CN102026203B - Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network - Google Patents

Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network Download PDF

Info

Publication number
CN102026203B
CN102026203B CN 201010592782 CN201010592782A CN102026203B CN 102026203 B CN102026203 B CN 102026203B CN 201010592782 CN201010592782 CN 201010592782 CN 201010592782 A CN201010592782 A CN 201010592782A CN 102026203 B CN102026203 B CN 102026203B
Authority
CN
China
Prior art keywords
subnet
node
network
reliability
path
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN 201010592782
Other languages
Chinese (zh)
Other versions
CN102026203A (en
Inventor
郭成城
李文翔
吴静
张洋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN 201010592782 priority Critical patent/CN102026203B/en
Publication of CN102026203A publication Critical patent/CN102026203A/en
Application granted granted Critical
Publication of CN102026203B publication Critical patent/CN102026203B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Mobile Radio Communication Systems (AREA)

Abstract

The invention relates to the technical field of combining wireless network technology and internet protocol (IP) speech technology, in particular to a multiple-session initiation protocol (SIP) server layout method in a wireless Mesh network. The method comprises the following steps of: measuring the communication quality of a wireless link in the wireless Mesh network; laying out SIP servers; and selecting routes so that a router accesses the SIP servers. The method has the following main advantages that: 1, the SIP servers are arranged in each partition region for providing calling access service as possible when the network is partitioned so as to ensure that speech communication functions in each region can be normally used; and 2, a group of reliable paths leading to the SIP servers which are laid out at different positions are selected for SIP agents on all the router nodes, so that the influence of single-point fault of the wireless link and the servers can be eliminated, and access loads are distributed on the selected server nodes. The method solves the problem that the instability of the wireless link has negative effect on an SIP-based speech calling service in the wireless Mesh network.

