CN106254032B - A Content-Aware-Based Resource Scheduling Method - Google Patents

Abstract

The invention discloses a kind of resource regulating methods based on perception of content, comprising steps of transmitting terminal counts each receiving end institute request content ID and associated channel condition information；Transmitting terminal is that mark maintains corresponding request queue, and calculates separately the number of users for the current service request that can satisfy under different request contents using different coding modulation system with content ID；Transmitting terminal according to user cache state in the content future temperature of prediction, coverage area with connect number, the future potential service number of users for currently dispatching certain content with certain code modulation mode is calculated, and this potential service number of users is added in dispatch weight；According to dispatch weight by content ordering, content is sent according to preset code modulation mode from high to low according to dispatch weight；Receiving end receives and decodes data of eating dishes without rice or wine, and for the good user of the channel of not sent request, optionally caches high attention rate content of eating dishes without rice or wine.The expected returns when present invention realizes scheduling resource under unit consumption resource are maximum.Due to channel, good user can be cached, and avoided the repetition transmission of Hot Contents, improved the efficiency of transmission of content, bring better experience to user.

Description

A Content-Aware-Based Resource Scheduling Method

technical field

The present invention relates to the field of resource scheduling in wireless communication, in particular to a content-aware-based resource scheduling method.

Background technique

With the ubiquity of mobile devices and the continuous enrichment of mobile applications, the demand for mobile data of users is showing an explosive growth trend, which brings great challenges to network operators. The massive data demand is mainly concentrated on multimedia content. It is predicted that multimedia content transmission will account for 72% of data traffic in 2019. Unlike voice calls or peer-to-peer messaging services, multimedia content has distinct popularity attributes, ie the same multimedia content can be requested by multiple users. According to statistics, although the multimedia content traffic requested by users is huge, most of the user requests are actually concentrated in a small part of high-interest content, and this part of the content can be effectively predicted by counting historical request data or the number of retrieval times before the content is generated. At the same time, with the rapid development of semiconductor devices, user terminals (smartphones, tablets, etc.) have a huge cache space, which can effectively pre-store hot content through broadcasting or WIFI. On the other hand, the traditional resource scheduling method, whether based on maximum rate scheduling or proportional fair scheduling, only focuses on the transmission rate of the information bit level during scheduling, but ignores the popularity attribute of the content requested by the user, resulting in duplication of content. send. Therefore, we start from the content level, comprehensively consider the channel state information (CSI), service information (popularity) and user terminal buffer space to make scheduling decisions. Among them, the unrequested users with good channels can receive and buffer the air interface transmission of the base station to the channel poor. of the requesting user's hot content. Through this joint scheduling, we are able to improve network transmission efficiency and user experience.

Regardless of whether the LTE resource scheduling algorithm in the prior art is based on maximum rate scheduling or proportional fair scheduling, it only focuses on the transmission rate at the information bit level during scheduling, ignoring the popularity attribute of content. In multimedia content, we should pay attention to the content transmission efficiency, rather than the simple physical layer bit transmission rate.

In the prior art, it has been considered to use the content popularity attribute for scheduling in the broadcast push service, but because it is a broadcast push service, not a real-time scheduling service, it cannot use the physical layer channel information requested by the user in real time to associate with the content attribute. Cooperative resource scheduling decisions.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art, and propose a content-aware-based resource scheduling method, which comprehensively considers channel state information (Channel State Information, CSI for short), service information (popularity) and user terminal buffer space to do Scheduling decisions. On the user terminal, the unsolicited users with good channels can receive and buffer the hot content transmitted by the base station air interface to the requesting users with poor channels, that is, serve potential future user groups at the same time without adding additional air interface overhead. At the same time, the current access and potential future access of the same content are satisfied through one multicast transmission, which can improve the system capacity and user experience.

Based on the above purpose, the technical solution adopted in the present invention comprises the following steps:

A content-aware resource scheduling method, comprising the following steps:

Step 1: The base station (not limited to base stations, applicable to various types of wireless transmission points) counts the requested content ID and related channel state information of each requesting user (not limited to users, but applicable to various types of information receiving points).

Step 2: The base station maintains the corresponding request queue with the content ID as an identifier, and separately calculates the number of users with current service requests that can be satisfied by using different coding and modulation modes under different request contents.

Step 3: The base station calculates the number of future potential service users who currently schedule a certain content in a certain coding and modulation method according to the predicted future popularity of the content, the cache status of users within the coverage area, and the number of connections, and adds the number of potential service users to the schedule. in weight. .

Step 4: Sort the content according to the scheduling weight, and send the content from high to low according to the scheduling weight in a preset coding and modulation manner.

Step 5: The user requesting the scheduled content receives and decodes the air interface data. For users with good channels who have not sent requests, content of high interest in the air interface (such as popular microblogs, videos) can be selectively cached.

