base station is solved by using an Upper Confidence Bound (UCB) algorithm based on terminal state perception.
The conventional MAB problem assumes that all terminals are available at every slot and is not applicable to scenarios where terminals have both active and dormant states. The present invention therefore considers an improved dynamic MAB problem, i.e. the set of active terminals is dynamically changing over time. However, under the fast uplink grant architecture, the base station cannot sense the states of all terminals, and when the base station selects a terminal in a dormant state in a time slot t, the terminal has no data transmission requirement, which may cause resource waste. Therefore, the invention considers that the base station has an active terminal prediction algorithm, and the prediction algorithm can predict the terminal u in each time slot according to the network flow modelkActive probability P ofk(t) and establishing an active terminal set. The access control algorithm (CLAC) based on context learning can be combined with various active terminal prediction algorithms for use, and has strong expansibility and compatibility.
Because the traditional UCB algorithm has limitation in solving the dynamic MAB problem, the invention improves the dynamic MAB algorithm by combining with a prediction algorithm, adds terminal state perception on the basis of the traditional UCB algorithm, and has the performance upper bound expression of
Figure BDA0002883076690000061
Wherein z isk,j(t) base station s to time slot tjSelecting terminal ukCumulative reward obtained, nk,j(t) denotes terminal u to time slot tkNumber of total time slots, t ', active and granted'kIndicating to time slot t terminal ukTotal number of slots in active state. Compared with the traditional UCB algorithm, the method only calculates the total active time slot number of the selected terminal, but not the total iterative times of the algorithm, and can ensure Vk,jThe calculation of (t) is more accurate. z is a radical ofk,j(t)、nk,j(t) and t'kAre respectively as follows
Figure BDA0002883076690000062
Compared with the prior art, the method has the following advantages and effects:
(1) the terminal state perception is improved, and the base station only selects the terminal authorization from the active terminal set based on the terminal state perception, so that the resource waste caused by distributing the authorization to the dormant terminal is avoided, and the network performance is reduced;
(2) the perception of access service quality requirement is improved, a base station dynamically optimizes a terminal authorization strategy based on the perception of access service quality requirement, and when a terminal ukWhen the actual access performance deviates seriously from the specified requirement, FkAnd (t) is gradually increased to force the base station to authorize the base station, so that the access performance of the base station is guaranteed.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate a certain embodiment of the present invention, and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is a terminal access model of the present invention based on fast uplink grant, wherein 1 — base station; 2-edge server; 3-active terminal; 4-dormant terminal; 5-authorizing the terminal; 6- -fast uplink grant;
FIG. 2 is an average energy efficiency in an embodiment of the invention;
FIG. 3 is a diagram of an average queue backlog in an embodiment of the present invention;
FIG. 4 is a diagram illustrating the ratio of terminals meeting quality of service requirements for access in an embodiment of the present invention;
FIG. 5 is a diagram illustrating average QoS requirement deficit accumulation according to an embodiment of the present invention;
FIG. 6 is a graph showing the variation of CLAC algorithm performance with prediction accuracy probability in an embodiment of the present invention;
fig. 7 is a diagram illustrating the effect of energy efficiency and access performance weights on the performance of the CLAC algorithm in an embodiment of the present invention.
Detailed Description
The present invention is further illustrated by the following detailed description.
The CLAC algorithm proposed by the invention comprises three stages, namely an initialization stage (2-3 lines), a decision stage (5-15 lines) and a learning stage (16-30 lines), as shown in an algorithm 1.
Algorithm 1CLAC algorithm
Figure BDA0002883076690000071
Figure BDA0002883076690000081
The invention carries out simulation experiment on the CLAC algorithm, and sets three baseline algorithms for comparing and verifying the performance, wherein the baseline algorithms are set as follows:
baseline algorithm 1: and the energy efficiency priority access control algorithm maximizes the total energy efficiency of the network based on a terminal state prediction algorithm without considering the long-term constraint of the access service quality requirement.
Baseline algorithm 2: the access control algorithm based on reinforcement learning maximizes the total energy efficiency of the network under the long-term constraint of the access service quality requirement, but does not consider a terminal state prediction algorithm.
