WO2016045329A1 - 小基站开关的控制方法及装置 - Google Patents

小基站开关的控制方法及装置 Download PDF

Info

Publication number
WO2016045329A1
WO2016045329A1 PCT/CN2015/074911 CN2015074911W WO2016045329A1 WO 2016045329 A1 WO2016045329 A1 WO 2016045329A1 CN 2015074911 W CN2015074911 W CN 2015074911W WO 2016045329 A1 WO2016045329 A1 WO 2016045329A1
Authority
WO
WIPO (PCT)
Prior art keywords
base station
small base
user
rate
value
Prior art date
Application number
PCT/CN2015/074911
Other languages
English (en)
French (fr)
Inventor
张冬英
甘小莺
王绍鹏
李楠
Original Assignee
中兴通讯股份有限公司
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 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2016045329A1 publication Critical patent/WO2016045329A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
    • H04W52/02—Power saving arrangements
    • 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

Definitions

  • the present invention relates to the field of communications, and more particularly to a method and apparatus for controlling a small base station switch.
  • a small base station also known as a small cell (small cell base station)
  • small cell base station can improve spectrum utilization by reducing the cell radius.
  • laying a small base station invisibly increases the energy consumption of the wireless system, so how to improve the energy efficiency of the wireless system has become an important issue in the research in this field.
  • the threshold needs to be recalculated. And the scheme considers that the arrival rate of the user is known, that is, the change of the user of the wireless communication system is known, which is also inconsistent with the actual situation.
  • the invention provides a method and a device for controlling a small base station switch to solve at least the above problems.
  • a method for controlling a small base station switch includes: counting user arrival rate in a predetermined time period; estimating a current user arrival rate according to a statistical result, and obtaining an estimated rate; The estimated rate adjusts a switching policy of the small base station to which the user belongs.
  • estimating a current user arrival rate according to the statistical result, and obtaining the estimated rate comprises: establishing a Markov model according to the statistical result; and obtaining the estimated rate according to the Markov model.
  • establishing a Markov model according to the statistical result comprises: establishing a first state set for the acquired user arrival rate as ⁇ 1 , ⁇ 2 , . . . , ⁇ n ⁇ , wherein the ⁇ 1 , The values of ⁇ 2 , . . . , ⁇ n represent the values of the user arrival rate at different times of the user, respectively.
  • the method before adjusting the switching policy of the small base station to which the user belongs according to the estimated rate, the method includes: determining, according to the first state set, a value range [ ⁇ min , ⁇ max ] of the user arrival rate, The ⁇ min is a minimum value of the first state set, and the ⁇ max is a maximum value of the first state set; the value range is quantized into a second state set according to a preset quantization precision. s 0 , s 1 , s 2 ... s m ⁇ , where s 0 ⁇ s 1 ⁇ s 2 ⁇ ... ⁇ s m , and s 0 ⁇ min ⁇ max ⁇ s m , the m is taken The value is determined by the quantization precision.
  • adjusting a switching policy of the small base station to which the user belongs according to the estimated rate including: according to a predetermined expected function Q and the user arrival rate Determining the value of a, where a is the number of small base stations that are enabled in the current time slot, and a ranges from [0, n], where n is the number of current small base stations; and is calculated according to the following formula When a is a value of 0, 1, 2, ..., the probability that the corresponding value of a is selected And according to the Calculate the current number of open base stations a t according to the following formula:
  • the expected function Q is updated according to the following formula: among them, Representing the expected function value at the current time slot t; s' is the user arrival rate corresponding to the next time slot t+1, Indicates the expected function value in the next time slot t+1, ⁇ represents the convergence speed of the algorithm, and the value ranges from 0 to 1; ⁇ is the discount factor, and the value ranges from 0 to 1; R t is the current time
  • the expected function Q is updated according to the following formula: among them, Representing the expected function value at the current time slot t; s' is the user arrival rate corresponding to the next time slot t+1, Indicates the expected function value in the next time slot t+1, ⁇ represents the convergence speed of the algorithm, and the value ranges from 0 to 1; ⁇ is the discount factor, and the value ranges from 0 to 1; R t is the current time
  • the expected function Q is updated according to the following formula: among them, Representing the expected function value at the current time slot t;
  • the plurality of small base stations are multiple, and the plurality of small base stations and one macro base station form a heterogeneous wireless network.
  • a control device for a small base station switch comprising: a statistics module configured to perform statistics on a user arrival rate within a predetermined time period; and a determining module configured to estimate based on the statistical result The current user arrival rate is obtained, and the adjustment module is configured to adjust a switching policy of the small base station to which the user belongs according to the estimated rate.
  • the determining module comprises: an establishing unit configured to establish a Markov model according to the statistical result; and an acquiring unit configured to acquire the estimated rate according to the Markov model.
  • the technical solution for calculating the current user arrival rate according to the user's rate in the predetermined time period and adjusting the switching rate of the small base station according to the statistical result is solved, and the related art is solved.
  • the energy efficiency of the base station sleeps the problem of deviation from the actual situation caused by the user's arrival rate change is not considered, and the sub-optimal solution of the system can be obtained, and the dynamic characteristics of the user are satisfied, which is more realistic.
  • FIG. 1 is a schematic diagram of a base station switch in the prior art
  • FIG. 2 is a flowchart of a method for controlling a small base station switch according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of networking in accordance with an embodiment of the present invention.
  • FIG. 5 is still another flowchart of a small base station switch according to a preferred embodiment of the present invention.
  • FIG. 6 is a structural block diagram of a control apparatus for a small base station switch according to an embodiment of the present invention.
  • FIG. 7 is a block diagram showing another structure of a control device for a small base station switch according to an embodiment of the present invention.
  • FIG. 8 is a flow chart of a small base station switch in accordance with a preferred embodiment of the present invention.
  • FIG. 9 is a schematic diagram of a model of a Markov Poisson process in accordance with a preferred embodiment of the present invention.
  • FIG. 10 is a schematic diagram of a cumulative revenue comparison between a dynamic sleep algorithm and a fixed sleep strategy according to an embodiment of the present invention
  • FIG. 11 is a schematic diagram of a convergence process of an algorithm according to an embodiment of the present invention.
  • FIG. 2 is a flowchart of a method for controlling a small base station switch according to an embodiment of the present invention. As shown in FIG. 2, the method includes the following steps. step:
  • Step S202 Perform statistics on the arrival rate of the user in the predetermined time period
  • Step S204 Estimating the current user arrival rate according to the statistical result, and obtaining an estimated rate
  • Step S206 Adjust a switching policy of the small base station to which the user belongs according to the foregoing estimated rate.
  • the technical solution for estimating the current user arrival rate according to the user's rate in the predetermined time period and adjusting the switching strategy of the small base station according to the statistical result is solved in the related art.
  • the problem of deviation from the actual situation caused by the user's arrival rate change is not considered, and the sub-optimal solution of the system can be obtained, and the dynamic characteristics of the user are satisfied, which is more realistic.
  • step S204 is various.
  • the method may be implemented in the following manner, but is not limited thereto: the Markov model is established according to the foregoing statistical result; and the estimated rate is obtained according to the Markov model.
