CN109673018B - A Novel Content Cache Distribution Optimization Method in Wireless Heterogeneous Networks - Google Patents

Abstract

A novel cache content placement and content cache distribution optimization method in a wireless heterogeneous network is characterized in that a wireless cache is sunk to an edge node and a user side of a wireless network, and when the network load is low, content possibly requested by a user is cached to the edge node and the user side of the network in advance and then distributed to a target user, so that network redundant flow is effectively reduced, and peak flow is effectively reduced. Meanwhile, a successful transmission probability expression is deduced, and the distribution of cache contents is optimized by taking the maximized successful transmission probability as a target. The optimized content cache distribution not only greatly improves the successful transmission probability, but also avoids the problems of network redundant flow, network congestion in peak periods, overhead waste and the like caused when the same content is repeatedly distributed.

Description

A Novel Content Cache Distribution Optimization Method in Wireless Heterogeneous Networks

technical field

The present invention relates to the technical field of wireless communication.

Background technique

With the rapid development of the Internet and the continuous growth of data traffic of intelligent terminals, traditional networks are gradually overwhelmed by the pressure of base station deployment overhead, operating costs, and user experience. In order to cope with this challenge, the wireless cache edge network, on the one hand, can "sink" network services to the wireless access network side closer to the user; obvious improvement. Therefore, the wireless network does not have to wait for the service request before responding and scheduling, but when the network load is low, the content that the user may request is cached in the edge node of the network in advance, and then distributed to the target user. Wireless edge caching can effectively reduce network redundant traffic, effectively reduce peak traffic and power consumption, overcome network backhaul link bottlenecks, and save network resources while meeting user needs. Caches deployed at the edge can store content with high popularity in advance, and these popular files can be delivered directly to the desired users, eliminating the need for content providers to repeatedly transmit these popular files through the backhaul link. In this case, content providers no longer need to transmit these popular files over the backhaul link, and the problem of limited backhaul capacity can be well addressed. However, the optimal content cache distribution is not only affected by content popularity, but also by other network parameters such as channel conditions, base station and user density, etc. Therefore, it is important to consider these network parameters and further optimize them to improve performance.

In this context, wireless heterogeneous networks are a very promising network architecture that can utilize wireless caching to effectively reduce peak traffic, power consumption, and lighten backhaul loads. At the same time, the cache deployed on the user side can also store content with high popularity in advance, and these popular files can be communicated directly through the Device-to-Device (D2D) technology and transmitted to nearby users.

In related art, "Analysis on cache-enabled wireless heterogeneous networks," IEEE Transactions on Wireless Communications, vol.15, no.1, pp.131–145, Jan 2016., the author assumes that in large-scale small-cell networks with cache In M, each small cell caches the most popular files, that is, all base stations store the same most popular files; while M "Probabilistic caching in wirelessd2d networks: Cache hit optimal vs. throughput optimal," IEEE Communications Letters, vol.21 , no.3, pp.584–587, 2016. The author studies probabilistic caching in a random wireless D2D caching network, that is, users in a D2D network cache content files in advance with a certain concept. However, the former only considers the caching mode and optimization at the base station, while the latter only considers the caching mode and optimization at the user end. Although in "Wireless contentcaching for small cell and d2d networks," IEEE Journal on Selected Areas in Communications, vol.34, no.5, pp.1222–1234, May 2016., the author considers both base station caching and user-side caching, But it only focuses on the performance comparison of the two caching methods, and does not consider the actual fusion caching method, that is, how to consider an effective caching method in a multi-layer network. Moreover, the caching methods in the existing literature are mainly popularity caching or average caching, etc., and there is no specific description on how to efficiently store multiple different files on each base station or user.

SUMMARY OF THE INVENTION

The purpose of the present invention is to solve the above problems, and to provide a novel cache content placement method and optimization method in a wireless heterogeneous network.

The technical solution is summarized as:

A novel cache content placement and content cache distribution optimization method in a wireless heterogeneous network, characterized in that it successively includes a first step of a cache content placement method and a second step of a content cache distribution optimization method; The cache sinks to the edge node (referring to the base station in the present invention) and the user terminal of the wireless network, and when the network load is low, the content that the user may request is cached to the edge node and the user terminal of the network in advance, and a corresponding cache configuration method is established; The second step: analyze the content distribution methods in different situations; at the same time, use random geometric knowledge to derive the expression of the probability of successful transmission; finally, with the goal of maximizing the probability of successful transmission, optimize the distribution of cached content; the above two steps are finally distributed to the target users.

