CN102325082B - Nomadic application-oriented network congestion control method - Google Patents

Abstract

The invention relates to a nomadic-application-oriented network congestion control method. The network congestion adaptive method provides a congestion adaptive control algorithm based on up semi-normal distribution-DARED (distributed adaptive random early detection). According to the DARED algorithm, the corresponding congestion control is performed by improving an original data packet discarding probability function curve, namely adjusting an existing linear discarding straight line of an RED (random early detection) algorithm to a smooth curve, and dividing into different network states according to the queue length. The DARED is characterized in that a real-time dynamic parameter Pmax is introduced on the basis of the traditional RED algorithm, the stability of a queue is kept and the controllability of network congestion is increased. Correspondingly, in the aspect of a nomadic application system, when applied to the aspect of a multimedia seamless migration system, the congestion control algorithm realizes the seamless migration strategy under different architectures, namely the migration strategy under a B/S (browser/server) architecture, the migration strategy under a P2P (peer-to-peer) architecture and the migration strategy under a C/S (client/server) architecture. By adopting the nomadic application-oriented network congestion control method, the continuity, the rapidity, the smoothness and the correctness of media migration can be effectively ensured.

Description

A kind of method for controlling network congestion towards nomadic application

[technical field]:

The invention belongs to the technical field that computer network, communication and Digital Media etc. combine.

[background technology]:

Fusion along with mobile communication and mobile computing technology, the progressively maturation of mobile technology, the application of mobile technology is mutual with the movement that development brings, for connection online, communication and information exchange whenever and wherever possible provides possibility, simultaneously for mobile working provides new opportunities and challenges.Along with the expansion of network application scale and increasing of bearer service, the increase of network load, the local appearance of network is congested, and throughput declines, and real-time Transmission audio frequency and video on the network under nomadic environment, must take to prevent the measure of network congestion.How effectively to avoid traffic congestion, improve network performance and operational efficiency, guarantee that the different demands of user on multimedia information become the focus that much research is paid close attention to simultaneously.

The subject matter of congestion control mechanism in communication environment is that the network of various mixing exists simultaneously, comprising coexisting of broadband networks, arrowband net, various local area network (LAN), wireless network, wired network, how to realize transmission data smooth between these networks, will become the important bottleneck of communications applications development.

The reason that network congestion control to produce is: when the available resources that provide when network can not meet user's demand, the transmission performance of network will decline, will produce so congested, because the network under nomadic application is the network of isomery, network environment more complicated, the inconsistency of the interface of various terminal equipments, diverse network and deposit, will inevitably cause the congested of network.The subnet in a network occurs that the grouping forwarding is too much, and the disposal ability of network is limited, that is to say that the load of network has surpassed the ability to bear of network, and network has occurred congested.Congested the following aspects that is mainly manifested in of network: propagation delay time increase end to end, the hydraulic performance decline of system are, the packet loss of network etc.Wherein the immediate cause of congested generation comprise following some:

(1) memory space inadequate: if several TCP stream sends data to a router side simultaneously, router side storage capacity is limited, the reception data capability of receiving terminal is limited simultaneously, at this moment, need to process these packets by setting up a queue, but owing to being subject to the restriction of the factors such as network equipment physics, the limited length of queue, some packet must abandon, and what for this situation, produce is congested, increases memory space and just can address this problem.

(2) bandwidth is too narrow: when several TCP stream sends data to a router simultaneously, when especially high-speed link is to low speed chain circuit transmission data, because the ability that link forwards is different, low speed chain circuit is easy to form bottleneck, be easy to like this produce congested, the bottleneck that this situation forms, can be resolved by increasing bottleneck bandwidth.

(3) processor disposal ability is limited: if the processing speed of CPU, lower than the transmission speed of high-speed link, also easily produces bottleneck, form congested.

