Base station content caching and collaborative content pre-fetching method based on maximizing throughput
By building a model that jointly considers Doppler effect and free space loss in the high-speed rail communication system, optimizing the base station and user cache strategies, the frequent connection switching and backhaul link congestion caused by high-speed rail trains are solved, and efficient content is achieved Cache and prefetch significantly improve user request response speed and network throughput.
Patent Information
- Application Number
- CN202210194211.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-01
- Publication Date
- 2024-11-01
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The high-speed movement of high-speed trains causes frequent switching between users and base stations, and short stay time, resulting in problems such as congestion of backhaul links and slow user request response.
Adopt the base station content cache and collaborative content prefetching method based on maximizing throughput, and by building a high-speed rail communication model that jointly considers Doppler effect and free space loss, optimize the base station and user cache strategies to maximize the network throughput that the base station can actually provide The greedy algorithm is used to solve the optimal cache placement.
It significantly reduces user request delay, improves file hit rate and network throughput, saves cache space, and improves system average delay and cache space utilization.
Smart Images

Figure CN114584953B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of high-speed rail communications, and relates to a base station content caching and collaborative content pre-fetching method based on a maximum available throughput. Background Art
[0002] With the continuous development of mobile networks, people tend to have greater bandwidth requirements while traveling on high-speed trains, and use mobile devices to request video content. However, when riding on a high-speed train, since the high-speed train moves at a high speed of nearly 300km / h, users will quickly cross each cell when requesting content, constantly establishing and disconnecting connections with base stations, and the requested data status is unstable; the path loss in free space and the Doppler frequency shift caused by high-speed movement will have a certain degree of impact on the quality of network signals, resulting in poor user service quality. During peak data traffic periods, the backhaul link is overloaded, the data center cannot normally meet user requests, and the same data is repeatedly requested multiple times, increasing the communication signaling overhead and also increasing the network burden on the data center.
[0003] Using cache at the wireless edge can effectively handle problems such as huge traffic of video playback, repeated request content, unstable transmission channels, etc. Predictive caching of user requests during off-peak hours allows users to directly request data locally during peak traffic hours. Since the requested content is sent directly from a server close to the user end, no backhaul cost is required. This method can relieve the pressure on the backhaul link, reduce network congestion, and reduce service latency. Pre-acquiring data is also a way to improve user service quality. The user's next request location is predicted based on the user's activity trajectory, and the base station at that location pre-fetches data from the data center based on the current request content.
[0004] For users moving at high speeds, the Doppler effect cannot be ignored. Currently, most communication models hope to eliminate this effect by acting on the physical layer. However, this method is relatively complicated and cannot completely offset this effect. In particular, in the future, when the operating frequency of base stations and the speed of trains are greatly increased, this effect will become more obvious.
[0005] Therefore, a new caching method is urgently needed to solve the problems of frequent switching of users' connections with base stations and short stay time due to the high mobility characteristics of high-speed rail, so as to alleviate the problems of backhaul link congestion and slow response to user requests. Summary of the invention
[0006] In view of this, the purpose of the present invention is to provide a base station content caching and collaborative content pre-fetching method based on maximizing throughput, so as to solve the problems of frequent switching of user connections with base stations and short stay time due to the high mobility characteristics of high-speed rail. The present invention is based on base station collaborative content pre-fetching and content caching solutions to adaptively meet the diverse service needs of users.
[0007] In order to achieve the above object, the present invention provides the following technical solutions:
[0008] A method for base station content caching and collaborative content pre-fetching based on maximizing throughput specifically comprises the following steps:
[0009] S1: Construct a high-speed railway communication model that jointly considers the Doppler effect and free space loss. It is a model for independent content caching and collaborative pre-fetching by track base stations (hereinafter referred to as base stations) during the operation of high-speed railway (HSR), including the HSR system model, the file popular distribution cache model and the system motion model.
[0010] S2: Construct cache placement optimization problem, that is, construct objective function: minimize user request delay as optimization goal;
[0011] S3: The optimization problem is divided into two parts: base station cache optimization and user cache optimization. That is, the objective function is divided into two sub-functions: maximizing the network throughput that the base station can actually provide and maximizing the user cache under the constraints of the base station cache. The two sub-functions are solved separately.