Specifically, in a cell in a wireless heterogeneous network, it is assumed that there is one base station and multiple users. The base station is provided with a cache space, which can store popular files in advance, so when a file requested by a user happens to be cached in the base station, the base station can directly serve the user with its own cached content. In addition, the user terminal is also provided with a cache, which can cache some popular files in advance, so that when the files requested by nearby users happen to be cached at the user, the users can also use the device-to-Device, D2D ) to communicate directly.

The first step is "cache content placement method", and its system model is divided into the establishment of the distribution model of the base station and the user, the modeling of the cache content, and the channel modeling.

其密度分别为ρλ_u；非活跃用户服从独立齐次泊松点过程分布

其密度分别为(1-ρ)λ_u。当用户为所述活跃用户时，表明这个用户正在请求文件，因而它不可以作为发送者服务周围用户；而所述非活跃用户则可以被视为潜在的D2D发送者。First, the distribution model of base stations and users is established: the total bandwidths allocated to base stations and users are W' and W respectively. At the same time, both the base station and the user are equipped with a transmit antenna, and the corresponding transmit powers are P _B and P _D respectively. The users obey the independent homogeneous Poisson point process distribution Φ _u , whose densities are λ _u respectively, and the probability of each user being active is ρ∈[0,1]. Active users obey an independent homogeneous Poisson process distribution

Their densities are ρλ _u respectively; inactive users obey the independent homogeneous Poisson point process distribution

Their densities are (1-ρ)λ _u , respectively. When a user is the active user, it indicates that the user is requesting files, so it cannot serve as a sender to serve surrounding users; and the inactive user can be regarded as a potential D2D sender.

是每个用户随机地请求文件n∈N的概率，假设a₁≥a₂...≥a_N。基站拥有所有用户请求的文件，而每个用户最多可以缓存M个文件。Second, the modeling of cache content: Assuming that all files are independent and of size 1, N@{1,2,...,N} represents the set of files N>M. The file popularity distribution is assumed to be a@(a _n ) _n∈N , where,

is the probability that each user randomly requests a file n∈N, assuming a ₁ ≥ a ₂ ... ≥ a _N . The base station owns all files requested by users, and each user can cache up to M files.

The probability that each user caches file n is p _n , satisfying:

0≤p _n ≤1,n∈N

The content cache distribution is defined as p@(p _n ) _n∈N .

Finally, channel modeling: In the channel conditions of this system, considering the random characteristics of channel fading, the attenuation factor of the transmitted signal is D - ^α for large-scale fading, where D is the transmission distance, and α>2 is the path loss index. For small-scale fading, Rayleigh fading is considered, and each small-scale fading channel is h: ^d CN(0,1). The second step "content cache distribution optimization method": analyze the content distribution methods in different situations; then use random geometric knowledge to derive the successful transmission probability expression; finally, with the goal of maximizing the successful transmission probability, optimize the cache content distribution;

First, analyze the content distribution methods in different situations, three situations: Step 2.1, service by its own cache: if the file n requested by user u ₀ is in its own cache, then u ₀ will be served by itself, otherwise go to step 2.2; 2.2 Steps, served by other users: when the file n requested by user u ₀ is not in its own cache, but other users cache file n within its communication radius R, then user u ₀ caches file n by the nearest user If the file n requested by user u ₀ is neither in its own cache nor in the caches of other users within its communication radius R, the base station will send the requested file n to the user.

