CN103746775A - Cache mechanism for improving wireless opportunity network coding gain - Google Patents

Abstract

The invention discloses a cache mechanism for improving a wireless opportunity network coding gain. The mechanism comprises the following steps of classifying data packets cached with the same time in a decoding packet pool into one type, and representing the number of the packet packets in the kth type of data packets with nk, thus converting the total decoding effectiveness target of the maximum cache system into the maximum cache system decoding efficiency; defining the distribution function, achieving the delay, of one coding packet as the probability, being smaller than or equal to a second, of the coding bag achieving the delay, when each data packet in the decoding packet pool is cached with an optimal cache time and achieves the maximum decoding efficiency, analyzing to obtain the maximum system decoding efficiency of the decoding packet pool. The specific realizing method of the distributed cache management mechanism is provided, the cache mechanism overcomes the deficiency on the research of the cache management of the opportunity network coding and decoding packet pool in the existing research, the decoding effectiveness of the cache data packet is improved, and the throughput rate performance gain in opportunity network coding is effectively enhanced.

Description

A kind of caching mechanism that promotes the gain of radio machine meeting network code

Technical field

The present invention relates to a kind of caching mechanism that promotes the gain of radio machine meeting network code.

Background technology

In order to improve wireless network throughput and energy efficiency, opportunistic network coding is applied as a promising technology, can significantly improve network performance.In its application, the packet that node need to be listened to or be all buffered in local Bao Chizhong by the packet of its transmission, the decoding possible for future needs.Due to the employing of opportunistic network coding is the coded system based on XOR, when coding nodes need to be by a plurality of data packet codings together time, each receiving node that need to guarantee this encoded packets has the every other packet except its needed packet, could decode and obtain needed packet.So caching mechanism of encoding in application to liking opportunistic network of the present invention's research.

Although the application of existing opportunistic network encoding mechanism all suppose the cache resources of node and disposal ability enough large, the management of this cache resources can have influence on the throughput performance gain of opportunistic network coding.For some resource-constrained wireless networks, the energy of node, spatial cache and disposal ability etc. are all limited.Such as, a large amount of packet of buffer memory is challenging to RFIDs and sensor device etc.; The equipment (as WLAN gateway) that while need to be processed mass data stream for some is also difficult.Under these scenes, from decoding angle, if the next-hop node of encoded packets cannot be collected all packets for decoding, will cause node decoder failure, cannot recover its required raw data packets, thereby have influence on the throughput performance gain of opportunistic network coding.The more important thing is, even if node resource is enough large, in existing opportunistic network encoding mechanism, need between neighbor node to receive the packet information in the mutual cache exchanging of form of report, thereby guarantee the solvability of coded data packet.Data cached Bao Yue is many, needs the amount of information of exchange mutually also just larger, and the network overhead being brought by reception report is also just larger, and then has influence on the throughput gain of opportunistic network coding.Due to throughput performance gain and code machine meeting positive correlation, the packet that buffer memory listens to is not that each is helpful to decoded data packets yet; For the cache-time of each packet, because meeting has influence on network overhead, also not the longer the better for cache-time.Research before most of all center of gravity is placed on to the meeting of increase code machine and encryption algorithm designs above, and ignored the caching mechanism and maximization buffer memory effectiveness of optimizing the Bao Chi that decodes.Along with advocating of global green communications, study a kind of caching mechanism efficient, main chance network code performance gain and seem particularly important.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, a kind of concrete methods of realizing of distributed cache management mechanism is provided, made up the deficiency to the research of opportunistic network coding and decoding bag pond cache management in existing research, can make full use of the cache resources of node, improve the decoding effectiveness of data cached bag, the lifting radio machine that has effectively promoted the throughput performance gain of opportunistic network coding is understood the caching mechanism of network code gain.