[0012] Furthermore, in step S1, the constructed HSR system model refers to a system in which the base station responds to user requests through content caching and content pre-fetching during the operation of the high-speed rail, and is composed of 1 macro base station (equivalent to a data center in the present invention), b track base stations (the base station set is represented by B = {1,2,…,b}) and a train; in order to alleviate the problem of communication signal penetration loss, the present invention deploys a mobile relay (MR) on the top of the train to collect K user (the user set is represented by K = {1,2,…,K}) requests and forward them to the track base station; the macro base station contains all possible user request files F = {1,…,f,,F}, assuming that the file sizes are the same, all Qbit, and the request probability obeys the zipf distribution. The file f is divided into N segments (represented by the segment set N = {1,2,…,N}) using MDS coding; the base stations considered in the present invention do not overlap in coverage on the train track, so there is no connection selection and interference problem. The base station and the user have a certain cache capacity, and the cache space is represented by SS1 = {S1,…,S i ,…,S b}、SS2={S1,…,S k ,…,S K}, where S i represents the cache space of the ith orbital base station, S k Represents the cache space of the kth user; the caches between base stations and users are independent of each other. The user caches after the base station and does not cache the same file segment as the base station.
[0013] Furthermore, in step S1, the cache model based on file popularity distribution is constructed, including: file popularity distribution is the user's preference for files. Generally, user preferences for files vary from person to person, but for all users, zipf distribution can roughly evaluate the popularity of a file. The probability of a user requesting the mth file f can be expressed as:
[0014]
[0015] Among them, α is the skewness factor of the zipf distribution, and m is the file number.
[0016] When a user requests content: first, it will search its own cache library and respond directly to the cached content; otherwise, the user will aggregate the request interest package (file index) to MR, and MR will forward the user request to the track base station; after the track base station receives the user's request interest package, it will check the local cache; if the content corresponding to the interest package already exists locally (cached or pre-fetched, in this invention, it is assumed that the base station will immediately discard the pre-fetched content once it is disconnected from the user, and does not occupy the cache space), the track base station will directly feedback the user's request content to MR, and MR will forward it to the corresponding user; otherwise, the track base station forwards the request interest package to the macro base station M, and the track base station downloads the user's request content through the backhaul link and hands it to MR (in this invention, it is assumed that sufficient and stable communication bandwidth can be provided between the user and MR, and its propagation delay will be ignored in subsequent research). Due to the processing delay and backhaul delay of the macro base station, the end-to-end delay from the macro base station to the user must be longer than the time it takes for the user to obtain the content locally from the track base station, and the link between the macro base station and the user is a multi-hop link, which is vulnerable to security threats. In order to improve user experience and enhance data security, more high-utilization content needs to be cached locally.
[0017] Furthermore, in step S1, the system motion model constructed specifically includes: given the train moving speed v and time t, the train travel distance is s(t)=v*t, given the distance between the base station and the train track is d, and the coverage range of the i-th base station is D i , then the connection range between base station i and the train is The relative travel distance of the train in the coverage area of base station i at time t is The distance between the train and the base station is The train moves at high speed relative to the base station, generating a Doppler frequency shift f D (t):
[0018]
[0019] Wherein, f' represents the base station operating frequency, c = 3*10^8 (m / s), and c represents the speed of light; represents the angle between the train running direction and the base station, then the actual operating frequency of the base station is f i' (t) = f' - fD (t);
[0020] When the rail base station communicates with the user as a signal transmitter, the signal quality will inevitably attenuate due to the diffusion of radiation energy. For simplicity, the present invention takes free space propagation loss as an example:
[0021] p iL (t) = 32.45 + 20logf i' (t)(Mhz)+20logd2 i (t)(km) (3)
[0022] The user signal-to-noise ratio at time t can be obtained by the following formula:
[0023]
[0024] Among them, δ 2 represents the Gaussian white noise variance, p ir (t) represents the user received power, p i represents the transmission power of the ith base station. The rate R that the orbital base station can provide to the user i (t) is:
[0025] R i (t) = Blog(1 + SNR i (t)) (5)
[0026] Where B is the channel bandwidth, SNR i (t) represents the user signal-to-noise ratio at time t.