Baseline algorithm 3: a fast uplink grant algorithm, which randomly distributes uplink grants to terminals,
terminal access quality of service requirement constraints, energy efficiency and state prediction are not considered.
The simulation parameters are set as follows:
Figure BDA0002883076690000091
fig. 2-5 show the average energy efficiency, average queue backlog, terminal proportion satisfying the access quality of service requirement, and average access quality of service requirement deficit backlog with time slot under four algorithms, respectively. As can be seen from fig. 2, since only energy efficiency optimization is considered, the energy efficiency performance of the baseline algorithm 1 is optimal, but as the time slot increases, the CLAC algorithm can gradually approach the performance of the baseline algorithm 1.
It can be seen from fig. 3 that the CLAC algorithm provided the best performance, and can maintain the queue backlog at a low level, which is 77.90%, 97.68% and 83.83% lower than baseline algorithm 1, baseline algorithm 2 and baseline algorithm 3, respectively.
As can be seen from fig. 4, the performance of the CLAC algorithm is significantly better than the other three algorithms, and compared with baseline algorithm 1, baseline algorithm 2, and baseline algorithm 3, the proportion of terminals meeting the access quality of service requirements is respectively improved by 15.07%, 77.46%, and 54.95%. It can be seen from fig. 2 and 4 that the CLAC algorithm achieves the balance between energy efficiency and access performance, and maximizes the total network energy efficiency as much as possible on the premise of ensuring the requirement of terminal access service quality, while the baseline algorithm 1 trades for higher energy efficiency by sacrificing the terminal access performance.
As can be seen from fig. 5, the CLAC algorithm can minimize the deficit backlog of the access quality of service requirement due to the terminal state awareness and the access quality of service requirement awareness. As can be seen from fig. 4 and 5, the baseline algorithm 2 has a low proportion of terminals meeting the access qos requirements, but has a small deficit backlog of the access qos requirements, because the baseline algorithm 2 considers the access qos requirement constraint and distributes the grant frequently to the terminals with high access qos requirements, while ignoring a large number of terminals with low access qos requirements. The CLAC algorithm has the best comprehensive performance as can be seen by combining the figures 2-5.
Figure 6 shows the variation of the average energy efficiency of terminals in the CLAC algorithm and the proportion of terminals meeting the requirements of access quality of service with the probability of prediction accuracy. It can be seen that, with the improvement of the prediction accuracy probability, the average energy efficiency of the terminal and the proportion of the terminal meeting the access service quality requirement are both increased. The reason is that when the prediction accuracy probability is low, the base station frequently distributes authorization to the dormant terminal, so that the active terminal cannot be accessed, resource waste is caused, and the overall performance of the network is reduced.
Fig. 7 shows the variation of the terminal average energy efficiency and the terminal scale to meet the access quality of service requirement with a, where a is defined as VEEAnd VFThe ratio of (a) to (b) is used for representing the attention of the terminal to the requirements of energy efficiency and access service quality. Simulation results show that as alpha increases, the terminal pays more attention to energy efficiency and ignores access performance, so that the average energy efficiency of the terminal gradually increases, and the proportion of the terminal meeting the access service quality requirement gradually decreases.
According to experimental data, the power internet of things terminal access control method adopting context-aware learning has the following technical effects: (1) the terminal state perception is improved, and the base station only selects the terminal authorization from the active terminal set based on the terminal state perception, so that the resource waste caused by distributing the authorization to the dormant terminal is avoided, and the network performance is reduced; (2) the perception of the access service quality requirement is improved, the base station dynamically optimizes the terminal authorization strategy based on the perception of the access service quality requirement, and when the terminal ukWhen the actual access performance deviates seriously from the specified requirement, Fk(t) will gradually increaseAnd the base station is forced to authorize the base station, and the access performance of the base station is guaranteed.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and it should be noted that, for those skilled in the art, many modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (1)

