CN111552886A

CN111552886A - Caching method, system and device in wireless network

Info

Publication number: CN111552886A
Application number: CN202010425918.6A
Authority: CN
Inventors: 徐瑨; 朱世鹏; 陈希; 陶小峰
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2020-05-19
Filing date: 2020-05-19
Publication date: 2020-08-18
Anticipated expiration: 2040-05-19
Also published as: CN111552886B

Abstract

The embodiment of the invention provides a caching method, a caching system and a caching device in a wireless network, relates to the technical field of communication, and can improve the cache hit rate of a base station. The scheme of the embodiment of the invention comprises the following steps: receiving an acquisition request sent by a terminal, wherein the acquisition request is used for requesting to acquire contents to be acquired; if the content to be acquired does not exist in the stored contents cached by the small base station, acquiring the content to be acquired from the macro base station to which the small base station belongs, and sending the content to be acquired to the terminal; updating the interest scores of the contents to be acquired, wherein the interest scores of the contents to be acquired are used for expressing the interest degree of the users in the contents to be acquired; determining the quality score of the content to be acquired according to the updated interest score, wherein the quality score is used for expressing the request quantity change trend of the content to be acquired; and if the quality scores of the contents to be acquired are greater than the lowest score in the quality scores of the stored contents, caching the contents to be acquired, and deleting the stored contents corresponding to the lowest score.

Description

Caching method, system and device in wireless network

Technical Field

The present invention relates to the field of communications technologies, and in particular, to a caching method, system, and apparatus in a wireless network.

Background

In a future wireless network, base stations which are densely deployed are connected with a core network through a backhaul link, and massive data required by a user is acquired from the core network, which puts high requirements on capacity, time delay and the like of the backhaul link and brings huge pressure and challenges to the backhaul link. At the edge of a wireless network, namely in a small cell, a cache strategy is applied to cache a part of content, when a user requests the part of content, the small cell does not need to acquire the part of content from a core network through a backhaul link, but directly acquires the part of content from the cache and returns the part of content to the user, so that the pressure of the backhaul link is reduced, and the response speed to the user request is improved.

The existing caching strategy generally only considers the content cached in the small base station from the content trend perspective, and the content trend of each content can be the change trend of the requested amount of the content in a past period of time. For example, if a content is requested 100 times in the first two hours and 50 times in the first hour, the content trend of the content may be considered to be decreasing.

However, only from the perspective of content trend, the caching policy is considered, some content cached in the caching tends to be upward, the user in the coverage area of the base station is not necessarily interested in the content, and the content interested by the user is not necessarily cached by the base station, if the content is not cached, the base station still needs to acquire the content from the core network through the backhaul link when the user requests the content from the base station. It can be seen that the cache hit rate of the base station using this method is low, where the cache hit rate refers to the proportion of the content requested by the user existing in the cache content of the base station.

Disclosure of Invention

Embodiments of the present invention provide a caching method, system and apparatus in a wireless network, so as to improve a cache hit rate of a base station. The specific technical scheme is as follows:

in a first aspect, an embodiment of the present invention provides a caching method in a wireless network, which is applied to a small cell, and the method includes:

receiving an acquisition request sent by a terminal, wherein the acquisition request is used for requesting to acquire contents to be acquired;

if the content to be acquired does not exist in the stored contents cached by the small base station, acquiring the content to be acquired from a macro base station to which the small base station belongs, and sending the content to be acquired to the terminal;

updating the interest scores of the contents to be obtained, wherein the interest scores of the contents to be obtained are used for expressing the interest degree of the users in the contents to be obtained;

determining the quality score of the content to be acquired according to the updated interest score, wherein the quality score is used for expressing the request quantity change trend of the content to be acquired;

and if the quality scores of the contents to be obtained are larger than the lowest score in the quality scores of the stored contents, caching the contents to be obtained, and deleting the stored contents corresponding to the lowest score.

Optionally, the updating the interest score of the content to be acquired includes:

determining attribute parameters according to the acquisition request, wherein the attribute parameters comprise user attribute parameters of a user using the terminal to log in and content attribute parameters of the content to be acquired;

inputting the attribute parameters into an interest evaluation model, and acquiring interest probability output by the interest evaluation model, wherein the interest probability is used for expressing the interest degree of the login user in the content to be acquired;

and taking the sum of the interest score of the content to be obtained before updating, a preset updating constant and the interest probability as the interest score of the updated content to be obtained.

Optionally, the interest evaluation model includes a user branch, a content branch and an output layer; the method comprises the steps that for each branch, the branch comprises an embedded layer and a plurality of full-connection layers, the embedded layer is used for determining historical characteristics of each input attribute parameter and splicing the historical characteristics end to obtain splicing characteristics, and each historical characteristic is a characteristic vector corresponding to one user attribute parameter or a characteristic vector corresponding to one content attribute parameter; the full connection layer is used for determining interest probability according to the splicing characteristics; the output layer is used for splicing the interest probability determined by the full connection layer of the user branch and the interest probability determined by the full connection layer of the content branch, and normalizing the spliced probabilities;

the inputting the attribute parameters into an interest evaluation model comprises:

and inputting the user attribute parameters into the user branch, and inputting the content attribute parameters into the content branch.