Description

Many sip servers layout method in a kind of Wireless Mesh network
Technical field
The present invention relates to the technical field that radio network technique combines with ip voice (VoIP) technology, relate in particular to many sip servers layout method in a kind of Wireless Mesh network, relate to specifically how at wireless mesh network (Wireless Mesh Network, high reliability is provided WMN), dialogue-based initiation protocol (Session Initiation Protocol, SIP) audio call access service, designed and how in network, to have chosen suitable node and lay sip server, and how the sip agent on every wireless router chooses the methods such as path of server in the network.
Background technology
Wireless Mesh netword combines the advantage of WLAN (wireless local area network) and mobile ad-hoc network.WMN is combined application with SIP, can for certain unit or department provide fast, flexibly, and have the voice communications services of high performance-price ratio.Particularly in the communications applications of particular surroundings and emergency management and rescue, good market prospects are arranged.
Raheleh B.Dilmaghani [1,2,3,4]Develop the medical treatment reaction manoeuvre when a hybrid WMN test platform is used for the crisis generation, and analyzed the reason that produces network bottleneck, causes network service to be blocked; Abdulrahman Yarali [5]Analyze the technical need and the background that are used for emergent and public safety communication, pointed out the technique influence of WMN.Asad Amir Pirzada [6]Point out that current Ad-hoc Routing Protocol is not suitable for hybrid WMN, can't set up path stable, high-throughput; Muddesar Iqbal [7]Proposed to be used for the design and implementation methods of WMN of the multicast dual-radio of emergency communication.
For the application that SIP is combined with wireless network, Sebastian Fudickar [8]Propose a kind of SIP nerve of a covering for mobile Ad hoc network, paid close attention to the availability of registration and positioning service, and managed to reduce the load of transmission of messages.Nilanjan Banerjee [9]Two kinds of solutions of setting up the SIP session in Ad hoc network have been proposed: 1) loose coupling scheme, the sip terminal discovery procedure separates with routing procedure; 2) close coupling scheme, adopt fully distributed, based on the Routing Protocol of cluster, generate the virtual topology of an energy high usage route, and combine with the terminal discovery procedure.Marcel Cavalcanti [10]Studied the problem of Application standard SIP service in being communicated to the MANET of Internet, analyzed the limited problem of SIP service extension, and proposed a kind of replacement scheme.Bo Rong [11]Studied the Session initiation Protocol (SIP) that is used for providing at WMN the IP network wireless voice service, point out relevant technical problem (calling establishment time delay, bandwidth prediction and reservation, Admission Control), proposed the design of an improved sip proxy server.
Above-mentioned work mainly is to conduct a research and innovate for aspects such as network system framework, channel quality, network capacity, route, multicast, signaling callings, for the research of communications network system reliability and self-organization aspect seldom, there is not special design.
At present, mostly concentrate on for the research of SIP service system reliability SIP is combined application with the P2P technology.Document [12] has proposed a kind of structure of introducing P2P bunch of shape net on the SIP basis, as the scheme that realizes the VoIP communication system at MANET, has analyzed main operation mechanism, has adopted the strategy as super node with CPU, internal memory, the maximum persons of bandwidth.Adopt in the document [13] based on the method for P2P and carry out SIP registration and resource location, design philosophy is to adopt as far as possible all operations of the sip message realization P2P of standard, and the system of having introduced mainly operates design and the test of (registration, deletion etc.) module.Document [14] has been developed a communication system of supporting voip conversation control based on JXTA platform and Session Initiation Protocol stack, and main feature is to use P2P issue and location mechanism to replace registration and the query script of centralized SIP.Document [15] and [16] have made up a network phone system based on SIP signaling and P2P framework, by the P2P technology being incorporated into single point failure and the performance bottleneck problem that solves centralized SIP system in the SIP networking telephone.Document [17] has been discussed in the fairly large P2P SIP network, the switching problem when terminal moves between a plurality of P2P overlay area.Document [18] utilizes the P2P technology to solve the problem of collaborative work between a plurality of sip servers.
Owing to affected by decline, interference, multipath effect, obstruct etc.; provisional communication quality usually can occur and descend or outage in Radio Link; sometimes the duration is lacked (tens seconds) to this phenomenon; sometimes the duration is grown (a few minutes~dozens of minutes), makes link fracture, network topological change.In case this situation occurs, will cause certain customers can't login sip server, and can not establish a communications link.One of its reason is that user terminal is gone on the path of sip server certain wireless links fault is arranged, and causes and can't establish call connection.Although adaptive routing algorithm can find new path, before the routing algorithm convergence, still can cause sip server unreachable, in cable network, also there is same problem.Former two is that if cut zone does not comprise the sip server node, then whole cut zone can't be worked after appearing at network and cutting apart therefore.
For this problem, method of expecting easily be with a SIP system configuration of having simplified in all nodes of network, consist of full distributed SIP system.But, the people such as Simone Leggio [19]Studies show that the full SIP system that distributes needs a large amount of broadcasting user log-on messages, network bandwidth consumption is very large, is only applicable to very small-scale network system.
List of references:
[1]Dilmaghani Raheleh B.,Rao Ramesh R.Hybrid wireless mesh network with application to emergency scenarios[J].Journal of Software,2008,3(2):52-60.
[2]Dilmaghani Raheleh B.,Rao Ramesh R.A reliable wireless mesh infrastructure deployment at crisis site[C].Proceedings of the IEEE International Performance,Computing,and Communications Conference,2007,579-581.
[3]Dilmaghani Raheleh B.,Rao Ramesh R.On designing communication networks for emergency situations[C].Proceedings of International Symposium on Technology and Society,2006.
[4]Dilmaghani Raheleh B.,Rao Ramesh R.Hybrid wireless mesh network deployment:A communication test bed for disaster scenarios[C].Proceedings of the 1st ACM International Workshop on Wireless Network Testbeds,Experimental Evaluation and Characterization,2006,90.
[5]Yarali Abdulrahman,Ahsant Babak,Rahman Saifur.Wireless mesh networking:A key solution for emergency & rural applications[C].Proceedings of the 2nd International Conference on Advances in Mesh Networks,2009,143-149.
[6]Pirzada Asad Amir,Portmann Marius,Wishart Ryan,et al.SafeMesh:A wireless mesh network routing protocol for incident area communications[J].Pervasive and Mobile Computing,2009,5(2):201-221.
[7]Iqbal Muddesar,Wang,Xingheng,Wertheim,David,et al.SwanMesh:A multicast enabled dual-radio wireless mesh network for emergency and disaster recovery services[J].Journal of Communications,2009,4(5):298-306.
[8]Fudickar Sebastian,Rebensburg Klaus,Schnor Bettina.MANETSip-A dependable SIP overlay network for MANET including presentity service[C].Proceedings of the 5th International Conference on Networking and Services,2009,314-319.
[9]Banerjee Nilanjan,Acharya Arup,Das Sajal K.Enabling SIP-based sessions in ad hoc networks[J].Wireless Networks,2007,13(4):461-479.
[10]De Castro Marcel Cavalcanti,Kassler Andreas J..Challenges of SIP in internet connected MANETs[C].Proceedings of 2nd International Symposium on Wireless Pervasive Computing,2007,447-452.
[11]Rong Bo,Qian Yi,Chen,Hsiao-Hwa.An enhanced SIP proxy server for wireless VoIP in wireless mesh networks[J].IEEE Communications Magazine,2008,46(1):108-113.
[12] Yu Fan, based on the application study of VoIP system in MANET of SIP, communication of power system, the 172nd phase of the 28th volume in 2007,41~44 pages.
[13] Bai Yu, flood fly, based on the instantaneous communication system of P2P Session Initiation Protocol, and computer system application, the 2nd phase in 2009,14~17 pages.
[14] virtuous, the Zhou Jiang in Wuzhong etc., the P2P-SIP design with realize telecommunications science, 10A phase in 2009,184~187 pages.
[15] Li Guanghua, Wang Chenwu, based on the research of the P2P-SIP networking telephone, Xian Institute of Posts and Telecoms's journal, the 1st phase of the 15th volume in 2010,127~129 pages.
[16] Wang Nan, Lu Jiande, based on P2P SIP systematic research and the design of Kademlia algorithm, microcomputer information, the 26th volume 4-3 phase in 2010,122~124 pages.
[17] Liu Tiancheng, Tao Jun etc., based on the cross-domain handover scheme of portable terminal of layering P2P SIP, computer science, the 36th volume o. 11th in 2009,47~51 pages.
[18] Shen Guoqing, Li Xiaoming etc., SIP multiregion system and call routing inquiry research thereof, military technology communication, the 1st phase of the 31st volume in 2010,36~42 pages.
[19]Simone Leggio,Jukka Manner,et al.Session Initiation Protocol Deployment in Ad-Hoc Networks:a Decentralized Approach.Processed in 2 nd International Workshops on Wireless Ad-hoc Networks(IWWAN),London,May,2005.
Summary of the invention
For the technical problem of above-mentioned existence, the purpose of this invention is to provide many sip servers layout method in a kind of Wireless Mesh network, to solve in the wireless Mesh netword negative effect of the unsteadiness of Radio Link to bringing based on the voice call service of SIP.
For achieving the above object, the present invention adopts following technical scheme:
The measurement of communication quality step of Radio Link in the wireless Mesh netword: this step adopts and initiatively sends the mode that the practical communication Packet probing was surveyed or utilized to unicast packet, and the communication quality of all Radio Links in the network is measured; And by statistical analysis, obtain long-time interior link-quality index, and come the design server layout scheme based on achievement data, described link-quality index is packet delivery rate PDR;
Server is laid step: this step is described the network that may occur with the method for minimal cut set and is cut apart for the wireless Mesh netword topological diagram, and network is divided according to cutting apart territory (subgraph); In each subnet that division obtains, select the highest network node of reliability to lay node as the sip server in this subnet, lay at least a server in each cut zone (subnet);
The call route selecting step: this step is set up process for the SIP call connection, based on the communication reliability computation model, design respectively answering first the formula routing scheme and answering first the formula routing scheme take call successful rate as optimization aim take calling speed as optimization aim, the described formula routing scheme of answering first take calling speed as optimization aim is provided at the service that realizes as early as possible access in the call attempt number of times of regulation, and has the load balancing effect; Describedly answer first the formula routing scheme to be provided to guarantee in the call attempt number of times of regulation the service that can access take call successful rate as optimization aim.
The measurement of communication quality step of Radio Link comprises following substep in the described wireless Mesh netword:
To certain given wireless links i, the total time length T of measuring is divided into N time slot, each time slot is a sense cycle;
If R IjBe j packet delivery rate PDR that sense cycle records on the i bar link, calculate the mean value E (R of packet delivery rate in the j cycle Ij) and the variance D (R of packet delivery rate Ij);
Contrast (administrative staff) predefined packet delivery rate mean value threshold value E ThWith variance threshold values D ThIf, E (R Ij)>E Th, and D (R Ij)<D Th, then this cycle link-quality is considered to well;
Boolean type N tuple A to i bar link link generation state i={ a Ij}={ a I1, a I2... a IN, if this link is good during cycle j, a then Ij=1, otherwise a Ij=0.
Described server is laid step and further be may further comprise the steps:
The network partiting step of cutting apart probability Network Based: calculate the probability that each cut set is cut apart network based on the Link State N tuple that acquires, and according to the height of this probable value edge cut set is sorted; Design is cut apart at given minimal network under the condition of probability threshold value, based on the figure dividing method of traversal minimal cut set network is divided, and makes to divide the subnet obtain and can embody the network that most probable occurs and cut apart situation;
Node selection step based on communication reliability: in each subnet that is partitioned into, successively with one or several node group in the subnet as laying both candidate nodes; Based on the Link State N tuple that obtains before, select non-loop paths to calculate all other nodes to the communication reliability of both candidate nodes; Select the highest both candidate nodes of reliability or node group as the laying node of this subnet, the sip server system is installed.
The step of laying described server adopts the network of cutting apart probability Network Based to divide, and wherein the given minimal network network of cutting apart probability is divided to calculate and be may further comprise the steps:
1. N state tuple corresponding to middle directed edge Network Based obtained the in twos edge cut set EC between the node of network iAnd network is cut apart probability P I (EC i), select to cut apart probability and be higher than defined threshold P eEdge cut set, the preferred order of arranging from high to low edge cut set by probability forms edge cut set array EdgeCut;
2. sub-netting index group SubNet of initialization is vertex set V;
If 3. also have edge cut set to attempt, select an element EC among the EdgeCut; Otherwise the subnet SubNet that output obtains and required Server number stop;
If 4. also have subnet to attempt, select an element S N among the SubNet iOtherwise, go to step 3.;
5. use EC sTo SN iCarry out subnet and divide, if two subnet SN that newly obtain aAnd SN bIn nodes be not less than MinSize, and SN aAnd SN bIn have at least one the end connected component, then with SN aAnd SN bAdd all in an interim subnet collection NewSubNet, otherwise with SN iAdd among the NewSubNet, and go to step 4.;
6. put each subnet among the NewSubNet in order, whether sign is main subnet or the end connection subnet of being communicated with, and all main subnets that is communicated with are merged into one;
7. check each subnet among the NewSubNet, if the interstitial content of each subnet between [MinSize, 2*MinSize], then all subnets can not divide again, MinSize is subnet node number lower limit;
8. according to each subnet among the NewSubNet, make Num iBe the nodes in i the subnet, calculate whole network required service device number N um,
Figure BDA0000038830150000061
MaxSize is subnet node number higher limit;
9. NewSubNet is assigned to SubNet, NewSubNet is emptied, if all subnets can not divide, then export SubNet and required Server number again, algorithm stops; Otherwise, go to step 3..
Node selecting method based on communication reliability in the described server laying step may further comprise the steps:
1. the subnet S to marking off, according to the number SK of both candidate nodes in the subnet, its all possible node combination has C SNum SKIndividual, take out these combinations, described SK is at least 1;
2. to the set of node T in each combination i, ask not at T iIn each node v pTo T iIn each node v qSet of paths, based on above-mentioned N state tuple, ask the Effective Probability of set of paths;
3. to S-T iIn node v p, calculate it to T iIn the maximum of 1 2-Terminal Reliability of node, be designated as Rel 1(v p, T);
If 4. S-T iIn also have node not take turns to, then 3. this node is carried out step, otherwise, the cumulative S-T that obtains iIn all nodes to T iThe peaked summation of 1 2-Terminal Reliability
Figure BDA0000038830150000062
If 5. there is the both candidate nodes combination not take turns to, then 2. this combination is carried out step, otherwise, finished the reliability of all combinations and calculated, relatively each reliability summation
Figure BDA0000038830150000063
Select the maximum corresponding T of summation iCollection is as the laying node of S subnet;
The node group T that all-ones subnet is selected iGather, form the set of node T of the sip server of laying in the whole network.
Described call route selecting step is further divided into: based on the calling route selection method of speed, based on the calling route selection method of success rate;
Described calling route selection method based on speed is: all non-server node j, according to the call attempt frequency n of setting, in leading to all paths of server set T, select the highest n paths of reliability, set up successively the transmission path of usefulness as node j call connection, to establish call connection by a paths as early as possible;
Described calling route selection method based on success rate is: to all non-server node j, according to the call attempt frequency n of setting, in leading to all paths of server set T, select the rear the highest path of total reliability of n bar combination, set up successively the transmission path of usefulness as node j call connection, in n time is attempted, successfully establish call connection guaranteeing.
Calling route selection method based on speed in the described call route selecting step may further comprise the steps:
1. to non-server node j, its all possible paths that arrive server node collection T are put into P (s j, T) collection calculates and concentrates each path P iReliability, determine the call attempt frequency n of regulation;
2. when n=1, the result path collection P of output S n(s j, a paths is only arranged in T).At P (s j, in all paths T), select to have the path P of high-reliability 1, P is arranged S 1(s j, T)={ P 1;
3. when n=2, to P (s j, T) in except P 1Outside every paths P i, calculate and work as P 1P during inefficacy iEffective conditional probability, namely
Figure BDA0000038830150000071
Selection has the path P of maximum probability value 2, P is arranged S 2(s j, T)={ P 1, P 2;
4. obtaining P S N-1(s j, T) afterwards, to P (s j, T) in except P 1, P 2..., P N-1Outside every paths P i, calculate and work as P 1, P 2..., P N-1P during inefficacy iEffective conditional probability, namely
Figure BDA0000038830150000072
Selection has the path P of maximum probability n, P is arranged S n(s j, T)={ P 1, P 2..., P n, so far the road collection makes up complete.
Calling route selection method based on success rate in the described call route selecting step may further comprise the steps:
1. to non-server node j, its all possible paths that arrive server node collection T are put into P (s j, T) collection calculates and concentrates each path P iReliability, determine the call attempt frequency n of regulation;
2. make that M is P (s j, the number of path in T) is tried to achieve each combination that contains the n paths, combination add up to C M n
3. to combination C t, random arrangement is the order of n paths wherein, uses C tIn the n paths, rather than P (s j, n the 1-K Terminal Reliability of all path computing in T);
4. select to have the combination of high-reliability, be institute's routing collection, path wherein can be obtained n time and attempt the highest interior power that connects into;
5. arrange the use order in these paths based on the path reliability: the path that at first has high-reliability is chosen as the first path P 1, afterwards in the m minor sort, to remaining the every paths P in the unsorted n-m+1 paths i, calculate the current path of choosing when all losing efficacy, P iEffective probability, i.e. conditional probability
Figure BDA0000038830150000073
Taking-up has the path P of maximum probability mCome the m position, with the order of the method arrangement n paths, the process of asking for of road collection finishes.
The present invention has the following advantages and good effect:
1) the present invention is by the analysis to the network cutting state that may occur in the wireless Mesh netword, and the most reliable node or node group are laid at least one sip server in cut zone in advance; Simultaneously, all determine call signaling transmission path one group of high-reliability, that lead to a plurality of sip servers for all-router.Like this, overcome as much as possible the negative effect that network voice communication is produced because of the frequent provisional fault that occurs of Radio Link, cut apart situation even network occurs, also still can guarantee the voice communications services in each cut zone, avoid the large tracts of land break in speech communications phenomenon that causes without sip server owing in some cut zone.
2) laying of multiple servers only relates to the network node of minority among the present invention, in the sip server log-on message do not need synchronously carry out the whole network diffusion, also less on the impact of network traffics, the feasibility of invention scheme implementation is strong.In addition, at each router node sip agent is installed and is responsible for call connection with user terminal, realize that by the function that increases the agency trusted path is selected and switch in the path again.So just guaranteed to greatest extent that the mobile phone of any support WiFi and SIP can directly use therein to the compatibility of existing SIP voice terminal device.
Description of drawings
Fig. 1 is in order to network is cut apart the exemplary plot that situation is described introduction among the present invention.
Fig. 2 is that given network is cut apart network partitioning method flow chart under the probability threshold value condition among the present invention.
Fig. 3 is subnet affair device system of selection flow chart for oral administration among the present invention.
Fig. 4 is based on the preferential call path collection method for solving flow chart of speed among the present invention.
Fig. 5 is based on the preferential call path collection method for solving flow chart of success rate among the present invention.
Embodiment
Forefathers have carried out a lot of researchs to aspects such as the architectural framework of wireless Mesh netword, channel quality, network capacity, route, multicast, signaling callings, but also have a lot of problems in this research field at present.
In the theoretical research and engineering practice of Wireless Mesh network, the problem that must face is exactly the stability problem of Radio Link.Wireless communication procedure is subject to the impact of the factors such as signal fadeout, channel disturbance, noise intensity, communication distance, weather condition and terrain environment easily, the Radio Link phenomenon off and on that is difficult to expect can occur.The time perseverance network topology change, even network is divided into the situation of a plurality of separate subnets, seriously affected the Stability and dependability of wireless communication system.In the provisional emergency communication environment that sets up, this impact may be more serious.Therefore, along with the day by day increase of people to the wireless communication technology demand, the reliability and stability guarantee of wireless communication has become the problem that network design, design and construction must emphasis in the stage be solved.
At present, mainly be by monitoring and analyze the random failure phenomenon for network node and link failure fault, in Networking Design and process of construction, take corresponding redundant technique and measure that the Reliability Assurance that meets certain requirements is provided.The netted networking structure of Wireless Mesh network has possessed the prerequisite of redundancy fault-tolerant.But up to now, wireless voice communication system based on wireless Mesh netword and SIP framework all is to adopt single sip server mode, some researchs of being combined with SIP about P2P mainly are to solve the large server performance bottleneck problem of bringing of customer volume, can not solve the network segmentation problem.
Based on the voice communication system of wireless Mesh netword and SIP, in case occur that network is cut apart or link failure, the user in divided zone just can't login sip server, causes large-area communication disruption.Therefore, the present invention proposes to lay the measure of a plurality of sip servers in Wireless Mesh network, guarantee as far as possible that also each cut zone has at least a sip server to provide subnet interior voice communications services even the generation network is cut apart, thereby reach the purpose that promotes the network global reliability.
In wireless Mesh netword, can lay sip server on any one router node, but select which node to lay the sip server Reliability Assurance that can offer the best, and what kind of server access path router select to have best fault-tolerant effect, also do not have at present effective scheme.The present invention is exactly the problem that will solve this respect.