1. A power Internet of things mass terminal access control method based on context learning is characterized by comprising the following steps: step (1) a system model is established, which is used for realizing that a base station predicts a terminal set in an active state and selects a terminal to distribute uplink permission, the authorized terminal can establish connection with the base station and carry out data transmission, and the access service quality is restrained; constructing a maximum network total energy efficiency model for realizing maximum network total energy efficiency under the long-term constraint of access service quality requirements; step (3), algorithm design, namely solving the problem of maximization of the accumulated income of the base station based on an upper signal boundary algorithm of terminal state perception;
the specific process of building the system model in the step S1 comprises the following steps:
step S101, an access time slot model is built, a total time period is divided into T time slots with equal length, the length of each time slot is tau, a total time slot set is represented as tau {1, …, T, …, T }, channel state information between a base station and a terminal in one time slot is unchanged and changes among the time slots, at the beginning of each time slot, the base station predicts a terminal set in an active state and selects the terminal to distribute uplink permission, and the terminal which obtains authorization can establish connection with the base station and perform data transmission;
step S102, a task transmission model is established, and each time slot reaches a terminal u initiallykHas a number of subtasks ofk(t) each subtask size is ρ, and all data is first stored in ukWhen u is in the local task cachekWhen the connection with the base station is established, the task data is unloaded to the edge server for calculation and is stored in ukNumber of tasks in local cacheIs modeled as a queue qkQueue backlog of Qk(t) the initial queue backlog at time t +1 time slot is
Qk(t+1)＝max{Qk(t)-Uk(t)+ρAk(t),0}，
Defining an active indicator variable as ak(t) when there is a data transmission need for the terminal, i.e. Qk(t) > 0, the terminal is in an active state and is represented as ak(t) ═ 1; otherwise, the system is in a dormant state and is denoted as ak(t) 0, defining an authorization indicator variable as xk,j(t)，xk,j(t) '1' indicates that terminal u is in time slot tkObtaining a base station sjOtherwise xk,j(t) ═ 0; terminal ukAnd base station sjThe signal-to-noise ratio of the data transmission between is expressed as
Figure FDA0003514572360000011
Wherein, PTX,kIs the transmission power, gk,j(t) is the t-th time slot ukAnd sjChannel gain between, Bk,jIs the transmission bandwidth, N0Is the noise power spectral density, the transmission rate is
Figure FDA0003514572360000012
Terminal ukTransmitting to base station s in the t time slotjIs expressed as
Uk,j(t)＝xk,j(t)min{Qk(t)+ρAk(t),τRk,j(t)}；
Terminal ukThe throughput at the t-th slot is expressed as
Figure FDA0003514572360000021
Step S103, an energy efficiency model is constructed, and in the t-th time slot,terminal ukOffloading task data to base station sjThe energy consumption of (1) is the product of transmission power and transmission delay, i.e.
Figure FDA0003514572360000022
ukOffloading task data to sjIs defined as throughput Uk,j(t) energy consumption Ek,j(t) the ratio, the size of the amount of data that can be transmitted per unit energy (bits/J), is expressed as
Figure FDA0003514572360000023
Step S104, constructing an access service quality demand model:
Xk,Tand T'kRespectively represent terminals ukThe total time slot number authorized by the base station and the total time slot number in the active state are obtained in T time slots,
Figure FDA0003514572360000024
Figure FDA0003514572360000025
the access quality of service requirement model is defined as:
Figure FDA0003514572360000026
wherein eta isk∈(0,1]Denotes ukAccess quality of service constraints;
step S2, constructing a maximum network total energy efficiency model for realizing maximum network total energy efficiency under the long-term constraint of access service quality requirement, wherein the maximum network total energy efficiency target is expressed as:
Figure FDA0003514572360000031
long term constraint of access quality of service requirements C3Converting into queue stability constraint;
virtual queue F for accessing qos requirementsk(t), the formula is as follows:
Figure FDA0003514572360000032
its meaning is terminal ukDeviation between actual access performance and specified requirements at the t-th time slot; defining vector psi (t) ═ Fk(t)]The lyapunov function is expressed as:
Figure FDA0003514572360000033
lyapunov drift is defined as the expected value of L (θ (t)) that varies over two consecutive time slots, expressed as:
ΔL(ψ(t))＝Ε[L(ψ(t+1))-L(ψ(t))|ψ(t)]；
under the long-term constraints of access quality of service requirements, a drift minus reward is defined to trade off minimizing drift or maximizing reward, i.e. minimizing access quality of service requirement deficit or maximizing energy efficiency, expressed as:
Figure FDA0003514572360000034
the upper bound for the drift minus reward is:
Figure FDA0003514572360000035
where C is a constant, P1 translates to an upper bound of minimizing the Drift minus the reward (or the inverse of the upper bound of maximizing the Drift minus the reward), expressed as
Figure FDA0003514572360000041
θk,j(t) is the inverse of the upper bound of the drift minus the reward, i.e. the weighted sum of energy efficiency and access performance, expressed as:
Figure FDA0003514572360000042
CN202110005745.7A 2021-01-05 2021-01-05 Access control method for mass terminals of power internet of things based on context learning Active CN112804728B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110005745.7A CN112804728B (en) 2021-01-05 2021-01-05 Access control method for mass terminals of power internet of things based on context learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110005745.7A CN112804728B (en) 2021-01-05 2021-01-05 Access control method for mass terminals of power internet of things based on context learning