[0027] Further, in step S2, a cache placement optimization problem is constructed, specifically including: firstly constructing a cache placement problem based on minimizing system delay: defining is an element of the track base station cache strategy C1; if This means that the nth segment of file f is in the cache of track base station i, otherwise, the segment is not cached in track base station i. Similarly, the elements of user cache strategy C2 Indicates the cache status of user k for the nth segment of file f; when It means that the nth segment of file f has been cached locally in user k, otherwise, the segment of file f is not cached in user k; Under the initial cache, the user and the orbital base station independently cache the content according to the distribution of file zipf under the constraints of their respective cache spaces; When the user requests file f, once all resources related to file f (pre-fetched and cached) in the orbital base station are fully transmitted to the user, the user is still in the coverage area of the current base station, and the orbital base station needs to request the relevant content from the macro base station (data center); During the whole process, if the wireless link rate R i Greater than the return link rate R m, the actual file transfer rate will change from R i Reduce to R m Otherwise, the user will use the rate R i Receive the file; since the high-speed rail stays in the coverage area of each base station for a short time, but the cache strategy of each track base station is relatively stable and does not change instantaneously over time, the track base station can provide the average rate to solve the cache (the track base station rates mentioned below refer to the average rate):
[0028]
[0029] The actual network throughput that track base station i can provide for file f can be expressed as:
[0030]
[0031] The first item represents the sum of pre-fetched and cached data of orbital base station i, and the second item represents the maximum network throughput that the orbital base station can provide; Indicates the time the user stays in the coverage area of orbital base station i; represents the data size of cache file f at track base station i; defines the user request service rate as r q , the macro base station provides a service rate R m ;
[0032] The total delay of the entire system, i.e. the optimization problem, can be expressed as:
[0033]
[0034] Where A1 represents the total delay caused by requesting file f on the base station, and A2 represents the delay that can be compensated by the user by caching file segments of f; the delay of requesting file f can be expressed as t f =A2-A1; q(f) represents the probability that the user requests file f.
[0035] From the optimization function, we can see that: when r q ≤min{R m ,R i}, the user's requested content can be served without delay; otherwise, the delay function shows that to minimize the system delay, the optimization can be considered from the user cache and the base station cache. Then the optimization problem can be divided into two sub-functions.
[0036] Further, in step S3, the base station cache optimization part: by optimizing the base station cache strategy, the rail base station provides the maximum network throughput to the terminal users connected within its coverage area, specifically through the following OP1 function optimization:
[0037]
[0038] The constraints are:
[0039]
[0040] Among them, st1 is the track base station cache space constraint, st2 is the track base station cache placement binary selection, and st3 represents the data size of the cache file f of the track base station i.
[0041] Furthermore, in step S3, a greedy algorithm is used to solve the optimal cache of the base station.
[0042] Further, in step S3, the user cache optimization part: let the user maximize the user cache data without repeatedly caching the file segments cached by the track base station, so as to maximize the user cache compensation delay, which is specifically optimized by the following OP2 function:
[0043]
[0044] The constraints are:
[0045]
[0046] Among them, st1 is the user cache space constraint, st2 is the track base station cache constraint under file diversity, st3 represents the Boolean cache decision of user k for the nth segment of file f, and st4 represents the cache data of user k for file f.
[0047] Further, in step S3, solving the objective function OP2 of user cache optimization includes: independently optimizing caches between users can be decoupled into K parallel sub-functions OP2.1:
[0048]
[0049] Constraints:
[0050] s.t1, s.t2, s.t3, s.t4
[0051] s.t5 And q'(f)<0
[0052] It can be seen from constraint condition st5 that the optimization of user cache strategy is a simple space allocation problem with different weights. Since the cache space is limited, when the user cache space cannot completely make up for the delay caused by base station transmission, the user needs to cache the file segments that the base station cannot satisfy the user's request file in order of file popularity to maximize the cache benefit.
[0053] It means that the current orbital base station can meet the user's demand for file f, so the user does not need to cache; there are:
[0054]
[0055] It means that the current track base station cannot meet the user's demand for file f; if the remaining amount of user cache space is sufficient to cache a segment of the file, and the track base station has not cached the segment, the user caches the segment, otherwise, the user does not cache the segment.