The below introduces in wireless Mesh netword, utilizes many sip servers that problem that high reliability guarantees service to be needed to solve, relevant theoretical foundation and concrete method step are provided, and the invention will be further described by reference to the accompanying drawings with the implementation example.
1, the measurement of radio link quality and state analysis
Because various fading effects, Radio Link ruptures often, makes the topology generation variation at random of network, can cause the transmission of message to seriously influence.Measure accurately link-quality and state with the phase-split network point-to-point transmission, deployment and Route Selection tool to network service have very important significance, i.e. probability and Time Distribution by the fracture of each bar link, for after server lay and Route Selection provides foundation.
At present, the method of carrying out the Radio Link measurement of communication quality is mainly: drive based on IEEE 802.11 wireless network cards and MadWifi, operating system is Linux, can adopt the mode that initiatively sends packet or passive monitored data bag to obtain wireless signal, packet can be clean culture, also can be broadcasting, gather the wireless signal in tens hours length, obtain link-quality index wherein.
The measurement index of four kinds of available main link-qualities is: received signal strength index (Received Signal Strength Indication, RSSI), ratio (the Signal-to-Noise Ratio of signal and noise intensity, SNR), packet delivery rate (Packet-Delivery Ratio, PDR), bit error rate (Bit-Error Rate, BER).Correlative study work is by the actual measurement under varying environment, different rates, the different pieces of information packet length condition, and the Main Conclusions that obtains has: 1) length of communication distance is not effective influencing factor of link-quality; 2) degree of correlation of the link-quality of reality is limited when RSSI and data communication, can't accurately embody the disturbed degree of link, but RSSI remains a kind of useful reference index; 3) SNR is difficult to accurate calculating, and when having more serious multipath fading, the link-quality evaluation effect of this index is undesirable; 4) PDR depends on data package size, selected transmission rate to estimating of link-quality, but PDR is easy to record, and can be similar to the quality that embodies link with it more easily, uses very general; 5) BER can reflect link-quality accurately on thinner granularity, but BER must measure in physical layer, is difficult to set up accurately mapping with the concrete data packet communication process in upper strata.
In general, in the invention process, we adopt PDR as the link evaluation index.Because fragmentary packet loss is a chance event, and continuous data-bag lost must be very large with the decline correlation of channel quality, and remarkable change can occur with the length of the duration of fading its probability distribution, can effectively characterize link-quality.
Because Radio Link has the multipath fading effect, and the sudden change of link-quality may occur at short notice.Therefore, need to analyze from thinner time granularity the state of Radio Link, and obtain the status data set that the interior Radio Link of long period section (such as 24 hours) engraves when of short duration, just have analysis significance.
The scheme that the present invention adopts is: establishing the link in network number is | E|, and the total time length of carrying out the radio link quality measurement is T (being approximately tens hours), and T is divided into N time slot, the time span of each slot is t 0(being approximately several seconds) represents statistic of attribute period of short-term Radio Link, wherein comprises the link-quality achievement data that several record.Calculating is at [j*t 0, (j+1) * t 0] in the period, link L iThe mathematic expectaion E (R of quality measure index (PDR) Ij) and variance D (R Ij), obtain like this all links | the numerical characteristic of E|*N short-term quality index.
Right | E|*N group achievement data (E (R Ij) and D (R Ij)), respectively mathematic expectaion E (R) and the variance D (R) of calculated population, and determine the threshold value E of mathematic expectaion with reference to this ThThreshold value D with variance ThIf, E (R Ij) greater than E Th, and D (R Ij) less than D Th, this link L then iIt is measured to be considered to matter in the period of correspondence j, do not lose efficacy, otherwise this link is considered to lose efficacy.Concrete quality index statistical value is converted to unified Boolean type data, to every link L iLink generation N state tuple A i={ a I1, a I2... a IN, if link-quality is considered to good during corresponding j time slot, a then Ij=1, otherwise a Ij=0, the foundation that this array is selected as work links such as back network design and Route Selection.To single link L i, the link average probability of failure on demand in whole Measuring Time section can be expressed as:
Figure BDA0000038830150000101
2, technology is selected in the position of many sip servers among the WMN
The reliability of every link is different, any the time, all may cause network to be divided into a plurality of subnets by the fracture of link.WMN under the given concrete physical environment, the layout scheme that has multiple sip server, need to select best sip server installation position based on effective theoretical method, a plurality of servers are placed on the suitable MR node, tackle the situation of not being communicated with of cutting apart that occurs at random in the network, make each router node at any time can in the subnet at own place, obtain the highest probability that is communicated to certain sip server.
At present the method selected of server has: based on lay cost (number of servers), based on node capacity (the connection number of degrees), based on propagation delay time (hop count), based on the compound system of selection of bandwidth and load (throughput) and above optimization aim.There is no the research that WMN environment link Problem of Failure is launched as a leading indicator.
The system of selection of server installation position comprises among the present invention:
In the WMN that the different condition Radio Link consists of, based on minimal cut set the network cutting state is described.
Cut apart under the Probability Condition at given minimal network, network partitioning method based on the traversal minimal cut set, make and divide the network division situation that the subnet that obtains represents appearance as much as possible, namely each subnet has higher connectedness, and the link between subnet is the edge cut set with low stability.
In each subnet that division obtains, stability according to link, theoretical and the 2-Terminal Reliability computational methods based on the connectivity of graph select best node to set up an office as the sip server cloth in this subnet in each subnet, so that this server can provide maximum Service Reliability for this zone.
2.1 the network of cutting apart probability Network Based is divided
(1) theoretical foundation
Based on graph theory and reliability theory, the WMN backbone network that is made of router can be described as a directed graph G (V, E), and wherein V represents the vertex set that is made of Wireless Mesh network router MR, the limit collection that the E representative is made of the Radio Link between MR.Given summit a and b, directed edge a-b may have different link failure probability with b-a.Adopt the limit failure model, suppose that the summit all is reliably, and the limit has had the failure probability of precognition.
Subnet dividing method among the present invention is based on the figure segmentation theory, goal description is: a given weighted undirected graph G (V, E), wherein there are weights on each summit and limit, 2-way figure segmentation problem is divided into 2 disjoint subsets with vertex set V, every subset has approximately identical summit total weight value (as the constraint of equilibrium), and minimizes the total weight value on the limit (such as cut edge) that different subset intersects.
The foundation of subnet division here is the failure probability as the limit of cut set, if a limit failure probability is large, then network separates easily from here.Select most probable and the edge cut set that network is cut apart occurs, need simultaneously to consider the requirement of load balancing, divide because only cut apart probability according to the network of edge cut set, undesirable network occurs easily and cut apart, the subnet of gained may be excessive or too small, even only contain a node.
Relevant concept is as follows:
[definition 1] export-oriented edge cut set.Connected set of points S among the given directed graph G, the directed edge from S to V-S forms an export-oriented edge cut set and closes { e 1, e 2..., e s, be designated as the export-oriented edge cut set EC of S O(S).
If at the export-oriented edge cut set { e of period j 1, e 1..., e sLost efficacy, then S is divided comes, the export-oriented limit e of each bar in the edge cut set iAll lost efficacy simultaneously.This requires relevant link in the N state tuple of period j, and each element must be 0, i.e. a 1j=a 2j=...=a Sj=0,
Figure BDA0000038830150000111
(containing N time slot) in whole link quality measurements time period T, all lost efficacy so that the statistical probability that network is cut apart is in the limit in the export-oriented edge cut set,
Figure BDA0000038830150000112
Export-oriented cut edge of each bar that this probable value equals intra-cluster over a long time lost efficacy simultaneously, this bunch outwards becomes the not probability of connected subgraph, the degree of reliability of the server bunch outward of the node visit in having embodied bunch.
[definition 2] is interior to edge cut set.Connected set of points S among the given directed graph G, the directed edge from V-S to S closes { e to edge cut set in forming one 1', e 2' ..., e t', be designated as the interior to edge cut set EC of S I(S).
If interior to edge cut set { e at period j 1', e 1' ..., e t' lost efficacy, then S is divided comes, and each bar in the edge cut set is interior to limit e i' all lost efficacy simultaneously.This requires relevant link in the N state tuple of period j, and each element must be 0, i.e. a 1j=a 2j=...=a Sj=0,
Figure BDA0000038830150000113
(containing N time slot) in whole link quality measurements time period T, all lost efficacy so that the statistical probability that network is cut apart is in interior limit in edge cut set,
Figure BDA0000038830150000114
This probable value equal in each bar of intra-cluster over a long time to cut edge lost efficacy simultaneously, this bunch inwardly become the not probability of connected subgraph, embodied the degree of reliability of the server in the node visit bunch bunch outward.
If one the failure probability on limit is lower, comprise so the edge cut set of this edge so that the probability that network is cut apart is just lower, therefore can set a limit failure probability threshold value Pr Th, only the Considering Failure probability is higher than the limit of threshold value and the cut set of formation thereof when analyzing, and namely produces the higher edge cut set of probability that network is cut apart.If all lost efficacy in interior limit in edge cut set or export-oriented edge cut set, then connected set of points S is not communicated with V-S, network occurred and has cut apart, and the network of this edge cut set is cut apart probability and is,
PI(EC i)=max(PI O(EC i),PI I(EC i))。
[definition 3] main subnet that is communicated with.Be higher than threshold value Pr by failure probability ThThe cut set set that consists of of limit in, if with point set S a plurality of cut sets of separated needs from G, then all limits of this point set and point set interior nodes consist of the main subnets that are communicated with.
[definition 4] end is communicated with subnet.Be higher than threshold value Pr by failure probability ThThe cut set set that consists of of limit in, if with point set S cut set of separated needs from G, then all limits of this point set and point set interior nodes consist of end and are communicated with subnets.
(2) implementation procedure
A given network is cut apart probability threshold value PI Th, all limits in the network are cut, get wherein network and cut apart probability and be higher than PI ThThe limit cut, by these edge cut sets network is divided into several subnets.Threshold value PI ThSelection should be so that the network that cut on selected limit be cut apart probability approaches, rather than have big difference.Shown in accompanying drawing 1, limit e 1, e 2... e 7Effective Probability different, the network of resulting edge cut set is cut apart probability and also can be very different, if { e 2, e 3And { e 5, e 6Joint failure probability will be much larger than { e 4, network node tends to be divided into three subset { v so 1, v 2, { v 3, v 4, { v 5, v 6, if { e 2, e 3, { e 5, e 6, { e 4Failure probability approach, network node tends to be divided into four subset { v so 1, v 1, { v 3, { v 4, { v 5, v 6.Under the circumstances, to each edge cut set that may occur in theory, cut apart probability based on the network on the Effective Probability theory of computation on each limit, from high to low edge cut set is sorted according to cutting apart probable value, select again suitable threshold value, so that the network of selected edge cut set is cut apart probable value is substantially approaching, divide with the network that embodiment matches.
