CN106454937A - Distribution control method for caching space of mobile terminal - Google Patents

Abstract

The invention discloses a distribution control method for a caching space of a mobile terminal. The distribution control method comprises the following steps: 1, starting a program, and setting a caching threshold value to be threshold, wherein the set caching threshold value is less than an available space of a mobile terminal node; 2, initializing the caching space of the mobile terminal node, wherein each node sets a caching space which is half of the threshold for APP for storage, carrying and forwarding of a message; 3, collecting network information, and counting the number of pieces of information reaching the node within unit time and the service time Tn-1 of previous n pieces of messages for each terminal node; 4, forecasting the size of the caching space Npre; and 5, adjudging the caching size. The distribution control method can dynamically distribute the caching space for storage, carrying and forwarding of the messages between the terminal nodes, so that the nodes can dynamically distribute the caching size to increase the cache utilization rate, and various applications can share the caching space to enhance the user experience; and the distribution control method can be cooperated with a strategy and an algorithm of the mobile terminal for use so as to complete forwarding of the messages between the mobile terminals.

Description

A kind of distribution control method of mobile terminal spatial cache

Technical field

The invention belongs to mobile terminal caching technology field, particularly to being based on Ma Erke in a kind of mobile ad-hoc network The distribution control method of the mobile terminal spatial cache of husband's model and the Forecasting Methodology of spatial cache size.

Background technology

Movable self-organization (MobileAd Hoc) network is the autonomous data of the multi-hop that a kind of mobile terminal (node) forms Transmission system, the infrastructure that whole network is not fixed, can cannot using or not convenient use network infrastructure In the case of (including base station and AP), provide the data communication services between mobile terminal.Mobile terminal tool in network There are route and packet forwarding, being communicated wirelessly by radio network interface, thus constituting corresponding network on demand Topology.Mobile ad-hoc network be widely used in Military Network, mobile conference, wireless sensor network, emergency services and Disaster recovery field.

Because mobile node memory space is relatively limited, it is real that mobile ad-hoc network needs efficient cache management strategy Apply storage, delivery and the discarding of message, to reduce the possibility of nodal cache spilling, it is to avoid network congestion.Existing movement is from group In knitmesh network, buffer memory management method is concentrated mainly in the scheduling to buffered message, can be divided into following three kinds：1)Drop The tradition buffer memory management method such as random, Drop head, Drop tail, this kind of method is simple, but in self-organizing network In performance very poor.2) the global buffer management method of global network state information, such as GBD (Global are utilized Knowledge based Drop), GBSD (Global Knowledge based Scheduling and Drop) etc..This kind of The performance of method preferably and provides an optimum framework, but because self-organizing network has the spy of dynamic topological structure Property, obtain network global state information more difficult, therefore such method is difficult to be applied to real network environment.3) profit Partial cache management with the remaining life cycle of partial network information such as message, the size of message, jumping figure, message copy number etc. Method, such as E-DROP (Equal drop), T-DROP (ThresholdDrop) etc..Such method is than global buffer management strategy It is easily achieved and more preferable than the performance of traditional cache management strategy, research is more at present.

The present invention finds in research process, and existing method is mostly to consider in the case that spatial cache is limited, such as What scheduling message is to improve delivery success rate, but in the case of nodal cache space relatively abundance, existing cache management is calculated Method mostly give tacit consent to by whole spaces be used for storage forward message (showing stronger greediness in resource bid and use), Lead to the other application on mobile terminal (APP) then cannot normally run, cause each application in terminal that spatial cache is used Unfair and significantly reduce Consumer's Experience.

Content of the invention

Goal of the invention：For problems of the prior art, the present invention provides in a kind of mobile ad-hoc network and is based on The distribution control method of the mobile terminal spatial cache of Markov model and the Forecasting Methodology of spatial cache size, Ke Yiwei Between terminal node, message stores, carries, forwarding dynamically distribution spatial cache, contributes to node dynamically distributes cache size to carry High level cache utilization rate, is easy to multiple Application share spatial caches to lift Consumer's Experience it is not necessary to change the message of mobile terminal Scheduling strategy and Related Routing Algorithm, can be used in combination with the strategy of mobile terminal and algorithm, and completion message is mobile whole Forwarding between end.

Technical scheme：For solving above-mentioned technical problem, the present invention provides a kind of distribution of mobile terminal spatial cache to control Method, comprises the steps：

Step one：Program starts and sets caching threshold values as threshold, and the caching threshold values of setting is less than mobile terminal section The free space of point；

Step 2：The spatial cache of mobile terminal node is initialized, each node arranges threshold/ for APP The storage that 2 spatial cache is used for message carries forwarding；

Step 3：Collecting network information, to reaching the Information Number λ of node and front in each terminal node statistical unit time The service time T of n bar message_n-1(service time of wherein front a piece of news is T₀, the service time of front two message is T₁, front n The service time of bar message is T_n-1)；

Step 4：Prediction spatial cache N_preSize；

Step 5：Adjustment cache size, the spatial cache N being predicted according to step 4_preAdjustment is currently used in message storage The spatial cache N forwarding, if N_preLess than N, then abandon part messages until remaining message is big using Drop head strategy Little no more than N_pre, then unnecessary spatial cache is returned system；If N_preIt is not more than N and then judge N_preWhether it is less than Threshold, if N_preMore than N and less than threshold, then apply for N again to system_preThe space of-N size；If N_preNo Less than threshold, then again to the space of system application threshold-N size.