  • the Markov model may be established in the following manner: establishing a first state set of the obtained user arrival rate as ⁇ 1 , ⁇ 2 , . . . , ⁇ n ⁇ , wherein The values of ⁇ 1 , ⁇ 2 , ..., ⁇ n above represent the values of the user arrival rate at different times of the user.
  • the following processing may be further performed: determining, according to the first state set, the value of the user arrival rate. a range [ ⁇ min , ⁇ max ], wherein the ⁇ min is a minimum value in the first state set, and the ⁇ max is a maximum value of the second state set; and the value range is quantized according to a preset quantization precision. s 0 , s 1 , s 2 ... s m ⁇ , where s 0 ⁇ s 1 ⁇ s 2 ⁇ ... ⁇ s m , and s 0 ⁇ min ⁇ max ⁇ s m , the value of the above m Determined by the above quantization precision.
  • a switching policy of the small base station to which the user belongs according to the foregoing estimated rate, specifically: according to a predetermined expected function Q and the user arrival rate Determining the value of a, where a is the number of small base stations that are enabled in the current time slot, and a ranges from [0, n], where n is the number of current small base stations; and a is calculated according to the following formula: When the value is 0, 1, 2, ..., the probability that the value corresponding to a above is selected And according to the above Calculate the current number of open base stations a t according to the following formula:
  • the above expected function Q is updated according to the following formula: among them, Representing the expected function value at the current time slot t; s' is the user arrival rate corresponding to the next time slot t+1, Indicates the expected function value in the next time slot t+1, ⁇ represents the convergence speed of the algorithm, and the value ranges from 0 to 1; ⁇ is the discount factor, and the value ranges from 0 to 1; R t is the current time
  • the above expected function Q is updated according to the following formula: among them, Representing the expected function value at the current time slot t; s' is the user arrival rate corresponding to the next time slot t+1, Indicates the expected function value in the next time slot t+1, ⁇ represents the convergence speed of the algorithm, and the value ranges from 0 to 1; ⁇ is the discount factor, and the value ranges from 0 to 1; R t is the current time The system benefits of the gap.
  • the plurality of small base stations are multiple, and the plurality of small base stations and one macro base station form a heterogeneous wireless network
  • the user arrival rate may refer to a rate at which the data packet reaches the small base station, which is simply referred to as a packet arrival rate. It can also be called the user packet arrival rate.
  • the embodiments of the present invention are directed to the above-mentioned prior art deficiencies, and propose a multi-base station dormancy scheme based on reinforcement learning, which models user dynamics through a Markov process and simultaneously estimates by statistical means.
  • the current user arrival rate is measured, and the base station switching policy is adjusted according to the arrival rate.
  • n small base stations and one macro base station together form a heterogeneous wireless network, as shown in FIG.
  • the user accepts the network service by sending a data packet, and specifies the wireless network service protocol.
  • the user performs wireless communication by sending a data packet
  • the user within the coverage of the macro base station can receive the wireless signal of the small base station and the macro base station in the working state, and the small base station in the dormant state cannot provide the service to the user.
  • the arrival process of all data packets is modeled using a Poisson process with a packet arrival rate of ⁇ .
  • the user's dynamic performance is represented by the change of the packet arrival rate.
  • the change of the packet arrival rate can be represented by a Markov process.
  • the state set of the Markov process is ⁇ 1 , ⁇ 2 , ... ⁇ L ⁇ , where L
  • the status indicates the L value of the packet arrival rate.
  • Packets are sent and received in a queue-first-served manner, and unserved packets are queued.
  • the rate at which the macro base station transmits data packets is ⁇ M
  • the rate at which a single small base station transmits data packets is ⁇ S .
  • the rate at which a single macro base station and a small base station transmit data packets is constant, and the system dynamically switches the base station to adjust the transmission rate.
  • the process can be seen in FIG. In each time slot, the number of data packets transmitted by the system is N s,p , and the length of the time slot is t s , then 0 ⁇ N s, p ⁇ t slot ( ⁇ M + n ⁇ S ).
  • the entire switching policy is scheduled by the control center of the macro base station.
  • the macro base station is always in working state, and the small base station performs dynamic switching by means of enhanced learning.
  • the switching strategy is implemented in the form of time slots, and the time slot length is t s .
  • the small base station dynamically selects whether to sleep according to the scheme.
  • the power consumed by the macro base station in each time slot can be expressed as follows
  • P M represents the power consumed by the macro base station
  • P M, 0 represents the constant power of the macro base station
  • P M, t represents the power consumed by the macro base station per unit of data transmission unit
  • ⁇ M represents the rate at which the macro base station transmits and receives data packets.
  • the power consumed by the small base station in each time slot can be expressed as follows:
  • P S represents the power consumed by the small base station
  • P S, 0 represents the constant power of the small base station
  • P S, t represents the power consumed by the small base station per unit of data transmission unit
  • ⁇ S represents the rate at which the small base station transmits and receives data packets.
  • the loss generated is ⁇ , and these losses are equally divided, so the loss of the small base station to turn on or off is ⁇ /2.
  • the system estimates the current state of the system by counting the number of users arriving within a certain time T s .
  • the state that the system needs to estimate is actually the user's arrival rate.
  • the second set of states ⁇ s 0 , s 1 , s 2 ... s m ⁇ in the embodiment, where s 0 ⁇ s 1 ⁇ s 2 ⁇ ... ⁇ s m and s 0 ⁇ min ⁇ max ⁇ s m .
  • the estimated state value is obtained by:
  • FIG. 5 Another flow chart of a small base station switch is also provided, and the specific implementation manner of the design is as follows:
  • Step S502 the control center of the macro base station determines a state set, and notifies all small base stations of the state set;
  • Step S504 at the beginning of each learning time slot, each small base station counts its own system arrival users.
  • the state of the cell is obtained by the third formula
  • Step S506 the status of the cell Compared with the previous statistical results, if not, go to step S508, if they are the same, go to step S504, that is, no signaling communication;
  • Step S508 the new statistical result is reported to the control center of the macro base station by signaling
  • Step S510 after receiving the information notified by the small base station, the control center updates the base station data, and integrates all the information to give the current state of the system, and the update mode is:
  • Step S512 counting the time when the small base station is not updated
  • Step S514 it is determined whether there is a small base station has not responded for a long time, if yes, go to step S516, if no, go to step S512;
  • Step S5166 the small base station if the report is not performed for a long time, then every T p slots, forces the center of the small base station control information reported, to avoid blocking signaling path leading to inaccurate information.
  • the Q value is obtained from the system benefit
  • the small base station switching strategy is dynamically selected according to the Q value system
  • the number of base stations is turned on.
  • the learner needs to set a Q value for each combination of state and behavior, indicating the expected function of obtaining the income, which is Q(s, a), where s represents the system state and a represents the system behavior.
  • the system state s is estimated by the second step.
  • the behavior a of the system refers to the number of small base stations that are turned on in the time slot, so the value range of a is Set all Q values to be 0 at the beginning.
  • Selecting the number of open a of the small base station is determined by the following method:
  • the number of open base stations is selected in a hybrid strategy, for each Give the probability of choosing it as:
  • the system generates a random number, and selects the number of open base stations a t in the time slot according to the probability obtained above.
  • Each time slot randomly selects a small base station to be turned on and off according to the obtained a t .
  • the Q value is updated.
  • the state of the system is estimated as s', and the update method is as follows:
  • N s,p represents the number of data packets transmitted and received by the system in the time slot
  • r p represents the revenue obtained by the system serving a single user
  • E c represents the energy loss of the base station, which can be obtained by Equations 1 and 2
  • E c ( P M + a t P S ) t slot
  • l t represents the queue length
  • ⁇ represents the influence factor of the system queue delay on the system performance
  • the larger ⁇ indicates that the system is more sensitive to the delay, and the user has higher requirements on the quality of service.
  • a t and a t-1 respectively indicate the number of small base stations that are opened in the time slot and the previous time slot.
  • the steps of the second step and the third step are repeated, and finally the system will converge to a stable state, and the strategy of the switching base station remains basically unchanged.
  • the user models the user according to the Markov Poisson process, and by using the algorithm of reinforcement learning, a macro base station small base station cooperative transmission dormancy scheme is proposed.
  • a macro base station small base station cooperative transmission dormancy scheme is proposed.
  • the invention effectively suppresses frequent switching of the base station by considering the switching loss, and obtains a suboptimal solution.
  • the algorithm complexity is simple.
  • FIG. 6 is a structural block diagram of a control apparatus of a small base station switch according to an embodiment of the present invention. As shown in Figure 6, the device includes:
  • the statistics module 60 is configured to perform statistics on the arrival rate of the user within a predetermined time period
  • the determining module 62 is connected to the statistic module 60 and configured to estimate the current user arrival rate according to the statistical result, and obtain an estimated rate;
  • the adjusting module 64 is connected to the determining module 62, and is configured to adjust a switching policy of the small base station to which the user belongs according to the foregoing estimated rate.
  • the technical solution based on the user's rate in the predetermined time period is used, and the current user arrival rate can be estimated according to the statistical result, thereby adjusting the switching strategy of the small base station, and the related solution is solved.
  • the problem that the user arrives at the rate change is deviated from the actual situation, and the sub-optimal solution of the system can be obtained, and the dynamic characteristics of the user are satisfied, which is more realistic.
  • the determining module 62 includes: an establishing unit 620, configured to establish a Markov model according to the foregoing statistical result; and an obtaining unit 622 connected to the establishing unit 620, configured to acquire the foregoing according to the Markov model Estimated rate.
  • the macro base station has a constant on power P M, and 0 is 30 W, which indicates that the power P M consumed by the macro base station per unit data packet is 10 W, which indicates that the rate M M of the macro base station transmitting and receiving data packets is 0.5.
  • the small base station has a constant on power P S, 0 is 7.5 W, which indicates that the power consumed by the small base station per unit of data packet P S, t is 2.5 W, and the rate of the small base station transmitting and receiving data packets ⁇ S is 2.
  • the discount factor ⁇ is 0.9 and the switching loss ⁇ is set to 50J.
  • the length of a single time slot t s is 30 s.
  • FIG. 8 is a flow chart of a small base station switch according to a preferred embodiment of the present invention, as shown in FIG.
  • Step S802 the two small base stations and one macro base station jointly form a heterogeneous wireless network.
  • the user accepts the network service by sending a data packet, and specifies the wireless network service protocol.
  • the user performs wireless communication by transmitting a data packet
  • the user within the coverage of the macro base station can receive the wireless signal of the small base station and the macro base station in the working state, and the small base station in the dormant state cannot provide the user with the data. service.
  • All arrivals of the data packet are modeled using a Poisson process with a packet arrival rate of ⁇ .
  • the user's dynamic performance is represented by the change of the packet arrival rate.
  • the change of the packet arrival rate can be represented by a Markov process.
  • the state set of the Markov process is ⁇ 1 , ⁇ 2 , ⁇ 3 ⁇ , 3 of which The status indicates three values of the packet arrival rate.
  • step S804 the system estimates the current state of the system by counting the number of users arriving within a certain time T s .
  • the state that the system needs to estimate is actually the user's arrival rate.
  • the system first needs to know the range of user packet arrival rate [0, 4], determine the state set ⁇ s 0 , s 1 , s 2 , ... s 4 ⁇ , where s 0 ⁇ s 1 ⁇ s 2 ⁇ ... ⁇ s 4 and take An integer from 0 to 4.
  • the specific operation method can be obtained by the second step in the preferred embodiment of the control process of the small base station switch described above.
  • Step S806 at the beginning of each time slot, according to the method of reinforcement learning, the Q value is obtained from the system benefit, the small base station switching strategy is dynamically selected according to the Q value system, and finally the number of base stations is turned on.
  • the learner needs to set a Q value for each combination of state and behavior, denoted as Q(s, a), where s represents the system state and a represents the system behavior.
  • the system state is estimated in the second step.
  • the behavior of the system refers to the number of small base stations that are turned on in the time slot, so the value range of a is Set all Q values to be 0 at the beginning.
  • the number of open a of the small base station is selected according to the state s of the system.
  • the number of open small base stations a is determined by Equation 6.
  • step S808 the system state S is updated, and at the end of each time slot, the Q value is updated according to Equation 7.
  • step S804 and S806 may be repeated after performing step S808, and finally the system will converge to a stable state, and the strategy of the switching base station remains substantially unchanged.
  • FIG. 9 is a schematic diagram of a model of a Markov Poisson process in accordance with a preferred embodiment of the present invention.
  • the ordinate represents the change in packet arrival rate over time and is continuously shifted between 1, 2, and 3.
  • FIG. 10 is a schematic diagram of a cumulative revenue comparison between a dynamic sleep algorithm and a fixed sleep policy according to an embodiment of the present invention. In a fixed sleep strategy, a small base station that is fixedly opened half is turned on. It can be seen that the system obtained by the dynamic dormancy scheme proposed by the present application has greater benefits.
  • FIG. 11 is a schematic diagram of a convergence process of an algorithm according to an embodiment of the present invention. The figure shows that different ⁇ values are obtained, and the algorithms converge after 50 time slots.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the embodiments of the present invention achieve the following beneficial effects: the problem of deviating from the actual situation caused by the user's arrival rate change is not considered in solving the energy efficiency of the base station sleep in the related art, and the system can be obtained.
  • the sub-optimal solution meets the dynamic characteristics of the user and is more realistic.
  • modules or steps of the present invention described above can be implemented by a general-purpose computing device that can be centralized on a single computing device or distributed across a network of multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device such that they may be stored in the storage device by the computing device and, in some cases, may be different from the order herein.
  • the steps shown or described are performed, or they are separately fabricated into individual integrated circuit modules, or a plurality of modules or steps thereof are fabricated as a single integrated circuit module.
  • the invention is not limited to any specific combination of hardware and software.
  • a technical solution for performing statistics on a user rate in a predetermined time period according to a user, and estimating a current user arrival rate according to the statistical result, and thereby adjusting a switching policy of the small base station is adopted.
  • the problem of deviating from the actual situation caused by the change of the user's arrival rate is not considered in solving the energy efficiency of the sleep of the base station in the related art, and the sub-optimal solution of the system can be obtained, and the dynamic characteristics of the user are satisfied.