Given minimal network is cut apart sub-clustering target and the principle that the network partitioning method of probability takes: the network of Given Graph G (V, E) and selected edge cut set is cut apart probability threshold value PI Th, network is divided into K subnet, so that V={C 1∪ C 2∪ Λ C K, C i∩ C j=φ, K≤L, the edge cut set EC between each subnet iThe edge cut set of preferentially cutting apart maximum probability by network consists of, and obtains 1 and mainly is communicated with subnet and K-1 end is communicated with subnet, at least one server of laying in each subnet, and satisfy the load restraint condition: contained nodes Num in each subnet iBe not less than MinSize, and distribute suitable number of servers according to the nodes in each subnet, satisfy each server and mostly be MaxSize node most service is provided, reach rational load Distribution.
It is as follows that design problem is expressed as the ILP model:
OBJECTIVES
Max (∑ PI (EC i))---select network to cut apart the edge cut set of maximum probability, the network that embodies the most probable appearance is cut apart
Max (K)---find out end as much as possible and be communicated with subnet, namely cut apart disconnected possibility situation
CONSTRAINTS
MinSize≤Num i---the nodes in each subnet can not be less than a lower limit
PI (EC i) 〉=PI Th---be used for cutting apart probability and can not being lower than threshold value PI of edge cut set that network cuts apart Th
Figure BDA0000038830150000131
---the number of server should be less than a lower limit in each subnet
K=∑ K i---server adds up to each subnet number of servers sum in the network.
Algorithm flow chart is seen accompanying drawing 2, specifically describes to be:
Input: network is cut apart probability threshold value P e, the contained minimum node of each subnet is counted MinSize, the N state tuple of each bar directed edge, the maximum node number MaxSize that server can be served
Output: each subnet, server sum.
Step:
The N state tuple that Stepl middle directed edge Network Based is corresponding is obtained the in twos edge cut set EC between the node of network iAnd network is cut apart probability P I (EC i).Selection is cut apart probability and is higher than P eThe limit cut, by probability the preferred order of edge cut set is set from high to low, form the EdgeCut array.
Step2 initialization subnet array SubNet is vertex set V.
If Step3 also has edge cut set to attempt, select an element EC among the EdgeCut s(being that cut on the current the highest limit of probability of cutting apart); Otherwise output SubNet and used Server number stop.
If Step4 also has subnet to attempt, select an element S N among the SubNet iOtherwise, go to 3).
Step5 EC sTo SN iCarrying out subnet divides.If two subnet SN of gained aAnd SN bIn nodes be not less than MinSize, and SN aAnd SN bIn have at least one the end connected component, then with SN aAnd SN bAdd among the NewSubNet, otherwise with SN iAdd among the NewSubNet, and go to 4).This processing mode so that the subnet that satisfies condition found out as much as possible.
Each subnet among the Step6 arrangement NewSubNet, whether sign is that main connection subnet or end are communicated with subnet, and all main subnets that is communicated with are merged into one.
Step7 checks each subnet among the NewSubNet, if the interstitial content of each subnet [MinSize, 2*MinSize) between, then all subnets are inseparable.
Step8 makes Num according to each subnet among the NewSubNet iBe the nodes in i the subnet, calculate whole network required service device number N um, for:
Figure BDA0000038830150000132
Step9 is assigned to SubNet with NewSubNet, and NewSubNet is emptied, if all subnets can not divide again, then exports SubNet and used Server number, and algorithm stops; Otherwise, go to 3).
2.2 the laying point selection method based on communication reliability
(1) theoretical foundation
Select node (group) in the subnet as the server both candidate nodes, calculate the reliability that other node is communicated by letter to this node (group) in the assessment subnet, selection have high-reliability as server.
The computational methods of the reliability that should be noted.The traditional reliability degree calculation method of node is considered " possibility " of connection based on all possible path, and considers " ageing " of connection, if finally success of the trial of a lot of bar routes of process also thinks between 2 it is reliable.When having loop among the figure, the path number of point-to-point transmission can increase sharply, even is the index law increase.In theory, for the node that is on the loop, can obtain one with traditional reliability degree calculation method and may fall far short with the result of emulation and actual measurement very near 1 reliability, not have reference significance.At present the research of communication reliability among wired network and the WMN all is based on all path computing reliabilitys.And the SIP call service is high to the requirement of real-time, the user expects that access success connects as early as possible, carry out at most the several times dial-up, otherwise adopt other communication mode with regard to the connection of possibility abandoned call, calculate so should adopt the path of limited number to carry out reliability here.
[definition 5] 1 2-Terminal Reliability Rel 1(s, T).Given all paths of gathering T from source point s to meeting point, the paths that the selection reliability is the highest is used for communication, and the reliability in this path is the 2-Terminal Reliability 1 time, embodies first trial and gets final product successful connection to the probability of some meeting points.
The reliability Pr (P) in [definition 6] path.The link L that given path P is contained 1, L 2..., L sLink State N tuple, whether path P effectively depends on the state of each bar link in period j, basis is Pr j(P)=a 1j∧ a 2j∧ ... ∧ a Sj, the reliability of path P then can be expressed as,
Figure BDA0000038830150000141
Select the paths with high-reliability, its reliability value is from node v pTo v q1 2-Terminal Reliability Rel 1(v p, v q), obtain node v p1 2-Terminal Reliability to T is Rel 1(v p, T)=max (Rel 1(v p, v q)), v q∈ T.
(2) implementation procedure
Make among the subnet S, the element number among the server set T is SK, and the selection problem of T is as follows with the ILP model representation:
OBJECTIVES
Figure BDA0000038830150000142
---to S-T, maximization node wherein is to 1 2-Terminal Reliability summation of T
CONSTRAINTS
v p∈ S-T, v q∈ T---calculating non-candidate server from subnet is gathered the reliability of candidate server set
| T|=SK---the size of candidate server set
The server selection algorithm is seen accompanying drawing 3, specifically describes to be:
Input: contain the subnet S of SNum node, number of servers SK, the N state tuple of each directed edge.
Output: a selected SK node.
Step:
The combination of Stepl candidate server node has C SNum SKIndividual, take out these combinations.
Step2 is to the set of node T in each combination i, ask S-T iIn each node v pTo T iIn each node v qSet of paths, based on the N state tuple on each limit, ask the reliability in each path.
Step3 is to S-T iIn node v p, ask the maximum of 1 2-Terminal Reliability of its node in the Ti, be designated as Rel 1(v p, T).
If Step4 is S-T iIn also have node not take turns to, then this node carry out step 3), otherwise, the cumulative S-T that obtains iIn all nodes to T iThe peaked summation of 1 2-Terminal Reliability
Figure BDA0000038830150000151
Combination does not take turns to if Step5 has the candidate server node, then this combination carry out step 2), otherwise, finished the reliability of all combinations and calculated, relatively each reliability summation
Figure BDA0000038830150000152
Select the maximum corresponding T of summation iGather as sip server.
3, the Route Selection technology of couple in router SIP call connection
WMN is the wireless multi-hop communication network, there are a variety of selections in path from the source point to the meeting point, the communication interface number of each router node of path process (the connection number of degrees), user's access frequency difference, the reliability of every link is also different, so that the communication reliability in these paths and communication delay differ from one another.Communication connection for limited number of attempt is called out, and needs to select best path and path preferred order, and the high reliability from router node to nearest sip server, the wireless connections of high real-time are provided.
The strategy of traditional Route Selection has multiple, and wherein the strategy based on minimum hop count is adopted by most of Routing Protocols in the wireless multi-hop network (such as AODV, DSR, DSDV).Its shortcoming is that this index is considered as all links in the network to be equal to, and does not consider the interference that link load, capacity and link are subject to.The expected transmission times in path (ETX) is defined as successfully and transmits the number of times that packet is required with certain paths as the common routing index of another one.In addition, also have a lot of route indexs to be based on the ETX improvement and obtain, such as ETT (Expected Transmission Time), WCETT (ETT of weighting) etc.Its advantage is that EXT is based on the transmission success rate, can embody link-quality, its shortcoming is: the transmission success rate of (1) packet need to depend on size and the transmission rate of packet in the MAC of ICP/IP protocol stack layer or more high-rise measurement, so this index is objective not; (2) this index is not considered the load of link, and the route that produces may be passed the heavier node of load, and it is unbalanced to cause resource to be used.
In the above index, jumping figure and RTT lay particular emphasis on and embody the ageing of route, and ETX lays particular emphasis on the success rate that embodies routing forwarding, and representative is effect on the link both direction, can't embody the one-way of link.The key index that the real time communication signaling is transmitted comprises transmission success rate and the transfer rate of message, because the signalling message data bag is less, occupied bandwidth is few, therefore needn't consider the factors such as load balancing and throughput, and need reliability with the path of unreliable wireless environment as main Route Selection index, the same success rate that can embody whole routing forwarding process with ETX of this index, but by analysis chain line state on thinner time granularity, can be embodied in the probability of in a short time path normal operation.
Among the present invention, every router comprises to the Route Selection technology of corresponding sip server:
● based on Reliability Theory, develop a kind of new, source point to the communication reliability computation model of one of a plurality of meeting points (one-to-' any-in-many ').
● for each node, selected server comprises the preferred server in the local subnet, and the backup server in a plurality of this locality and the foreign subnet.Real-time SIP call connection request for limited number of attempt is set up process, based on new Reliability Calculation Model, what design maximized respectively call successful rate and called out speed answers first the formula routing scheme, is used for efficiently signaling message being sent to from initial router node any one of server of a plurality of distributions.
(1) theoretical foundation
On K MR node, server be will lay, Tolerate and redundance and high-performance will be used to provide.As long as signaling message is forwarded to wherein any one server, namely represents message and transmit successfully.Traditional 2-Terminal Reliability Rel 2(s, t) model description be 1 s to the communication reliability of 1 t (one-to-one), and K Terminal Reliability Rel KWhat (s, T) described is that 1 s is to containing K the every bit t among the set T that puts i(one-to-many) communication reliability, these models can't be described in the multiserver communication pattern (one-to-' any-in-many ') of any one.Corresponding communication reliability should be described below.