[0056]
[0057] Then the objective function OP solves the optimal cache placement through two sub-objective functions respectively, and the optimal cache placement of the system can be obtained.
[0058] The beneficial effects of the present invention are:
[0059] (1) The present invention considers a cache model in a mobile high-speed rail scenario where multiple base stations with certain cache capabilities cover the area. In order to better provide services to traveling users, the base stations can collaborate to pre-fetch user request data.
[0060] (2) The present invention proposes a system for minimizing user request delays, taking into account the Doppler shift generated during the movement of high-speed rail and the signal loss from the rail base station transmitter to the user terminal. In order to ensure full utilization of user cache space and ensure file diversity, users do not cache the same file segments as the base station.
[0061] (3) The present invention divides the minimization of user request delay into two parts: base station and user. The base station can be transformed into maximizing the actual network throughput provided by the base station, and a greedy algorithm is used to solve it. The present invention has significant improvements in system average delay, file hit rate, base station throughput and cache space occupancy.
[0062] (4) The present invention can not only provide more network throughput, significantly reduce system latency, effectively improve file hit rate, but also save cache space.
[0063] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:
[0065] Figure 1 Collaborative prefetching of content structures for base station caches in high-speed rail environments;
[0066] Figure 2 Request a file model for the user;
[0067] Figure 3 is the relative position of the high-speed rail motion model and the base station. DETAILED DESCRIPTION
[0068] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.
[0069] See also Figure 1 to Figure 3 The present invention provides a base station content caching and collaborative content pre-fetching method based on maximizing throughput, comprising the following steps:
[0070] Step 1: Establish a high-speed rail high-speed mobility model, and use different base station coverage and train running speed to build a relative motion model between users and base stations. The entire model mainly consists of three parts: HSR system model, file popular distribution cache model, and system motion model.
[0071] 1) HSR system model
[0072] The HSR system model refers to the framework in which the base station responds to user requests through content caching and content pre-fetching during the operation of the high-speed rail. Figure 1 . The entire model consists of 1 macro base station (equivalent to a data center in the present invention), b track base stations (the base station set is represented by B = {1,2,…,b}) and a train; in order to alleviate the problem of communication signal penetration loss, the present invention deploys a mobile relay (MR) on the top of the train to collect K user (user set is represented by K = {1,2,…,K}) requests and forward them to the track base station; the macro base station contains all possible user request files F = {1,…,f,…,F}, assuming that the file sizes are the same, all Qbit, and the request probability obeys the zipf distribution. MDS coding is used to divide the file f into N segments (represented by the segment set N = {1,2,…,N}); the base stations considered in the present invention do not overlap in coverage on the train track, so there is no connection selection and interference problem. The base station and the user have a certain cache capacity, and the cache space is represented by SS1 = {S1,…,S i ,…,S b}、SS2={S1,…,S k ,…,S K}, where S irepresents the cache space of the ith orbital base station, S k Represents the cache space of the kth user; the caches between base stations and users are independent of each other. The user caches after the base station and does not cache the same file segment as the base station.
[0073] 2) Based on the file popularity distribution cache model
[0074] File popularity distribution refers to the user's preference for a file. Usually, user preferences for files vary from person to person, but for all users, the zipf distribution can roughly evaluate the popularity of a file. The probability that a user requests the mth file f can be expressed as:
[0075]
[0076] Among them, α is the skewness factor of the zipf distribution, and m is the file number.
[0077] When a user requests content: first, it will search its own cache library and respond directly to the cached content; otherwise, the user will aggregate the request interest package (file index) to MR, and MR will forward the user request to the track base station; after the track base station receives the user request interest package, it will check the local cache. If the content corresponding to the interest package already exists locally (cached or pre-fetched, in the present invention, it is assumed that the base station will immediately discard the pre-fetched content once it is disconnected from the user, and does not occupy cache space), the track base station will directly feed back the user's requested content to MR, and MR will forward it to the corresponding user. Otherwise, the track base station forwards the request interest package to the macro base station M, and the track base station downloads the user's requested content through the return link to MR (in the present invention, it is assumed that sufficient and stable communication bandwidth can be provided between the user and MR, and its propagation delay will be ignored in subsequent research). Due to the processing delay and return delay of the macro base station, the end-to-end delay from the macro base station to the user must be longer than the time it takes for the user to obtain the content locally from the track base station, and the link between the macro base station and the user is a multi-hop link, which is vulnerable to security threats. In order to improve user experience and enhance data security, more high-utilization content needs to be cached locally. The process of user k requesting file f is as follows Figure 2 .