Landscapes

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

Abstract

本发明提供了一种小基站开关的控制方法及装置，其中所述控制方法包括：对预定时间段内的用户到达速率进行统计；根据统计结果估测当前的用户到达速率，得到预估速率；根据所述预估速率调整所述用户所归属的小基站的开关策略。采用本发明实施例提供的上述技术方案，解决了相关技术中在解决基站休眠的能效时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。

Description

小基站开关的控制方法及装置 技术领域
本发明涉及通信领域，更具体地说，涉及一种小基站开关的控制方法及装置。
背景技术
随着无线通信业务量的不断增加，无线网络对系统通信容量的要求越来越高，传统的单层蜂窝网络已经难以满足日益增长的流量需求。在增加系统容量的方法中，铺设小基站是最有效的方法之一。小基站，又称为小小区(small cell基站)，可以通过减小小区半径来提高频谱利用率。但是铺设小基站无形之中增加了无线系统的能量消耗，所以如何提高无线系统的能量效率成为了该领域研究中的重要问题。
相关技术中，对无线通信网络基站休眠的能效问题进行了深入的研究，经过对现有的技术文献检索发现，Jian Wu，Sheng Zhou，Zhisheng Niu在发表在IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS上的“Traffic-Aware Base Station Sleeping Control and Power Matching for Energy-Delay Tradeoffs in Green Cellular Networks”将基站休眠问题与排队论相结合，提出了一种基于队列长度门限的开关方案。在该方案中，系统根据用户到达速率确定基站开关队列门限，控制基站开关，如图1所示。图中系统的状态由基站开关状态{0,1}和队列长度{0,1,...N...}来表示，根据用户到达速率λ求解出队列长度的最佳开关门限N。
但其没有考虑用户的动态性，一旦用户到达速率发生变化，需要重新计算门限。并且该方案下考虑用户的到达速率是已知的，即无线通信系统已知用户的变化情况，这也与实际情况不符。
针对相关技术中在解决基站休眠的能效问题时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，目前尚未提出有效的解决方案。
发明内容
本发明提供了一种小基站开关的控制方法及装置，以至少解决上述问题。
根据本发明的一个实施例，提供了一种小基站开关的控制方法，包括：对预定时间段内的用户到达速率进行统计；根据统计结果估测当前的用户到达速率，得到预估速率；根据所述预估速率调整所述用户所归属的小基站的开关策略。
优选地，根据统计结果估测当前的用户到达速率，得到预估速率包括：根据所述统计结果建立马尔可夫模型；根据马尔可夫模型获取所述预估速率。
优选地，根据所述统计结果建立马尔可夫模型，包括：对获取的所述用户到达速率建立第一状态集为{λ1，λ2，……，λn}，其中所述λ1，λ2，……，λn的取值分别表示所述用户在不同时刻用户到达速率的取值。
优选地，根据所述预估速率调整所述用户所归属的小基站的开关策略之前，包括：根据所述第一状态集确定所述用户到达速率的取值范围[λmin，λmax]，其中，所述λmin为所述第一状态集中的最小值，所述λmax为所述第一状态集的最大值；根据预设量化精度对所述取值范围量化为第二状态集{s0，s1，s2……sm}，其中，s0＜s1＜s2＜……＜sm，且，s0＜λmin＜λmax＜sm，所述m的取值由所述量化精度确定。
优选地，根据统计结果估测当前的用户到达速率，得到预估速率，包括：获取当前时刻到之前Ts时刻的所述小基站的当前用户数量
Figure PCTCN2015074911-appb-000001
根据以下公式确定所述当前用户到达速率
Figure PCTCN2015074911-appb-000002
Figure PCTCN2015074911-appb-000003
其中，所述sK取值自所述第二状态集，K＝0,1，……m。
优选地，根据所述预估速率调整所述用户所归属的小基站的开关策略，包括：根据预先确定的期望函数Q以及所述用户到达速率
Figure PCTCN2015074911-appb-000004
确定a的取值，其中，所述a为当前时隙下开启的小基站的数量，a取值范围为[0,n]，其中，n为当前小基站的数量；根据以下公式依次计算出a取值为0,1,2,……n时，所述a对应的取值所被选择的概率
Figure PCTCN2015074911-appb-000005
Figure PCTCN2015074911-appb-000006
并根据所述
Figure PCTCN2015074911-appb-000007
按照以下公式计算当前小基站的开启数量at：
Figure PCTCN2015074911-appb-000008
优选地，在到达当前时隙t结束时间时，根据以下公式对所述期望函数Q进行更新：
Figure PCTCN2015074911-appb-000009
其中，
Figure PCTCN2015074911-appb-000010
表示所述当前时隙t下的期望函数值；s'为下一时隙t+1所对应的用户到达速率，
Figure PCTCN2015074911-appb-000011
表示所述下一时隙t+1下的期望函数值，η表征算法的收敛速度，取值范围为0～1；γ为折扣因子，取值范围为0～1；Rt为所述当前时隙的系统收益。
优选地，所述小基站为多个，并且所述多个小基站与一个宏基站组成异构无线网络。
根据本发明的另一个实施例，还提供了一种小基站开关的控制装置，包括：统计模块，设置为对预定时间段内的用户到达速率进行统计；确定模块，设置为根据统计结果估测当前的用户到达速率，得到预估速率；调整模块，设置为根据所述预估速率调整所述用户所归属的小基站的开关策略。
优选地，所述确定模块包括：建立单元，设置为根据所述统计结果建立马尔可夫模型；获取单元，设置为根据马尔可夫模型获取所述预估速率。
通过本发明，采用根据用户在预定时间段内的用户速率进行统计，并根据统计结果能够估测到当前的用户到达速率进而能够调整小基站的开关策略的技术方案，解决了相关技术中在解决基站休眠的能效问题时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。
附图说明
此处所说明的附图用来提供对本发明的进一步理解，构成本申请的一部分，本发明的示意性实施例及其说明用于解释本发明，并不构成对本发明的不当限定。在附图中：
图1现有技术中基站开关的说明示意图；
图2为根据本发明实施例的小基站开关的控制方法的流程图；
图3为根据本发明实施例的组网示意图；
图4为根据本发明优选实施例的系统状态估测的流程图；
图5为根据本发明优选实施例的小基站开关的又一流程图
图6为根据本发明实施例的小基站开关的控制装置的结构框图；
图7为根据本发明实施例的小基站开关的控制装置的另一结构框图；
图8为根据本发明优选实施例的小基站开关的流程图；
图9为根据本发明优选实施例的马尔可夫泊松过程的模型示意图；
图10为根据本发明实施例的动态休眠算法与固定休眠策略的累计收益比较示意图；
图11为根据本发明实施例的算法的收敛过程示意图。
具体实施方式
下文中将参考附图并结合实施例来详细说明本发明。需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。
本发明的其它特征和优点将在随后的说明书中阐述，并且，部分地从说明书中变得显而易见，或者通过实施本发明而了解。本发明的目的和其他优点可通过在所写的说明书、权利要求书、以及附图中所特别指出的结构来实现和获得。
为了使本技术领域的人员更好地理解本发明方案，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分的实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本发明保护的范围。
为了解决上述技术问题，本发明实施例提供了一种小基站开关的控制方法，图2为根据本发明实施例的小基站开关的控制方法的流程图，如图2所示，该方法包括如下步骤：
步骤S202：对预定时间段内的用户到达速率进行统计；
步骤S204：根据统计结果估测当前的用户到达速率，得到预估速率；
步骤S206：根据上述预估速率调整上述用户所归属的小基站的开关策略。