Network has the existence of a large amount of destabilizing factors in data transmission procedure, data due to Internet have sudden simultaneously, be prone to a large amount of pending packets, and Internet of Things is as a kind of emerging, complicated latticed form, sudden inevitable especially, therefore, want to solve sudden under nomadic environment, must find a suitable algorithm, this algorithm must be able to process a large amount of burst data bag of automatic network.

[summary of the invention]:

The smooth problem of communication that the present invention seeks to solve the network congestion under nomadic application model and apply under multimedia seamless migration system, to guarantee the smooth clog-free operation of network, promote the development and application of nomadic communication, thereby proposed congested adaptive control DARED (the Distributed Adaptive RED) algorithm based on rising half normal distribution, and analyzed theoretically the system feasibility of this algorithm under nomadic applied environment.

The concrete steps of the method for controlling network congestion towards nomadic application provided by the invention:

1st,, when having grouping to arrive router, adopt RED algorithm to calculate average queue length Q according to low pass filter _av, i.e. the mean value of the queue length on the unit interval.Can be obtained by formula below:

Q _av＝(1-w _q)×Q _av+w _q×q

Wherein, w _qbe expressed as the time coefficient of being determined by low pass filter, q is expressed as the length of active queue while measuring.

2nd, the average queue length Q the 1st step being calculated _avmax threshold threshold values Q with drop probability curve _maxwith minimum threshold threshold values Q _mincompare, according to comparative result, current Congestion Level SPCC be divided into three kinds of states, be respectively:

Between, average queue length is in Q _minand Q _maxbetween, i.e. queue is in stable state, and states is set between;

Above, average queue length is greater than Q _max, the state of state is set to above state;

Bellow, average queue length is less than Q _min, the state of states is set to bellow;

3rd, according to the definite Congestion Level SPCC of the 2nd step, carry out dynamic self adaptation adjustment, take corresponding data packet discarding probability, hold queue length, in stable scope, is avoided unnecessary shake, and concrete operations are:

When continuous 2 time intervals are during in below state, think that the utilance of link is low, by adjusting p _maxvalue improves the utilance of link;

When continuous 2 time intervals are during in above state, think that link occurs congested, dynamically adjust p _maxvalue, increase and to send congested notice to source, to make source reduce transmission rate, thereby alleviate congested.

Described method for controlling network congestion is to realize by the seamless migration fast of the media under different frameworks, and concrete grammar is as follows:

Migration strategy under B/S framework: use Ajax technology as the technology of front end data migration.Ajax is the fusion of several technology, i.e. fusion between asynchronous JavaScript and XML.For the pushing module of service end, realize data-pushing module that we have adopted Comet art designs simultaneously.

Migration strategy under P2P framework: used emphatically the ripe like this technology of JMF to be realized, utilize the SessionManager interface that this technology provides to realize well the congestion control in Streaming Media transition process simultaneously, excavation based on key frame and the method for migration have been studied on this basis again, the key code of having designed receiver module and action migration event module, has guaranteed QoS;

Seamless migration strategy under C/S framework: used emphatically the ripe like this control of ActiveMovie to construct player, and used location-based cutting shift function and congestion control in audio frequency and video file transfer, guaranteed the efficiency of migration.

Advantage of the present invention and good effect:

Proposed a kind of network congestion adaptive control algorithm based on rising half normal distribution herein, by improving original data packet discarding probability function and maximum drop probability parameter value, controlled accordingly.Utilize NS2 simulation software to carry out emulation, experimental result shows, network congestion adaptive control algorithm based on rising half normal distribution can adapt to the variation of the network congestion environment of Internet of Things complexity, the stability that has kept queue length, is better than traditional RED algorithm aspect in the variation of data packet discarding probability.

Inquired into a kind of mechanism that can accurately distinguish efficiently data-bag lost reason herein, and taked different measures according to different congestion states, effectively Suppression network is congested to utilize this mechanism, improves network performance.

Propose simultaneously and designed the seamless migration strategy of three kinds of different frameworks based on this congestion control mechanism herein, be seamless migration strategy, the seamless migration strategy under P2P framework and the seamless migration strategy under C/S framework under B/S framework, guarantee efficiency, the qualitative and validity of seamless migration, thereby there is feasibility under nomadic system.