Optionally, the determining the quality score of the content to be acquired according to the updated interest score includes:

determining a quality score of the content to be acquired according to the following formula:

qual＝ζ(time,t_H)*inte

wherein qual is quality score, time is the sending time of the acquisition request, inte is the interest score of the updated content to be acquired, ζ is a trend evaluation function, t_HAnd sending the obtaining request of the content to be obtained.

Optionally, the method further includes:

updating the interest scores of the cached contents which are not requested to be obtained in the current caching period by the following formula:

inte＝inte_old*exp(-γ(C-rank))

wherein, aiming at the cached content which is not requested to be obtained in each current caching period, inte is the interest score after the cached content is updated, inte_oldAnd the interest score before updating the cached content is determined, gamma is an adjusting coefficient, C is the number of the cached content, and rank is the ranking of the cached content after the cached content is sorted from big to small.

Optionally, the method further includes:

taking the attribute parameters included in the acquisition request received in the current cache period as attribute parameters to be acquired;

sending a feature request to other small base stations and/or the macro base station under the macro base station to which the small base station belongs, wherein the feature request comprises the attribute parameters to be acquired;

if a characteristic response is received, extracting the usage amount of the historical characteristics to be acquired corresponding to the attribute parameters to be acquired, which are included in the characteristic response;

and fusing the historical characteristics to be acquired according to the usage amount, and taking the fused historical characteristics to be acquired as the historical characteristics corresponding to the attribute parameters to be acquired.

Optionally, the fusing the historical features to be obtained according to the usage amount includes:

fusing the historical characteristics to be acquired according to the following formula:

wherein M is the number of historical features included in the feature response, w_iFor historical features included in the feature response, r_iIs the amount of usage of the historical feature included in the feature response.

Optionally, the method further includes:

determining data to be uploaded, wherein the data to be uploaded comprises: the method comprises the steps that contents to be shared, the requested times of the contents to be shared, historical characteristics to be shared and the usage amount of the historical characteristics to be shared, wherein the times of the contents to be shared, the requested times of the contents to be shared, the usage amount of the historical characteristics to be shared are more than a preset threshold value in a current cache period;

and uploading the data to be uploaded to a macro base station to which the small base station belongs.

In a second aspect, an embodiment of the present invention provides a cache system in a wireless network, where the system includes: the system comprises a terminal, a small base station, a macro base station and core network equipment;

the small base station is used for receiving an acquisition request sent by a terminal, wherein the acquisition request is used for requesting to acquire contents to be acquired; if the content to be acquired does not exist in the stored contents cached by the small base station, forwarding the acquisition request to a macro base station to which the small base station belongs, receiving the content to be acquired sent by the macro base station, and sending the content to be acquired to the terminal; updating the interest scores of the contents to be obtained, wherein the interest scores of the contents to be obtained are used for expressing the interest degree of the users in the contents to be obtained; determining the quality score of the content to be acquired according to the updated interest score, wherein the quality score is used for expressing the request quantity change trend of the content to be acquired; if the quality scores of the contents to be obtained are larger than the lowest score in the quality scores of the stored contents, caching the contents to be obtained, and deleting the stored contents corresponding to the lowest score;

the macro base station is used for sending the content to be obtained to the small base station if the content to be obtained exists in each stored content cached by the macro base station when the obtaining request sent by the small base station is received; if the content to be acquired does not exist in the stored contents cached by the macro base station, forwarding the acquisition request to core network equipment, receiving the content to be acquired sent by the core network equipment, and sending the content to be acquired to the small base station;

the core network device is configured to acquire the content to be acquired when receiving the acquisition request sent by the macro base station, and send the content to be acquired to the macro base station.

In a third aspect, an embodiment of the present invention provides a caching apparatus in a wireless network, where the caching apparatus is applied to a small cell, and the apparatus includes:

the terminal comprises a receiving module, a sending module and a receiving module, wherein the receiving module is used for receiving an acquisition request sent by the terminal, and the acquisition request is used for requesting to acquire the content to be acquired;

a sending module, configured to, if the content to be acquired does not exist in the stored contents cached by the small base station, acquire the content to be acquired from the macro base station, and send the content to be acquired to the terminal;

the updating module is used for updating the interest scores of the contents to be obtained, and the interest scores of the contents to be obtained are used for expressing the interest degree of the users in the contents to be obtained;

the determining module is used for determining the quality score of the content to be obtained according to the updated interest score, wherein the quality score is used for representing the request quantity change trend of the content to be obtained;

and the caching module is used for caching the content to be acquired and deleting the stored content corresponding to the lowest score if the quality score of the content to be acquired is greater than the lowest score in the quality scores of the stored contents.

Optionally, the update module is specifically configured to:

the update module is specifically configured to:

Optionally, the determining module is specifically configured to:

qual＝ζ(time,t_H)*inte

Optionally, the update module is further configured to:

inte＝inte_old*exp(-γ(C-rank))

Optionally, the apparatus further comprises: an extraction module and a fusion module;

the determining module is further configured to use an attribute parameter included in the acquisition request received in the current cache cycle as an attribute parameter to be acquired;

the sending module is further configured to send a feature request to other small base stations and/or the macro base station under the macro base station to which the small base station belongs, where the feature request includes the attribute parameter to be acquired;

the extraction module is used for extracting the usage amount of the historical feature to be acquired corresponding to the attribute parameter to be acquired, which is included in the feature response, if the feature response is received;

and the fusion module is used for fusing the historical characteristics to be acquired according to the usage amount and taking the fused historical characteristics to be acquired as the historical characteristics corresponding to the attribute parameters to be acquired.