通过上述各个步骤，采用根据用户在预定时间段内的用户速率进行统计，并根据统计结果能够估测到当前的用户到达速率进而能够调整小基站的开关策略的技术方案，解决了相关技术中在解决基站休眠的能效问题时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。
步骤S204的实现方式有多种，在一个优选实施方式中可以采用以下方式实现，但不限于此：根据上述统计结果建立马尔可夫模型；根据马尔可夫模型获取上述预估速率。
在本发明实施例的一个可选实施例中，可以以下方式建立马尔可夫模型：对获取的上述用户到达速率建立第一状态集为{λ1，λ2，……，λn}，其中上述λ1，λ2，……，λn的取值分别表示上述用户的不同时刻用户到达速率的取值。
可选地，在步骤S206之前，即在根据上述预估速率调整上述用户所归属的小基站的开关策略之前，还可以执行以下处理过程：根据上述第一状态集确定上述用户到达速率的取值范围[λmin，λmax]，其中，上述λmin为上述第一状态集中的最小值，上述λmax为上述第二状态集的最大值；根据预设量化精度对上述取值范围量化为{s0，s1，s2……sm}，其中，s0＜s1＜s2＜……＜sm，且，s0＜λmin＜λmax＜sm，上述m的取值由上述量化精度确定。
在本发明实施例中，步骤S204还可通过以下过程实现，但不限于此：获取当前时刻到之前Ts时刻的上述小基站的当前用户数量
Figure PCTCN2015074911-appb-000012
根据以下公式确定上述当前用户到达速率
Figure PCTCN2015074911-appb-000013
Figure PCTCN2015074911-appb-000014
其中，上述sK取值自上述第二状态集，K＝0,1，……m。
在确定了当前用户到达速率
Figure PCTCN2015074911-appb-000015
之后，根据上述预估速率调整上述用户所归属的小基站的开关策略，具体地：根据预先确定的期望函数Q以及上述用户到达速率
Figure PCTCN2015074911-appb-000016
确定a的取值，其中，上述a为当前时隙下开启的小基站的数量，a取值范围为[0,n]，其中，n为当前小基站的数量；根据以下公式依次计算出a取值为0,1,2,……n时，上述a对 应的取值所被选择的概率
Figure PCTCN2015074911-appb-000017
Figure PCTCN2015074911-appb-000018
并根据上述
Figure PCTCN2015074911-appb-000019
按照以下公式计算当前小基站的开启数量at：
Figure PCTCN2015074911-appb-000020
可选地，在到达当前时隙结束时间时，根据以下公式对上述期望函数Q进行更新：
Figure PCTCN2015074911-appb-000021
其中，
Figure PCTCN2015074911-appb-000022
表示所述当前时隙t下的期望函数值；s'为下一时隙t+1所对应的用户到达速率，
Figure PCTCN2015074911-appb-000023
表示所述下一时隙t+1下的期望函数值，η表征算法的收敛速度，取值范围为0～1；γ为折扣因子，取值范围为0～1；Rt为所述当前时隙的系统收益。
需要说明的是，上述小基站为多个，并且上述多个小基站与一个宏基站组成异构无线网络，上述用户到达速率可以指的是数据包达到小基站的速率，简称为包到达速率，也可称为用户包到达速率。
综上所述，实际上本发明实施例针对上述现有技术不足，提出了一种基于强化学习的多基站休眠方案，该方案将用户动态性通过马尔可夫过程建模，同时通过统计方式估测当前用户到达速率，根据到达速率调整基站开关策略。
为了更好的理解上述小基站开关的控制过程，以下结合一个优选实施例进行说明，但不用于限定本发明实施例。
第一步，n个小基站和一个宏基站共同组成异构无线网络，如图3所示。用户以发送数据包的方式接受网络服务，规定无线网络服务协议。
本方案中用户以发送数据包的方式进行无线通信，宏基站覆盖范围内的用户可以接收到处于工作状态的小基站和宏基站的无线信号，处于休眠状态的小基站无法为用户提供服务。所有数据包的到达过程用泊松过程建模，其包到达率为λ。用户的动态性通过包到达率的变化表现，包到达率的变化可以用一个马尔可夫过程表示，该马尔可夫过程的状态集为{λ1,λ2,…λL}，其中的L个状态表示包到达率的L种取值。数据包的收发按照队列方式先到先服务，未服务的数据包排列在队列中。宏基站发送数据包的速率为μM，单个小基站发送数据包的速率为μS。单个宏基站和小基站发送数据包的速率是恒定不变的，系统动态开关基站调整发送速率，该过程可见图4。在每个时隙内，系统发送数据包的个数为Ns,p，时隙的长度为ts，则有0≤Ns,p<tslot(μM+nμS)。
在本发明实施例中，整个开关策略由宏基站的控制中心进行调度。宏基站一直处于工作状态，小基站则通过强化学习的方式进行动态的开关。开关策略以时隙的方式实现，时隙长度为ts。在每个时隙开始时，小基站根据方案动态选择是否休眠。每个时隙内，宏基站的消耗的功率可以表示如下
PM＝PM,0+μMPM,t；       式一
其中PM表示宏基站消耗的功率，PM,0表示宏基站恒定开启功率，PM,t表示宏基站每发送单位数据包消耗的功率，μM表示宏基站收发数据包的速率。
同样的，每个时隙内，小基站消耗的功率可以表示如下：
PS＝PS,0+μSPS,t；       式二
其中PS表示小基站消耗的功率，PS,0表示小基站恒定开启功率，PS,t表示小基站每发送单位数据包消耗的功率，μS表示小基站收发数据包的速率。
每次开启关闭小基站一次会产生的损耗为β，将这些损耗做均分，所以小基站进行开启或关闭的损耗为β/2。
第二步，系统通过统计一定时间Ts内到达用户数，估计系统的当前状态。
系统需要估计的状态实际是用户的到达速率。系统首先需要知道用户包到达率的范围[λmin,λmax]，对其重新量化，根据量化精度要求确定状态集S＝{s0,s1,s2,…sm}(相当于上述实施例中的第二状态集{s0，s1，s2……sm})，其中s0<s1<s2<…<sm且s0<λmin<λmax<sm。然后统计从当前时间到之前Ts时刻系统到达的用户数
Figure PCTCN2015074911-appb-000024
则估测的状态值可由下式得到：
Figure PCTCN2015074911-appb-000025
0≤k≤m；       式三
为了减少信令开销，如图5所示，还提供了一种小基站开关的又一流程图，其设计的具体的实现方式如下：
步骤S502，宏基站的控制中心确定状态集，并将状态集通知所有小基站；
步骤S504，在每个学习时隙开始时，每个小基站统计自己的系统到达用户数
Figure PCTCN2015074911-appb-000026
由式三得出该小区的状态
Figure PCTCN2015074911-appb-000027
步骤S506，将小区的状态
Figure PCTCN2015074911-appb-000028
与之前的统计结果进行比较，如果不相同，则转步骤S508，如果相同，则转步骤S504，即不进行信令通信；
步骤S508，将新的统计结果通过信令报知给宏基站的控制中心；
步骤S510，在接收小基站告知的信息后，控制中心更新该基站数据，并综合所有信息给出系统当前状态，更新方式为：
Figure PCTCN2015074911-appb-000029
0≤k≤m；       式四
其中，θ＝1.5时
Figure PCTCN2015074911-appb-000030
表示宏基站统计的状态值，1≤i≤n时
Figure PCTCN2015074911-appb-000031
表示第i个小基站统计的状态值；
在控制中心通知各基站状态集S后，还可执行以下过程：
步骤S512，统计小基站未更新时间；
步骤S514，判断是否有小基站长时间未响应，如果是，则转到步骤S516，如果否，则转到步骤S512；
步骤S516，如果小基站长时间不进行汇报，则每过Tp个时隙，控制中心强制该小基站上报信息，避免信令通道阻塞导致信息不准确。
第三步，在每个时隙开始时，根据强化学习地方法，由系统收益得出Q值，根据Q值系统动态选择小基站开关策略，最后得出开启基站的数量。
首先，需要确定Q值。强化学习中，学习者需要为每一个状态和行为的组合设定一个Q值，表示获得收益的期望函数，为Q(s,a)，其中s表示系统状态，a表示系统行为。系统状态s由第二步估测得到，系统的行为a是指在该时隙中，小基站的开启数量，所以a的取值范围为
Figure PCTCN2015074911-appb-000032
设定所有的Q值初始均为0。
在第t个时隙开始时，根据系统的状态