[accompanying drawing explanation]:

Fig. 1 is network analog topological diagram;

Fig. 2 is RED average queue length and packet loss graph of a relation and the probability dropping based on rising half normal distribution;

Fig. 3 is three kinds of state transition graphs;

Fig. 4 is program flow diagram;

Fig. 5 is that while being sent as 10k/s under RED and DRED algorithm, queue length changes;

Fig. 6 be RED and DRED algorithm transmission rate while being 10k/s packet loss change;

Fig. 7 is the variation of RED and DRED algorithm transmission rate queue length while being 10m/s;

Fig. 8 be RED and DRED algorithm transmission rate while being 10m/s packet loss change;

Fig. 9 is that when under RED and DRED algorithm, speed changes suddenly, queue length changes;

Figure 10 is that when under RED and DRED algorithm, transmission rate changes suddenly, packet loss changes;

Figure 11 is TCP throughput and the actual bandwidth comparison of DRED and other several algorithms;

Figure 12 is the Voice & Video seamless migration design sketch under B/S framework;

Figure 13 is the video seamless migration design sketch under P2P framework;

Figure 14 is the video seamless migration design sketch under C/S framework.

[embodiment]:

The concrete technical scheme of the method for controlling network congestion towards nomadic application provided by the invention is as follows

It is exactly to be average queue length Q by existing parameter that the 1st DRED algorithm calculates _av, max threshold threshold values Q _maxwith minimum threshold threshold values Q _minaccording to the new probability formula that requires substitution data packet discarding (1) of fuzzy theory normal distribution membership function, thereby change linearity and abandon curve, can carry out parameter configuration according to the congestion situation of the network environment under nomadic pattern simultaneously, reduced the sensitiveness that algorithm arranges for parameter, be max threshold threshold values and minimum threshold threshold values, by queue length control between max threshold threshold value and minimum threshold threshold value.As shown in Figure 2, the straight line of linear change is become to the round and smooth luminance curve that rises half cosine normal distribution.

The new probability formula of data packet discarding is as follows:

$p=\left\{\begin{array}{cc}0& 0\&le;Q_{\mathrm{av}}<Q_{\min }\\ P_{\max }\&times;\frac{Q_{\mathrm{av}}-Q_{\min }}{Q_{\max }-Q_{\min }}& Q_{\min }\&le;Q_{\mathrm{av}}<Q_{\max }\\ 1& Q_{\max }\&le;Q_{\mathrm{av}}<B\end{array}\right.---\left(1\right)$

First

Rise half normal distribution and transform and obtain by normal distribution, the same with normal distribution, be to there is two parameter μ and σ ²continuous type random distribution, wherein μ is the average of normally distributed random variable, σ ²be variance of a random variable, rise half normal distribution formula as follows:

Setting μ is average queue queue Q _av; β is minimum threshold threshold values Q _min; By Q _avand Q _minwe obtain data packet discarding new probability formula substitution formula (2):

$P\left(Q_{\mathrm{av}}\right)=\left\{\begin{array}{cc}0& Q_{\mathrm{av}}\&le;Q_{\min }\\ 1-e^{-\mathrm{\&alpha;}{(Q_{\mathrm{av}}-Q_{\min })}^2}& Q_{\mathrm{av}}>Q_{\min }\end{array}\right.---\left(3\right)$

For convenience of calculation, we set k is max threshold threshold values Q _maxwith minimum threshold threshold values Q _minpoor, following formula:

k＝Q _max-Q _min (4)

According to the characteristic of RED algorithm, we know, when queue length is max threshold threshold values Q _maxtime, the drop probability of packet is p _max, can obtain α and be:

$\mathrm{\&alpha;}=-\frac{\ln (1-P_{\max })}{k^2}---\left(5\right)$

From formula, we can find out, α value is by p _maxand determine.