[0078] 3) This embodiment considers the following mobility model: given the train speed v and time t, the train travel distance s(t) = v*t, given the distance between the base station and the train track d, the coverage range of the i-th base station is D i , then the connection range between base station i and the train is The relative travel distance of the train in the coverage area of base station i at time t is The distance between the train and the base station is
[0079] The train moves at high speed relative to the base station, generating a Doppler frequency shift f D (t):
[0080]
[0081] Wherein, f' represents the base station operating frequency, c = 3*10^8 (m / s), and c represents the speed of light; represents the angle between the train running direction and the base station, then the actual operating frequency of the base station is f i' (t) = f' - f D (t);
[0082] When the rail base station communicates with the user as a signal transmitter, the signal quality will inevitably attenuate due to the diffusion of radiation energy. For simplicity, the present invention takes free space propagation loss as an example:
[0083] p iL (t) = 32.45 + 20logf i' (t)(Mhz)+20logd2 i (t)(km) (3)
[0084] The user signal-to-noise ratio at time t can be obtained by the following formula:
[0085]
[0086] Among them, δ 2 represents the Gaussian white noise variance, p ir (t) represents the user received power, p i represents the transmission power of the ith base station. The rate R that the orbital base station can provide to the user i (t) is:
[0087] R i (t) = Blog(1 + SNR i (t)) (5)
[0088] Where B is the channel bandwidth, SNR i (t) represents the user signal-to-noise ratio at time t.
[0089] Step 2: Build a cache placement plan;
[0090] In view of the problems of Doppler frequency shift and frequent connection switching in high-speed mobile scenarios, the present invention applies edge caching technology to the HSR communication model based on the collaborative pre-fetching of content by adjacent base stations, aiming to reduce user request delay and improve user service quality. Based on this, the present invention proposes to minimize system delay by optimizing the placement of base stations and user caches.
[0091] 1) Build cache placement optimization problem
[0092] First, we construct a cache placement problem based on minimizing system latency. Definition is an element of the track base station cache strategy C1; if This means that the nth segment of file f is in the cache of track base station i, otherwise, the segment is not cached in track base station i. Similarly, the elements of user cache strategy C2 Indicates the cache status of user k for the nth segment of file f; when It means that the nth segment of file f has been cached locally in user k, otherwise, the segment of file f is not cached in user k; Under the initial cache, the user and the orbital base station independently cache the content according to the distribution of file zipf under the constraints of their respective cache spaces; When the user requests file f, once all resources related to file f (pre-fetched and cached) in the orbital base station are fully transmitted to the user, the user is still in the coverage area of the current base station, and the orbital base station needs to request the relevant content from the macro base station (data center); During the whole process, if the wireless link rate R i Greater than the return link rate R m , the actual file transfer rate will change from R i Reduce to R m Otherwise, the user will use the rate R i Receive the file; since the high-speed rail stays in the coverage area of each base station for a short time, but the cache strategy of each track base station is relatively stable and does not change instantaneously over time, the track base station can provide the average rate to solve the cache (the track base station rates mentioned below refer to the average rate):
[0093]
[0094] The actual network throughput that track base station i can provide for file f can be expressed as:
[0095]
[0096] The first item represents the sum of pre-fetched and cached data of orbital base station i, and the second item represents the maximum network throughput that the orbital base station can provide; Indicates the time the user stays in the coverage area of orbital base station i; represents the data size of cache file f at track base station i; defines the user request service rate as r q , the macro base station provides a service rate R m ;
[0097] The total delay of the entire system, i.e. the optimization problem, can be expressed as:
[0098]
[0099] Where A1 represents the total delay caused by requesting file f on the base station, and A2 represents the delay that can be compensated by the user by caching file segments of f; the delay of requesting file f can be expressed as t f =A2-A1; q(f) represents the probability that the user requests file f.