Then, the expression for the total successful transmission probability of the entire network is as follows:

为用户u₀在它的通信半径R内找到缓存了文件n的其他用户的概率,q_d(p)为步骤2.2时的成功传输概率,q_b(p)为步骤2.3时的成功传输概率。最后，以最大化成功传输概率为目标，优化无线异构网络中的新型缓存内容放置方法，具体为：in,

is the probability that user u ₀ finds other users who have cached file n within its communication radius R, q _d (p) is the probability of successful transmission in step 2.2, and q _b (p) is the probability of successful transmission in step 2.3. Finally, with the goal of maximizing the probability of successful transmission, a novel cache content placement method in wireless heterogeneous networks is optimized, specifically:

Build an optimization model:

st0≤p _n ≤1,n∈N

Optimize cache distribution:

By rewriting the objective function as a Lagrangian function, a dual factor is introduced to eliminate the equality constraint;

where η(t) and μ(t) are defined as duality factors;

Solve a suboptimization problem for Lagrangian functions

Use the gradient descent method to solve the minimum value of the above Lagrangian function, and obtain the corresponding value of p _n that makes the Lagrangian function get the minimum; and eliminate the inequality constraint by the maximum value function:

p _n (t)=max{argmin _n∈N [L],0}

In the above first iterative sub-algorithm, the minimum value in the above Lagrangian descent method is iteratively solved; in the second iterative sub-algorithm, the dual factor is iteratively optimized according to the following rules:

η(t+1)=max{η(t)+c ₁ (p _n (t)-1),0}

Among them, c ₁ and c ₂ are the corresponding step sizes in the iterative optimization of dual factors η(t) and μ(t), respectively. In the t-th iteration, p _n (t) is iteratively updated using η(t) and μ(t) obtained after (t-1) iterations, and the two-layer sub-iteration algorithm cyclically updates the iterations until the condition is satisfied: p _n (t+1)-p _n (t)<δ(δ is the threshold value that can be set by the user), the updated optimal solution is obtained after

Beneficial effects of the present invention:

(1) The novel cache content placement method in the present invention solves the problem of explosive growth of the demand for content-based services.

(2) The novel cache content placement method in the present invention avoids the problems of redundant network traffic, network congestion during peak periods, and waste of overhead caused when the same content is repeatedly distributed.

(3) The content cache distribution obtained after optimization in the present invention can greatly improve the probability of successful transmission.

In general, the present invention obtains a better content cache distribution scheme through the design and optimization of a novel cache content placement method, which is a favorable condition for the further development of the next generation wireless communication technology.

Description of drawings

1 is a network architecture diagram of the present invention;

Figure 2 is a flow chart of the method of the present invention.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings.

其密度分别为ρλ_u；非活跃用户服从独立齐次泊松点过程分布

其密度分别为(1-ρ)λ_u。当用户为活跃用户时，表明这个用户正在请求文件，而非活跃用户则可以被视为潜在的D2D发送者。需要注意的是，当一个用户是活跃用户时，表明它正在请求文件，因而它不可以作为发送者服务周围用户。分配给基站、用户的总带宽分别是W'和W。同时，基站、用户都配有一根发射天线，相应的发送功率分别为P_B和P_D，每个移动用户都配备一个接收天线。最后，考虑信道衰落的随机特性，对于大尺度衰落，传输信号的衰减因子为D^-α，其中D是传输距离，α＞2是路径损耗指数。对于小尺度衰落，考虑瑞利衰落，每个小尺度衰落信道为

FIG. 1 is a network architecture diagram of the present invention. The present invention considers a wireless heterogeneous network. In a cell of the network, it is assumed that there is a base station and a part of mobile users. Mobile users obey the independent homogeneous Poisson point process distribution Φ _u , whose densities are λ _u respectively, and the probability of each user being active is ρ∈[0,1]. Active users obey an independent homogeneous Poisson process distribution

Their densities are ρλ _u respectively; inactive users obey the independent homogeneous Poisson point process distribution

Their densities are (1-ρ)λ _u , respectively. When a user is an active user, it indicates that the user is requesting files, while an inactive user can be regarded as a potential D2D sender. Note that when a user is an active user, it is requesting files, so it cannot serve as a sender to surrounding users. The total bandwidths allocated to base stations and users are W' and W, respectively. At the same time, the base station and the user are equipped with a transmitting antenna, and the corresponding transmission powers are P _B and _PD respectively, and each mobile user is equipped with a receiving antenna. Finally, considering the random characteristics of channel fading, for large-scale fading, the attenuation factor of the transmitted signal is D - ^α , where D is the transmission distance, and α > 2 is the path loss index. For small-scale fading, considering Rayleigh fading, each small-scale fading channel is

FIG. 2 is a flow chart of the method of the present invention.