Bring the value of parameter alpha into just can obtain based on rising half normal distribution grouping and abandon distribution probability function:

$P\left(Q_{\mathrm{av}}\right)=\left\{\begin{array}{cc}0& Q_{\mathrm{av}}\&le;Q_{\min }\\ 1-{(1-P_{\max })}^{-{\left(\frac{Q_{\mathrm{av}}-Q_{\min }}{k}\right)}^2}& Q_{\mathrm{av}}>Q_{\min }\end{array}\right.---\left(6\right)$

From formula, can find out, the data packet discarding probability function based on rising half normal distribution is with max threshold threshold values Q _maxwith minimum threshold threshold values Q _minrelevant, we find out simultaneously, along with waiting for the variation of forwarding queue average length, improve the hybrid network environment that algorithm more can adapt to Internet of Things complexity.

Second

When grouping arrives router, setting maximum queue length is max threshold threshold values Q _maxwith minimum queue length be minimum threshold threshold values Q _min, by the mean value Q of formula queue length _av=(1-w _q) * Q _av+ w _q* q, obtains average queue length Q _avif queue length is between max threshold threshold value and minimum threshold threshold value, in stable state, there is not network congestion in network state.

As average queue length Q _avbe not in Q _minand Q _maxbetween time, we consider in two kinds of situation:

1. average queue length is less than minimum threshold threshold value, and drop probability is less than 0.5:

If queue length is less than minimum queue length, increase transmission rate; If queue length is greater than minimum queue length, reduces to abandon speed p and keep the stable transmission rate of source.

2. average queue length is greater than minimum threshold threshold value, and drop probability is greater than 0.5:

If queue length surpasses maximum queue length, increase and abandon speed p; If queue length is less than maximum queue length, keep stable queue length.

As average queue length Q _avat Q _minand Q _maxbetween time, we consider in two kinds of situation:

1. as average queue length Q _avmore near Q _minwhen neighbouring,

the running status of network is good, now should reduce the value of drop probability p, increases or keep the transmission rate of source, increases the utilance of bandwidth;

2. as average queue length Q _avmore near Q _maxwhen neighbouring,

the load of the network operation starts to increase, and now should increase data packet discarding Probability p value, can effectively prevent like this generation of network congestion.

So we are divided into three kinds of states by congestion control: (queue length is in Q to be respectively Between _minand Q _maxbetween), (queue length is greater than Q to Above _max), (queue length is less than Q to Bellow _min).As shown in Figure 3.

One, Between: queue length is in Q _minand Q _maxbetween time, the transmission of network is in good condition, keeps transmission rate;

Two, Above: queue length is greater than Q _max, the value of increase data packet discarding probability, notice source reduces transmission rate;

Three, Bellow: queue length is less than Q _min, the network operation is in good condition, and bandwidth is idle in a large number, and notice source increases transmission rate;

The 3rd

According to above analysis, we have designed based on rising the adaptive DARED algorithm of half normal distribution, and Fig. 4 is flow chart, and specific algorithm is analyzed as follows:

Meaning of parameters:

Qav: queue length;

P _max: maximum drop probabilities value;

BufferSize: current buffer size.

2nd, by improved RED algorithm application at multimedia seamless migration system aspects:

Along with the development of Internet technology explosion type, user needs network that a greater variety of services are provided, and guarantees good communication performance.Therefore, in the situations such as resource migration, network congestion or the system failure, how providing continuously and stably high-quality service, is the hot issue of current internet area research.Seamless migration technology proposes in order to overcome the above problems, for stability and the reliability of internet communication provides good basis.The present invention proposes the seamless migration mechanism based on improved RED algorithm, and before and after migration, server can have different application protocols and host-host protocol, and communication can continue to carry out, unaffected.It is in order to guarantee the continuity of resource acquisition in communication process that the control of network congestion is applied in multimedia seamless migration system, has solved due to the communication disruption problem of bringing in resource migration process.User's impression, less than the migration connecting in communication process, guarantees stability and the reliability of service.The test of platform by experiment, shows that this scheme meets the functional requirement of network service migration substantially.