Further, predict spatial cache size N in step 4_preMethod as follows：

(1) the service time T needed for each node calculates (n+1)th message_nAverage：

(2) the service time T needed for each node calculates (n+1)th message_nVariance：

(3) each node calculates the average queue length of message：

(4) each node, according to the message size m setting in advance, calculates the spatial cache needed for subsequent time：

N_pre=m × L_s

Compared with prior art, it is an advantage of the current invention that：

1st, the present invention can make each mobile terminal node with reference to network operation conditions predict and arrange node caching big Little, improve Buffer Utilization and forward success rate, and avoid message to overflow.

2nd, the present invention enables each terminal node dynamically distributes cache size, is easy to multiple Application share spatial caches, carries Rise Consumer's Experience.

3rd, the present invention does not need to change scheduling message strategy and the Related Routing Algorithm of mobile terminal, can be with above-mentioned strategy It is used in combination with algorithm, forwarding between mobile terminal for the completion message.

Brief description

Fig. 1 is the flow chart of the present invention；

Fig. 2 is M | G | 1 queuing model schematic diagram in specific embodiment；

Fig. 3 is the state transition diagram of M | G | 1 embedded Markov chain in specific embodiment；

Fig. 4 is that have storage in mobile terminal when algorithm runs in specific embodiment to forward the APP occupancy of message function slow Deposit variation diagram.

Specific embodiment

With reference to the accompanying drawings and detailed description, it is further elucidated with the present invention.

It is mobile terminal node dynamic prediction distribution caching workflow as shown in Figure 1.

Step 1, sets cache threshold.After program starts, user is to have the APP setting use that message stores forwarding capability Capacity-threshold threshold, if the threshold value of setting is more than the free space of mobile terminal node, prompting user's " threshold of warning Value arranges excessive, insufficient space ".

Step 2, spatial cache initializes.The spatial cache of mobile terminal node is initialized, each node is APP The storage that the spatial cache of setting is used for message carries forwarding.

Step 3, collecting network information.The message count of node and front n bar is reached in each terminal node statistical unit time The service time of message.

Step 4, predicts cache size.Each point calculates the average of service time and the variance needed for (n+1)th message, and According to the message size m setting in advance, calculate the spatial cache needed for subsequent time.

Step 5, adjusts cache size.Each node is currently used in message according to the cache size adjustment that step 4 predicts The spatial cache N that storage forwards.

What Fig. 2 and Fig. 3 was respectively described is M | G | 1 queuing model and the state transition diagram of M | G | 1 embedded Markov chain.Right In each mobile terminal node, the arrival process of message is modeled as poisson arrival process, the service time of message obeys typically Distribution G, and think that only a piece of news can accept service every time, it is medium to be serviced that the message that other reach is saved in caching. Accept the period of service in (n+1)th message, when the new message count entering nodal cache is solely dependent upon the service of (n+1)th message Between and unrelated with the time before.

Specifically, message arrival process is modeled as poisson arrival process, the service time of message is typically distribution G, whole Individual system modelling becomes queuing model

t_n：Represent that nth bar message accepts the moment that service completes.

X_n：Represent when nth bar message accepts service and completes, the message count in nodal cache.

T_n：Represent and leave nodal cache, the service time needed for (n+1)th message when nth bar message.

Y_n：Represent the period accepting service in (n+1)th message, the new message count entering nodal cache.

So, we can represent system queuing's situation with Fig. 2.

As seen from Figure 2

If making a_j=P (Y_n=j) ＞ 0, wherein a_jRepresent the period accepting service in (n+1)th message, newly enter ingress and delay The message count deposited is the probability of j, and j is any nonnegative integer, then may certify that { X_nConstitute a Markov chain, commonly referred to as Embedded Markov chain.Additionally, Y_nIt is solely dependent upon the service time of (n+1)th message and unrelated with the time before.Therefore, X_n It is discrete-time Markovian chain.Note P_ij=P (X_n+1=j | X_n=i), then

P_0j=P (X_n+1=j | X_n=0)=P (Y_n=j)=a_j,j≥0 (1)

Thus

I.e. the Matrix of shifting of a step of this Markov chain is

Its state transition diagram thus can be drawn, as shown in Figure 2.