Optionally, the fusion module is specifically configured to:

Optionally, the apparatus further comprises: an uploading module;

the determining module is further configured to determine data to be uploaded, where the data to be uploaded includes: the method comprises the steps that contents to be shared, the requested times of the contents to be shared, historical characteristics to be shared and the usage amount of the historical characteristics to be shared, wherein the times of the contents to be shared, the requested times of the contents to be shared, the usage amount of the historical characteristics to be shared are more than a preset threshold value in a current cache period;

and the uploading module is used for uploading the data to be uploaded to a macro base station to which the small base station belongs.

In a fourth aspect, an embodiment of the present invention provides a small cell, including a processor, a communication interface, a memory, and a communication bus, where the processor and the communication interface complete communication between the processors and the memory through the communication bus;

a memory for storing a computer program;

and the processor is used for realizing the steps of the caching method in any wireless network when executing the program stored in the memory.

In a fifth aspect, an embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the computer program implements the steps of the caching method in any wireless network.

In a sixth aspect, an embodiment of the present invention further provides a computer program product containing instructions, which when run on a computer, causes the computer to execute the caching method in any of the wireless networks described above.

The technical scheme of the embodiment of the invention can at least bring the following beneficial effects: the embodiment of the invention can determine the quality score according to the interest score of the content to be obtained, and further determine whether to cache the content to be obtained according to the quality score, and the interest score can represent the interest of the user in the content to be obtained, and the quality score can represent the request quantity change trend of the content to be cached, so that the caching strategy provided by the embodiment of the invention can comprehensively consider the interest of the user and the content trend, and compared with a mode of only considering the content trend, the caching hit rate of the embodiment of the invention is higher.

Of course, not all of the advantages described above need to be achieved at the same time in the practice of any one product or method of the invention.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a schematic structural diagram of a cache system in a wireless network according to an embodiment of the present invention;

fig. 2 is a flowchart of a caching method in a wireless network according to an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of an interest model according to an embodiment of the present invention;

fig. 4 is a signaling flowchart of a caching method in a wireless network according to an embodiment of the present invention;

fig. 5 is a signaling flowchart of an information sharing process according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of a cache device in a wireless network according to an embodiment of the present invention;

fig. 7 is a schematic structural diagram of a small cell base station according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1, a cache system in a wireless network according to an embodiment of the present invention includes: core network equipment 101, a macro base station 102, a small cell 103, and a terminal 104. Wherein:

the core network device 101 may communicate with a plurality of macro base stations 102, and a large amount of content is stored in the core network device 101.

A plurality of small base stations 103 exist in the signal coverage area of the macro base station 102, the macro base station 102 can communicate with the small base stations 103 in the signal coverage area, and different macro base stations 102 communicate with each other through an X2 interface.

The small cell 103 may communicate with the terminals 104 within its signal coverage area, and different small cells under the same macro cell 102 communicate with each other through an X2 interface.

In order to improve the cache hit rate of the base station, an embodiment of the present invention provides a caching method in a wireless network, which is applied to the small base station in fig. 1, and referring to fig. 2, the method includes the following steps.

Step 201, receiving an acquisition request sent by a terminal.

The obtaining request is used for requesting to obtain the content to be obtained. For example, the content to be acquired may be video, audio, pictures, documents, and the like.

Optionally, the obtaining request may include: a timestamp, an Identity Document (ID), an ID of the content to be obtained, and a source of the content. The timestamp may be the time when the terminal sends the acquisition request; the content source may be a Uniform Resource Locator (URL) of the content to be obtained.

For example, the request message may include [10,1,2,3], where 10 is a timestamp, 1 is a user ID, 2 is a content ID, and 3 is a content source.

Step 202, if there is no content to be acquired in each stored content cached by the small base station, acquiring the content to be acquired from the macro base station to which the small base station belongs, and sending the content to be acquired to the terminal.

In one embodiment, the small cell may request the macro base station to acquire the content to be acquired, and if the macro base station caches the content to be acquired, the small cell directly returns the content to be acquired; and if the macro base station does not cache the content to be acquired, requesting the core network equipment to acquire the content to be acquired. The core network equipment acquires the content to be acquired from the content source of the content to be acquired, sends the content to the macro base station and forwards the content to the small base station by the macro base station. And after acquiring the content to be acquired, the small base station sends the content to be acquired to the terminal.

And step 203, updating the interest scores of the contents to be acquired.

The interest score of the content to be acquired is used for representing the interest degree of the user in the content to be acquired.

And step 204, determining the quality score of the content to be acquired according to the updated interest score.

The quality score is used for representing the request quantity variation trend of the content to be acquired.

In step 205, if the quality score of the content to be acquired is greater than the lowest score among the quality scores of the stored contents, the content to be acquired is cached, and the stored content corresponding to the lowest score is deleted.

In the embodiment of the present invention, the manner for updating the interest score of the content to be acquired in step 103 may include the following three steps.

Step one, determining attribute parameters according to the acquisition request. The attribute parameters comprise attribute parameters of a terminal login user and attribute parameters of the content to be acquired.

In one embodiment, the user ID and the content ID are extracted from the acquisition request, and the attribute parameters are acquired from the operator database according to the user ID and the content ID. For example, the attribute parameters of the login user include: user ID, age, gender, etc. The attribute parameters of the content to be acquired include: content ID, format and title information, etc.