The realization of the multimedia seamless migration strategy under B/S framework

At Browser end, the function of audio frequency and video control module mainly concentrates on the control of the audio player to embedding in webpage.Control operation mainly comprises play operation, pausing operation, obtains break-point operation, migration operation.The playing module that we use is the MediaPlayer webpage player based on embedding in webpage.The self-defining control function relevant to this playing module has: play (), pause () and stop () function; The relevant parameter of obtaining audio frequency and video play position, we are defined as currentPosition.Below just the code of corresponding control function is shown as follows:

Play () is for realizing the broadcasting of the designated tone video media of assigned address.Code is as follows.

In order to complete the function of the pushing module of service end, we must be at Browser end definition function of the same name.The function of the same name is here update (Position, Name).The code of its realization is as follows.

Update () function is for realizing the realization that coordinates service end pushing module.The value of giving relevant parameter is media name and breakpoint size, makes to play.

In order to realize corresponding function by button click, need to use <inpu/> label to create corresponding required button element, and then, be further integrated into Browser end.Arrive this, it is complete that Browser holds important modularity function to be created.Need to coach, use <div> label to realize the demonstration of position position sign.<iframe> label is for showing other pages.

In order to realize, play all audio frequency and video, we have specially studied again RealPlayer playing module.RealPlayer playing module code omits.Similar with MediaPlayer webpage player.Need this player code to be embedded in <body></bodyGrea tT.GreaT.GT label.Have a look its main control function code.Main function code is as follows.

Play () function plays for realizing in the position of appointment the audio-video frequency media arranging.This power function code is as follows.

GetPosition () function is used for obtaining breakpoint location, display media breakpoint location on the page.This power function code is as follows.

Equally, in order to send POST request and to realize propelling movement function performance, respectively definition: CometTransfer () function and Update (Position, Name) function.Its function code and MediaPlayer playing module are similar, at this, no longer provide.

The realization of file without seam transferring module mainly relies on <iframe></iframe > and two labels of <form></formGrea tT.GreaT.GT.Meanwhile, the building of environment that also needs jspsmartupload.jar bag and Apache Server.First, used <iframe> label that the webpage needing is embedded, by this interface, carry out upload file, by CometTransfer () functional based method, send POST request afterwards, finally, in the browser of Client end, by <Ahref> label, created hyperlink, so that Client end is downloaded.

Its main function code or algorithm are as follows.

Update (Position, Name) algorithm is achieved as follows.

Create corresponding hyperlink.

CometTransfer () function is for transfer parameter and parameter value.Code is as follows.

Certainly, for the seamless migration of file, also can be upload and download Function Integration Mechanism simultaneously.So only need, the function of uploading download just can be realized in the address of the webpage that migration is uploaded, and then realizes the function of seamless migration.

Under B/S framework, Comet ^[34]pushing module is the most difficult realization.The implication of CometContext is context, can acquire change and the communal space that can migration data when environmental context changes; The implication of CometEvent is event, can comprise the object of CometContext, and for response, steering procedure is processed the generation of some events; The implication of CometHandler is handle, can realize the interface of mutual CometContext.The step that realizes Comet pushing module is as follows.

The first step, the contextual establishment of Comet or registration.In an application program, create a new Comet context, or register to a Comet context having existed.The object of doing is like this to allow application program shared environment share context.

Second step, establishment handle.Be generally each client or service end and connect a Handler handle of establishment, and this handle is joined and shared in context, the object of doing is like this in order to utilize handle operation to share contextual data.Handle is similar with the handle meaning defining in VC++.

The 3rd step, client create long connection.Be generally each client and create one to sharing contextual one " long connection ".The object of doing is like this to keep " long connection ", realizes the real-time update, propelling movement of data etc.

The interaction of the 4th step, client.When a Client end carries out the renewal of data, just by length, connect to push data into and share in context environmental, other Client end is updated thereupon.