Figure PCTCN2015074911-appb-000033
选择小基站的开启数量a。选择小基站开启数量a具体由如下方法确定：
小基站的开启数量的选择采取混合策略的方式，对于每一个
Figure PCTCN2015074911-appb-000034
给出选择它的概率为：
Figure PCTCN2015074911-appb-000035
       式五
然后通过系统生成随机数，按照上面得到的概率选出该时隙下小基站的开启数量at为
Figure PCTCN2015074911-appb-000036
       式六
每个时隙根据得到的at随机地选择要开启和关闭的小基站。在每个时隙结束时，对Q值进行更新。此时再次根据第二步估测系统的状态为s'，更新方法如下：
Figure PCTCN2015074911-appb-000037
       式七
其中
Figure PCTCN2015074911-appb-000038
表示更新后的对应于状态
Figure PCTCN2015074911-appb-000039
和行为at的Q值，Qt(s,at)表示更新前的对应于状态
Figure PCTCN2015074911-appb-000040
和行为at的值；η为学习因子，主要决定算法的收敛速度，取值范围为0～1；γ为折扣因子，取值范围为0～1；Rt为该时隙的系统收益，具体形式如下：
Figure PCTCN2015074911-appb-000041
       式八
其中,Ns,p表示该时隙中系统收发数据包的数量，rp表示系统服务单个用户得到的收益，Ec表示基站能量损耗，该值可由式一和式二得到，Ec＝(PM+atPS)tslot；lt表示队列长度，ω表示系统队列延迟对系统性能的影响因子，ω越大表示系统对延迟越敏感，用户对服务质量的要求越高。为了抑制基站频繁开关造成过大损耗，式中最后减去开关小基站带来的损耗，at和at-1分别表示该时隙和上一时隙开启小基站的数量。
第四步，重复第二步和第三步的步骤，最终系统会收敛到稳定状态，此时开关基站的策略基本保持不变。
本发明上述优选实施例提供的上述技术方案，将用户根据马尔可夫泊松过程对用户进行建模，通过运用强化学习的算法，提出了一种宏基站小基站协作传输休眠方案。在用户流量未知的情况下，能有效的满足用户需求，并减小能量损耗。本发明通过考虑开关损耗，有效抑制基站频繁开关，并得到次优解决方案。算法复杂度简单。
在本实施例中还提供了一种小基站开关的控制装置，用于实现上述实施例及优选实施方式，已经进行过说明的不再赘述，下面对该装置中涉及到的模块进行说明。如以下所使用的，术语“模块”可以实现预定功能的软件和/或硬件的组合。尽管以下实施例所描述的装置较佳地以软件来实现，但是硬件，或者软件和硬件的组合的实现也是可能并被构想的。图6为根据本发明实施例的小基站开关的控制装置的结构框图。如图6所示，该装置包括：
统计模块60，设置为对预定时间段内的用户到达速率进行统计；
确定模块62，与统计模块60连接，设置为根据统计结果估测当前的用户到达速率，得到预估速率；
调整模块64，与确定模块62连接，设置为根据上述预估速率调整上述用户所归属的小基站的开关策略。
通过上述各个模块的综合作用，采用根据用户在预定时间段内的用户速率进行统计，并根据统计结果能够估测到当前的用户到达速率进而能够调整小基站的开关策略的技术方案，解决了相关技术中在解决基站休眠的能效时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。
可选地，如图7所示，确定模块62包括：建立单元620，设置为根据上述统计结果建立马尔可夫模型；获取单元622，与建立单元620连接，设置为根据马尔可夫模型获取上述预估速率。
为了更好的理解上述小基站开关的控制方法及装置的工作流程，以下再结合一个优选实施例进行说明：但本发明的保护范围不限于下述的实施例。
宏基站恒定开启功率PM,0为30W，表示宏基站每发送单位数据包消耗的功率PM,t为10W，表示宏基站收发数据包的速率μM为0.5。小基站恒定开启功率PS,0为7.5W，表示小基站每发送单位数据包消耗的功率PS,t为2.5W，小基站收发数据包的速率μS为 2。学习因子η取值按时隙t变化，
Figure PCTCN2015074911-appb-000042
θ＝1.5。折扣因子γ为0.9，开关损耗β设为50J。单个时隙长度ts为30s。
图8为根据本发明优选实施例的小基站开关的流程图，如图8所示：
步骤S802，2个小基站和一个宏基站共同组成异构无线网络。用户以发送数据包的方式接受网络服务，规定无线网络服务协议。
本发明优选实施例中用户以发送数据包的方式进行无线通信，宏基站覆盖范围内的用户可以接收到处于工作状态的小基站和宏基站的无线信号，处于休眠状态的小基站无法为用户提供服务。所有被数据包的到达过程用泊松过程建模，其包到达率为λ。用户的动态性通过包到达率的变化表现，包到达率的变化可以用一个马尔可夫过程表示，该马尔可夫过程的状态集为{λ1,λ2,λ3}，其中的3个状态表示包到达率的3种取值。
步骤S804，系统通过统计一定时间Ts内到达用户数，估计系统的当前状态。
系统需要估计的状态实际是用户的到达速率。系统首先需要知道用户包到达率的范围[0,4]，确定状态集{s0,s1,s2,…s4}，其中s0<s1<s2<…<s4并取0到4的整数。具体操作方法可由上述小基站开关的控制过程优选实施例中的第二步得到。
步骤S806，每个时隙开始时，根据强化学习地方法，由系统收益得出Q值，根据Q值系统动态选择小基站开关策略，最后得出开启基站的数量。
首先，需要确定Q值。强化学习中，学习者需要为每一个状态和行为的组合设定一个Q值，表示为Q(s,a)，其中s表示系统状态，a表示系统行为。系统状态又第二步估测得到，系统的行为是指在该时隙中，小基站的开启数量，所以a的取值范围为
Figure PCTCN2015074911-appb-000043
设定所有的Q值初始均为0。
在第t个时隙开始时，根据系统的状态s选择小基站的开启数量a。选择小基站开启数量a由式六决定。
步骤S808，更新系统状态S，并在每个时隙结束时，根据式七对Q值进行更新。
在实际应用中，在执行步骤S808之后可重复S804和S806的步骤，最终系统会收敛到稳定状态，此时开关基站的策略基本保持不变。
图9为根据本发明优选实施例的马尔可夫泊松过程的模型示意图，纵坐标表示的是数据包到达速率随时间的变化，在1,2,3之间不断转移。图10为根据本发明实施例的动态休眠算法与固定休眠策略的累计收益比较示意图，在固定休眠策略中，开启固定开启半数的小基站。比较看出，本申请提出的动态休眠方案得到的系统收益更大。图11为根据本发明实施例的算法的收敛过程示意图，图中看出取不同的θ值，算法均在50个时隙后收敛。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
需要说明的是，本发明的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的对象在适当情况下可以互换，以便这里描述的本发明的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
综上所述，本发明实施例实现了以下有益效果：解决了相关技术中在解决基站休眠的能效时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。
显然，本领域的技术人员应该明白，上述的本发明的各模块或各步骤可以用通用的计算装置来实现，它们可以集中在单个的计算装置上，或者分布在多个计算装置所组成的网络上，可选地，它们可以用计算装置可执行的程序代码来实现，从而，可以将它们存储在存储装置中由计算装置来执行，并且在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤，或者将它们分别制作成各个集成电路模块，或者将它们中的多个模块或步骤制作成单个集成电路模块来实现。这样，本发明不限制于任何特定的硬件和软件结合。
以上仅为本发明的优选实施例而已，并不用于限制本发明，对于本领域的技术人员来说，本发明可以有各种更改和变化。凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。
工业实用性
基于本发明实施例提供的上述技术方案，采用根据用户在预定时间段内的用户速率进行统计，并根据统计结果能够估测到当前的用户到达速率进而能够调整小基站的开关策略的技术方案，解决了相关技术中在解决基站休眠的能效时，没有考虑用户到达速率变化的而导致的与实际情况有偏差的问题，既能得到系统的次优解决方案，又满足了用户的动态特性，更加符合实际。