And step two, inputting the attribute parameters into the interest evaluation model, and acquiring the interest probability output by the interest evaluation model. The interest probability is used for expressing the interest degree of the login user in the content to be acquired.

In an embodiment of the present invention, referring to fig. 3, the interest evaluation model includes a user branch, a content branch, and an output layer; the method comprises the steps that for each branch, the branch comprises an embedded layer and a plurality of full-connection layers, the embedded layer is used for determining historical characteristics of each input attribute parameter and splicing the historical characteristics end to obtain splicing characteristics, and each historical characteristic is a characteristic vector corresponding to one user attribute parameter or a characteristic vector corresponding to one content attribute parameter. And the full connection layer is used for determining interest probability according to the splicing characteristics. The output layer is used for splicing the interest probability determined by the full connection layer of the user branch and the interest probability determined by the full connection layer of the content branch, and normalizing the spliced probabilities. Wherein numerals in parentheses represent characteristic dimensions.

In one embodiment, the user attribute parameters may be entered into a user branch and the content attribute parameters may be entered into a content branch.

It can be understood that the format of the input attribute parameter conforms to the format obtained from the operator database, and the output of the interest evaluation model is a value in the range of [0,1], the closer the value is to 0, the less interested the login user is in the content to be obtained, and the closer the value is to 1, the more interested the login user is in the content to be obtained.

In a possible implementation, each branch includes one embedded layer and two fully connected layers, where the embedded layer may be understood as a dictionary and records the corresponding relationship between each user attribute parameter and a feature vector or the corresponding relationship between each content attribute parameter and a feature vector. After the attribute parameters are input into the embedding layer, the embedding layer firstly obtains the characteristic vectors corresponding to the user attribute parameters or the characteristic vectors corresponding to the content attribute parameters, and then splices the characteristic vectors corresponding to the user attribute parameters or splices the characteristic vectors corresponding to the content attribute parameters to obtain splicing characteristics. Wherein, the length of the history characteristics can be set according to actual needs.

The historical characteristics output by the embedding layer are input into the first full connection layer, and the output of the first full connection layer is the input of the second full connection layer. And finally, after the output result of the second-layer full-connection layer of the user branch is spliced with the output result of the second-layer full-connection layer of the content branch, converting the output result into a number in a range of [0,1] through an excitation function (softmax) included by the output layer and outputting the number.

And step three, taking the sum of the interest score of the content to be acquired before updating, the preset updating constant and the interest probability as the interest score of the updated content to be acquired.

In one embodiment, the updated interest score of the content to be acquired is obtained through formula (1).

inte＝inte_old+β+P(r) (1)

Wherein, the inte is the interest score of the updated content to be obtained, the inte_oldFor the interest score before updating of the content to be acquired, β is a preset updating constant, and p (r) is an interest probability.

In an embodiment of the present invention, inte_oldIs 0. For example, when the small cell receives a request for obtaining a content for the first time, the inte of the content_oldIf 0 is set according to the scene, β is set to 0.2 in advance, p (r) output by the interest assessment model is 0.8, and the interest score inte after the content update is 1.

In one embodiment, the method for determining the quality score in step 104 comprises: and determining the quality score of the content to be acquired by using the formula (2).

qual＝ζ(time,t_H)*inte (2)

Wherein qual is quality score, time is sending time of the acquisition request, inte is interest score of the updated content to be acquired, zeta is a trend evaluation function, and t is_HThe sending time of the acquisition request for requesting to acquire the content to be acquired.

It is understood that time may be extracted from the get request.

In the embodiment of the invention, for any cached content, the small cell records the additional information vector [ inte, qual, t ] of the content_H]Where int represents the interest score of the content, qual represents the quality score of the content, and t_HThe time queue with the maximum length of H is used for storing the sending time of the acquisition request for acquiring the content. In the embodiment of the present invention, the value of H may be 5, i.e., t_HThe transmission time of the acquisition request for acquiring the content for the last 5 requests. In the embodiment of the invention, if the time included in the time queue is less than H, zero is filled in the time queue. For example, for a content, if the small cell receives only 3 acquisition requests for acquiring the content, t of the content is_HIncluding the transmission times of the three acquisition requests, and two 0 s.

Alternatively, the trend evaluation function ζ is formula (3).

Wherein H is a queue t_HLength of (d), t_iIs t_HThe ith timestamp α is a preset popularity decay factor.

For example, α is 0.2, the quality score of the content to be acquired is calculated as 1, and the quality scores of the stored contents cached by the base station are [1.5,1.2,0.7], respectively. And if the lowest score in the quality scores of the stored contents is 0.7<1, deleting the stored contents corresponding to 0.7 and caching the contents to be cached.

It can be understood that, while determining the quality score of the content to be obtained, the quality score qual ═ ζ (time, t) of each cached content may also be calculated by using the formula (2) for the cached content_H′)*inte′，t_H'is the sending time of the obtaining request for obtaining the cached content, inte' is the latest updated interest score of the cached content, and time is the sending time of the obtaining request for obtaining the content to be obtained.

The technical scheme of the embodiment of the invention can also bring the following beneficial effects: according to the embodiment of the invention, the quality scores of the content to be acquired and the cached content can be updated in real time according to the acquisition request, so that the cached content is updated in real time, the complex change of the user request is effectively responded, and the cache hit rate is improved.