In the step 1: when a user generates a content request, first check whether the local storage space is completely cached, and if so, play it locally; otherwise, report the content request to the base station. The base station can obtain the content identifier and channel information in the following manner: the service content can be marked and distinguished, such as generating a unique identifier through a HASH function. When a user sends a content request, the base station side identifies the requested content ID. User CSI information can be fed back through user downlink estimation, or acquired by the base station side through uplink measurement in a channel reciprocity system.

In the step 2, the base station does not maintain a user-centered request queue, but maintains a request queue with content as an identifier, and calculates the number of users currently serving requests that can be satisfied under different coding and modulation modes. Specifically, it is assumed that the total number of contents is M, and there are K types of modulation and coding methods (corresponding to the coding and modulation methods with spectral efficiency from high to low), in which users with good channels can decode and transmit to users with poor channels. Information. The base station side maintains an M×K request matrix, where the matrix elements refer to the number of users that can be served under the jth coding and modulation scheme for the ith content, denoted as Qi _,j (t).

In the step 3: the way of obtaining the future popularity of the content can be predicted by the server's statistics of historical request data of the content or the recent search popularity of the content; the user cache status and the number of connections within the coverage can be obtained in various ways: a) through the user application Layer software statistics are fed back to the base station; b) the base station caches the number of user terminals within the coverage of user access information/user access base station signaling. According to the future heat and user cache status, we can calculate the expected number of future services N _i,j (t) that uses j-coded modulation to schedule content i. Specifically, assuming that the scheduling content i adopts the j-coded modulation method, the expected number of future services that can be obtained is where U is the set of users, p _k,i (t) is the probability that user k requests content i in the future, and I _k,i,j (t) indicates that user k can successfully buffer the air interface and transmit content i in j-coded modulation mode , which is determined by the remaining buffer space and the current channel state. The base station integrates the current number of requested users Q _i,j (t) and the number of potential future services N _i,j (t), and considers the scheduling resource overhead to calculate the scheduling weight of the scheduling content i of the j-coded modulation scheme The numerator term is the potential benefit of sending content i by using the coding modulation method with spectral efficiency c _j , including the number of users that can be served at the current scheduling moment and the number of users that can be scheduled in the future, where ρ∈[0,1] is the weight factor, which is used to measure the weight ratio of the number of future services in the current service revenue, which can be determined by service attributes and network load; the denominator term is the bandwidth-time product resource consumed by the jth coding and modulation mode c _j scheduling, where S _i is the transmission file size of the scheduling content i; the physical meaning of w _i,j (t) is the scheduling benefit that can be obtained under the unit resource consumption at this time, and ρ=1 is the expected number of users that can be served under the unit resource consumption.

In the step 4: the base station will Sort from high to low and schedule them in turn. Assuming that the user terminal does not have storage space and does not recognize the requested content ID, the scheduling criterion is user-centered, and the weight of the j-th user (spectral efficiency is c _j ) is scheduled at this time. That is, it degenerates into the maximum rate scheduling criterion.

In described step 5: at the moment of requesting content scheduling, if Q _i,j (t) > 1, then c _j is the multicast service requesting user group, and the fountain code or the user terminal request retransmission mode can be used to restore the error packet; At the time of content scheduling, the base station will first transmit the content identification ID. For users with good channels within the coverage area, they can decode the content identification ID. At the same time, the application software predicts the future access probability of this content through the user's historical request information, and comprehensively considers factors such as terminal power and remaining storage space to decide whether to receive this content. If it is determined to cache the content, the base station side will be notified of the cache status when interacting with the base station next time, and the base station side will update the maintained user cache status at the same time. At the same time, since the unsolicited users with good channels are not the service subjects of this scheduling, they just piggyback and buffer the content transmitted by the receiving base station to users with poor channels by using the broadcast characteristics of wireless transmission. Therefore, considering the fading of the wireless channel and the mobility of the user, even if there is an error packet at this time, it will not be obtained through retransmission immediately, but only when the user clicks to watch the cached content with lost packets in the future. The base station side performs data recovery.

Compared with the prior art, the present invention has the following beneficial effects

1. Higher network utilization efficiency: Different from the traditional physical layer resource allocation, the proposed invention can make use of the content popularity attribute, comprehensively consider the number of currently scheduled users and the number of potential future services, and serve current user requests without introducing transmission overhead. It can serve more users in the case of scheduling resources, that is, it realizes the maximum expected return under the unit consumption of resources when scheduling resources. Since users with good channels can cache, repeated transmission of hot content is avoided, and content transmission efficiency is improved.

2. Better user experience: Compared with the prior art, without introducing additional transmission overhead, it enables unrequested users with good channels to cache hot content, and when they click on this hot content, it can be played directly locally , which can bring a better user experience to the user.