The 5th step, context environmental change, and client data changes.When context generation Data Update, all there is corresponding variation in the long Client end connecting of all maintenances.

Relevant container function in Comet application pushing module: doPost (), onEvent ().Its succinct implementation algorithm is as follows respectively:

DoPost () function is mainly used in intercepting and capturing parameter, thereby obtains parameter value, moreover is to wake onEvent () function up.Its implementation algorithm is as follows.

Character code mode is set;

Obtain position and resource name information;

Initialized location and resource name information;

Define a CometEngine object;

Define a CometContext object;

Define a CometSelector object.

OnEvent () function is for being pushed to client by data message.Its implementation algorithm is as follows.

Define a PrintWriter object;

Realize the propelling movement of object data;

Refresh PrintWriter object.

The realization of the Streaming Media seamless migration strategy based on P2P framework

We adopt the audio frequency and video seamless migration strategy based on JMF technology.The class that we realize has been inherited JFrame frame clsss, and has realized the monitors such as ActionListener, ControllerListener and ItemListener.Wherein, comprised response window event monitor, use adapter adapter for to the response of the X button of System menu, the response that menu is entered, use e.getActionCommand () .equals (" ") to realize the judgement of click event and respond etc. to operate.The interface of the playing module that JMF is used is Player, and this object flows into audio frequency and video as data flow, subsequently, this audio/video flow is led in corresponding audio & video equipment, and we are by sound equipment, screen as an example here.

Player needs just can carry out the broadcasting of audio frequency and video through six kinds of states in the use.We are existing is listed below these six kinds of states, respectively is Unrealized, Realizing, Realized, Prefetching, Prefetched and Stated.Wherein, every kind of state is representing its literal meaning as its title.As Stated state representation is called after start () function, this Player object just enters the state of being ready to and can play at any time the audio/video flow of loading.

Under P2P framework, for the realization of audio-video frequency media seamless migration, we only provide the key code of receiver module and action migration event module.Key code is as follows.

Receiver module key code is as follows.

Action migration event module key code is as follows.

In order to arrange, to obtain and transit time object, we have defined TimeS class, for realizing the serialization of object.

The realization of the Streaming Media seamless migration strategy based on C/S framework

Under C/S framework, we also need the function realizing to have congestion control, the transmission based on small documents and Socket transfer function.

Transferring module implementation algorithm is as follows.

Realize the shift function of data;

The type of conversion breakpoint data is that breakpoint location is converted to CString type by Float type;

Create Socket socket;

Realize File cutting;

Realize fritter file migration.

Receiver module implementation algorithm is as follows.

Realize the reception of data message;

Character string and the operation to dependent variable are keyed in exchange interface;

For scheduling thread.

Thread module implementation algorithm is as follows.

Definition CActiveMovie3 type pointer variable;

Create socket, first parameter represents the region that designated communication occurs; Second parameter represents Socket type;

Realize the function of Socket receiving data information;

Resolve the character string that client transmits;

Carry out the control of playing module;

Played file.

DARED algorithm based on rising half normal distribution has been proposed herein, by changing the data packet discarding probability curve of the random earlier detection RED algorithm of tradition, utilize existing luminance curve to replace original linear change curve, thereby increased the utilance of bandwidth, alleviated the congestion condition of network, from point of theory, proved that the DARED algorithm based on rising half normal distribution can meet the requirement of the network environment nomadic application.At system aspects, the media migration strategy under B/S framework has been used congestion control new method and has combined the part that Ajax technology realized Asynchronous Request, webpage and refreshed, and takes full advantage of Internet resources, has reduced the waste of bandwidth.Realize the migration of type file that browser is supported, as text type, Doctype, realized the seamless migration of wireless wide area network, indirectly improved the performance of mobile device.Under C/S framework, we are again from the consistent wrong-doing that transmits whole file of Angulation changes of cutting file, and especially for large file, its transmission efficiency can be very significantly improved.In load mode, there is the function of breakpoint transmission.Under P2P pattern, use congestion control new method-Dynamic Acquisition network condition information to change buffer size, improved efficiency of transmission.At hardware aspect, we have used the mechanism of many network interface cards and have adopted hard disk texture storage technology.Finally, realized the seamless migration strategy under three kinds of heterogeneous networks frameworks.

From Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9 and Figure 10, we can find out: it is better that the DARED algorithm based on rising half normal distribution shows in the stability of hold queue length than traditional RED algorithm when TCP transmission rate is 10kb/s; In the situation that transmission rate is 10mb/s, the situation of network faces will be severeer, what we found that improved DARED algorithm can be very fast adjusts data packet discarding probability according to change of network environment, notification source transmitting terminal changes transmission rate timely, adjust the various parameters of network, can be faster, better hold queue length, prevent congested generation.Thereby effectively prevented the generation of a large amount of packet loss events, can meet the network complex environment of Internet of Things, kept the good throughput of network, improved the performance of network.

Figure 12, Figure 13 and Figure 14 are system aspects Voice & Video seamless migration design sketchs, have all shown accuracy and the uniformity of information before and after migration.The improved Congestion Control Algorithm DRED that this patent adopts, by controlling the delay inequality of relevant propagation delay time shake and the migration of adjacent data bag, rapidity, flatness and correctness while having guaranteed media migration, test test case has shown that this kind of seamless migration strategy is effectively, and corresponding several designs have made up the deficiency of legacy network mechanism effectively.The realization of the method will solve the migration of the media of nomadic application, serves better pervasive family, mobile learning, mobile working, Mobile business etc.

Concrete application

From application experiment result, in the heterogeneous network environment that this DARED congestion control algorithm based on rising half normal distribution coexists at wireline-wireless, show better performance, we can find comparison by table 11, the throughput of this algorithm is the highest, actual average bandwidth is the highest, network resource utilization is high, change steadily, the transmission speed that network is described is more stable, the network traffics of burst have been made to congestion control in time to be processed, met the performance requirement of nomadic application, can illustrate that in sum network is in bellow, when between and tri-kinds of states of above, DRED algorithm is all cashed out better performance.

Under B/S framework, with mobile phone as use-case, for picture, text, we can describe its seamless migration process like this: after directly obtaining the Information Monitoring after processing, derive network address, and open this webpage, touch transfer button, the control module in self-organizing module just can be obtained contextual information, and picture or text are moved on the display device of another appointment.

When client buys clothes in market, rfid interrogator has obtained the information of these commodity, just can on the browser of having specified, play the video of relevant this kind of clothes, as video " Show ".Client is in order to allow oneself and household's satisfaction just can use this strategy to carry out the migration of corresponding video.Be similar to mobile phone floor, the migration of the scene photograph based on Web page; Be similar to the head portrait network legal power authentication based on Web page.

Audio and video data streams under P2P framework can be applied to following situation: when the video flowing that surveillant obtains by camera or detectaphone in concern or audio stream, when seeing or the suspicion part of hearing just carries out, by video flowing other surveillants that lead again, being conducive to like this improve investigation speed.Because be subject to the restriction of space size, so just realized no matter monitor in which room all can watch or uppick suspicion part, be convenient to the breakthrough of case.Be similar to, can carry out the mutual biography of file, information simultaneously, reach the effect that the chat of mutual biography video segment limit is seen on a kind of limit.

Under C/S framework, for the migration situation under environment of internet of things, we have designed such migration situation: by camera, obtain video flowing, pass through Intemet, in data center, carry out cloud computing, store, after just " guiding " application end play, when needs move, just the service end that audio-video document led, continues to play.