Publications (2)

Publication Number Publication Date
CN112804728A CN112804728A (en) 2021-05-14
CN112804728B true CN112804728B (en) 2022-05-03

Family

ID=75807967

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110005745.7A Active CN112804728B (en) 2021-01-05 2021-01-05 Access control method for mass terminals of power internet of things based on context learning

Country Status (1)

Country Link
CN (1) CN112804728B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113747514B (en) * 2021-08-27 2022-08-16 爱浦路网络技术(南京)有限公司 Service quality parameter allocation method, computer device, and storage medium
CN113905347B (en) * 2021-09-29 2024-04-16 华北电力大学 Cloud edge end cooperation method for air-ground integrated power Internet of things
CN114629804A (en) * 2022-04-19 2022-06-14 华北电力大学 5G slice arrangement method adaptive to millisecond-level source network load shedding load
CN115580577A (en) * 2022-09-23 2023-01-06 华北电力大学 Intelligent park multi-state routing protocol optimization method based on time synchronization perception

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102932913B (en) * 2012-10-24 2018-10-26 南京中兴新软件有限责任公司 Obtain method, server, gateway and the system of MAB terminal access positions
CN109905334B (en) * 2019-03-01 2021-06-08 华北电力大学 Access control and resource allocation method for power Internet of things mass terminal
CN110493826B (en) * 2019-08-28 2022-04-12 重庆邮电大学 Heterogeneous cloud wireless access network resource allocation method based on deep reinforcement learning
CN111182509B (en) * 2020-01-07 2020-11-24 华北电力大学 Ubiquitous power Internet of things access method based on context-aware learning

Also Published As

Publication number Publication date
CN112804728A (en) 2021-05-14

Similar Documents

Publication Publication Date Title
CN112804728B (en) Access control method for mass terminals of power internet of things based on context learning
CN113242568B (en) Task unloading and resource allocation method in uncertain network environment
CN113612843B (en) MEC task unloading and resource allocation method based on deep reinforcement learning
Dai et al. Joint load balancing and offloading in vehicular edge computing and networks
CN110493360B (en) Mobile edge computing unloading method for reducing system energy consumption under multiple servers
CN111953758B (en) Edge network computing unloading and task migration method and device
CN112600921A (en) Heterogeneous mobile edge network-oriented dynamic task unloading method
US11831708B2 (en) Distributed computation offloading method based on computation-network collaboration in stochastic network
CN111954236A (en) Hierarchical edge calculation unloading method based on priority
CN112188627B (en) Dynamic resource allocation strategy based on state prediction
US9369925B2 (en) System and methods to achieve optimum efficiency-jain fairness in wireless systems
CN111885147A (en) Dynamic resource pricing method in edge calculation
CN114650228B (en) Federal learning scheduling method based on calculation unloading in heterogeneous network
CN114760311A (en) Optimized service caching and calculation unloading method for mobile edge network system
CN112860429A (en) Cost-efficiency optimization system and method for task unloading in mobile edge computing system
Xing et al. Task classification unloading algorithm for mobile edge computing in smart grid
Yao et al. Data-driven resource allocation with traffic load prediction
Zhang et al. Endogenous security-aware resource management for digital twin and 6G edge intelligence integrated smart park
WO2021083230A1 (en) Power adjusting method and access network device
Mei et al. Semi-decentralized network slicing for reliable V2V service provisioning: A model-free deep reinforcement learning approach
US7924777B2 (en) Method for deciding transmission priority of non-realtime data and apparatus and method for controlling interference between cells using the same
Messaoud et al. In-depth performance evaluation of network slicing strategies in large scale industry 4.0
Wang et al. Offloading Strategies for Mobile Edge Computing Based on Multi-Attribute Preferences in Smart Grids
Ma Communication Resource Allocation Strategy of Internet of Vehicles Based on MEC.
Liao et al. Task migration and resource allocation scheme in iovs with roadside unit

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