[0100] From the optimization function, we can see that: when r q ≤min{R m ,R i}, the user's requested content can be served without delay; otherwise, the delay function shows that to minimize the system delay, the optimization can be considered from the user cache and the base station cache. Then the optimization problem can be divided into two sub-functions.
[0101] 2) Base station cache placement
[0102] For base stations: Base station cache strategies can be optimized to allow base stations to provide maximum network throughput to end users connected within their coverage area. Base station cache placement can be optimized through the OP1 function:
[0103]
[0104] The constraints are:
[0105]
[0106] Among them, st1 is the track base station cache space constraint, st2 is the track base station cache placement binary selection, and st3 represents the data size of the track base station i cache file f; From OP1, it can be seen that when (ΔT i +ΔT i-1 )R m -R i ΔT i ≥0, In other words, when the rate of the macro base station is higher than that of the orbital base station, When the base station does not cache, the actual network throughput can reach the maximum value that can be provided. This paper mainly considers the ratio of macro base stations and orbital base stations. Base station cache placement strategy under the condition of .
[0107] When the base station cache is large enough, the base station cache does not need to consider the probability of user request files (characterized by file popularity in the present invention) when providing local services for all user requests through content pre-fetching and content caching. All files are cached indiscriminately.
[0108]
[0109] However, the cache space of the base station is limited relative to the number of files. Therefore, the following will consider that the base station provides local cache for all files, and it is necessary to use the file request probability to cache the file segments with higher utilization. Analysis of problem OP1 shows that it can be regarded as a monotone submodular function maximization problem under pseudo-matrix constraints. According to the characteristics of the objective function, the present invention proposes to use a greedy algorithm to solve the optimal cache of the base station.
[0110] First, define the base station cache placement location set as ξ:
[0111]
[0112] ξ can be divided into b independent subsets ξ 1 ,ξ 2 ,…,ξ b ,in Can be divided into F independent subsets in Represents all possible base station cache strategies C1 placed on the track i The set of file segments on ;
[0113] It can be proved that the orbital base station cache space constraint is a partitioned pseudo-matrix of the cache position ξ of the orbital base station, and the objective function OP1 is a monotone submodular function. Based on the properties of the objective function, a greedy algorithm is used to solve the optimal cache of the base station.
[0114] 3) User cache placement
[0115] For user cache, it is necessary to maximize the user cache data without repeatedly caching the file segments that the base station has cached, so as to maximize the user cache compensation delay. User cache placement can be optimized through the OP2 function:
[0116]
[0117] The constraints are:
[0118]
[0119] Among them, st1 is the user cache space constraint, st2 is the track base station cache constraint under file diversity, st3 represents the Boolean cache decision of user k for the nth segment of file f, and st4 represents the cache data of user k for file f.
[0120] For the objective function OP2, including: independent optimization cache between users can be decoupled into K parallel sub-functions OP2.1:
[0121]
[0122] Constraints:
[0123] s.t1, s.t2, s.t3, s.t4
[0124] s.t5 And q'(f)<0
[0125] It can be seen from constraint condition st5 that the optimization of user cache strategy is a simple space allocation problem with different weights. Since the cache space is limited, when the user cache space cannot completely make up for the delay caused by base station transmission, the user needs to cache the file segments that the base station cannot satisfy the user's request file in order of file popularity to maximize the cache benefit.
[0126] It means that the current orbital base station can meet the user's demand for file f, so the user does not need to cache; there are:
[0127]
[0128] It means that the current track base station cannot meet the user's demand for file f; if the remaining amount of user cache space is sufficient to cache a segment of the file, and the track base station has not cached the segment, the user caches the segment, otherwise, the user does not cache the segment.
[0129]
[0130] Then the objective function OP solves the optimal cache placement through two sub-objective functions respectively, and the optimal cache placement of the system can be obtained.
[0131] Embodiment 1:
[0132] This embodiment provides a cache placement method, in which the base station and the user cache content according to zipf in the initial cache. The cache strategies are defined as C1 and C2 respectively, and SS1 = {S1, ..., Si, ... Sb} and SS2 = {S1, ..., Sk, ... SK} represent the base station and user cache space. The base station coverage can be calculated according to the base station transmission power and the minimum communication SNR.
