CN103249006B - The pre-method for pushing of a kind of network data based on multicast - Google Patents

Abstract

Translated fromChinese

本发明基于组播的网络数据预推送传输方法，属于无线网络数据传输技术领域，在无线网络数据传输中，当有一个用户请求网络数据时，通过对该业务请求的分析及当前的网络的信道状态，由基站决定是否对其覆盖范围内及邻小区内有相似网络使用习惯的用户使用多播的方式预推送该业务，通过对之前积累的用户对网络使用习惯的先验概率函数，当前用户的信道状态，计算出参与多播预推送的备选用户方案，多小区协同传输方案，调制编码方案，计算出使用多播方式可以获得的期望收益，并与传统单播方式比较，若获得收益比大于设置的阈值，则使用本方法进行推送，否则使用传统方式推送，本方法能够有效调高无线频谱利用率，能量利用率，从整体上优化网络传输性能。

The multicast-based network data pre-push transmission method of the present invention belongs to the technical field of wireless network data transmission. In wireless network data transmission, when a user requests network data, the service request is analyzed and the current network channel State, the base station decides whether to pre-push the service by multicast to users with similar network usage habits in its coverage area and neighboring cells. Through the prior probability function of the previously accumulated users' network usage habits, current users channel status, calculate the alternative user schemes participating in multicast pre-push, multi-cell coordinated transmission schemes, modulation and coding schemes, calculate the expected benefits that can be obtained by using the multicast method, and compare it with the traditional unicast method. If the ratio is greater than the set threshold, use this method to push, otherwise use the traditional method to push. This method can effectively increase the wireless spectrum utilization rate and energy utilization rate, and optimize the network transmission performance as a whole.

Description

Translated fromChinese

一种基于多播的网络数据预推送方法A multicast-based network data pre-push method

技术领域technical field

本发明涉及一种基于多播的网络数据预推送方法，属于无线网络数据传输技术领域。The invention relates to a multicast-based network data pre-push method, belonging to the technical field of wireless network data transmission.

背景技术Background technique

随着智能手机，手持移动终端数目的增加，未来移动通信网络数据流量将会大幅度增长，而其中绝大部分的数据流量将来自于web浏览及网络视频类业务。为了在有限的物理资源下，满足高速增长的业务需求，未来的无线移动通信系统必须具有更高的传输速率，更高的频谱利用率，更低的能耗。With the increase in the number of smart phones and handheld mobile terminals, the future mobile communication network data traffic will increase significantly, and most of the data traffic will come from web browsing and network video services. In order to meet the rapidly growing service demands with limited physical resources, future wireless mobile communication systems must have higher transmission rates, higher spectrum utilization, and lower energy consumption.

在传统的无线数据传输中，有网络业务需求的终端获得服务的一般流程如下：In traditional wireless data transmission, the general process for terminals with network service requirements to obtain services is as follows:

1．终端读取网络系统信息及随机接入资源1. The terminal reads network system information and randomly accesses resources

2．终端随机接入，并在随机接入竞争失败时重新发起2. The terminal randomly accesses, and re-initiates when the random access competition fails

3．同网络基站建立连接3. Establish a connection with a network base station

4．网络根据当前负载及邻小区干扰状况进行无线资源分配4. The network allocates radio resources according to the current load and the interference status of adjacent cells

5．终端同网络进行上下行数据传输5. The terminal performs uplink and downlink data transmission with the network

6．终端同网络释放连接，完成业务数据的传输6. The terminal releases the connection with the network to complete the transmission of business data

据统计分析显示，不同用户对于网络业务的需求具有一定的相关性，某些热点的网络资源可能被不同用户所需求。这种相关性同业务的内容，用户的上网的行为特征，资源创建的时间，用户地点分布等属性均有一定的联系。在传统的网络业务传输流程中，并没有考虑网络服务的相关性，不同的用户请求需要分别按照上述步骤各自建立网络连接，独立的进行业务的传输，因此具有相同的内容的业务，可能被多次重复传输，造成物理资源的浪费。According to statistical analysis, different users have certain correlations in their demands for network services, and certain hotspot network resources may be required by different users. This correlation is related to attributes such as business content, user's online behavior characteristics, resource creation time, and user location distribution. In the traditional network service transmission process, the relevance of network services is not considered. Different user requests need to establish network connections according to the above steps, and carry out service transmission independently. Therefore, services with the same content may be used by multiple users. Repeated transmission, resulting in a waste of physical resources.

发明内容Contents of the invention

为了克服上述现有技术的缺点，本发明的目的是提出一种基于多播的网络数据预推送方法，能够利用网络业务的相关性特性，根据之前积累的用户对网络使用习惯的先验概率函数和当前用户的信道状态，充分利用物理资源，提高传输效率，降低功耗。In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to propose a network data pre-push method based on multicast, which can utilize the correlation characteristics of network services, according to the prior probability function of users' habits of network usage accumulated before and the channel status of the current user, make full use of physical resources, improve transmission efficiency, and reduce power consumption.

为实现上述目的，本发明采取如下技术方案：To achieve the above object, the present invention takes the following technical solutions:

一种基于多播的网络数据预推送方法，在网络侧积累其覆盖内终端的用户行为数据，当某一用户终端请求网络数据时，网络侧根据该积累数据以及该次请求的业务内容，如果判断出覆盖内其余终端对于此项业务内容的需求概率大于预先设定的值，则使用多播的方式预推送数据，在向当前用户发送数据的同时，完成多个终端未来的业务需求。A multicast-based network data pre-push method, which accumulates user behavior data of terminals within its coverage on the network side. When a user terminal requests network data, the network side, based on the accumulated data and the service content of the request, if If it is judged that the demand probability of the remaining terminals in the coverage for this service content is greater than the preset value, the multicast method is used to pre-push data, and while sending data to the current user, the future service needs of multiple terminals are fulfilled.

当一个用户终端请求网络数据时，先根据请求的内容搜索预推送数据缓存，寻找是否已经在之前某时刻已经通过预推送的方式接收到了此次请求的数据内容，如果没有收到，或者该终端需要更新并接收新的数据，该用户终端生成上行数据，将自己的业务请求的信息放入上行消息中，并附着一个标志是否需要接收新数据的Flag标记位；用户终端同网络侧建立链接后，发送请求数据，网络侧检测请求内容，用户终端ID，利用请求内容更新业务热度数据库，根据用户行为更新用户行为数据库，检测Flag标记位，若此用户终端不需要请求新的数据，则此次会话结束，断开网络连接；若此用户终端需要进行新的数据发送时，网络侧再进行覆盖内其余终端对于此项业务内容的需求概率计算。When a user terminal requests network data, it first searches the pre-push data cache according to the requested content to find out whether the requested data content has been received by pre-push at a certain point before, if not received, or the terminal Need to update and receive new data, the user terminal generates uplink data, puts the information of its own service request into the uplink message, and attaches a Flag flag indicating whether it needs to receive new data; after the user terminal establishes a link with the network side , send the request data, the network side detects the request content, the user terminal ID, uses the request content to update the business popularity database, updates the user behavior database according to the user behavior, and detects the Flag bit. If the user terminal does not need to request new data, then this time When the session ends, the network connection is disconnected; if the user terminal needs to send new data, the network side calculates the probability of demand for this service content by other terminals within the coverage.

其中网络侧对覆盖内其余终端对于此项业务内容的需求概率P_UE通过如下公式计算：Among them, the demand probability P_UE of the other terminals in the coverage on the network side for this service content is calculated by the following formula:

P_UE=P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)P_UE =P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)

其中，UE_ID表示用户，UE_POS表示用户位置，TASK_CLASS表示业务类型，TASK_TIME表示业务发起时间，TASK_LIFE表示业务生命，省略号表示附加属性。Among them, UE_ID indicates user, UE_POS indicates user location, TASK_CLASS indicates service type, TASK_TIME indicates service initiation time, TASK_LIFE indicates service life, and ellipsis indicates additional attribute.

网络侧根据业务内容，以及请求业务终端所在小区和邻小区覆盖下其它终端用户习惯，信道状态，计算出小区调度策略及终端调度策略，根据信道状态计算出调制编码策略，推算出使用本方法优化后的发送期望速率Rmax，不使用本方法的单播发送速率Rsingle，以及使用本方法后的收益Rmax/Rsingle：The network side calculates the cell scheduling strategy and terminal scheduling strategy according to the service content, the habits of other terminal users under the coverage of the cell where the requesting service terminal is located and the adjacent cell, and the channel state, and calculates the modulation and coding strategy according to the channel state, and calculates the optimization using this method. Expected sending rate Rmax, unicast sending rate Rsingle without using this method, and revenue Rmax/Rsingle after using this method:

若网络使用多播的方式所获得的预测收益小于预先设定的期望收益时，使用传统单播的方式发送数据；If the predicted income obtained by the network using the multicast method is less than the preset expected income, the traditional unicast method is used to send data;

若网络侧预测使用多播方式预推送数据获得收益大于预先设定的最小期望收益时，则使用多播预推送的方式发送数据。If the network side predicts that the profit obtained by using the multicast pre-push method is greater than the preset minimum expected profit, then the multicast pre-push method is used to send the data.

根据所述优化策略，选择调度参与多播推送的备选小区和在这些小区中的备选终端，由核心网向备选小区发送多播推送通知，通知各备选小区覆盖内参与多播推送的备选终端ID，物理资源分配情况，调制编码方式，由各个备选小区各自调度小区覆盖范围内的备选终端。According to the optimization strategy, select and schedule candidate cells participating in multicast push and candidate terminals in these cells, and the core network sends a multicast push notification to the candidate cells, notifying each candidate cell to participate in multicast push within coverage Alternative terminal IDs, physical resource allocation, modulation and coding schemes, and each candidate cell schedules candidate terminals within the coverage of the cell.

由于网络侧覆盖内终端可能处于不同状态，因此针对它们的状态分为以下情况进行调度：Since the terminals in the coverage of the network side may be in different states, their states are divided into the following situations for scheduling:

a)对于处在连接态(connected)，激活状态/连续接收状态(active)的终端，由于其持续监听控制信道消息，因此可以直接在控制信道上通知下行数据到来，下行数据物理资源位置，调制编码策略等信息，并保持connected_active状态；a) For a terminal in the connected state (connected), activated state/continuous reception state (active), since it continuously monitors the control channel message, it can directly notify the arrival of downlink data, the location of the physical resource of the downlink data, and the modulation on the control channel Coding strategy and other information, and keep connected_active state;

b)对于处在连接态，非连续接收状态的终端，由于终端只有在其非连续监听周期到来时才读取控制信道，因此网络侧在各个终端的非连续监听周期到来时，在终端的控制信道发送调度信息，激活终端，使其进入connected_active状态；b) For a terminal in a connected state and a discontinuous reception state, since the terminal reads the control channel only when its discontinuous monitoring period arrives, the network side will control the terminal when the discontinuous monitoring period of each terminal arrives. The channel sends scheduling information, activates the terminal, and makes it enter the connected_active state;

c)对于处在空闲状态的终端，需要网络侧首先对终端，发起寻呼，触发终端建立连接，在终端初始连接建立完成后，终端直接进入connected_active状态。c) For a terminal in an idle state, the network side first needs to initiate paging to the terminal to trigger the terminal to establish a connection. After the initial connection establishment of the terminal is completed, the terminal directly enters the connected_active state.

当全部参加多播预推送数据的终端进入connected_active状态后，各个终端连续监听自己的控制信道，网络侧在控制信道中发送多播数据的物理资源位置，调制编码方式，各个终端接收控制信令后，开始在网络侧指定的物理资源位置接收数据。When all the terminals participating in the multicast pre-push data enter the connected_active state, each terminal continuously monitors its own control channel, and the network side sends the physical resource location, modulation and coding method of the multicast data in the control channel, and each terminal receives the control signaling. , start receiving data at the physical resource location specified by the network side.

根据终端QoS需求，网络侧决定终端进行HARQ和数据重传时机，数据发送完毕后，此次网络数据多播预推送完成，中断连接。According to the QoS requirements of the terminal, the network side determines the timing of HARQ and data retransmission for the terminal. After the data is sent, the network data multicast pre-push is completed and the connection is interrupted.

本发明基于多播的网络数据预推送方法，利用网络数据业务的相关性，每当一个终端请求网络数据时，网络侧根据以往积累的用户行为数据，该次请求的业务内容，决策出当前业务是否适用于多播推送。结合当前网络负载状况，小区覆盖内其余终端信道状态，网络侧调度合适的小区及终端，使用多播的方式预推送数据。当一个终端成功使用了预推送数据，它会为网络侧提供一个反馈，用以优化用户行为分析。本发明中的多播推送方式，充分的利用了在网络中数据传输的冗余性，将可能被多个用户使用到的数据，在其第一次被请求时通过多播的方式发送给的对该业务有相似需求的其它终端，通过一次业务传输资源的消耗，完成多个终端未来的业务需求。本发明克服了传统网络数据业务中，部分热点业务被重复传输的缺点，在不影响发起业务请求的终端服务的条件下，提高了系统资源利用率。The multicast-based network data pre-push method of the present invention utilizes the correlation of network data services, and whenever a terminal requests network data, the network side decides the current service according to the user behavior data accumulated in the past and the service content of the request. Whether to apply to multicast push. Combined with the current network load status and the channel status of other terminals within the coverage of the cell, the network side schedules suitable cells and terminals, and uses multicast to pre-push data. When a terminal successfully uses the pre-push data, it will provide a feedback to the network side to optimize user behavior analysis. The multicast push method in the present invention fully utilizes the redundancy of data transmission in the network, and the data that may be used by multiple users is sent to the user through multicast when it is requested for the first time. Other terminals that have similar requirements for this service can fulfill the future service requirements of multiple terminals through the consumption of one service transmission resource. The invention overcomes the shortcoming of repeated transmission of some hot spots in traditional network data services, and improves the utilization rate of system resources without affecting the service of the terminal that initiates the service request.

附图说明Description of drawings

图1为本发明中用户行为数据库更新及业务需求概率计算流程框图。Fig. 1 is a flow chart of user behavior database update and service demand probability calculation in the present invention.

图2为本发明提出的基于多播的网络数据预推送方法流程框图。FIG. 2 is a block diagram of a multicast-based network data pre-push method proposed by the present invention.

具体实施方式Detailed ways

下面结合附图和实施例对本发明进行进一步描述。The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

本发明提出的基于多播的网络数据预推送方法，其流程框图如图1所示，包括以下步骤：The network data pre-push method based on multicast that the present invention proposes, its flow chart as shown in Figure 1, comprises the following steps:

(1)当一个支持网络数据预推送的终端有网络数据请求时，根据请求的内容搜索预推送数据缓存，寻找是否已经在之前某时刻已经通过预推送的方式接收到了此次请求的数据内容。(1) When a terminal that supports network data pre-push has a network data request, it searches the pre-push data cache according to the content of the request to find out whether the requested data content has been received through pre-push at a certain point before.