The object of the invention is to be achieved through the following technical solutions: a kind of caching mechanism that promotes the gain of radio machine meeting network code, it comprises the following steps:

S1: the packet of decoding Bao Chizhong buffer memory same time is classified as to a class, and uses n _kthe number that represents packet in k class packet, thus convert the total decoding effectiveness target of largest buffered system to largest buffered system decodes efficiency Y _b:

$\begin{array}{c}\begin{array}{cc}\mathrm{Maximize}& Y_B=\underset{k}{\mathrm{\Σ}}(\frac{n_k{t}_k}{B\×T}\·\frac{D_k\left(t_k\right)}{s\×t_k})=\underset{k}{\mathrm{\Σ}}{a}_k{Y}_k\left(t_k\right);\end{array}\\ a_k=\frac{n_k{t}_k}{B\×T}\end{array}$

Wherein, a _k, Y _k(t _k) be respectively the ratio that decoding Bao Chizhong k class packet accounts for caching system global solution code efficiency, data pack buffer time t _kdecoding efficiency;

S2: definition encoded packets arrives the distribution function F (t) postponing and postpones to be less than or equal to the t probability of second for encoded packets arrives; The probability density function that this distribution function is corresponding is f (t), and for a packet of decoding Bao Chizhong, from this bag enters decoding bag pond, the contribution amount of this packet of decoding Bao Chizhong can be expressed as

D(t)＝s·F(t)

Corresponding decoding efficiency is expressed as

$Y\left(t\right)=\frac{D\left(t\right)}{s\·t}=\frac{F\left(t\right)}{t}$

Optimum cache-time t of each data pack buffer as decoding Bao Chizhong _oand while obtaining maximal solution code efficiency, the system decodes efficiency of decoding Bao Chi also just obtains maximum.

The computational methods of the optimum cache-time of described decoding Bao Chizhong packet adopt ODCT algorithm, guarantee all δ of buffer memory Yi Ge chronomere at least of every group of packet, and iterative computation finishes rear basis to t _owith the comparison of δ, obtain optimum decoding bag pond cache-time: work as t _owhile equating with δ, cache-time and the decoding efficiency of packet are respectively T ₀and Y (T ₀); Work as t _owhen unequal with δ, the cache-time of packet is with probability α ₀be made as T ₀, with probability α ₁be made as T ₁, system decodes efficiency is two kinds of weighted results in probability situation, is:

Y _B＝α ₀(λT ₀/B)Y(T ₀)+α ₁(λT ₁/B)Y(T ₁)。

The invention has the beneficial effects as follows:

1, according to the concrete feature of opportunistic network coding, the concrete methods of realizing of distributed cache management mechanism is proposed, made up the deficiency to the research of opportunistic network coding and decoding bag pond cache management in existing research;

2, the cache model based on decoding efficiency, can make full use of the cache resources of node, improves the decoding effectiveness of data cached bag, reduces network overhead, has effectively promoted the throughput performance gain of opportunistic network coding.

Accompanying drawing explanation

Fig. 1 is several typical network code topological diagrams;

Fig. 2 is to be the decoding efficiency figure under different cache policies;

Fig. 3 is network code caching mechanism of the present invention (NCP) and the network throughput comparison diagram of typical opportunistic network coding framework COPE under different cache size conditions.

Embodiment

Below in conjunction with accompanying drawing, further illustrate technical scheme of the present invention, but the content that the present invention protects is not limited to the following stated.

A kind of caching mechanism that promotes the gain of radio machine meeting network code, the replacement of combining cache-time and the packet of having considered packet, the basic application scenarios of this caching mechanism is the cache management mechanism of each node in the situation lower network for maximum coding degree is 2 in opportunistic network coding to decoding Bao Chi, comprise " X " type topology (X-topology) as shown in Figure 1, " Bob-Alice " type topology (Bob-Alice-topology), " Y " type topology (Y-topology) and " Cross " type topology (Cross-topology); It comprises the following steps:

Cache management target is the total decoding effectiveness D of largest buffered system _b:

$\begin{array}{c}\mathrm{Maximize}{D}_B=\underset{i}{\mathrm{\Σ}}{D}_i\left(t_i\right),\\ s.t.\underset{i}{\mathrm{\Σ}}\left(s\×t_i\right)\≤s\×B\×T.\end{array}$

S wherein, B, T, t _i, D _i(t _i) be respectively the size of bag, the cache size of Yi Baowei unit, caching system operating time, the cache-time of packet i, packet i buffer memory t _ithe decoding effectiveness of time.