[0133] 1) Update the base station cache.
[0134] (1) Base station content pre-fetching (base stations communicate in full-duplex mode: they communicate with the red base station as a receiver while responding to the user). The train enters the current base station and sends a request interest packet (the index of the file segment that needs the base station to respond) to the base station. The base station is recorded as the first base station i through which the user requests the file. The base station stands immediately according to the actual network throughput that can be provided. Compared with the user demand index, once it is found that the actual network throughput cannot meet the user demand, the processed user demand index (the difference between the user demand and the demand processed by base station i, that is, the file segment that needs to be requested from the next base station) is immediately sent to the next base station i+1. After base station i+1 receives the demand index (at this time, the user is still in the coverage area of base station i), the base station first checks the local cache if Base station i+1 does not perform any operation and directly responds to the user after the user moves to its coverage area; otherwise, base station i+1 responds to the user within ΔT. i Pre-fetch content from the macro base station within a certain time period, and calculate the actual network throughput that can be provided when the user enters base station i+1 Then the demand is compared with the actual network throughput. If the actual throughput of base station i+1 is greater than the user demand, the user is directly responded to; otherwise, the base station will forward the processed user demand index to the next base station while responding to the user.
[0135] (2) Evaluate cache benefits. When the cache space of the base station is limited, it is impossible to provide all files with zero-delay service. It is necessary to evaluate the cache benefit of each file segment to maximize its cache utilization.
[0136] (3) Implement caching. Under the current caching state, calculate the difference between the maximum throughput that the base station can provide and the actual network throughput of each file. Multiply it by the file popularity to approximate the benefits that can be brought by caching file segments of each file under the current cache. Sort them in descending order as the cache priority. When the base station has remaining cache space to cache file segments, cache the file segments according to the priority.
[0137] 2) Update the user cache.
[0138] The user cache strategy is used as a supplementary cache after all base station caches are updated.
[0139] For files whose actual throughput still cannot reach the maximum throughput after the base station cache is updated, users use file probability as cache priority and give priority to caching file segments that are highly popular and not cached by the base station.
[0140] After the base station and user caches are updated, the current cache placement is the optimal placement of the file cache at this time, which can minimize the user request response delay in the system.
[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.
Claims
1. A method for base station content caching and collaborative content pre-fetching based on maximizing throughput, characterized in that: The method specifically comprises the following steps: S1: Construct a high-speed railway communication model that jointly considers the Doppler effect and free space loss, including a high-speed railway (HSR) system model, a file popular distribution cache model, and a system motion model; In step S1, the constructed HSR system model refers to a system in which the base station responds to user requests through content caching and content pre-fetching during the operation of the high-speed rail. It consists of 1 macro base station, b track base stations and trains. In order to alleviate the problem of communication signal penetration loss, a mobile relay MR is deployed on the top of the train to collect K user requests and forward them to the track base station. The macro base station contains all the user-requested files F = {1, ..., f, ..., F}. Assuming that the file sizes are the same, all are Qbit, and the request probability obeys the zipf distribution, the file f is divided into N segments using MDS coding. The base station and the user have a certain cache capacity, and the cache space is represented as SS1 = {S1, ..., S i ,…,S b }、SS2={S1,…,S k ,…,S K }, where S i represents the cache space of the ith orbital base station, S k Represents the cache space of the kth user; the caches between base stations and users are independent of each other. The user caches after the base station and does not cache the same file segment as the base station; The constructed cache model based on file popular distribution includes: when a user requests content: first, search its own cache library and directly respond to the cached content; otherwise, the user aggregates the request interest packet to MR, and MR forwards the user request to the track base station; after the track base station receives the user request interest packet, it checks the local cache; if the content corresponding to the interest packet already exists locally, the track base station directly feeds back the user request content to MR, and MR forwards it to the corresponding user; otherwise, the track base station forwards the request interest packet to the macro base station M, and the track base station downloads the user request content through the backhaul link and hands it over to MR; The constructed system motion model specifically includes: given the train moving speed v and time t, the train travel distance is s(t) = v*t, given the distance between the base station and the train track is d, and the coverage