是每个用户随机地请求文件n∈N的概率，假设a₁≥a₂...≥a_N。First, caches are configured at the base station and the user respectively (corresponding to step ① in Figure 2). It is assumed that the base station has all the files requested by the users, and each user can cache up to M files. Here, we assume that all files are independent and have the same size as 1, and N@{1,2,...,N} represents the set of files N>M. The file popularity distribution is assumed to be a@(a _n ) _n∈N , where,

is the probability that each user randomly requests a file n∈N, assuming a ₁ ≥ a ₂ ... ≥ a _N .

The probability that each user caches file n is p _n , let p _n satisfy:

0≤p _n ≤1,n∈N

The second constraint is caused by the limited cache capacity M on the client side. Therefore, the content cache distribution is defined as p@(p _n ) _n∈N .

Secondly, as can be seen from Figure 1 and Figure 2, for a typical user u ₀ , when user u ₀ randomly requests file n, the following three situations may occur (corresponding to step ② in Figure 2) :

2.1 Step, serve by its own cache: if the file n requested by user u ₀ is in its own cache, then u ₀ is served by itself, otherwise, go to step 2.2;

2.2 Steps, served by other users: When the file n requested by user u ₀ is not in its own cache, but other users cache file n within its communication radius R, then user u ₀ caches the file by the nearest user n user service, otherwise go to step 2.3;

2.3 Step, served by the base station: when the file n requested by user u ₀ is neither in its own cache, nor in the caches of other users within its communication radius R, the base station will send the requested file n to the user.

Then, the following will deduce the probability of successful transmission (corresponding to step ③ in Figure 2), we think that when the transmission rate reaches a critical value, that is, the corresponding channel capacity is greater than the critical value, the file n can be correctly transmitted at user u ₀ Decoding, this process is called successful transmission, and the corresponding probability is called successful transmission probability.

For the first case (step 2.1), since the user is served by its own cache, the probability of successful transmission is 1. For case step 2.2, first, the probability that user u ₀ finds other users who have cached file n within its communication radius R is:

Secondly, in the second case (step 2.2), the probability of successful transfer of file n is:

q _d (p)=Pr[Wlog ₂ (1+SINR _k,0 )≥τ]

So in the same way, in the third case (step 2.3), the probability of successful transmission of file n is:

q _b (p)=Pr[W'log ₂ (1+SINR _b,0 )≥τ']

In the above, τ and τ' are the critical values described above, respectively. When the transmission rate reaches a critical value, that is, after the corresponding channel capacity is greater than the critical value, we consider the transmission to be successful. Finally, the expression to obtain the total successful transmission probability of the entire network is as follows:

Below, with the goal of maximizing the probability of successful transmission, the novel cache content placement method in the wireless heterogeneous network of the present invention is optimized, and an optimization model is first established (corresponding to step 4 in Fig. 2 ):

st0≤p _n ≤1,n∈N

Next, optimize the cache distribution (corresponding to step ⑤ in Figure 2), because the above problem is a non-convex optimization problem, and n variables are restricted by both equality constraints and inequality constraints, so it is difficult to obtain a closed-form solution. Therefore, by rewriting the objective function into a Lagrangian function form, a dual factor is introduced to remove the equality constraint;

where η(t) and μ(t) are defined as duality factors.

Next, solve the suboptimization problem for the Lagrangian function

This is an unconstrained optimization problem. The gradient descent method is used to solve the minimum value of the above Lagrangian function, and the corresponding value of p _n is obtained so that the Lagrangian function can obtain the minimum value; and the maximum value function is used to eliminate the Inequality constraints:

p _n (t)=max{argmin _n∈N [L],0}

In the first iterative sub-algorithm above, we iteratively solve the minimum in the Lagrangian descent method above; in the second iterative sub-algorithm, we iteratively optimize the dual factor following the following rules:

η(t+1)=max{η(t)+c ₁ (p _n (t)-1),0}

Among them, c ₁ and c ₂ are the corresponding step sizes in the iterative optimization of dual factors η(t) and μ(t), respectively. In the t-th iteration, p _n (t) is iteratively updated using η(t) and μ(t) obtained after (t-1) iterations, and finally, the two-layer sub-iteration algorithm loops and updates the iterations until the condition is satisfied: |p _n (t+1)-p _n (t)|＜δ (δ is the threshold value that can be set by the user) to get the updated optimized solution

The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above-mentioned embodiments, and the descriptions in the above-mentioned embodiments and the description are only to illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have Various changes and modifications fall within the scope of the claimed invention. The claimed scope of the present invention is defined by the appended claims and their equivalents.