A caching mechanism that promotes the gain of radio machine meeting network code, it comprises the following steps:

S1: the packet of decoding Bao Chizhong buffer memory same time is classified as to a class, and uses n _kthe number that represents packet in k class packet, thus convert the total decoding effectiveness target of largest buffered system to largest buffered system decodes efficiency Y _b:

$\begin{array}{c}\begin{array}{cc}\mathrm{Maximize}& Y_B=\underset{k}{\mathrm{\Σ}}(\frac{n_k{t}_k}{B\×T}\·\frac{D_k\left(t_k\right)}{s\×t_k})=\underset{k}{\mathrm{\Σ}}{a}_k{Y}_k\left(t_k\right);\end{array}\\ a_k=\frac{n_k{t}_k}{B\×T}\end{array}$

Wherein, a _k, Y _k(t _k) be respectively the ratio that decoding Bao Chizhong k class packet accounts for caching system global solution code efficiency, data pack buffer time t _kdecoding efficiency;

S2: definition encoded packets arrives the distribution function F (t) postponing and postpones to be less than or equal to the t probability of second for encoded packets arrives; The probability density function that this distribution function is corresponding is f (t), a packet for decoding Bao Chizhong, from this bag enters decoding bag pond, if it is in the t encoded packets that degree of coding is 2 that is used in second decode, the contribution amount of this packet of Bao Chizhong of decoding can be expressed as

D(t)＝s·F(t)

Corresponding decoding efficiency is expressed as

$Y\left(t\right)=\frac{D\left(t\right)}{s\·t}=\frac{F\left(t\right)}{t}$

Optimum cache-time t of each data pack buffer as decoding Bao Chizhong _oand while obtaining maximal solution code efficiency, the system decodes efficiency of decoding Bao Chi also just obtains maximum.

The computational methods of the described optimum cache-time of decoding Bao Chizhong packet adopt ODCT algorithm.

ODCT algorithm:

In this algorithm, need to guarantee all δ of buffer memory Yi Ge chronomere at least of every group of packet, iterative computation finishes rear basis to t _owith the comparison of δ, obtain optimum decoding bag pond cache-time: work as t _owhile equating with δ, cache-time and the decoding efficiency of packet are respectively T ₀and Y (T ₀); Work as t _owhen unequal with δ, the cache-time of packet is with probability α ₀be made as T ₀, with probability α ₁be made as T ₁, system decodes efficiency is two kinds of weighted results in probability situation, is:

Y _B＝α ₀(λT ₀/B)Y(T ₀)+α ₁(λT ₁/B)Y(T ₁)。

How specific explanations realizes the decoding cache management strategy NCP in opportunistic network coding once below.

1, encoded packets arrives the mensuration that postpones distribution function F (t): first, set equably the serial desired value f that encoded packets arrives the distribution function F (t) postponing _i∈ [0,1], and find and meet F (t _i)=f _itime t _i, in algorithm 2, we are each f in k desired value _i∈ [0,1], 1≤i≤k is provided with 2 variablees, is respectively t _iand h _i, h _ipreserve search t _itime F (t _x) value, wherein, each desired value f _i∈ [0,1] is set to i/k.First, we arrange h _i=0.5, t _ifor arriving, first encoded packets postpones.Often obtain a new encoded packets and arrive delay l _coded, by l _codedwith t _icompare, according to following formula, upgrade h _ivalue.

$h_i^{\′}=\left\{\begin{array}{cc}(1-\mathrm{\α})h_i+\mathrm{\α}& \mathrm{if}{l}_{\mathrm{coded}}\≤t\\ (1-\mathrm{\α})h_i& \mathrm{if}{l}_{\mathrm{coded}>t_i}\end{array}\right.$

If upgrade the h obtaining _ivalue is little and t than the desired value of setting _ibe less than the encoded packets newly obtaining and arrive delay, by t _ivalue increase α * (l _coded-t _i), but according to the character of distribution function, need to guarantee the t newly obtaining _ishould be less than or equal to t _i+1value, same, for the another kind of situation that may occur: if h _ivalue is large and t than the desired value of setting _ibe greater than the encoded packets newly obtaining and arrive delay, by t _ivalue reduce α * (t _i-l _coded), in like manner, the t newly obtaining _ishould be more than or equal to t _i-1, here, we set k=50, and α=0.01 preferably approach coding bag arrives the distribution function F (t) postponing.

The estimation algorithm of distribution function F (t) is:

2, decoding cache management strategy NCP concrete methods of realizing is divided into following several situation:

Situation one: once decoding Bao Chiman, for newly arrived packet, now needs to determine that this packet is be buffered or should abandon.First, find out the longest packet of decoding Bao Chizhong cache-time, and by its cache-time τ _maxwith T ₁compare.If τ _max>=T ₁, this bag can be cleared out from decoding Bao Chizhong; Otherwise, check decoding Bao Chi, check and whether exist cache-time to be greater than T ₀packet, if present, abandoned.If all do not belong to above situation, by the cache-time τ of the longest packet of cache-time of decoding Bao Chizhong _maxwith the cache-time t that can obtain maximal solution code efficiency _omake comparisons, if cache-time τ _maxbe more than or equal to t _o, we are just τ by cache-time _maxthe oldest data packet discarding, otherwise, just newly arrived packet is abandoned.More than operate as shown in Case1 in table 1.