In the embodiment of the invention, the small base station can also periodically update the interest score. The buffer cycles (Ru) may be configured according to a specific scenario. In an embodiment of the present invention, Ru may be 10 minutes.

In one embodiment, the interest score of the cached content that is not requested to be obtained in the current caching period is updated by formula (4).

inte＝inte_old*exp(-γ(C-rank)) (4)

Wherein, aiming at the cached content which is not requested to be obtained in each current caching period, inte is the interest score after the cached content is updated, inte_oldAnd gamma is an adjustment coefficient, C is the number of the cached contents, and rank is the ranking of the cached contents after the cached contents are sorted from large to small.

It is understood that γ may be configured according to a specific scenario, for example, γ may be 0.02; each buffer cycle may be 10 minutes. Illustratively, the cached content ID that was not requested to be retrieved in the first 10 minutes is 2, and its interest score is updated to 0.4.

The technical scheme of the embodiment of the invention can also bring the following beneficial effects: the embodiment of the invention can update the quality score of the cached content periodically, so that the quality score of the cached content has timeliness, and the cache hit rate is further improved.

In the embodiment of the present invention, the interest evaluation model may also be updated, where updating the interest evaluation model includes: updating the embedding layer and updating the training interest evaluation model.

In one embodiment, updating the embedding layer includes the following steps.

Step one, taking the attribute parameters included in the acquisition request received in the current cache period as the attribute parameters to be acquired.

It will be appreciated that each attribute corresponds to one or more attribute parameters. For example, the attribute parameter corresponding to the gender attribute may be 1 or 2, where 1 indicates that the gender is female and 2 indicates that the gender is male.

For example, the acquisition request received in the week includes an attribute parameter 30 indicating age 30, an attribute parameter 10 indicating age 10, and an attribute parameter 1 indicating gender as female, these attribute parameters are used as attribute parameters to be acquired, and the history features corresponding to these attribute parameters are used as history features to be acquired.

And step two, sending a feature request to other small base stations and/or macro base stations under the macro base station to which the small base station belongs, wherein the feature request comprises attribute parameters to be acquired.

In one implementation mode, a feature request is sent to other small base stations under a macro base station to which the small base station belongs, the request is recorded as req1, and after the other small base stations receive req1, whether historical features corresponding to attribute parameters to be acquired exist or not is inquired in an embedding layer of an interest evaluation model of the other small base stations. And if so, sending the feature information of the attribute parameters to be acquired back to the small base station. If not, no return is made. Wherein, the characteristic information may include: the attribute parameters to be acquired, the historical characteristics to be acquired corresponding to the attribute parameters to be acquired and the usage amount of the historical characteristics to be acquired.

For the feature information of the attribute parameters to be acquired, which is not obtained by the small base station from other small base stations, the small base station may send a feature request for requesting the feature information of the attribute parameters to be acquired to the macro base station to which the small base station belongs, and note that the request is req2, and after receiving req2, the macro base station forwards req2 to other macro base stations. After receiving the req2, the other macro base station determines the feature information requested by req2 from its database and returns the feature information to the macro base station. And the macro base station forwards the characteristic information to the small base station after receiving the characteristic information.

For the feature information of the attribute parameters to be acquired, which is not acquired from the macro base station by the small base station, the small base station may send a feature request for requesting the feature information of the attribute parameters to be acquired to the macro base station, and note that the request is req3, and after receiving req3, the macro base station forwards req3 to the core network device. After receiving the req3, the core network device determines the feature information requested by the req3 from its database, and returns the feature information to the macro base station. And the macro base station forwards the characteristic information to the small base station after receiving the characteristic information.

For example, the feature request includes { age: [10,30], gender: [1] }, which indicates that acquisition of the attribute to be acquired parameter 10 indicating age 10, the attribute to be acquired parameter 30 indicating age 30, and the attribute to be acquired parameter 1 indicating gender as woman is requested.

And step three, if the characteristic response is received, extracting the usage amount of the historical characteristics to be acquired corresponding to the attribute parameters to be acquired included in the characteristic response.

In one embodiment, if the feature request of the small cell includes an attribute parameter to be acquired, the feature response includes the usage amount of the historical feature to be acquired; and if the feature request of the small base station comprises various attribute parameters to be acquired, the feature response comprises various historical features to be acquired and the usage amount of each historical feature to be acquired.

In the embodiment of the present invention, a history feature usage counter may be set, and each time the interest evaluation model in the small cell is used, the history feature counter corresponding to the input attribute parameter is incremented by 1, that is, the usage of the history feature is incremented by 1.

For example, the attribute parameters of the first input interest assessment model are [1,20], which means that the attribute parameter 1 representing gender as female and the attribute parameter 20 representing age as 20 are input, wherein the attribute parameter 1 corresponds to the history vector v1, and the attribute parameter 20 corresponds to the history vector v 2. The usage of v1 and v2 is increased by 1, and assuming that the initial usage of v1 and v2 is 0, the usage of v1 and v2 is 1 after 1 is increased. The attribute parameters input into the interest assessment model for the second time are [1,30], which indicates that attribute parameter 1 of female gender and attribute parameter 30 of age 30 are input, and the attribute parameter 30 corresponds to the history vector v3, so that the usage of v1 and v3 is increased by 1, assuming that the initial usage of v3 is 0, the usage of v1 after the increase is 2, and the usage of v3 is 1.