Claims (10)

  1. 一种小基站开关的控制方法，包括：
    对预定时间段内的用户到达速率进行统计；
    根据统计结果估测当前的用户到达速率，得到预估速率；
    根据所述预估速率调整所述用户所归属的小基站的开关策略。
  2. 根据权利要求1所述的方法，其中，根据统计结果估测当前的用户到达速率，得到预估速率包括：
    根据所述统计结果建立马尔可夫模型；
    根据马尔可夫模型获取所述预估速率。
  3. 根据权利要求2所述的方法，其中，根据所述统计结果建立马尔可夫模型，包括：
    对获取的所述用户到达速率建立第一状态集为{λ1，λ2，……，λn}，其中所述λ1，λ2，……，λn的取值分别表示所述用户在不同时刻用户到达速率的取值。
  4. 根据权利要求2所述的方法，其中，根据所述预估速率调整所述用户所归属的小基站的开关策略之前，包括：
    根据所述第一状态集确定所述用户到达速率的取值范围[λmin，λmax]，其中，所述λmin为所述第一状态集中的最小值，所述λmax为所述第一状态集的最大值；
    根据预设量化精度对所述取值范围量化为第二状态集{s0，s1，s2……sm}，其中，s0＜s1＜s2＜……＜sm，且，s0＜λmin＜λmax＜sm，所述m的取值由所述量化精度确定。
  5. 根据权利要求4所述的方法，其中，根据统计结果估测当前的用户到达速率，得到预估速率，包括：
    获取当前时刻到之前Ts时刻的所述小基站的当前用户数量
    Figure PCTCN2015074911-appb-100001
    根据以下公式确定所述当前用户到达速率
    Figure PCTCN2015074911-appb-100002
    Figure PCTCN2015074911-appb-100003
    其中，所述sK取值自所述第二状态集，K＝0,1，……m。
  6. 根据权利要求5所述的方法，其中，根据所述预估速率调整所述用户所归属的小基站的开关策略，包括：
    根据预先确定的期望函数Q以及所述用户到达速率
    Figure PCTCN2015074911-appb-100004
    确定a的取值，其中，所述a为当前时隙下开启的小基站的数量，a取值范围为[0,n]，其中，n为当前小基站的数量；
    根据以下公式依次计算出a取值为0,1,2,……n时，所述a对应的取值所被选择的概率
    Figure PCTCN2015074911-appb-100005
    Figure PCTCN2015074911-appb-100006
    并根据所述
    Figure PCTCN2015074911-appb-100007
    按照以下公式计算当前小基站的开启数量at：
    Figure PCTCN2015074911-appb-100008
  7. 根据权利要求6所述的方法，其中，
    在到达当前时隙t结束时间时，根据以下公式对所述期望函数Q进行更新：
    Figure PCTCN2015074911-appb-100009
    其中，
    Figure PCTCN2015074911-appb-100010
    表示所述当前时隙t下的期望函数值；s'为下一时隙t+1所对应的用户到达速率，
    Figure PCTCN2015074911-appb-100011
    表示所述下一时隙t+1下的期望函数值，η表征算法的收敛速度，取值范围为0～1；γ为折扣因子，取值范围为0～1；Rt为所述当前时隙的系统收益。
  8. 根据权利要求1-7任一项所述的方法，其中，所述小基站为多个，并且所述多个小基站与一个宏基站组成异构无线网络。
  9. 一种小基站开关的控制装置，包括：
    统计模块，设置为对预定时间段内的用户到达速率进行统计；
    确定模块，设置为根据统计结果估测当前的用户到达速率，得到预估速率；
    调整模块，设置为根据所述预估速率调整所述用户所归属的小基站的开关策略。
  10. 根据权利要求9所述的装置，其中，所述确定模块包括：
    建立单元，设置为根据所述统计结果建立马尔可夫模型；
    获取单元，设置为根据马尔可夫模型获取所述预估速率。
PCT/CN2015/074911 2014-09-25 2015-03-23 小基站开关的控制方法及装置 WO2016045329A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410498691.2A CN105517120A (zh) 2014-09-25 2014-09-25 小基站开关的控制方法及装置
CN201410498691.2 2014-09-25