Situation two: to participating in the packet of decoding, when decoding Bao Chiman, to wherein having participated in the packet of decoding, removed, with other packets of buffer memory, otherwise, continue this packet of buffer memory.More than operate as shown in the Case2 in table 1.

Situation three: when a packet arrives, if through search decoding Bao Chi, there is this packet; Meanwhile, the ttl field value in this newly arrived packet IP packet header is less than the ttl field value in decoding that packet IP packet header being buffered of Bao Chizhong, abandons that packet being buffered of decoding Bao Chizhong; Simultaneously also no longer to newly arrived this identical data packet buffer memory in addition.Otherwise, for a newly arrived packet, if find decoding Bao Chizhong buffer memory a same packet, and there is identical ttl field in the IP head of these two packets, will be no longer to this newly arrived packet buffer memory in addition.More than operate as shown in the Case3 in table 1.

Table 1 opportunistic network coding cache policy

CASE1.

CASE2.

CASE3.

Fig. 2 is to be the decoding efficiency figure under different cache policies: the cache policy (COPE-TBCP) based on timer (0.5 second) under typical opportunistic network coding framework COPE and optimum decoding cache-time strategy (ODCT);

Fig. 3 is network code caching mechanism of the present invention (NCP) and the network throughput comparison diagram of typical opportunistic network coding framework COPE under different cache size conditions, as can be seen from the figure, use the effectively throughput performance gain of elevator meeting network code of method of the present invention.

Claims (2)

1. a caching mechanism that promotes the gain of radio machine meeting network code, is characterized in that: it comprises the following steps:

S1: the packet of decoding Bao Chizhong buffer memory same time is classified as to a class, and uses n _kthe number that represents packet in k class packet, thus convert the total decoding effectiveness target of largest buffered system to largest buffered system decodes efficiency Y _b:

$\begin{array}{c}\begin{array}{cc}\mathrm{Maximize}& Y_B=\underset{k}{\mathrm{\Σ}}(\frac{n_k{t}_k}{B\×T}\·\frac{D_k\left(t_k\right)}{s\×t_k})=\underset{k}{\mathrm{\Σ}}{a}_k{Y}_k\left(t_k\right);\end{array}\\ a_k=\frac{n_k{t}_k}{B\×T}\end{array}$

Wherein, a _k, Y _k(t _k) be respectively the ratio that decoding Bao Chizhong k class packet accounts for caching system global solution code efficiency, data pack buffer time t _kdecoding efficiency;

S2: definition encoded packets arrives the distribution function F (t) postponing and postpones to be less than or equal to the t probability of second for encoded packets arrives; The probability density function that this distribution function is corresponding is f (t), and for a packet of decoding Bao Chizhong, from this bag enters decoding bag pond, the contribution amount of this packet of decoding Bao Chizhong can be expressed as

D(t)＝s·F(t)

Corresponding decoding efficiency is expressed as

$Y\left(t\right)=\frac{D\left(t\right)}{s\·t}=\frac{F\left(t\right)}{t}$

Optimum cache-time t of each data pack buffer as decoding Bao Chizhong _oand while obtaining maximal solution code efficiency, the system decodes efficiency of decoding Bao Chi also just obtains maximum.

2. a kind of radio machine that promotes according to claim 1 is understood the caching mechanism that network code gains, it is characterized in that: the computational methods of the described optimum cache-time of decoding Bao Chizhong packet adopt ODCT algorithm, guarantee all δ of buffer memory Yi Ge chronomere at least of every group of packet, iterative computation finishes rear basis to t _owith the comparison of δ, obtain optimum decoding bag pond cache-time: work as t _owhile equating with δ, cache-time and the decoding efficiency of packet are respectively T ₀and Y (T ₀); Work as t _owhen unequal with δ, the cache-time of packet is with probability α ₀be made as T ₀, with probability α ₁be made as T ₁, system decodes efficiency is two kinds of weighted results in probability situation, is:

Y _B＝α ₀(λT ₀/B)Y(T ₀)+α ₁(λT ₁/B)Y(T ₁)。

Images