range of the i-th base station is D i , then the connection range between base station i and the train is The relative travel distance of the train in the coverage area of base station i at time t is The distance between the train and the base station is The train moves at high speed relative to the base station, generating a Doppler frequency shift fD(t): Where f' represents the base station operating frequency, and c represents the speed of light; represents the angle between the train running direction and the base station, then the actual operating frequency of the base station is f i' (t) = f' - f D (t); The rate R provided by the orbital base station to the user i (t) is: R i (t)=Blog(1+SNR i (t)) (5) Where B is the channel bandwidth, SNR i (t) represents the user signal-to-noise ratio at time t; S2: Construct a cache placement optimization problem, that is, construct an objective function: minimize the user request delay as the optimization goal, specifically including: first construct a cache placement problem based on minimizing system delay: define is an element of the track base station cache strategy C1; if This means that the nth segment of file f is in the cache of track base station i, otherwise, the segment is not cached in track base station i. Similarly, the elements of user cache strategy C2 Indicates the cache status of user k for the nth segment of file f; when It means that the nth segment of file f has been cached locally in user k, otherwise, the segment of file f is not cached in user k. Under the initial cache, the user and the orbital base station independently cache the content according to the distribution of file zipf under the constraints of their own cache space. When the user requests file f, once all the resources related to file f in the orbital base station are transmitted to the user, the user is still in the coverage area of the current base station, and the orbital base station needs to request the relevant content from the macro base station. During the whole process, if the wireless link rate R i Greater than the return link rate R m , the actual file transfer rate will change from R i Reduce to R m Otherwise, the user will use the rate R i Receive the file; use the average rate provided by the orbital base station to solve the cache: The actual network throughput that track base station i can provide for file f is expressed as: in, represents the sum of pre-fetched and cached data of track base station i, R i ΔT represents the maximum network throughput that the orbital base station can provide; Indicates the time the user stays in the coverage area of orbital base station i; represents the data size of cache file f at track base station i; defines the user request service rate as r q , the macro base station provides a service rate R m ; The total delay of the entire system, i.e. the optimization problem, is expressed as: Where A1 represents the total delay caused by requesting file f on the base station, and A2 represents the delay compensated by the user by caching file segments of f; the delay of requesting file f is expressed as t f =A2-A1; q(f) represents the probability that the user requests file f; S3: Split the optimization problem into two parts: base station cache optimization and user cache optimization. That is, split the objective function into two sub-functions: maximizing the network throughput that the base station can actually provide and maximizing the user cache under the base station cache constraint. Solve the two sub-functions separately. Base station cache optimization part: By optimizing the base station cache strategy, the rail base station can provide the maximum network throughput to the terminal users connected within its coverage area. Specifically, the following OP1 function optimization is performed: The constraints are: Among them, st1 is the track base station cache space constraint, st2 is the track base station cache placement binary selection, and st3 represents the data size of the cache file f of the track base station i; User cache optimization: This allows users to maximize user cache data without repeatedly caching file segments that have already been cached by the orbital base station, so that the user cache compensation delay is maximized. This is specifically optimized through the following OP2 function: The constraints are: Among them, st1 is the user cache space constraint, st2 is the track base station cache constraint under file diversity, st3 represents the Boolean cache decision of user k for the nth segment of file f, and st4 represents the cache data of user k for file f.
2. The base station content caching and collaborative content pre-fetching method according to claim 1, characterized in that: In step S3, a greedy algorithm is used to find the optimal cache of the base station.
3. The base station content caching and collaborative content pre-fetching method according to claim 1, characterized in that: In step S3, solving the objective function OP2 of user cache optimization includes: independently optimizing caches between users can be decoupled into K parallel sub-functions OP2.1: Constraints: s.t1, s.t2, s.t3, s.t4 And q'(f)<0 When the user cache space cannot fully compensate for the delay caused by the base station transmission, the user needs to cache the file segments that the base station cannot satisfy the user's request for the file in order of file popularity to maximize the cache benefit; It means that the current orbit base station can meet the user's demand for file f, so the user does not need to cache; It means that the current track base station cannot meet the user's demand for file f; if the remaining amount of user cache space is sufficient to cache a segment of the file, and the track base station has not cached the segment, the user caches the segment, otherwise, the user does not cache the segment.