By T_nDefinition understand, service time { T_n, n >=1 } and it is independent identically distributed sequence of random variables, remember its public point Cloth function is G (t)=P (T_n≤t).Then

Wherein, P (Y_n=j | T_n=t) represent (0, t) (i.e. service time needed for (n+1)th message in time interval Interior) the new probability for j for the message number entering nodal cache.Because message is to reach according to Poisson flow, so should have

Bring formula (3) into, obtain

A is understood by formula (1)₀=P₀₀＞ 0, and each state of Markov chain is intercommunication, therefore this Markov chain is non-week Phase is irreducible, and calculates in a period of (n+1)th message sink service, new message count Y entering nodal cache_nEqual Value E (Y_n)

Then, the period accepting service in (n+1)th message, new message count Y entering nodal cache are calculated_nVariance D (Y_n)

D(Y_n)=E (Y_n ²)-[E(Y_n)]²=ρ+λ²D(T_n)

Can verify, as ρ ＜ 1, this Markov chain is traversal, therefore there is Stationary Distribution { p_j, j >=0 }, and { p_j} Must meet

Go to solve p with generating function below_j, order

Using formula (4) and formula (2), then have

Above-mentioned all formulas are added, just can get

Thus obtain the generating function of message team leader distribution in system

To seek P again₀, because

P (1)=1, A (1)=1

Therefore obtained by L'Hospital rule

Then

p₀=1- ρ

Substituted into formula (5), had

Utilize by above formula and twice L'Hospital rule, then in system message average queue length L_sFor：

Assume that the size of every message is the same, be m, then the spatial cache N needed for the subsequent time predicting_pre=m × L_s.

Fig. 4 describes has the APP occupancy caching change that storage forwards message function in mobile terminal when algorithm runs Situation.After user sets the use capacity-threshold threshold of an APP, system at most can distribute threshold size Space to this application.Spatial cache N when the prediction of node subsequent time_preLess than current cache space N, then using Drop Head strategy abandons part messages until remaining message size is not more than N_pre, unnecessary spatial cache is returned system；If N_preMore than N and less than threshold, then apply for N again to system_preThe space of-N size；If N_preMore than threshold, then Apply for the space of threshold-N size again.

The foregoing is only embodiments of the invention, be not limited to the present invention.All principles in the present invention Within, the equivalent made, should be included within the scope of the present invention.The content that the present invention is not elaborated belongs to In prior art known to this professional domain technical staff.

Claims (5)

1. a kind of distribution control method of mobile terminal spatial cache it is characterised in that：Comprise the steps：

Step one：Program starts and sets caching threshold values as threshold；

Step 2：The spatial cache of mobile terminal node is initialized, each node arranges threshold/2's for APP The storage that spatial cache is used for message carries forwarding；

Step 3：Collecting network information, to the Information Number λ of arrival node and front n bar in each terminal node statistical unit time The service time T of message_n-1；

Step 4：Prediction spatial cache N_preSize；

Step 5：Adjustment cache size.

2. a kind of mobile terminal spatial cache according to claim 1 distribution control method it is characterised in that：Described step Spatial cache N is predicted in rapid four_preThe method of size is as follows：

Step 4.1：Calculate the service time T needed for (n+1)th message of each node_nAverage E (T_n)：

$E\left(T_n\right)=\frac{T_0+T_1+...+T_{n-1}}{n}$

Step 4.2：Calculate the service time T needed for (n+1)th message of each node_nVariance D (T_n)：

$D\left(T_n\right)=\frac{{\[T_0-E\left(T_n\right)\]}^2+{\[T_1-E\left(T_n\right)\]}^2+...+{\[T_{n-1}-E\left(T_n\right)\]}^2}{n}$

Step 4.3：Calculate average queue length L of each node messages_s：

$L_s=\mathrm{\ρ}+\frac{{\mathrm{\λ}}^{2}D\left(T_n\right)+{\mathrm{\ρ}}^2}{2(1-\mathrm{\ρ})}=\mathrm{\λ}E\left(T_n\right)+\frac{{\mathrm{\λ}}^{2}D\left(T_n\right)+{\[\mathrm{\λ}E\left(T_n\right)\]}^2}{2\[1-\mathrm{\λ}E\left(T_n\right)\]}$

Step 4.4：Each node, according to the message size m setting in advance, calculates the spatial cache N needed for subsequent time_pre

N_pre=m × L_s.

3. a kind of distribution control method of mobile terminal spatial cache according to claim 1 is it is characterised in that described step In rapid five, adjustment cache size comprises the following steps that：The spatial cache N being predicted according to step 4_preAdjustment is currently used in and disappears The spatial cache N that breath storage forwards, if N_preLess than N, then abandon part messages until remaining disappear using Drophead strategy Breath size is not more than N_pre, then unnecessary spatial cache is returned system；If N_preIt is not more than N and then judge N_preWhether it is less than Threshold, if N_preMore than N and less than threshold, then apply for N again to system_preThe space of-N size；If N_preNo Less than threshold, then apply for the space of threshold-N size again to system.

4. the distribution control method of a kind of mobile terminal spatial cache according to one of claims 1 to 3, its feature exists In：In described step one, the caching threshold values threshold of setting is less than the free space of mobile terminal node.

5. the distribution control method of a kind of mobile terminal spatial cache according to one of claims 1 to 3, its feature exists In：When threshold value threshold of setting is more than the free space of mobile terminal node in described step one, then warn prompting user " threshold value arranges excessive, insufficient space ".