Claims (1)

1. a novel content cache distribution optimization method in a wireless heterogeneous network, it is characterized in that, successively comprise the cache content placement method of the first step, the content cache distribution optimization method of the second step; Step 1: Deploy the wireless cache to the edge nodes and clients of the wireless network, where the edge nodes refer to base stations, which cache the content that users may request to the edge nodes and clients of the network in advance when the network load is low. Establish the corresponding cache configuration method; The second step: analyze the content distribution methods in different situations; at the same time, use random geometric knowledge to derive the expression of the probability of successful transmission; finally, optimize the distribution of cached content with the goal of maximizing the probability of successful transmission; The above two steps are finally realized and distributed to the target users; The first step: "cache content placement method", the system model is divided into the establishment of the distribution model of the base station and the user, the modeling of the cache content, and the channel modeling; The establishment of the distribution model of the base station and the user: the total bandwidth allocated to the base station and the user is W' and W respectively, and the corresponding transmit power is P _B and P _D respectively; The modeling of the cached content: assuming that all files are independent and the size is 1, N@{1,2,...,N} represents the set of files N>M; the file popularity distribution is assumed to be a @(a _n ) _n∈N , where,

is the probability that each user randomly requests a file n∈N, assuming a ₁ ≥ a ₂ ... ≥ a _N ; the base station owns the files requested by all users, and each user can cache up to M files; The probability that each user caches file n is p _n , satisfying: 0≤p _n ≤1,n∈N

The content cache distribution is defined as p@(p _n ) _n∈N ; The channel modeling: in the channel conditions, the attenuation factor of the transmitted signal is D - ^α for large-scale fading, where D is the transmission distance, and α>2 is the path loss index; for small-scale fading, Rayleigh fading is considered, and each small-scale fading considers Rayleigh fading. The scale fading channel is

Step 2: According to the analysis of content distribution methods in different situations, there are three situations: 2.1 Step, serve by its own cache: if the file n requested by user u ₀ is in its own cache, then u ₀ is served by itself, otherwise, go to step 2.2; 2.2 Steps, served by other users: When the file n requested by user u ₀ is not in its own cache, but other users cache file n within its communication radius R, then user u ₀ caches the file by the nearest user n user service, otherwise go to step 2.3; 2.3 Steps, served by the base station: when the file n requested by the user u ₀ is neither in its own cache, nor in the cache of other users within its communication radius R, the base station will send the requested file n to the user; The expression for the total successful transmission probability of the entire network is as follows:

in,

is the probability that user u ₀ finds other users who have cached file n within its communication radius R, q _d (p) is the probability of successful transmission in step 2.2, and q _b (p) is the probability of successful transmission in step 2.3; With the goal of maximizing the probability of successful transmission, a novel cache content placement method in wireless heterogeneous networks is optimized, specifically: Build an optimization model:

st0≤p _n ≤1,n∈N

Optimize cache distribution: By rewriting the objective function as a Lagrangian function, a dual factor is introduced to eliminate the equality constraint;

where η(t) and μ(t) are defined as duality factors; Solve a suboptimization problem for Lagrangian functions

Use the gradient descent method to solve the minimum value of the above Lagrangian function, and obtain the corresponding value of p _n that makes the Lagrangian function get the minimum; and eliminate the inequality constraint by the maximum value function: p _n (t)=max{arg min _n∈N [L],0} In the first iterative sub-algorithm, gradient descent is used to iteratively solve the corresponding p _n that minimizes the above Lagrangian function; in the second iterative sub-algorithm, the dual factor is iteratively optimized according to the following rules: η(t+1)=max{η(t)+c ₁ (p _n (t)-1),0}

Among them, c ₁ and c ₂ are the corresponding step sizes in the iterative optimization of the dual factor η(t) and μ(t), respectively; in the t-th iteration, use the η(t) obtained after (t-1) iterations and μ(t) iteratively update p _n (t), the two-layer sub-iteration algorithm loops and iterates until the condition is satisfied: |p _n (t+1)-p _n (t)|<δ, δ can be set by the user Threshold value, and then get the updated optimized solution

Images