And 4, fusing the historical characteristics to be acquired according to the usage amount, and taking the fused historical characteristics to be acquired as the historical characteristics corresponding to the attribute parameters to be acquired.

In one embodiment, the historical features to be obtained may be fused according to equation (5).

Where M is the number of historical features included in the feature response, w_iFor historical features i, r included in the feature response_iIs the amount of usage of the historical signature i included in the signature response.

Optionally, for the attribute parameter to be acquired that is not acquired from any of the other small base stations, the macro base station, and the core network device, the all-zero vector may be used as the historical feature of the attribute parameter to be acquired.

The technical scheme of the embodiment of the invention can also bring the following beneficial effects: the small base station in the embodiment of the invention effectively relieves the problem of incomplete data acquired by a single small base station due to user mobility by sharing historical characteristics with surrounding base stations, improves the estimation accuracy of user interest and is beneficial to improving the cache hit rate.

In the embodiment of the present invention, the method for updating the training interest evaluation model includes: and generating a training sample by using the attribute parameters included in the acquisition request received in the current cache period. The training sample consists of attribute parameters and labels of the attribute parameters, and the labels are divided into a positive type and a negative type. The data part of the positive label sample is the attribute parameters carried in the real acquisition request, and the data part of the negative label sample is obtained by randomly replacing the attribute parameters in the real acquisition request with the content which is not requested by the user.

And then training the interest evaluation model by using the training samples until the training is finished. The training method of the model is the prior art, and all the training methods adopted in the invention are general existing model training methods, so that the details are not repeated.

The technical scheme of the embodiment of the invention can also bring the following beneficial effects: the caching method of the embodiment of the invention reduces the problems of network redundant flow, network congestion and overhead waste and the like caused by repeated distribution of the same content during peak periods.

In the embodiment of the invention, data sharing can be realized among the base stations. The method comprises the following two steps.

Step one, data to be uploaded is determined. Wherein, waiting to upload data includes: the content to be shared, the requested times of the content to be shared, the historical characteristics to be shared and the usage amount of the historical characteristics to be shared, wherein the times of the content to be shared and the usage amount of the historical characteristics to be shared are more than the preset times of the content to be shared, and the usage amount of the historical characteristics to be shared, and the times of the content to be shared and the usage amount of the historical characteristics to.

And step two, uploading the content to be uploaded to a macro base station to which the small base station belongs.

In order to more clearly explain the caching method in the wireless network provided by the embodiment of the present invention, fig. 4 shows an interaction flow between devices included in the caching system in the wireless network, where the interaction flow includes the following steps.

Step 401, the terminal sends an acquisition request to the small cell, and correspondingly, the small cell receives the acquisition request sent by the terminal. The obtaining request is used for requesting to obtain the content to be obtained.

Step 402, if there is no content to be acquired in each stored content cached by the small cell, the small cell forwards an acquisition request to the macro base station to which the small cell belongs.

In one implementation mode, the small base station extracts the ID of the content to be acquired from the acquisition request, and checks whether the ID of the content to be acquired exists in the stored content ID of the small base station; and if the content ID to be acquired does not exist in the stored content ID of the small base station, the small base station adds the self identification to the acquisition request and forwards the acquisition request to the macro base station to which the small base station belongs.

In the embodiment of the invention, if the content to be acquired exists in the stored contents cached by the small base station, the small base station directly acquires the content to be acquired from the cache and returns the content to be acquired to the terminal.

In step 403, when the macro base station receives the acquisition request sent by the small base station, if the content to be acquired exists in the stored contents cached by the macro base station, the macro base station sends the content to be acquired to the small base station. And proceeds to step 408.

In step 404, if there is no content to be acquired in each stored content cached by the macro base station, the macro base station forwards an acquisition request to the core network device.

With reference to step 403 and step 404, in an embodiment, after receiving the acquisition request sent by the small base station, the macro base station checks whether a to-be-acquired content ID exists in the stored content IDs stored by the macro base station, and if so, sends the to-be-acquired content to the terminal through the small base station; and if the request does not exist, forwarding the acquisition request sent by the small base station to the core network equipment.

It should be noted that steps 403 and 404 are only for distinguishing two steps, and these two steps are not both executed.

Step 405, when the core network device receives an acquisition request sent by the macro base station, the core network device acquires content to be acquired.

In one embodiment, after receiving the acquisition request, the core network device acquires the content to be acquired from the source of the content to be acquired.

In step 406, the core network device sends the content to be acquired to the macro base station. Correspondingly, the macro base station receives the content to be acquired, which is sent by the core network equipment.

Step 407, the macro base station sends the content to be acquired to the small base station.

And step 408, the small base station receives the content to be acquired sent by the macro base station and sends the content to be acquired to the terminal.

And 409, updating the interest scores of the contents to be acquired by the small base station, determining the quality scores of the contents to be acquired according to the updated interest scores, caching the contents to be acquired by the small base station if the quality scores of the contents to be acquired are greater than the lowest score in the quality scores of the stored contents, and deleting the stored contents corresponding to the lowest score.

The interest scores of the contents to be acquired are used for representing the interest degree of the users in the contents to be acquired, and the quality score of one content user represents the request quantity change trend of the content.

In the embodiment of the present invention, information sharing may also be implemented between devices included in a cache system in a wireless network, and a process of information sharing is shown in fig. 5, and includes the following steps.