Network congestion control and system table based on nomadic application model in this paper revealed good flexibility and the stability under different network congestion degree, computation complexity is not high, compare and there is less network resource consumption with traditional algorithm, do not affect efficiency of transmission.Emulation experiment has also proved the validity of this algorithm, in performance and data, can aspect property, all than traditional method, be improved to some extent.This congestion avoidance algorithm is applied to mobile communication system, proof has feasibility by experiment, the Multimedia Task that can guarantee user can be carried out seamless migration pellucidly whenever and wherever possible between distinct device, and typical application comprises: mobile office (E-working), mobile learning (E-learning), mobile working (E-working), Mobile business (E-bussiness), mobile government (E-Office), stock exchange (E-Exchanging), intelligence commander (E-commanding) etc.In office what is the need for of user can be selected the online stream medium data of access mutual with it with various available devices when wanting, and typical application comprises: video request program, online game, vision security monitoring.

Claims (2)

1. towards a method for controlling network congestion for nomadic application, it is characterized in that the concrete steps of the method comprise:

Existing linearity is abandoned to straight line and be adjusted into round and smooth curve, can carry out parameter configuration according to the congestion situation of the network environment under Internet of Things pattern, reduce the sensitiveness that algorithm arranges for parameter simultaneously;

DARED introduces real-time dynamic parameter Pmax to replace original static parameter on the basis of traditional RED algorithm, and the stability of hold queue, has eliminated the drawback of original RED algorithm, thereby has increased the congested controllability of network; DARED algorithm design is as follows: existing parameter, according to the substitution that requires of fuzzy theory normal distribution membership function, is abandoned to curve thereby change linearity;

Wherein the membership function of fuzzy liter half normal distribution rises the membership function of half normal distribution as formula:

$u_{^}\left(\mathrm{\&mu;}\right)=\left\{\begin{array}{cc}0& \mathrm{\&mu;}\&le;\mathrm{\&alpha;}\\ 1-e^{-\mathrm{\&alpha;}{(\mathrm{\&mu;}-\mathrm{\&beta;})}^2}& \mathrm{\&mu;}\&GreaterEqual;\mathrm{\&alpha;},\mathrm{\&alpha;}>0,\mathrm{\&beta;}<0\end{array}\right.$

Setting μ is average queue queue Qav; β is minimum threshold threshold values Qmin, sets:

k=Q _max-Q _min

According to the characteristic of RED algorithm, when queue length is max threshold threshold values Qmax, the drop probability of packet is Pmax, and α is:

$\mathrm{\&alpha;}=-\frac{\mathrm{In}(1-P_{\max })}{k^2}$

Grouping based on rising half normal distribution abandons distribution probability function:

$P\left(Q_{\mathrm{av}}\right)=\left\{\begin{array}{cc}0& Q_{\mathrm{av}}\&le;Q_{\min }\\ 1-(1-P_{\max })-{\left(\frac{Q_{\mathrm{av}}-Q_{\min }}{k}\right)}^2& Q_{\mathrm{av}}>Q_{\min }\end{array}\right..$

2. method according to claim 1, is characterized in that described method for controlling network congestion is to embody by the seamless migration fast of the media under different frameworks, and concrete grammar is as follows:

Migration strategy under B/S framework: use Ajax technology as the technology of front end data migration; Ajax is the fusion of several technology, i.e. fusion between asynchronous JavaScript and XML; For the pushing module of service end, realize data-pushing module that we have adopted Comet art designs simultaneously; In the Comet application, we provide realized with lower module: doPost () module and onEvent () module; DoPost () function algorithm: be mainly used in intercepting and capturing parameter, wake onEvent () function module up;

Migration strategy under P2P framework: used emphatically the ripe like this technology of JMF to be realized, and realized ActionListener, ControllerListener and ItemListener monitor; Utilize the SessionManager interface that this technology provides to realize well the congestion control in Streaming Media transition process simultaneously, excavation based on key frame and the method for migration have been studied on this basis again, design receiver module and action migration event module, guaranteed QoS;

Seamless migration strategy under C/S framework: used emphatically the control that ActiveMovie is ripe like this to construct player, the function of realization has a congestion control, based on File cutting fritter file transfer, Socket transfer function; And in audio frequency and video file transfer, designed the function of location-based cutting shift function and congestion control, guaranteed the efficiency of migration.

Images