[definition 7] 1/K link road collection SP (s j, T).Be expressed as from s jTo the minimal path collection of all meeting point set T, namely from s jTo t iThe union of all minimal path collection, the node in the wherein each paths should not contain the summit among the T, the terminal point that only has the path just can be the summit among the T.
Make the set that does not contain all other summits of terminal point in MidNode (P) the expression path P, each union to the road collection between source point-meeting point contains the N paths altogether, is expressed as: ∪ SP (s j, t i)={ P 1, P 2, P 3... P N, SP (s then j, T) can be expressed as ∪ SP (s j, t i) in, the path intermediate point does not contain the path P on summit among the T iUnion, have
Figure BDA0000038830150000161
[definition 8] 1/K Terminal Reliability Rel 1/K(s j, T).Based on SP (s j, T), source point s jCan be connected to the probability of T, namely connect the probability of any one rather than each meeting point.
This reliability can be by not handing over and method (SDP) is obtained: make P i(i=1,2 ..., N) be SP (s j, a minimal path in T), A i(i=1,2 ..., N) be P iIn all limits normal event all, then the 1/K Terminal Reliability is not with handing over and be expressed as,
Calculate Rel 1/K(s j, T) used road collection is than calculating Rel K(s j, T) the used contained path of road collection is less, than calculating single Rel 2(s j, t i) used path wants many, but be different from Rel K(s j, T) ask the thought of the common factor that is communicated with probability, Rel 1/K(s j, T) only need to calculate the probability that reaches at least one meeting point, employing be the thought of asking the union that is communicated with probability, so Rel 1/K(s j, T) to be higher than Rel K(s j, T), also be higher than any Rel 2(s j, t i).
Communication for limited number of times is attempted, and relevant reliability is defined as follows:
[definition 9] n 1/K Terminal Reliability Rel 1/K n(s j, T): by the 1/K Terminal Reliability that contains link road collection that the n paths consists of and calculate.
In this reliability definition, the n paths is attempted corresponding to n communication, and each trial selects a paths to send signaling message.The selection scheme of n paths has a variety of, and the path of How to choose the best and path preferred order are the problems of a key.
Theorem 1 has provided Rel 1/K 1(s j, lower bound T).
[theorem 1] to each node to s jAnd t i, find P (s j, t i) in have a paths of high-reliability, be designated as Rel based on the 2-Terminal Reliability of this path computing gained 2 1(s j, t i), Rel then 1/K 1(s j, T) more than or equal to Rel arbitrarily 2 1(s j, t i), t i∈ T, i.e. Max (Rel 2 1(s j, t i))≤Rel 1/K 1(s j, T), when K=1, equal sign is set up.
Card: when K=1, T={t i, Rel 2 1(s j, t i) and Rel 1/K 1(s j, meeting point T) is all identical respectively with the path, so equal sign is set up.
When K>1, suppose Rel 2 1(s j, t q) be all 2-Terminal Reliability Rel 2 1(s j, t i) in the highest, from s jTo t qRoad collection P (s j, t q), calculate Rel 2 1(s j, t i) a unique paths be [P 1] qFrom s j1-K link road to T integrates the (s as P j, T)=P (s j, t 1) ∪ P (s j, t 2) ∪ ... ∪ P (s j, t q) ∪ ... ∪ P (s j, t K), assumed calculation Rel 1/K 1(s j, unique paths T) is [P 1] p, this path is P (s j, T) middle reliability is the highest.Because
Figure BDA0000038830150000173
, path reliability Pr ([P 1] q)≤Pr ([P 1] p), Max (Rel is arranged 2 1(s j, t i))≤Rel 1/K 1(s j, T).
Theorem 2 has provided Rel 1/K n(s j, upper bound T).
[theorem 2] Rel 1/K n(s j, T)≤Rel 1/K(s j, T), as P (s j, when the path number in T) was not more than n, equal sign was set up.
Card: calculate Rel 1/K n(s j, T) need P (s j, the n paths in T), and calculate Rel 1/K(s j, T) need P (s j, the All Paths in T).Owing to there is more path that the Route Selection that connects, Rel are provided 1/K(s j, T) more than or equal to Rel 1/K n(s j, T),
Figure BDA0000038830150000174
Figure BDA0000038830150000175
As P (s j, when the path number in T) is no more than n, P is arranged n(s j, T)=P (s j, T), equal sign is set up.
Traditional SIP service is on the selecting sequence in path or do not make regulation, or based on shortest path, and does not consider the difference of path reliability.The SIP call service is high to the requirement of real-time, and the user expects that access success connects as early as possible, carries out at most the several times dial-up, otherwise adopt other communication mode with regard to the connection of possibility abandoned call, therefore, in the selection of call path, should preferentially select the then high path of reliability.
Improve the reliability degree calculation method of original all routes of consideration, the basic thought that proposition is set up for real-time SIP service call is: adopt a plurality of servers that Tolerate and redundance is provided, consider to calculate path number (the corresponding number of calls) and these two factors of path reliability (corresponding single call success rate) that reliability adopts, the set of paths of certain condition is satisfied in selection, but not All Paths calculates reliability, and the selecting sequence in path is sorted according to reliability.
(2) implementation procedure
To each node, calculate active path and the path reliability of its each sip server in the network, based on 1-K Terminal Reliability computational methods, select best several paths and ordering, be respectively the first via by, secondary route, Third Road by ..., corresponding route terminal point is respectively at first server and a plurality of backup server, when unsuccessful with previous route call try, just use current route, make up like this and call out the road collection, reach the highest so that this node can obtain the probability of the SIP service in the network.
Signaling message is sent to an adjacent MR from Wireless Mesh network user terminal MC, and MR selects a route that signaling is sent to corresponding server, and a server meeting point has mulitpath from source point MR to K, and the tendency selection has the path of short jumping figure usually.Real time communication is strict to time requirement, the user wishes to connect as early as possible, and the at short notice explosive increase of the data volume of emergency communication, often once communication is attempted being difficult to successfully connect, therefore communication mode is: if attempt the signaling retransmission failure with a paths, can be after this trial be overtime, again attempt communication to server corresponding to new route with a new path immediately, until server successful return messages in attempting time-out time, or reaching maximum attempts abandons communication.But existing scheme or do not consider the difference of path on reliability, or only preferentially select the little path of jumping figure, and do not pay the utmost attention to the height of reliability.
Therefore the key issue of research is: based on new Reliability Model, select to satisfy n path of performance requirement (such as the fastest forwarding speed, the highest forwarding success rate), and arrange the use order in these paths, be used for each trial of transmitting.
Be used for optimizing routing scheme---n the preferential 1-K link road of speed collection P of message forwarding speed S n(s j, T).
This road collection is optimized for the message forwarding speed, tries hard to as soon as possible successful transmitting signaling message.Based on P S n(s j, T), in each the trial, current available path with high-reliability is selected to be transmitted for message.Ask the concrete grammar of this road collection to be:
Stepl is to P (s j, each path P in T) iAsk the path reliability.
Step2 when n=1, P S n(s j, a paths is only arranged in T).At P (s j, in all paths T), select to have the path P of high-reliability 1, P is arranged S 1(s j, T)={ P 1.
Step3 is when n=2, to P (s j, T) in except P 1Outside every paths P i, calculate and work as P 1P during inefficacy iEffective conditional probability, namely
Figure BDA0000038830150000181
Selection has the path P of maximum probability value 2, P is arranged S 2(s j, T)={ P 1, P 2.
Step4 is similar, is obtaining P S N-1(s j, T) afterwards, to P (s j, T) in except P 1, P 2..., P N-1Outside every paths P i, calculate and work as P 1, P 2..., P N-1P during inefficacy iEffective conditional probability, namely
Figure BDA0000038830150000182
Selection has the path P of maximum probability n, P is arranged S n(s j, T)={ P 1, P 2..., P n, so far the road collection makes up complete.
Be used for optimizing routing scheme---n the preferential 1-K link road of effect collection P that message is transmitted success rate E n(s j, T).
Total forwarding success rate in this road collection is attempted for n time is optimized.The first path that concentrate on the road not necessarily provides the highest reliability.Ask the concrete grammar of this road collection to be:
Step1 is to P (s j, each path P in T) iAsk the path reliability.
Step2 makes that M is P (s j, the number of path in T) obtains each combination that contains the n paths, combination add up to C M n
Step3 is to combination C t, random arrangement is the order of n paths wherein, uses C tIn the n paths, rather than P (s j, n the 1-K Terminal Reliability of all path computing in T).
Step4 selects to have the combination of high-reliability, is institute's routing collection, and path wherein can be obtained n time and attempt the highest interior power that connects into.
Step5 arranges the use order in these paths at last based on the path reliability.The path that at first has high-reliability is chosen as the first path P 1Afterwards in the m minor sort, to remaining the every paths P in the unsorted n-m+1 paths i, calculate when before the path chosen when all losing efficacy, P iEffective probability, i.e. conditional probability Taking-up has the path P of maximum probability mCome the m position.In this way arrange the order of n paths, the process of asking for of road collection finishes.
Based on the corresponding road of above Different Optimization target collection, provide different Reliability Model, be respectively:
[Rel 1/K n(s j, T)] S: s jUse P S n(s j, T) in maximum n trials, be connected to the probability of T.
[Rel 1/K n(s j, T)] E: s jUse P E n(s j, T) in maximum n trials, be connected to the probability of T.
When n=1, P S 1(s j, T)=P E 1(s j, T), so [Rel 1/K 1(s j, T)] S=[Rel 1/K 1(s j, T)] EWhen n>1, P S n(s j, T) in the reliability in the first path more than or equal to P E n(s j, T) in the reliability in the first path, i.e. [Pr (P 1)] S>[Pr (P 1)] E
Can think P S n(s j, T) be dominant in the forwarding success rate index of attempting first, because P S n(s j, the first path in T) is at P (s j, have the highest reliability in all paths T), and in each the trial, select current available reliable routing, successfully transmit to realize as soon as possible signaling.And P E n(s j, T) in n time is attempted, transmit this index of success rate and be dominant, because P E n(s j, the n paths in T) provides the highest overall path reliability.Adopt conditional probability to carry out routing, therefore, the impact in the inefficacy path before the current path is taken into account, and disjoint limit is contained in the path of gained as much as possible.
Along with the increase of maximum attempts, total call successful rate can promote, but call delay can increase, and need to be when network link quality is the not good careful maximum call attempt number of times of balance is to realize the compromise of two performance index.
Four, sum up
The target of method provided by the invention is: the one, as far as possible so that when network occuring cut apart, have sip server that the calling access service is provided in each cut zone, thereby guarantee that the function of voice communication in the regional still can normally use; The 2nd, choose one group of trusted path that leads to sip server for the sip agent on each router node, both can overcome the Single Point of Faliure impact of Radio Link and server, will access again load Distribution on all server nodes of selecting simultaneously.
In wireless Mesh netword, we deal with the impact that network is cut apart by the Nodes that a plurality of sip servers is placed in correct position, solve the problem that trusted path is selected and switch in the path by the function that increases sip agent on each router node.Our implementation does not relate to subscriber terminal equipment, and this mainly is in order to guarantee that to greatest extent the mobile phone of any support WiFi and SIP can directly use therein to the compatibility of existing terminal equipment.
More than implement example only for the present invention is described, but not limitation of the present invention, person skilled in the relevant technique in the situation that does not break away from the spirit and scope of the present invention, can also be made various conversion or modification.Therefore, the technical scheme that all are equal to all falls into protection scope of the present invention.

Claims (8)

1. many sip servers layout method in the Wireless Mesh network is characterized in that, may further comprise the steps:
The measurement of communication quality step of Radio Link in the wireless Mesh netword: this step adopts and initiatively sends the mode that the practical communication Packet probing was surveyed or utilized to unicast packet, and the communication quality of all Radio Links in the network is measured; And by statistical analysis, obtain long-time interior link-quality index, and come the design server layout scheme based on achievement data, described link-quality index is packet delivery rate PDR;
Server is laid step: this step is described the network that may occur with the method for minimal cut set and is cut apart for the wireless Mesh netword topological diagram, and network is divided according to cutting apart the territory; In each subnet that division obtains, select the highest network node of reliability to lay node as the sip server in this subnet, lay at least a server in each cut zone;
The call route selecting step: this step is set up process for the SIP call connection, based on the communication reliability computation model, design respectively answering first the formula routing scheme and answering first the formula routing scheme take call successful rate as optimization aim take calling speed as optimization aim, the described formula routing scheme of answering first take calling speed as optimization aim is provided at the service that realizes as early as possible access in the call attempt number of times of regulation, and has the load balancing effect; Describedly answer first the formula routing scheme to be provided to guarantee in the call attempt number of times of regulation the service that can access take call successful rate as optimization aim.
2. many sip servers layout method in the Wireless Mesh network according to claim 1 is characterized in that:
The measurement of communication quality step of Radio Link comprises following substep in the described wireless Mesh netword:
To certain given wireless links i, the total time length T of measuring is divided into N time slot, each time slot is a sense cycle;
If R IjBe j packet delivery rate PDR that sense cycle records on the i bar link, calculate the mean value E (R of packet delivery rate in the j cycle Ij) and the variance D (R of packet delivery rate Ij);
Contrast predefined packet delivery rate mean value threshold value E ThWith variance threshold values D ThIf, E (R Ij) E Th, and D (R Ij)<D Th, then this cycle link-quality is considered to well;
Boolean type N tuple A to i bar link link generation state i={ a Ij}={ a I1, a I2... a IN, if this link is good during cycle j, a then Ij=1, otherwise a Ij=0.
3. many sip servers layout method in the Wireless Mesh network according to claim 1 is characterized in that:
Described server is laid step and further be may further comprise the steps:
The network partiting step of cutting apart probability Network Based: calculate the probability that each cut set is cut apart network based on the Link State N tuple that acquires, and according to the height of this probable value edge cut set is sorted; Design is cut apart at given minimal network under the condition of probability threshold value, based on the figure dividing method of traversal minimal cut set network is divided, and makes to divide the subnet obtain and can embody the network that most probable occurs and cut apart situation;
Node selection step based on communication reliability: in each subnet that is partitioned into, successively with one or several node group in the subnet as laying both candidate nodes; Based on the Link State N tuple that obtains before, select non-loop paths to calculate all other nodes to the communication reliability of both candidate nodes; Select the highest both candidate nodes of reliability or node group as the laying node of this subnet, the sip server system is installed.
4. many sip servers layout method in the Wireless Mesh network according to claim 3 is characterized in that:
The step of laying described server adopts the network of cutting apart probability Network Based to divide, and wherein the given minimal network network of cutting apart probability is divided to calculate and be may further comprise the steps:
1. N state tuple corresponding to middle directed edge Network Based obtained the in twos edge cut set EC between the node of network iAnd network is cut apart probability P I (EC i), select to cut apart probability and be higher than defined threshold P eEdge cut set, the preferred order of arranging from high to low edge cut set by probability forms edge cut set array EdgeCut;
2. sub-netting index group SubNet of initialization is vertex set V;
If 3. also have edge cut set to attempt, select an element EC among the EdgeCut; Otherwise the subnet SubNet that output obtains and required number of servers stop;
If 4. also have subnet to attempt, select an element S N among the SubNet iOtherwise, go to step 3.;
5. cut EC with the current the highest limit of probability of cutting apart sTo SN iCarry out subnet and divide, if two subnet SN that newly obtain aAnd SN bIn nodes be not less than MinSize, and SN aAnd SN bIn have at least one the end connected component, then with SN aAnd SN bAdd all in an interim subnet collection NewSubNet, otherwise with SN iAdd among the NewSubNet, and go to step 4.;
6. put each subnet among the NewSubNet in order, whether sign is main subnet or the end connection subnet of being communicated with, and all main subnets that is communicated with are merged into one;
7. check each subnet among the NewSubNet, if the interstitial content of each subnet between [MinSize, 2*MinSize], then all subnets can not divide again, MinSize is subnet node number lower limit;
8. according to each subnet among the NewSubNet, make Num iBe the nodes in i the subnet, calculate whole network required service device number N um,
Figure FDA00002514733600021
MaxSize is subnet node number higher limit;
9. NewSubNet is assigned to SubNet, NewSubNet is emptied, if all subnets can not divide, then export SubNet and required Server number again, algorithm stops; Otherwise, go to step 3..
5. many sip servers layout method in the Wireless Mesh network according to claim 3 is characterized in that:
Node selecting method based on communication reliability in the described server laying step may further comprise the steps:
1 couple of subnet S that marks off, according to the number SK of both candidate nodes in the subnet, its all possible node combination has C SNum SKIndividual, take out these combinations, described SK is at least the number that 1, SNum is node among the subnet S;
2. to the set of node T in each combination i, ask not at T iIn each node v pTo T iIn each node v qSet of paths, based on above-mentioned N state tuple, ask the Effective Probability of set of paths;
3. to S-T iIn node v p, calculate it to T iIn the maximum of 1 2-Terminal Reliability of node, be designated as Rel 1(v p, T);
If 4. S-T iIn also have node not take turns to, then 3. this node is carried out step, otherwise, the cumulative S-T that obtains iIn all nodes to T iThe peaked summation of 1 2-Terminal Reliability
Figure FDA00002514733600031
If 5. there is the both candidate nodes combination not take turns to, then 2. this combination is carried out step, otherwise, finished the reliability of all combinations and calculated, relatively each reliability summation
Figure FDA00002514733600032
Select the maximum corresponding T of summation iCollection is as the laying node of S subnet;
The node group T that all-ones subnet is selected iGather, form the set of node T of the sip server of laying in the whole network.
6. many sip servers layout method in the Wireless Mesh network according to claim 1 is characterized in that:
Described call route selecting step is further divided into: based on the calling route selection method of speed, based on the calling route selection method of success rate;
Described calling route selection method based on speed is: all non-server node j, according to the call attempt frequency n of setting, in leading to all paths of server set T, select the highest n paths of reliability, set up successively the transmission path of usefulness as node j call connection, to establish call connection by a paths as early as possible;
Described calling route selection method based on success rate is: to all non-server node j, according to the call attempt frequency n of setting, in leading to all paths of server set T, select the rear the highest path of total reliability of n bar combination, set up successively the transmission path of usefulness as node j call connection, in n time is attempted, successfully establish call connection guaranteeing.
7. many sip servers layout method in the Wireless Mesh network according to claim 6 is characterized in that:
Calling route selection method based on speed in the described call route selecting step may further comprise the steps:
1. to non-server node j, its all possible paths that arrive server node collection T are put into P (s j, T) collection calculates and concentrates each path P iReliability, determine the call attempt frequency n of regulation;
2. when n=1, the result path collection PS of output n(s j, a paths is only arranged in T), at P (s j, in all paths T), select to have the path P of high-reliability 1, P is arranged S 1(s j, T)={ P 1;
3. when n=2, to P (s j, T) in except P 1Outside every paths P i, calculate and work as P 1P during inefficacy iEffective conditional probability, namely
Figure FDA00002514733600033
Selection has the path P of maximum probability value 2, P is arranged S 2(s j, T)={ P 1, P 2;
4. obtaining P S N-1(s j, T) afterwards, to P (s j, T) in except P 1, P 2..., P N-1Outside every paths P i, calculate and work as P 1, P 2..., P N-1P during inefficacy iEffective conditional probability, namely
Figure FDA00002514733600034
Selection has the path P of maximum probability n, P is arranged S n(s j, T)={ P 1, P 2..., P n, so far the road collection makes up complete.
8. many sip servers layout method in the Wireless Mesh network according to claim 6 is characterized in that:
Calling route selection method based on success rate in the described call route selecting step may further comprise the steps:
1. to non-server node j, its all possible paths that arrive server node collection T are put into P (s j, T) collection calculates and concentrates each path P iReliability, determine the call attempt frequency n of regulation;
2. make that M is P (s j, the number of path in T) is tried to achieve each combination that contains the n paths, combination add up to C M n
3. to combination C t, random arrangement is the order of n paths wherein, uses C tIn the n paths, rather than P (s j, n the 1-K Terminal Reliability of all path computing in T);
4. select to have the combination of high-reliability, be institute's routing collection, path wherein can be obtained n time and attempt the highest interior power that connects into;
5. arrange the use order in these paths based on the path reliability: the path that at first has high-reliability is chosen as the first path P 1, afterwards in the m minor sort, to remaining the every paths P in the unsorted n-m+1 paths i, calculate the current path of choosing when all losing efficacy, P iEffective probability, i.e. conditional probability
Figure FDA00002514733600041
Taking-up has the path P of maximum probability mCome the m position, with the order of the method arrangement n paths, the process of asking for of road collection finishes.
CN 201010592782 2010-12-17 2010-12-17 Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network Expired - Fee Related CN102026203B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010592782 CN102026203B (en) 2010-12-17 2010-12-17 Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010592782 CN102026203B (en) 2010-12-17 2010-12-17 Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network

Publications (2)

Publication Number Publication Date
CN102026203A CN102026203A (en) 2011-04-20
CN102026203B true CN102026203B (en) 2013-04-10

Family

ID=43866916

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010592782 Expired - Fee Related CN102026203B (en) 2010-12-17 2010-12-17 Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network

Country Status (1)

Country Link
CN (1) CN102026203B (en)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103684928A (en) * 2013-12-26 2014-03-26 广州华多网络科技有限公司 VoIP (Voice over Internet Protocol) call quality monitoring method and device
US9204320B1 (en) * 2014-05-06 2015-12-01 Tyco Fire & Security Gmbh End node personal definition and management
CN103929737A (en) * 2014-05-08 2014-07-16 国家电网公司 Data backup and recovery method applied to mobile digital network
CN106210089A (en) * 2016-07-18 2016-12-07 浪潮集团有限公司 Media stream separation transmission method based on SPI and P2P network
CN108134772B (en) * 2017-11-06 2020-11-27 杭州电子科技大学 Safe routing method realized by adopting AODV (Ad hoc on demand distance vector) or DSDV (direct distance vector) protocol
CN109152065B (en) * 2018-10-31 2022-03-25 重庆邮电大学 IPv 6-based industrial wireless network multi-node secure network access method
CN110247826B (en) * 2019-07-10 2022-03-25 上海理工大学 Point-to-point network connectivity test method
CN112671937B (en) * 2021-03-18 2021-06-01 杭州金线连科技有限公司 Chat data receiving and sending method
CN113543365B (en) * 2021-06-23 2023-05-30 北京小米移动软件有限公司 Networking method, networking device and storage medium
US11929907B2 (en) 2022-03-08 2024-03-12 T-Mobile Usa, Inc. Endpoint assisted selection of routing paths over multiple networks

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101207537A (en) * 2006-12-22 2008-06-25 中兴通讯股份有限公司 Method for searching stabilizing path in wireless self-organizing network
CN101494593A (en) * 2008-01-25 2009-07-29 上海摩波彼克半导体有限公司 Method for implementing optimization unite connection allowance and route control of wireless mesh network
CN101616431A (en) * 2009-07-30 2009-12-30 迈玺(北京)科技有限公司 The implementation method of network in wireless mesh network and node

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2893476B1 (en) * 2005-11-14 2008-01-25 Radiotelephone Sfr METHOD AND SYSTEM FOR SIMULATION AND RESOURCE MANAGEMENT OF A MOBILE TELEPHONY NETWORK.

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101207537A (en) * 2006-12-22 2008-06-25 中兴通讯股份有限公司 Method for searching stabilizing path in wireless self-organizing network
CN101494593A (en) * 2008-01-25 2009-07-29 上海摩波彼克半导体有限公司 Method for implementing optimization unite connection allowance and route control of wireless mesh network
CN101616431A (en) * 2009-07-30 2009-12-30 迈玺(北京)科技有限公司 The implementation method of network in wireless mesh network and node

Also Published As

Publication number Publication date
CN102026203A (en) 2011-04-20

Similar Documents

Publication Publication Date Title
CN102026203B (en) Multiple-session initiation protocol (SIP) server layout method in wireless Mesh network
Hou et al. Green survivable collaborative edge computing in smart cities
CN101652959B (en) Arrangement and method relating to network management
ES2636889T3 (en) Method and device for coordinating access points for backlink aggregation in a telecommunications network
CN108512760A (en) The method for routing of QoS of survice is ensured based on SDN
Zhang et al. A dynamic trust establishment and management framework for wireless sensor networks
Mazaheri et al. Qos based and energy aware multi-path hierarchical routing algorithm in wsns
Battat et al. Monitoring in mobile ad hoc networks: A survey
CN112492583A (en) Software defined wireless sensor network management method based on cloud edge-side cooperation
Kumar et al. An energy efficient and trust management based opportunistic routing metric for wireless sensor networks
Xu et al. Fiber-wireless network virtual resource embedding method based on load balancing and priority
CN114615706A (en) Method for increasing communication capacity of micropower wireless network in smart power grid
CN110113262A (en) A kind of routing policy determines method, apparatus, SDN controller and heterogeneous network
Khan et al. Gradient cost establishment (grace) for an energy-aware routing in wireless sensor networks
CN108696880A (en) A kind of orientation spanning tree cross-layer communication method based on power control
CN109309944B (en) A kind of wireless multi-hop network topology discovery method
Xiang et al. Self-adjustable trust-based energy efficient routing for smart grid systems
Maheswari et al. Repetitive node categorization technique based reliable clustering and energy efficient communication in P2P wireless sensor network
Wu et al. Extending the lifetime of heterogeneous sensor networks using a two-level topology
Diaz et al. Dynamic gallager-humblet-spira algorithm for wireless sensor networks
CN103701622B (en) Multicast processing method and device based on IRF systems
An et al. Analysis and Comparison of Routing Protocol of Wireless Sensor Network.
Saqaeeyan et al. IEATH: Improved energy aware and two hop multipath routing protocol in wireless sensor networks
Burgos et al. Evaluation of a fault-tolerant wsn routing algorithm based on link quality
Liu et al. Collision-constrained minimum energy node-disjoint multipath routing in ad hoc networks

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C17 Cessation of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20130410

Termination date: 20131217