Step 501, the small base station determines data to be uploaded. Wherein, waiting to upload data includes: the content to be shared, the requested times of the content to be shared, the historical characteristics to be shared and the usage amount of the historical characteristics to be shared, wherein the times of the content to be shared and the usage amount of the historical characteristics to be shared are more than the preset times of the content to be shared, and the usage amount of the historical characteristics to be shared, and the times of the content to be shared and the usage amount of the historical characteristics to.

Step 502, the small cell uploads data to be uploaded to a macro base station to which the small cell belongs.

In step 503, the macro base station checks whether the content to be shared exists in its own cache. If yes, executing step 504, otherwise, directly assigning the received historical characteristics to the historical characteristics to be shared, and accumulating the requested times of the content to be shared aiming at each received content to be shared.

Step 504, the macro base station updates the requested times of the contents to be shared, sequences the contents cached by the macro base station and the contents to be shared according to the sequence of the requested times from high to low, and takes the first D contents as new cache contents. And D is the maximum number of cache contents preset by the macro base station.

In one embodiment, the number of times updates are requested may be: and accumulating the received requested times of the contents to be shared and the requested times of the contents to be shared stored by the user.

And step 505, the macro base station updates the historical characteristics to be shared.

In one embodiment, the history feature to be shared is updated using equation (6).

Wherein w is the updated history feature to be shared,

the historical characteristics cached for the macro base station,

for historical characteristics uploaded by the small base station, K₁The number of the small base stations in the communication range of the macro base station.

Optionally, a first reset period Rn is set, and the macro base station clears the requested number of times and the history feature of the cache after Rn reset periods. Rn may be configured according to a specific scenario, and may be 50 times of a buffering period, for example.

Step 506, the macro base station uploads the historical features to be uploaded to the core network device. The historical characteristics to be uploaded are historical characteristics of which the usage amount in the current cache period is larger than a preset threshold value.

And step 507, the core network equipment checks whether the historical characteristics to be uploaded exist in the cache of the core network equipment, and if so, the historical characteristics to be uploaded are updated.

In one embodiment, the historical characteristics to be uploaded are updated using equation (7).

Wherein w is the updated historical characteristics to be uploaded,

the historical characteristics cached for the core network device,

historical characteristics uploaded for macro base stations, K₂The number of macro base stations in the communication range of the core network equipment.

In the embodiment of the invention, if the historical characteristics to be uploaded do not exist in the cache of the core network device, the core network device directly assigns the received historical characteristics to the historical characteristics to be uploaded.

Optionally, a second reset period Rm is set, and the core network device clears the history feature of the cache after Rm reset periods. Rm may be configured according to a specific scenario, and may be, for example, 50 times the cache cycle.

Based on the same technical concept, corresponding to the above method embodiment, an embodiment of the present invention further provides a cache system in a wireless network, where the cache system includes: core network equipment, a macro base station, a small base station and a terminal.

The small base station is used for receiving an acquisition request sent by the terminal, and the acquisition request is used for requesting to acquire the content to be acquired; if the content to be acquired does not exist in the stored contents cached by the small base station, forwarding an acquisition request to a macro base station to which the small base station belongs, receiving the content to be acquired sent by the macro base station, and sending the content to be acquired to the terminal; updating the interest scores of the contents to be acquired, wherein the interest scores of the contents to be acquired are used for expressing the interest degree of the users in the contents to be acquired; determining the quality score of the content to be acquired according to the updated interest score, wherein the quality score is used for expressing the request quantity change trend of the content to be acquired; if the quality scores of the contents to be acquired are larger than the lowest score in the quality scores of the stored contents, caching the contents to be acquired, and deleting the stored contents corresponding to the lowest score;

the macro base station is used for sending the content to be acquired to the small base station if the content to be acquired exists in the stored contents cached by the macro base station when the acquisition request sent by the small base station is received; if the content to be acquired does not exist in the stored contents cached by the macro base station, forwarding an acquisition request to the core network equipment, receiving the content to be acquired sent by the core network equipment, and sending the content to be acquired to the small base station;

and the core network equipment is used for acquiring the content to be acquired when receiving the acquisition request sent by the macro base station and sending the content to be acquired to the macro base station.

Optionally, the small cell is configured to perform step 401, step 402, step 408, and step 409, the macro cell is configured to perform step 403, step 404, and step 407, and the core network device is configured to perform step 405 and step 406.

Optionally, the small cell is further configured to perform step 501 and step 502, the macro base station is further configured to perform step 503 to step 506, and the core network device is further configured to perform step 507.

Based on the same technical concept, corresponding to the above method embodiment, an embodiment of the present invention further provides a caching apparatus in a wireless network, which is applied to a small base station, and referring to fig. 6, the apparatus includes: a receiving module 601, a sending module 602, an updating module 603, a determining module 604 and a buffering module 605;

a receiving module 601, configured to receive an acquisition request sent by a terminal, where the acquisition request is used to request to acquire a content to be acquired;

a sending module 602, configured to, if there is no content to be acquired in each stored content cached by the small base station, acquire the content to be acquired from the macro base station, and send the content to be acquired to the terminal;

the updating module 603 is configured to update the interest score of the content to be acquired, where the interest score of the content to be acquired is used to represent the interest degree of the user in the content to be acquired;

a determining module 604, configured to determine a quality score of the content to be acquired according to the updated interest score, where the quality score is used to indicate a request quantity change trend of the content to be acquired;

the caching module 605 is configured to cache the content to be acquired and delete the stored content corresponding to the lowest score if the quality score of the content to be acquired is greater than the lowest score in the quality scores of the stored contents.