(2)请求业务的终端生成上行数据，将自己的业务请求的信息放入上行消息中。根据此次请求的业务数据是否已经通过预推送的方式成功接收，或者该终端需要更新数据，由终端决策是否需要接收新的数据，将决策结果在上行消息中附着一个Flag标记位，标志终端是否需要接收新数据。(2) The terminal requesting the service generates uplink data, and puts the information of its own service request into the uplink message. According to whether the business data requested this time has been successfully received through the pre-push method, or the terminal needs to update the data, the terminal decides whether to receive new data, and attaches a Flag bit to the uplink message of the decision result to indicate whether the terminal is Need to receive new data.

(3)终端同网络侧建立链接，网络侧接收用户业务请求。由于不同用户对于业务的需求具有相关性，这种相关性同用户（UE_ID），用户位置（UE_POS），业务类型（TASK_CLASS），业务发起时间（TASK_TIME），该业务生命（TASK_LIFE）以及其它可能属性相关，因此用户对于业务的需求概率是与这些属性相关的条件概率：(3) The terminal establishes a link with the network side, and the network side receives user service requests. Because the needs of different users for services are related, this correlation is the same as user (UE_ID), user location (UE_POS), service type (TASK_CLASS), service initiation time (TASK_TIME), service life (TASK_LIFE) and other possible attributes Correlation, so the user's demand probability for the service is the conditional probability related to these attributes:

P_UE=P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)P_UE =P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)

对于接收到的业务请求，网络侧可以通过该公式计算出其它用户对于该业务的需求，从而判断其它用户对于此项业务需求的可能性。步骤3的具体操作流程如下：For the service request received, the network side can calculate the demand of other users for the service through the formula, so as to judge the possibility of other users' demand for the service. The specific operation process of step 3 is as follows:

a)网络侧接收用户业务请求。a) The network side receives the user service request.

b)分析业务中的时间，位置信息等属性，为计算其它用户对此项业务的需求情况，需要将本次业务信息中的属性进行离散化。首次业务发起的时间将作为该业务时间属性（TASK_TIME）的起始点，后面相同的业务均需要以此作为自己的时间属性。对于时间信息，可以根据实际情况精确到不同的时间等级，如将不同的日期作为不同的时间属性或者将每天分为几个时间段作为业务发起的时间属性。对于业务的生命时长(TASK_LIFE)，即计算当前时间（CUR_TASK_TIME），同起始时间的时间间隔：TASK_LIFE=CUR_TASK_TIME-TASK_TIME，并根据需要对其进行离散化，例如：3天以内，3天-6天…。对于业务的区域属性（UE_POS），可以按照用户发起业务时的物理坐标作为划分依据，或者以所在小区作为划分依据，为了降低计算量，可以合并一个或若干个地区将其视作相同地区。对于业务类型（TASK_CLASS），根据业务实际情况进行分类，如体育类，娱乐类等分类方式。通过分析此次业务的请求信息，获得本次业务的属性集：b) Analyze attributes such as time and location information in the service. In order to calculate the demand of other users for this service, it is necessary to discretize the attributes in this service information. The time when the first service is initiated will be used as the starting point of the service time attribute (TASK_TIME), and the subsequent same services need to use this as their own time attribute. For time information, it can be accurate to different time levels according to the actual situation, such as using different dates as different time attributes or dividing each day into several time periods as time attributes for service initiation. For the life span of the business (TASK_LIFE), that is to calculate the current time (CUR_TASK_TIME), the time interval from the start time: TASK_LIFE=CUR_TASK_TIME-TASK_TIME, and discretize it as needed, for example: within 3 days, 3 days-6 sky…. For the area attribute (UE_POS) of the service, the physical coordinates when the user initiates the service can be used as the basis for division, or the cell where the user is located can be used as the basis for division. In order to reduce the amount of calculation, one or several areas can be combined and regarded as the same area. For the business type (TASK_CLASS), classify according to the actual situation of the business, such as sports, entertainment and other classification methods. By analyzing the request information of this business, the attribute set of this business is obtained:

S={UE_ID，UE_POS，TASK_CLASS，TASK_TIME，TASK_LIFE…}。S={UE_ID, UE_POS, TASK_CLASS, TASK_TIME, TASK_LIFE...}.

c)网络侧检测Flag标记位，若该次业务需要发起新的数据传输则进入步骤d），否则进入步骤e）c) The network side detects the flag bit, if the service needs to initiate a new data transmission, go to step d), otherwise go to step e)

d)计算该小区及邻小区对于该次业务的需求概率：d) Calculate the demand probability of the cell and neighboring cells for the service:

P_UE=P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)P_UE =P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)

具体计算方法如下：The specific calculation method is as follows:

${\mathrm{<span><span>P</span>P</span>}}_{\mathrm{<span><span>UE</span>UE</span>}}<span><span>=</span>=</span>\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>|</span>|</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>ID</span>ID</span>}<span><span>,</span>,</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>POS</span>POS</span>}<span><span>,</span>,</span>\mathrm{<span><span>TASK</span>TASK</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>CLASS</span>CLASS</span>}<span><span>,</span>,</span>\mathrm{<span><span>TASK</span>TASK</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>TIME</span>time</span>}<span><span>,</span>,</span>\mathrm{<span><span>TASK</span>TASK</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>LIFE</span>LIFE</span>}<span><span>\&CenterDot;</span>\&CenterDot;</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span><span><span>\&CenterDot;</span>\&CenterDot;</span><span><span>)</span>)</span>$

$<span><span>=</span>=</span>\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>|</span>|</span>\mathrm{<span><span>S</span>S</span>}<span><span>)</span>)</span>$

$<span><span>=</span>=</span>\frac{\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>,</span>,</span>\mathrm{<span><span>S</span>S</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>S</span>S</span>}<span><span>)</span>)</span>}$

$<span><span>=</span>=</span>\frac{\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>S</span>S</span>}<span><span>|</span>|</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span>\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>S</span>S</span>}<span><span>)</span>)</span>}$

假设总的业务请求记录条数为N，其中以往被UE需求的业务总数为COUNT(UE)，则P(UE)计算公式为：Assuming that the total number of service request records is N, and the total number of services requested by UE in the past is COUNT(UE), then the calculation formula of P(UE) is:

$\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\frac{\mathrm{<span><span>COUNT</span>COUNT</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>N</span>N</span>}}$

P(S|UE)=P(UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…|UE)假设各个属性之间无相关性，则上式可以表示为：P(S|UE)=P(UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE...|UE) Assuming that there is no correlation between each attribute, the above formula can be expressed as:

P(S|UE)=P(UE_ID|UE)P(UE_POS|UE)P(TASK_CLASS|UE)...P(S|UE)=P(UE_ID|UE)P(UE_POS|UE)P(TASK_CLASS|UE)...

以P(UE_POS|UE)的计算为例：Take the calculation of P(UE_POS|UE) as an example:

COUNT(UE_POS，UE)为该UE在UE_POS处发起业务请求的总数COUNT(UE_POS, UE) is the total number of service requests initiated by the UE at UE_POS

$\mathrm{<span><span>P</span>P</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>POS</span>POS</span>}<span><span>|</span>|</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\frac{\mathrm{<span><span>COUNT</span>COUNT</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>\_</span>\_</span>\mathrm{<span><span>POS</span>POS</span>}<span><span>,</span>,</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>COUNT</span>COUNT</span>}<span><span>(</span>(</span>\mathrm{<span><span>UE</span>UE</span>}<span><span>)</span>)</span>}$

对于P（S）是一个只与属性相关的变量，可以假设其为恒定值，不参与计算。For P(S) is a variable only related to attributes, it can be assumed to be a constant value and not involved in the calculation.

e)将此次业务获得的新信息更新原有旧数据库，更新方法如下：e) Update the original old database with the new information obtained in this business, the update method is as follows:

以此次业务（TASK_ID）为索引向数据库中添加一项新元组，该元组的属性为此次请求的UE_ID，重新计算UE之间的相关性，即重新搜索之前对该TASK_ID发起需求的UE，对称更新两个相关UE：COUNT（UE_ID，UE）=COUTN(UE_ID，UE)+1，重新计算同时刷新该UE的其它相关属性：Add a new tuple to the database with this service (TASK_ID) as the index. The attribute of this tuple is the UE_ID of this request, and recalculate the correlation between UEs, that is, to re-search the previously requested TASK_ID. UE, symmetrically update two related UEs: COUNT (UE_ID, UE) = COUTN (UE_ID, UE) + 1, recalculate and refresh other related attributes of the UE at the same time:

对于计算P(UE)，需要更新：N=N+1，COUNT(UE_ID)=COUNT(UE_ID+1)；For calculating P(UE), it needs to be updated: N=N+1, COUNT(UE_ID)=COUNT(UE_ID+1);

对于其它条件概率，需要更新：COUNT(UE_POS，UE_ID)=COUNT(UE_POS，UE_ID)+1…For other conditional probabilities, it needs to be updated: COUNT(UE_POS, UE_ID)=COUNT(UE_POS, UE_ID)+1…

f)检测Flag标记位，若该次业务仅仅是一次对预推送内容使用的上报消息，则此次业务终止，断开连接。否则继续进行步骤4。f) Detect the flag bit, if the service is only a report message for the use of the pre-push content, the service is terminated and the connection is disconnected. Otherwise proceed to step 4.

(4)当检测到终端需要进行新的数据发送时，网络侧根据业务内容，以及请求业务终端所在小区和邻小区覆盖下其它终端用户习惯，信道状态，计算出小区调度策略及终端调度策略，根据信道状态计算出调制编码策略，优化出期望收益最大的发送方式，并与多播发送的最小期望收益作比较。(4) When it is detected that the terminal needs to send new data, the network side calculates the cell scheduling strategy and the terminal scheduling strategy based on the service content, the habits of other terminal users under the coverage of the cell where the terminal is located and the neighbor cell, and the channel status. Calculate the modulation and coding strategy according to the channel state, optimize the transmission mode with the maximum expected revenue, and compare it with the minimum expected revenue of multicast transmission.

通过步骤3已经获得了不同终端对于该业务的需求情况，假设同时参加多播的UE最大数目为Nmax，多播范围内全部UE数目为Ncur，则该问题为一个组合优化问题：即从Ncur个UE中挑选至多Nmax使得发送数据总吞吐量期望最大化。The requirements of different terminals for this service have been obtained through step 3. Assuming that the maximum number of UEs participating in multicast at the same time is Nmax, and the number of all UEs in the multicast range is Ncur, this problem is a combinatorial optimization problem: that is, from Ncur At most Nmax is chosen in the UE such that the total throughput of transmitted data is expected to be maximized.

该问题可以表示为：The problem can be expressed as:

a)Rmax=max{min(R_UE0,R_UE1,...)*[1+Σf(P_UEi)]}a)Rmax=max{min(R_UE0 ,R_UE1 ,...)*[1+_Σf (P UEi )]}

b)s.t：UE∈Ncur，UE总数<Nmaxb) s.t: UE ∈ Ncur, the total number of UEs < Nmax

c)Power>max(POWER(UEi))c)Power>max(POWER(UEi))

其中f(P_UEi)为P_UEi的单调增函数，取值范围[0，1]，作为优化目标函数的加权系数。Power为基站最大发射功率，max(POWER(UEi))表示，为了达到Rmax，对发送功率要求最大的UEi所需要的发射功率，要保证每个基站发射总功率小于它的最大发射功率。Where f(P_UEi ) is a monotonously increasing function of P_UEi , with a value range of [0, 1], which is used as the weighting coefficient of the optimization objective function. Power is the maximum transmission power of the base station, and max(POWER(UEi)) indicates that in order to achieve Rmax, the transmission power required by UEi, which requires the largest transmission power, must ensure that the total transmission power of each base station is less than its maximum transmission power.

由香农信道容量公式可知，对于单个终端所能支持的理论最大速率为：According to the Shannon channel capacity formula, the theoretical maximum rate that can be supported by a single terminal is:

Rsingle=w*log(1+SNR)[b/Hz]Rsingle=w*log(1+SNR)[b/Hz]

对于R在单个小区时信噪比（SNR）计算较为简单：For R in a single cell, the calculation of the signal-to-noise ratio (SNR) is relatively simple:

假设小区发送功率为P，带宽内噪声功率为P_N，单小区内SNR为：Suppose the transmission power of the cell is P, the noise power in the bandwidth is P_N , and the SNR in a single cell is:

$\mathrm{<span><span>SNR</span>SNR</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>P</span>P</span>}}{{\mathrm{<span><span>P</span>P</span>}}_{\mathrm{<span><span>N</span>N</span>}}}$

对于使用多个小区协同向终端发送数据的情况，价格第i个小区发送功率为P_i，当他到达终端的路损位PL_i，则其接收端信噪比为：For the case of using multiple cells to send data to the terminal cooperatively, the transmission power of the i-th cell is P_i , and when it reaches the path loss point PL_i of the terminal, the signal-to-noise ratio at the receiving end is:

$\mathrm{<span><span>SNR</span>SNR</span>}<span><span>=</span>=</span>\frac{\underset{\mathrm{<span><span>i</span>i</span>}<span><span>\&Element;</span>\&Element;</span>\mathrm{<span><span>\&Gamma;</span>\&Gamma;</span>}}{\mathrm{<span><span>\&Sigma;</span>\&Sigma;</span>}}{\mathrm{<span><span>P</span>P</span>}}_{\mathrm{<span><span>i</span>i</span>}}<span><span>*</span>*</span>{\mathrm{<span><span>PL</span>PL</span>}}_{\mathrm{<span><span>i</span>i</span>}}}{{\mathrm{<span><span>P</span>P</span>}}_{\mathrm{<span><span>N</span>N</span>}}}$

通过将SNR计算公式带入上面的优化问题，可以求解出上述问题最优解的Rmax，通过计算进行判断，若大于设定阈值，则使用多播方式进行预推送，否则使用传统方式单播发送。By bringing the SNR calculation formula into the above optimization problem, the Rmax of the optimal solution to the above problem can be solved. By calculating Make a judgment, if it is greater than the set threshold, use the multicast method for pre-push, otherwise use the traditional method for unicast transmission.

对于实际系统而言，网络侧最终决策的发送速率通常根据终端的SNR查表获得，他对应于终端的调制方式，以及编码速率。在这里统一将最终的发送速率设为R_max。For the actual system, the transmission rate finally decided by the network side is usually obtained according to the SNR look-up table of the terminal, which corresponds to the modulation mode and coding rate of the terminal. Here, the final sending rate is uniformly set as R_max .

(5)若网络使用多播的方式所获得的预测收益小于预先设定的期望收益时，使用传统单播的方式发送数据。若网络侧预测使用多播方式预推送数据获得收益大于预先设定的最小期望收益时，则使用多播预推送的方式发送数据。(5) If the predicted income obtained by the network using the multicast method is less than the preset expected income, the traditional unicast method is used to send data. If the network side predicts that the profit obtained by using the multicast pre-push method is greater than the preset minimum expected profit, then the multicast pre-push method is used to send the data.

(6)网络侧按照步骤（4）中的优化策略，选择调度参与多播推送的备选小区和在这些小区中的备选终端，由核心网向备选小区发送多播推送通知，通知各备选小区覆盖内参与多播推送的备选终端ID，物理资源分配情况，调制编码方式，由各个备选小区各自调度小区覆盖范围内的备选终端。(6) According to the optimization strategy in step (4), the network side selects and schedules the candidate cells participating in the multicast push and the candidate terminals in these cells, and the core network sends a multicast push notification to the candidate cells to notify each IDs of candidate terminals participating in multicast push within the coverage of the candidate cell, physical resource allocation, modulation and coding methods, and each candidate cell schedules candidate terminals within the coverage area of the cell.

(7)由于小区内终端可能处于不同状态，因此针对它们的状态分为以下情况进行调度：(7) Since the terminals in the cell may be in different states, their states are divided into the following situations for scheduling:

a)对于处在连接态(connected)，激活状态/连续接收状态(active)的终端，由于其持续监听控制信道消息，因此可以直接在控制信道上通知下行数据到来，下行数据物理资源位置，调制编码策略等信息，并保持connected_active状态。a) For a terminal in the connected state (connected), activated state/continuous reception state (active), since it continuously monitors the control channel message, it can directly notify the arrival of downlink data, the location of the physical resource of the downlink data, and the modulation on the control channel Encode policy and other information, and keep the connected_active state.

b)对于处在连接态，非连续接收状态的终端，由于终端只有在其非连续监听周期到来时才读取控制信道，因此网络侧在各个终端的非连续监听周期到来时，在终端的控制信道发送调度信息，激活终端，使其进入connected_active状态。由于终端处在连接态，因此在此情况下，省去了RRC建链等时间，终端可以较快的转入到连续接收状态，终端监听时刻计算公式如下：b) For a terminal in a connected state and a discontinuous reception state, since the terminal reads the control channel only when its discontinuous monitoring period arrives, the network side will control the terminal when the discontinuous monitoring period of each terminal arrives. The channel sends scheduling information, activates the terminal, and makes it enter the connected_active state. Since the terminal is in the connected state, in this case, the time such as RRC link establishment is saved, and the terminal can quickly transfer to the continuous receiving state. The calculation formula of the terminal monitoring time is as follows:

短周期：Short cycle:

[(SFN*10)+subframe number]%(shortDRX_Cycle)=(drxStartOffset)%(shortDRX_Cycle)[(SFN*10)+subframe number]%(shortDRX_Cycle)=(drxStartOffset)%(shortDRX_Cycle)

长周期：Long cycle:

[(SFN*10)+subframe number]%(longDRX_Cycle)=(drxStartOffset)%(longDRX_Cycle)[(SFN*10)+subframe number]%(longDRX_Cycle)=(drxStartOffset)%(longDRX_Cycle)

以上参数均由RRC层完成配置The above parameters are configured by the RRC layer

c)对于处在空闲状态的终端，需要网络侧首先对终端，发起寻呼，触发终端建立连接，在终端初始连接建立完成后，终端直接进入connected_active状态。对于处在空闲态的终端，此时已经没有了RRC连接和分配给用户的专有资源，终端需要监听寻呼信道及广播控制信道，此时监听的时间周期通过NAS消息控制，以LTE系统为例，终端监听寻呼信道时刻计算公式如下：c) For a terminal in an idle state, the network side first needs to initiate paging to the terminal to trigger the terminal to establish a connection. After the initial connection establishment of the terminal is completed, the terminal directly enters the connected_active state. For a terminal in an idle state, there is no RRC connection and dedicated resources allocated to the user at this time. The terminal needs to monitor the paging channel and the broadcast control channel. At this time, the monitoring time period is controlled by NAS messages. The LTE system is used as For example, the calculation formula for the time when the terminal monitors the paging channel is as follows:

监听帧号计算Monitor frame number calculation

SFN%T=(T/N)*(UE_ID%N)SFN%T=(T/N)*(UE_ID%N)

监听子帧计算Monitor subframe calculation

T=min(Tc（小区指定默认寻呼周期）,Tue（终端默认寻呼周期）)T=min(Tc (the default paging cycle specified by the cell), Tue (the default paging cycle of the terminal))

N=min(T,每帧中监听的子帧数′T)N=min(T, the number of subframes monitored in each frame'T)

Ns=max(1,每帧中监听的子帧数)Ns=max(1, the number of subframes monitored in each frame)

其中T的时长为32-256个子帧，整个寻呼可能引入平均时延小于500ms。The duration of T is 32-256 subframes, and the average delay may be introduced by the entire paging to be less than 500ms.

在终端接收到寻呼后为了与网络建立RRC连接需要发起随机接入，对于LTE系统而言每个时隙最多可以配置64个随机接入资源，，具体每秒允许随机接入次数与系统配置帧格式有关，其中FDD帧格式：6400-64000次，TDD帧格式：3200-38400次。通过仿真可知，总共产生延时<1s。After receiving the paging, the terminal needs to initiate random access in order to establish an RRC connection with the network. For the LTE system, a maximum of 64 random access resources can be configured in each time slot. The number of random accesses allowed per second depends on the system configuration. The frame format is related, among which FDD frame format: 6400-64000 times, TDD frame format: 3200-38400 times. It can be seen through simulation that the total delay is <1s.

(8)当全部参加多播预推送数据的终端进入connected_active状态后，各个终端连续监听自己的控制信道，网络侧在控制信道中发送多播数据的物理资源位置，调制编码方式，各个终端接收控制信令后，开始在网络侧指定的物理资源位置接收数据。(8) When all the terminals participating in the multicast pre-push data enter the connected_active state, each terminal continuously monitors its own control channel, and the network side sends the physical resource location of the multicast data in the control channel, modulation and coding methods, and each terminal receives control After signaling, start to receive data at the physical resource location specified by the network side.

(9)根据终端QoS需求，网络侧决定终端进行HARQ和数据重传时机，重传结束后继续（8）。(9) According to the QoS requirements of the terminal, the network side determines the timing of HARQ and data retransmission for the terminal, and continues to (8) after the retransmission is completed.

(10)数据发送完毕后，此次网络数据多播预推送完成，中断连接。(10) After the data is sent, the network data multicast pre-push is completed, and the connection is interrupted.

Claims (7)

1. the pre-method for pushing of the network data based on multicast, the user behavior data of terminal in its covering is accumulated at network side, when a certain user terminal requests network data, network side is according to the business tine of these accumulation data and this request, if judge that covering all the other terminals interior are greater than for the probability of demand of this business tine the value preset, then use the pre-propelling data of the mode of multicast, while sending data to active user, complete the business demand in multiple terminal future, it is characterized in that, when a user terminal requests network data, first according to the pre-propelling data buffer memory of content search of request, find whether certain moment have received the data content of this time asking by the pre-mode pushed before, if do not received, or this terminal needs upgrade and receive new data, this user terminal generates upstream data, the information of the service request of oneself is put into upstream message, and adhere to a mark the need of the Flag marker bit receiving new data, after user terminal establishes the link with network side, send request data, network side detects request content, user terminal ID, utilizes request content more new business temperature database, upgrades user behavior data storehouse according to user behavior, detect Flag marker bit, if this user terminal does not need the data that please look for novelty, then this conversation end, disconnects network and connects, if this user terminal needs to carry out new data when sending, in network side carries out covering again, all the other terminals calculate for the probability of demand of this business tine.

2. according to claim 1 based on the pre-method for pushing of network data of multicast, it is characterized in that, network side is to covering the probability of demand P of all the other terminals interior for this business tine_uEby following formulae discovery:

P_UE＝P(UE|UE_ID,UE_POS,TASK_CLASS,TASK_TIME,TASK_LIFE…)

Wherein, UE_ID represents user, and UE_POS represents customer location, and TASK_CLASS represents type of service, and TASK_TIME represents that business initiates the time, and TASK_LIFE represents business life, and ellipsis represents adeditive attribute.

3. according to claim 1 based on the pre-method for pushing of network data of multicast, it is characterized in that, network side is according to business tine, and community, requested service terminal place and adjacent cell cover lower other terminal use custom, channel status, calculate cell scheduling strategy and terminal scheduling strategy, modulating-coding strategy is calculated according to channel status, extrapolate the transmission expected rate Rmax after using this method to optimize, do not use the clean culture transmission rate Rsingle of this method, and the income Rmax/Rsingle after using we to send out:

If when the prediction income that the mode of Web vector graphic multicast obtains is less than the expected revenus preset, the mode of traditional unicast is used to send data;

If when network side prediction uses multicast pre-propelling data acquisition income to be greater than the minimum expectation income preset, then the mode using multicast to push in advance sends data.

4. according to claim 3 based on the pre-method for pushing of network data of multicast, it is characterized in that, according to described optimisation strategy, selection scheduling participates in the alternative cell of multicast propelling movement and the alternative terminal in these communities, send multicast sending out notice by core net to alternative cell, notify that each alternative cell covers the interior alternative Termination ID participating in multicast and push, physical source distributing situation, modulation coding mode, by the alternative terminal in each alternative cell separately dispatching cell coverage.

5. according to claim 3 based on the pre-method for pushing of network data of multicast, it is characterized in that, may be in different conditions because network side covers interior terminal, the state therefore for them is divided into following situation to dispatch:

A) for being in connected state (connected), the terminal of state of activation/continuous accepting state (active), due to its lasting monitoring control channel message, therefore directly can notify that downlink data arrives on a control channel, downlink data physical resource position, the information such as modulating-coding strategy, and keep connected_active state;

B) for being in connected state, the terminal of discontinuous receiving state, due to the terminal only ability reading control signal channel when its discontinuous listening period arrives, therefore network side is when the discontinuous listening period of each terminal arrives, schedule information is sent at the control channel of terminal, activated terminals, makes it enter connected_active state;

C) for the terminal being in idle condition, need network side first to terminal, initiate paging, triggering terminal connects, and after terminal initial connection establishment completes, terminal directly enters connected_active state.

6. according to claim 5 based on the pre-method for pushing of network data of multicast, it is characterized in that, after the terminal of the pre-propelling data of full entry multicast enters connected_active state, each terminal monitors oneself control channel continuously, network side sends the physical resource position of multi-case data in control channel, modulation coding mode, after each terminal receives control signal, the physical resource position starting to specify at network side receives data.

7., according to claim 5 based on the pre-method for pushing of network data of multicast, it is characterized in that, according to terminal QoS demand, network side determines that terminal carries out HARQ and data re-transmission opportunity, after data are sent, this time network data multicast has pushed in advance, middle connection breaking.