Publications (1)

Publication Number Publication Date
WO2016045329A1 true WO2016045329A1 (zh) 2016-03-31

Family

ID=55580222

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/074911 WO2016045329A1 (zh) 2014-09-25 2015-03-23 小基站开关的控制方法及装置

Country Status (2)

Country Link
CN (1) CN105517120A (zh)
WO (1) WO2016045329A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106376063A (zh) * 2016-08-30 2017-02-01 重庆邮电大学 基于宏基站协作处理微基站部分数据业务的微基站节能方法
CN107222875B (zh) * 2017-07-10 2019-10-01 东南大学 基于大数据分析的小基站开关控制方法
CN108337682B (zh) * 2018-01-30 2021-12-07 北京佰才邦技术股份有限公司 一种控制基站开启的方法及装置
CN109819522B (zh) * 2019-03-15 2021-08-24 电子科技大学 一种平衡能耗与用户服务质量的用户带宽资源分配方法
CN115066006A (zh) * 2022-05-28 2022-09-16 哈尔滨工业大学 一种基于强化学习的基站休眠方法、设备和介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102917446A (zh) * 2012-09-29 2013-02-06 北京邮电大学 一种面向绿色节能的动态小区休眠方法
CN103391604A (zh) * 2013-07-17 2013-11-13 浙江大学 小基站休眠的控制方法
CN103957532A (zh) * 2014-04-15 2014-07-30 电子科技大学 一种提高异构蜂窝网络能效的方法
CN103997756A (zh) * 2014-05-09 2014-08-20 宇龙计算机通信科技(深圳)有限公司 用于微小区快速开关的传输方法和装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102111816A (zh) * 2011-03-22 2011-06-29 北京邮电大学 节能的毫微微网络系统及节能方法
CN102665259B (zh) * 2012-05-07 2014-12-24 中国联合网络通信集团有限公司 基站节能模式控制方法、装置和系统
US9450695B2 (en) * 2012-11-02 2016-09-20 Blackberry Limited Wireless communication in heterogeneous networks
CN103957527A (zh) * 2014-04-22 2014-07-30 常州市西屋自动化有限公司 一种动态分配用户网络带宽的无线接入设备及方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102917446A (zh) * 2012-09-29 2013-02-06 北京邮电大学 一种面向绿色节能的动态小区休眠方法
CN103391604A (zh) * 2013-07-17 2013-11-13 浙江大学 小基站休眠的控制方法
CN103957532A (zh) * 2014-04-15 2014-07-30 电子科技大学 一种提高异构蜂窝网络能效的方法
CN103997756A (zh) * 2014-05-09 2014-08-20 宇龙计算机通信科技(深圳)有限公司 用于微小区快速开关的传输方法和装置

Also Published As

Publication number Publication date
CN105517120A (zh) 2016-04-20

Similar Documents

Publication Publication Date Title
CN110493826B (zh) 一种基于深度强化学习的异构云无线接入网资源分配方法
JP4267963B2 (ja) 無線コンピューティングデバイス内のネットワークインターフェースモジュールの電力消費を管理する方法およびシステム
WO2016045329A1 (zh) 小基站开关的控制方法及装置
Mohanty et al. Energy efficient structure-free data aggregation and delivery in WSN
JP2009518967A (ja) アドホックモードの８０２．１１ネットワーク用の最適ａｔｉｍサイズセットアップの方法及び装置
US8804589B2 (en) Adaptive awake window
CN103906207A (zh) 基于自适应按需唤醒技术的无线传感器网络数据传输方法
TWI484850B (zh) 用於長期演進技術中省電資料排程系統及其方法
Papadopoulos et al. T-AAD: Lightweight traffic auto-adaptations for low-power MAC protocols
Tewari et al. Performance study of SEP and DEC hierarchical clustering algorithm for heterogeneous WSN
Niyato et al. Competitive wireless energy transfer bidding: A game theoretic approach
Rostami et al. Wake-up scheduling for energy-efficient mobile devices
CN110121206A (zh) 一种基于有限队列长度排队模型的多跳无线唤醒方法
Sheth et al. EAPS: Edge-assisted predictive sleep scheduling for 802.11 IoT stations
CN102202344B (zh) 无线Mesh网中基于博弈论的跨层拥塞控制方法
Pal et al. Network adaptive round-time clustering algorithm for wireless sensor networks
Deng et al. Rate-constrained uniform data collection in wireless sensor networks
Bhandari et al. Study on improving the network life time maximazation for wireless sensor network using cross layer approach
Liu et al. Pushback: A hidden Markov model based scheme for energy efficient data transmission in sensor networks
Hwang et al. Offloading optimization with delay distribution in the 3-tier federated cloud, edge, and fog systems
Raval et al. Improving energy estimation based clustering with energy threshold for wireless sensor networks
Sundararaju et al. Novel C-DRX mechanism for multi SIM multi standby UEs in 5G and B5G networks
Lee et al. Phonepool: On energy-efficient mobile network collaboration with provider aggregation
Yang et al. Modeling power saving for GAN and UMTS interworking
Zhiqun et al. Adaptive WiFi-offloading algorithm based on attractor selection in heterogeneous wireless networks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15843377

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15843377

Country of ref document: EP

Kind code of ref document: A1