Description of drawings

Figure 1 is a schematic diagram of the system;

FIG. 2 is a flowchart of a content-aware-based resource scheduling method according to the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings.

Figure 1 is a schematic diagram of the system. As shown in the figure, two users, UE1 and UE5, send content requests to the base station at a certain moment. It is assumed that the base station can only schedule one user in this time slot. Among them, UE1 is located near the base station and requests non-hot content; UE5 is located at the edge of a cell and requests hot content. According to the traditional maximum rate scheduling criterion, the UE1 in a good channel will be preferentially scheduled, ignoring the content popularity attribute of the user. If the content popularity attribute is considered, because UE5 requests hot content, by scheduling UE5, idle users (UE1 to UE4) with good remaining channels in the cell who have not sent data requests can also cache the hot content, so that when other users When this hot content data request is generated, it can be directly satisfied locally. In this example, by scheduling edge users requesting high-popularity content, most users with good channels can piggyback and cache the hot content, avoiding sending it when re-requesting. Different user requests at different times are concentrated in one multicast transmission, which improves the transmission efficiency of the network. At the same time, since the hot content can be piggybacked and cached on the user terminal with a good channel, when the user terminal with the hot content cached generates a data demand, it can be played directly locally and smoothly, which improves the user experience.

2 is a flowchart of the content-aware resource scheduling method of the present invention. As can be seen from the figure, first, the user generates a content request, and checks whether the requested content has been completely stored locally, and if so, browse and watch directly locally. If not, the user reports the content to the base station, which includes the following steps:

Step 1: The base station counts the content ID and related channel state information requested by each requesting user. As shown in FIG. 2 , UE1 with good channel requests non-hot content F_i, and UE5 with poor channel at the cell edge requests hot content F_j.

Step 2: The base station maintains the corresponding request queue with the content ID as an identifier, and separately calculates the number of users with current service requests that can be satisfied by using different coding and modulation modes under different request contents. For the convenience of explanation, considering large-scale fading, it is assumed that there are three different coding and modulation methods, and the corresponding spectral efficiencies from high to low are c ₁ , c ₂ and c ₃ respectively. Among them, UE1 and UE2 that serve the innermost channel best use the spectrum. The most efficient coding and modulation mode of c ₁ ; serving UE3 and UE4 need to use the coding and modulation mode of frequency efficiency c ₂ , and the service edge user UE5 needs to use the coding and modulation mode of frequency efficiency c ₃ .

Step 3: The base station calculates the weight of the content hotspot according to the future popularity of the content and the user cache status and the number of connections within the coverage area, and adds the content hotspot weight to the scheduling criterion. Assuming that the size of the files to be transmitted for F_j and F_i is the unit length, and UE1 to UE4 have a 50% chance to access the content F_j in the future, and all have excess cache space and sufficient power, the potential future of the content F_j is scheduled in the c ₃ method. The number of service users is 4×0.5=2. Assuming that the number of future service users and the current number of users have the same weight, that is, ρ=1, the expected benefit of scheduling content F_j in the c ₃ method is 3, that is, the total expected number of service users, corresponding to The scheduling weight is 3c ₃ ; the expected number of people to serve the content F_i in the c ₁ mode is 1, and the corresponding scheduling weight is c ₁ .

Step 4: Sort the contents according to the scheduling weight, and install the preset coding and modulation modes to send the contents from high to low according to the scheduling weight. The scheduling weight of the content F_i scheduled in the c ₁ mode is c ₁ , and the scheduling weight of the content F_j in the c ₃ mode is 3c ₃ . Assuming that c ₁ <3c ₃ , at this time, the edge user UE 5 that requests the hot content F_j should be scheduled preferentially. However, in the traditional maximum rate scheduling criterion, it prioritizes users with good channels, and does not pay attention to the difference in the popularity of the requested content.

Step 5: The user requesting the scheduled content receives and decodes the air interface data. For users with good channels that do not send requests, high-interest content on the air interface can be selectively cached. The content F_j is sent in a coded modulation scheme with spectral efficiency c ₃ , the requesting user UE5 on the edge receives the content, and the request is satisfied. At the same time, unrequested users with good channels can also successfully decode data, such as UE1 to UE4, and determine whether to receive the cache according to their own historical access requests, remaining storage space and power and other factors. Assuming that a packet error occurs in the receiving process (such as fast channel fading, user movement), the data is not immediately requested for retransmission recovery, but the data is recovered when the user clicks to view the request next time.

The wireless transmission point side of the present invention comprehensively considers the wireless channel information of the information receiving point, the content popularity attribute and the buffering capability of the client to perform resource allocation. Unrequested users with good channels can receive and cache the hot content transmitted by the air interface of the base station to requesting users with poor channels; whether to cache is determined by their own historical request information, remaining power and storage space.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should The solution can be modified or equivalently replaced without departing from the spirit and scope of the technical solution of the present invention, and it should be included in the scope of the claims of the present invention.

Claims (8)

1. a kind of resource regulating method based on perception of content, which is characterized in that this method comprises the following steps:

Step 1: transmitting terminal counts each receiving end institute request content ID and associated channel condition information；

Step 2: transmitting terminal is that mark maintains corresponding request queue, and calculates separately and use under different request contents with content ID The number of users for the current service request that different coding modulation system can satisfy；

Step 3: transmitting terminal according to user cache state in the content future temperature of prediction, coverage area with connect number, calculate Currently dispatch the future potential service number of users of certain content with certain code modulation mode, and by this potential service number of users It is added in dispatch weight；

Step 4: according to dispatch weight by content ordering, according to dispatch weight by content from high to low according to preset coded modulation Mode is sent；

Step 5: receiving end receives and decodes data of eating dishes without rice or wine, alternative for the good user of the channel of not sent request Caching eat dishes without rice or wine high attention rate content.

2. the resource regulating method according to claim 1 based on perception of content, which is characterized in that in the step 1: connecing When receiving end generates content requests, first check for whether local storage space completely caches, if any then local to play；Conversely, will be interior Hold request and reports transmitting terminal.

3. the resource regulating method according to claim 1 based on perception of content, which is characterized in that the step 2 is specific It is:

Total content number is M, and alternative modulation coding scheme is K kind, and wherein the good receiving end of channel can decode transmission To the information of the receiving end of bad channel；Transmitting terminal maintains M × K request matrix, and wherein matrix element is that i-th of content is used The number of users that can be serviced under jth kind coded modulation mechanism, is denoted as Q_i,j(t), t is scheduling instance.

4. the resource regulating method according to claim 1 based on perception of content, which is characterized in that pre- in the step 3 The acquisition modes of the content future temperature of survey are by server statistics content utilization request data or to pass through server statistics Content searches for temperature in the recent period.

5. the resource regulating method according to claim 1 based on perception of content, which is characterized in that covered in the step 3 User cache state includes: with the acquisition modes for connecting number within the scope of lid

A) transmitting terminal is fed back to by receiving end application layer software statistics；

B) transmitting terminal is according to user's history access information counting user buffer status；

C) the user terminal number with spatial cache in transmitting terminal statistics access transmitting terminal signaling coverage area.

6. the resource regulating method according to claim 1 based on perception of content, which is characterized in that sent out in the step 3 End is penetrated according to content future temperature, the user cache state of prediction, calculates the future for using j code modulation mode Scheduling content i Desired service number N_i,j(t), specifically Scheduling content i uses j code modulation mode, the following desired service number that can be obtained Mesh isWherein U is user's set, p_k,i(t) be user's k further request content i probability, I_k,i,j(t) what is shown is that user k can be cached successfully and be eated dishes without rice or wine with the mark of j code modulation mode transferring content i, slow by residue It deposits space and is determined with channel status is presently in.

7. the resource regulating method according to claim 6 based on perception of content, which is characterized in that in the step 3:

Transmitting terminal integrates current request number of users Q_i,j(t) with the following potential service number N_i,j(t), and consider that dispatching resource opens Pin calculates the dispatch weight of j code modulation mode Scheduling content i

Wherein, molecule item is to use spectrum efficiency for c_jCode modulation mode send the obtained potential income of content i, including work as The number of users at the time of number of users and future that preceding scheduling instance can service can be dispatched, wherein ρ ∈ [0,1] be weight because Son is determined for measuring future service number in the weight proportion of current service income by service attribute, network load；Point Female item is with jth kind code modulation mode c_jLower consumed bandwidth time product resource is dispatched, wherein S_iFor the biography of Scheduling content i Defeated file size；w_i,j(t) physical significance is the lower scheduling benefits that can be obtained of unit resource consumption at this time, is in ρ=1 The lower expectation number of users that can be serviced of unit resource consumption.

8. the resource regulating method according to claim 3 based on perception of content, which is characterized in that in the step 5:

In request content scheduling instance, if the number of users Q serviced_i,j(t) 1 >, then modulation system c_jFor multicast service User group is requested, restores accidentally to wrap using fountain codes or user terminal request retransmission mode；In content scheduling instance, transmitting terminal can be first Transferring content identifies ID；

Good for the channel in coverage area does not request user, can decode content identification ID, while application software passes through use This content future access probability of family historical requests information prediction, and considered according to terminal power and residual memory space combined factors Decide whether to receive this content；If it is determined that caching this content, then buffer status is accused when being interacted with transmitting terminal next time Know transmitting terminal, while transmitting terminal updates the user cache state maintained.