Optionally, the updating module 603 is specifically configured to:

inputting the attribute parameters into an interest evaluation model, and acquiring interest probability output by the interest evaluation model, wherein the interest probability is used for expressing the interest degree of a login user to the content to be acquired;

and taking the sum of the interest score of the content to be acquired before updating, the preset updating constant and the interest probability as the interest score of the updated content to be acquired.

Optionally, the interest evaluation model includes a user branch, a content branch and an output layer; the method comprises the steps that for each branch, the branch comprises an embedded layer and a plurality of full-connection layers, the embedded layer is used for determining historical characteristics of each input attribute parameter and splicing the historical characteristics end to obtain splicing characteristics, and each historical characteristic is a characteristic vector corresponding to a user attribute parameter or a characteristic vector corresponding to a content attribute parameter; the full connection layer is used for determining interest probability according to the splicing characteristics; the output layer is used for splicing the interest probability determined by the full connection layer of the user branch and the interest probability determined by the full connection layer of the content branch, and normalizing the spliced probabilities;

the update module 603 is specifically configured to:

the user attribute parameters are input into the user branch and the content attribute parameters are input into the content branch.

Optionally, the determining module is specifically configured to:

qual＝ζ(time,t_H)*inte

Optionally, the updating module 603 is further configured to:

inte＝inte_old*exp(-γ(C-rank))

wherein, aiming at the cached content which is not requested to be obtained in each current caching period, the inte is the interest score after the cached content is updated, and the inte_oldAnd the interest score before updating the cached content is determined, gamma is an adjusting coefficient, C is the number of the cached content, and rank is the ranking of the cached content after the cached content is sorted from large to small.

the determining module 604 is further configured to use an attribute parameter included in the acquisition request received in the current cache cycle as an attribute parameter to be acquired;

the sending module 602 is further configured to send a feature request to other small base stations and/or macro base stations under the macro base station to which the small base station belongs, where the feature request includes an attribute parameter to be acquired;

the extraction module is used for extracting the usage amount of the historical characteristics to be acquired corresponding to the attribute parameters to be acquired and included in the characteristic response if the characteristic response is received;

Optionally, the fusion module is specifically configured to:

where M is the number of historical features included in the feature response, w_iFor historical features included in the feature response, r_iIs the amount of usage of the historical signature included in the signature response.

Optionally, the apparatus further comprises: an uploading module;

the determining module 602 is further configured to determine data to be uploaded, where the data to be uploaded includes: the method comprises the steps that contents to be shared, the requested times of the contents to be shared, historical characteristics to be shared and the usage amount of the historical characteristics to be shared, wherein the times of the contents to be shared, the requested times of the contents to be shared, the usage amount of the historical characteristics to be shared are more than a preset threshold value in a current cache period;

and the uploading module is used for uploading the data to be uploaded to the macro base station to which the small base station belongs.

The embodiment of the present invention further provides a small cell, as shown in fig. 7, including a processor 701, a communication interface 702, a memory 703 and a communication bus 704, where the processor 701, the communication interface 702, and the memory 703 complete mutual communication through the communication bus 704,

a memory 703 for storing a computer program;

the processor 701 is configured to implement the steps executed by the small cell in the foregoing embodiments when executing the program stored in the memory 703.

The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.

The communication interface is used for communication between the electronic equipment and other equipment.

The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.

The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component.

In yet another embodiment of the present invention, a computer-readable storage medium is further provided, in which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the caching method in any of the wireless networks described above.

In yet another embodiment, a computer program product containing instructions is provided, which when run on a computer, causes the computer to perform the caching method in any of the wireless networks described above.

In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for other embodiments, since they are substantially similar to the method embodiments, the description is simple, and for relevant points, reference may be made to part of the description of the method embodiments.

The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims

1. A buffering method in a wireless network is applied to a small cell, and the method comprises the following steps:

2. The method of claim 1, wherein the updating the interest score of the content to be obtained comprises:

3. The method of claim 2, wherein the interest assessment model comprises a user branch, a content branch, and an output layer; the method comprises the steps that for each branch, the branch comprises an embedded layer and a plurality of full-connection layers, the embedded layer is used for determining historical characteristics of each input attribute parameter and splicing the historical characteristics end to obtain splicing characteristics, and each historical characteristic is a characteristic vector corresponding to one user attribute parameter or a characteristic vector corresponding to one content attribute parameter; the full connection layer is used for determining interest probability according to the splicing characteristics; the output layer is used for splicing the interest probability determined by the full connection layer of the user branch and the interest probability determined by the full connection layer of the content branch, and normalizing the spliced probabilities;

4. The method of claim 1, wherein determining the quality score of the content to be obtained according to the updated interest score comprises:

qual＝ζ(time,t_H)*inte

5. The method of claim 1, further comprising:

inte＝inte_old*exp(-γ(C-rank))

6. The method of claim 3, further comprising:

7. The method according to claim 6, wherein fusing the historical features to be obtained according to the usage amount comprises:

8. The method of claim 1, further comprising:

9. A caching system in a wireless network, the system comprising: the system comprises a terminal, a small base station, a macro base station and core network equipment;

10. A buffering apparatus in a wireless network, applied to a small cell, the apparatus comprising: