CN103209434A - Fuzzy-control-based TCP (transmission control protocol) congestion control method in vehicle-mounted communication network - Google Patents

Abstract

The invention discloses a fuzzy-control-based TCP congestion control method in a vehicle-mounted communication network and relates to the field of wireless communication. The method includes utilizing fuzzy control and introducing relative speed of neighboring nodes, distance between neighboring nodes, relative round-trip-delay change rate, short-term throughout change rate as the input parameters of fuzzy inference to establish a fuzzy control rule library; and determining a packet loss reason through the fuzzy inference, and improving TCP congestion control strategies and route switching strategies in combination with cross-layer schemes to take different processing countermeasures for different packet loss reasons, so that the route stability can be enhanced, the transmission interruptions caused by route switching can be reduced, and the system throughput and the average delay can be improved. The fuzzy-control-based TCP congestion control method in the vehicle-mounted communication network is used for improving the TCP transmission performance in the vehicle-mounted communication network, and is particularly applicable to conditions of high-speed node motion, frequent topological change and poor channel quality.

Description

In a kind of vehicle-carrying communication net based on the TCP jamming control method of fuzzy control

Technical field

The present invention relates generally to wireless communication field, especially about the problem of TCP transmission performance in the vehicle-carrying communication net.

Background technology

In recent years, communication network and transportation network are combined closely in the explosive growth of vehicle and ubiquitous information requirement, and the vehicle-carrying communication net arises at the historic moment thus, and become a research focus.The vehicle-carrying communication net is divided into two parts: RVC(Roadto Vehicle Communications, communicate by letter between bus) and IVC(Inter-Vehicle Communication, inter-vehicular communication).RVC makes vehicle insert the Internet by road infrastructure or mobile base station, obtains Internet resources; And IVC makes and can form self-organizing network by the mode of multi-hop between the vehicle by wireless telecommunications system is installed at vehicle, i.e. VANET(Vehicular Ad hoc Network, vehicle-mounted self-organizing network).

VANET is the inter-vehicular communication network of a kind of self-organizing, structure opening, not only have characteristics such as mobile ad-hoc network is dynamic, distributed, multi-hop, also has self exclusive attribute, high speed, network topology change as vehicle movement is frequent, running orbit is measurable, no tangible hardware constraints etc., these attributes directly influence the transmission performance in the VANET environment, make packet loss increase, postpone to increase.Under the VANET environment, vehicle movement is retrained by road conditions, and its running orbit is measurable, and vehicle can obtain abundant outer secondary supplementary information, in conjunction with these advantages can design more efficiently, transmission control protocol reliably.

The relevant application of many security information is arranged among the VANET, be necessary to select reliable host-host protocol for these application, as TCP(Transmission Control Protocol, transmission control protocol).Yet traditional TCP is based on wire message way design, because the transmission medium reliability is very high under the cable environment, the various improvement versions of traditional TCP and TCP as TCPReno, TCPNewReno, TCPVegas etc., think that all packet loss is because due to the network congestion.And in the wireless network, because channel quality is not good, the packet loss rate height if adopt wired TCP congestion avoidance algorithm, all is considered as the indication of network congestion with packet loss, must cause the serious reduction of whole network transmission performance.The improvement project of TCP under a lot of wireless environments has now been proposed, as TCP Veno(ChengPeng Fu, SC.Liew " TCP Veno:TCP enhancement for transmission over wireless access networks " IEEE Journal of Selected Areas in Communication.2003, Vol.21 (2): 216-228) be the TCP improvement project that proposes at solving high bit-error problem in the wireless network environment specially, a large amount of evidences is in cell cellular net and 3G network, and the network transmission performance of TCP Veno has bigger improvement.

Under the VANET environment, translational speed is fast because its transmission range is short, and link switches very frequent, and to switch the packet loss that causes be the major reason that influences the TCP transmission performance to route frequently.When causing route, the node fast moving interrupts, can cause a plurality of continuous T CP data-bag losts, if transmitting terminal can not receive ACK(Acknowledgment, confirmation signal), need to wait for RTO(Retransmission Time Out, the overtime re-transmission time) begin the re-transmission task after, may trigger unnecessary congested control process simultaneously.This shows, switch the packet loss that causes if can not effectively handle route, will cause the increase of propagation delay time, and bring irrational congested control and re-transmission.At document (Marc Bechler, Sven Jaap, Lars Wolf " An Optimized TCP for Internet Access of Vehicular Ad Hoc Networks " Lecture Notes in Computer Science, 2005, Volume 3462, NETWORKING2005:869-880) proposed in based on TCP prioritization scheme under the VANET environment of agency mechanism: MCTP(Mobile Control Transport Protocol, mobile transmission control protocol), this scheme is separated by agency's TCP connection end to end is set, and in the agency, the TCP transmission performance is optimized, but this scheme is owing to need to increase the agency, realize comparatively complicated, compatibility is also good inadequately, and this scheme is also carried out combined optimization in conjunction with network layer not in conjunction with characteristics such as vehicle-carrying communication net movement tendency are measurable.

Under the VANET environment, the reason and the parameter that influence packet loss are a lot, the speed that moves as communication distance, vehicle, the fluctuation of data transmission delay, end-to-end change in throughput etc., yet, corresponding relation between these available parameters and the Link State is difficult to show by setting up a clear and definite functional relation, therefore, the method by fuzzy control (Fuzzy Control) is a kind of good idea.The TCP transmission performance that some documents have applied to fuzzy control theory in the mobile ad-hoc network is improved, as document (Kai Shi, Yantai Shu, Qingfeng Song " Receiver Centric Fuzzy Logic Congestion Control for TCP Throughput Improvement over Wireless Networks " The Sixth International Conference on Fuzzy Systems and Knowledge Discovery, 2009:146-150) carry out fuzzy control in conjunction with bandwidth availability ratio and packet loss, to adjust the size of congested control window, improve the TCP transmission performance.Document (WANGXiaoBo, YANGYuLiang, AN Jian Wei " Multi-Metric Routing Decisions in VANET " The Eighth IEEE International Conference on Dependable, Autonomic and Secure Computing, 2009:551-556) fuzzy control theory is incorporated into improves Routing Protocol in the vehicle-carrying communication net, in conjunction with the characteristic of vehicle-carrying communication net, take all factors into consideration vehicle heading and the route life cycle is carried out routing decision.Above achievement in research is used reference for providing based on the TCP improvement in performance of fuzzy control in the vehicle-mounted communication network of design.

Summary of the invention

The present invention is directed under the VANET environment, the corresponding relation that influences between a lot of and Link State of the reason of packet loss and parameter is difficult to by setting up a clear and definite functional relation, the above-mentioned defective that exists in the prior art, a kind of packet loss differentiating algorithm based on fuzzy control has been proposed, and introduce and to stride layer thought a kind of TCP congestion avoidance algorithm and routing protocol based on fuzzy control is provided, with overcome that route in the vehicle-carrying communication net is switched and the error code packet loss to the influence of TCP transmission performance.

The technical scheme that the present invention addresses the above problem is, a kind of TCP jamming control method VF-TCP(Vehicular Fuzzy-control Transmission Control Protocol based on fuzzy control in the vehicle-carrying communication net is proposed, vehicle-mounted fuzzy control TCP) and improved Routing Protocol VF-DSR(Vehicular Fuzzy-control Dynamic Source Routing, vehicle-mounted fuzzy control DSR), adopt fuzzy control method, relative velocity according to adjacent node, the distance of adjacent node, relative round-trip delay rate of change, short-term throughput rate of change is as the input parameter of fuzzy reasoning, set up fuzzy rule base, judge the packet loss reason by fuzzy reasoning, and improve TCP congestion control policy and route switchover policy in conjunction with striding layered scheme.Based on the TCP jamming control method of fuzzy control, the first fuzzy control subsystem FC1 carries out fuzzy control according to car-mounted terminal relative velocity and distance, obtains the fuzzy set about routing stability in a kind of vehicle-carrying communication net; Routing stability fuzzy set, relative round-trip delay rate of change, short-term throughput rate of change are imported the second fuzzy control subsystem FC2; FC2 obtains the packet loss reason according to fuzzy rule base through fuzzy reasoning, and determines Link State according to the packet loss reason, takes corresponding TCP congestion control policy according to Link State; Carry out in route foundation and the reconstruction according to the VF-DSR Routing Protocol, source node upgrades routing table, obtains all routes that arrive destination node, and selects optimum route based on fuzzy control.

Can reflect a plurality of parametric synthesis utilizations of Link State by fuzzy control, comprise the distance of relative velocity, the adjacent node of adjacent node, round-trip delay rate of change, short-term throughput rate of change relatively, propose a kind of effective packet loss differentiating algorithm thus.

Packet loss differentiating algorithmic system based on fuzzy control comprises two fuzzy control subsystem FC1 and FC2.The effect of FC1 is to carry out fuzzy control according to relative velocity and distance, obtains one about the fuzzy set of routing stability; Routing stability fuzzy set and relative round-trip delay rate of change, short-term throughput rate of change together as the output parameter of FC2, obtain the packet loss reason through fuzzy reasoning, and control the transmission control procedures of TCP according to final output result.

After packet loss took place, the TCP transmitting terminal was distinguished the packet loss reason and is taked congested control measure by fuzzy control, if judge that packet loss is former because link congestion changes common congested state of a control (abbreviating congested state of a control as) over to; If judge that packet loss is former because the link error code changes the congested state of a control (abbreviating " bad " channel status as) under " bad " channel over to; If judge that packet loss is former because link disconnects, change over to wait and stop state, waiting for that the route layer is finished switches back re-transmission packet loss.If judge that packet loss is former because link disconnects, trigger to wait and stop state, TCP transmitting terminal replacement ssthresh(starts thresholding slowly), the cwnd(congestion window), the rwnd(receive window) value, and time-out sends data, after treating that route is rebuild successfully, the network layer transmitting terminal rebulids success by striding a layer communication notifications TCP transmitting terminal route, and ssthresh, cwnd, rwnd and the SYN(sequence number of record were synchronous before the TCP transmitting terminal recover to be reset) value, and reduce cwnd and send data behind the 4/5cwnd; If judge that packet loss is former because the link error code, indicating channel is in " bad " channel status, the ssthresh value is revised as (4/5) * cwnd, cwnd resets to ssthresh+3, retransmission of lost packets then, when whenever receiving a repeat ACK, ssthresh adds 1 and send 1 message segment, when the new ACK of the next one arrived, putting cwnd was ssthresh; If judge that packet loss is former because link congestion enters congested state of a control, retransmit fast and fast quick-recovery according to the TCPReno congestion control policy.

Adopt the VF-DSR agreement, in route foundation and process of reconstruction, source node at first upgrades routing table, obtains all routes that arrive destination node, and selects optimum route based on fuzzy control.VF-DSR selects the criterion of optimum route to be:

1. selecting routing stability is the route of " highly stable " or " more stable ";

2. not selecting routing stability is the route of " very unstable ";

3. exist many stability preferably during route, select the minimum route of jumping figure.

Adopt TCP congestion control mechanism of the present invention and route improvement strategy (VF-TCP﹠amp; VF-DSR), transmission performance is greatly improved under vehicle-carrying communication net environment, effectively solves in the vehicle-carrying communication net owing to not distinguishing the TCP transmission performance reduction problem that link congestion, route switching and error code packet loss cause.Throughput of system is improved and end-to-end time delay decreases, and especially under the not good environment of node high-speed mobile, topological frequent variations, channel quality, its advantage is more obvious.

Description of drawings

Fig. 1 is VF-TCP fuzzy control overall framework;

Fig. 2 is the fuzzy control process schematic diagram;

Fig. 3 is based on VF-TCP four state transition diagrams of fuzzy control;

Fig. 4 is that VF-TCP and TCPVeno average throughput are with the velocity variations correlation curve;

Fig. 5 is that the average end-to-end time delay of VF-TCP and TCPVeno is with the velocity variations correlation curve.

Embodiment

In order to overcome route switching and error code packet loss to the influence of TCP transmission performance in the vehicle-carrying communication net, the present invention proposes in a kind of vehicle-carrying communication net the TCP jamming control method VF-TCP(VehicularFuzzy-controlTransmissionControlProtocol based on fuzzy control, vehicle-mounted fuzzy control TCP) and improved Routing Protocol VF-DSR(VehicularFuzzy-controlDynamicSourceRouting, vehicle-mounted fuzzy control DSR), adopt fuzzy control and stride a layer thought, a kind of effective packet loss differentiating algorithm is provided, and improve TCP congestion control policy and route switchover policy, to solve in the vehicle-carrying communication net owing to can not distinguish link congestion, route is switched and error code packet loss and the TCP transmission performance that causes reduces.

Below by accompanying drawing and instantiation enforcement of the present invention is further described in detail.Comprise as the lower part: based on the packet loss differentiating algorithm of fuzzy control, based on fuzzy control and stride the TCP congestion control policy of layer design, based on the route switchover policy of fuzzy control.

One, based on the packet loss differentiating algorithm of fuzzy control

Be illustrated in figure 1 as the overall system framework of packet loss differentiating algorithm of the present invention, this system comprises two fuzzy control subsystem FC1 and FC2.The first fuzzy control subsystem FC1 carries out fuzzy control according to vehicle mobile terminals relative velocity and distance, sets up the routing stability fuzzy set.With routing stability fuzzy set and relative round-trip delay rate of change, the short-term throughput rate of change input parameter as the second fuzzy control subsystem FC2, according to the distance of the relative velocity of adjacent node, adjacent node, round-trip delay rate of change, short-term throughput rate of change be as the input parameter of fuzzy reasoning relatively, set up fuzzy rule base, judge packet loss reason (Reason) through the FC2 fuzzy reasoning, the packet loss reason is used for the transmission control procedures of control TCP.And improve TCP congestion control policy and route switchover policy in conjunction with striding layered scheme.

The process that FC1 and FC2 carry out fuzzy control comprises obfuscation, fuzzy knowledge base, fuzzy reasoning and reverse gelatinization as shown in Figure 2.Further specify below.

Fuzzy processing

Acquisition node { i, the speed of j} According to formula:

(1)

Calculate adjacent node { i, the relative velocity of j}.Be [0,60] as domain, unit is meter per second, and the value set of speed variable (Velocity) is divided, and it is divided into three fuzzy sets: and SV(is at a slow speed), the MV(middling speed), FV(is quick).

According to node coordinate calculate adjacent node i, the distance of j}:

$S=\sqrt{{(x_i^{\′}-x_j^{\′})}^2+{(y_i^{\′}-y_j^{\′})}^2}$ （2）

Wherein,

Be respectively node { i, the coordinate of j}.

Be [0,300] as domain, unit is rice, will be divided into four fuzzy sets apart from (Distance):

The SS(short distance), and distance among the MS(), the long distance of LS() and, the very long distance of VLS()

According to formula:

(3) calculate relative round-trip delay rate of change.Wherein, subscript n represents to report code (down together).

Domain is [∞ ,+∞], and time delay rate of change (dRTT) is divided into three fuzzy sets: the SR(reduction), and MR(keeps), LR(increases) }.Calling following formula calculates:

SRTT _n=(1-α)·SRTT _n-1+α·RTT _n （4）

Wherein, RTT _nBe the sampled value of current message segment RTT, SRTT _nBe level and smooth round-trip delay, α is the smoothing factor between 0 and 1, and according to experiment, the value optimum of α gets 1/8.

Short-term throughput rate of change is:

${\mathrm{dSTT}}_n=\frac{{\mathrm{STT}}_n-{\mathrm{STT}}_{n-1}}{{\mathrm{STT}}_n}$ （5）

Domain is [∞ ,+∞], and throughput rate of change (dSTT) is divided into three fuzzy sets: the SR(reduction), and MR(keeps), LR(increases) }.

Wherein, according to formula:

${\mathrm{STT}}_n=\frac{\mathrm{Packet}\_\mathrm{num}\×\mathrm{Packet}\_\mathrm{size}}{T}$ （6）

Calculate STT _nHere, T is measurement period, and Packet_num is data packets for transmission number in the statistic period T, and Packet_size is the packet size.

Routing stability is got the minimum stability of jumping as the whole piece route of stability in the multi-hop route as the output parameter of fuzzy reasoning subsystem FC1.Itself also be a fuzzy quantity, as the input parameter of subsystem FC2.If domain is [0,1], routing stability (Route-Stability) is divided into four fuzzy sets: VLR(is very unstable), LR(is unstable), MR(is generally stable), HR(is highly stable).Adopt the Sugeno inference pattern, output parameter should be single order multinomial or constant, can select constant for use here.As the FC1 output parameter, VLR, LR, MR, HR corresponding constant respectively can be 0,0.3,0.6,0.9.

Finally be output as the packet loss reason by what subsystem FC2 fuzzy reasoning drew, take different congestion control policies according to the packet loss reason.The congestion control policy domain is [0,1], and packet loss reason (Reason) is divided into three fuzzy sets: and the disconnection of LB(link), LE(link error code), the LC(link congestion).Equally, can adopt the output parameter of Sugeno inference pattern, RH, TE, LC corresponding constant 0,0.5,1 respectively is set.

Characteristics according to fuzzy object are selected suitable membership function respectively for use, use matlab blur tool case according to the membership function (routing stability is as the input parameter of FC2) of five input parameters, obtain the membership function of relative velocity, distance, routing stability, dRTT and dSTT.

For convenient, the membership function of an I/O parameter is designated as:

Wherein, u _iBe the fuzzy set that this parameter comprises, contained fuzzy set adds up to I, x _iBe u _iCorresponding degree of membership.

Fuzzy rule base

Set up fuzzy rule base, wherein, the fuzzy inference rule table is represented the derivation relationship between adjacent node relative velocity, adjacent node distance and the routing stability, and " if-then " inference rule of its FC1 is as follows:

1.If(Velocity is SV)and(Distance is SS)then(Route-Stability is HR)

2.If(Velocity is SV)and(Distance is MS)then(Route-Stability is HR)

3.If(Velocity is SV)and(Distance is LS)then(Route-Stability is MR)

4.If(Velocity is SV)and(Distance is VLS)then(Route-Stability is VLR)

5.If(Velocity is MV)and(Distance is SS)then(Route-Stability is HR)

6.If(Velocity is MV)and(Distance is MS)then(Route-Stability is HR)

7.If(Velocity is MV)and(Distance is LS)then(Route-Stability is LR)

8.If(Velocity is MV)and(Distance is VLS)then(Route-Stability is VLR)

9.If(Velocity is FV)and(Distance is SS)then(Route-Stability is HR)

10.If(Velocity is FV)and(Distance is MS)then(Route-Stability is MR)

11.If(Velocity is FV)and(Distance is LS)then(Route-Stability is LR)

12.If(Velocity is FV)and(Distance is VLS)then(Route-Stability is VLR)

Be example with above-mentioned the 1st rule, rule 1 is: when the adjacent node relative velocity is very low, and the adjacent node distance very in short-term, and two internodal routing stabilitys are very high.In like manner, other interpretation of rules are similar to rule 1.

" if-then " fuzzy inference rule following (part) of FC2:

1.If(Route-Stability is VLR)and(dRTT is LRTT)and(dSTT is LSTT)then(Reason is LB)

2.If(Route-Stability is VLR)and(dRTT is LRTT)and(dSTT is MSTT)then(Reason is LB)

3.If(Route-Stability is VLR)and(dRTT is LRTT)and(dSTT is HSTT)then(Reason is LB)

4.If(Route-Stability is VLR)and(dRTT is MRTT)and(dSTT is LSTT)then(Reason is LB)

5.If(Route-Stability is VLR)and(dRTT is MRTT)and(dSTT is MSTT)then(Reason is LB)

6.If(Route-Stability is VLR)and(dRTT is MRTT)and(dSTT is HSTT)then(Reason is LB)

7.If(Route-Stability is VLR)and(dRTT is HRTT)and(dSTT is LSTT)then(Reason is LB)

8.If(Route-Stability is VLR)and(dRTT is HRTT)and(dSTT is MSTT)then(Reason is LB)

…………

30.If(Route-Stability is HR)and(dRTT is LRTT)and(dSTT is HSTT)then(Reason is LE)

31.If(Route-Stability is HR)and(dRTT is MRTT)and(dSTT is LSTT)then(Reason is LC)

32.If(Route-Stability is HR)and(dRTT is MRTT)and(dSTT is MSTT)then(Reason is LE)

33.If(Route-Stability is HR)and(dRTT is MRTT)and(dSTT is HSTT)then(Reason is LE)

34.If(Route-Stability is HR)and(dRTT is HRTT)and(dSTT is LSTT)then(Reason is LC)

35.If(Route-Stability is HR)and(dRTT is HRTT)and(dSTT is MSTT)then(Reason is LC)

36.If(Route-Stability is HR)and(dRTT is HRTT)and(dSTT is HSTT)then(Reason is LE)

Defuzzification

Adopting above-mentioned membership function and rule base, is that example is explained the defuzzification process with the FC1 subsystem.As if V=40, S=80, calculate the membership function of input parameter V, S respectively according to degree of membership:

$V=\{\frac{0}{\mathrm{SV}}+\frac{0.6}{\mathrm{MV}}+\frac{0.2}{\mathrm{FV}}\},$ $S=\{\frac{0.2}{\mathrm{SS}}+\frac{0.8}{\mathrm{MS}}+\frac{0}{\mathrm{LS}}+\frac{0}{\mathrm{VLS}}\}$

Inference rule according to FC1 in the fuzzy rule base obtains result as shown in table 1.Here " prod " method is adopted in " and " operation, is about to input element and multiplies each other.

Table 1 adopts FC1 inference rule table to carry out reasoning

Rule	Velocity	Distance	Route-Stability
				1	SV：0	SS：0.2	HR：0
2	SV：0	MS：0.8	HR：0
				3	SV：0	LS：0	MR：0
4	SV：0	VLS：0	VLR：0
				5	MV：0.6	SS：0.2	HR：0.12

6	MV：0.6	MS：0.8	HR：0.48
				7	MV：0.6	LS：0	LR：0
8	MV：0.6	VLS：0	VLR：0
				9	FV：0.2	SS：0.2	HR：0.04
10	FV：0.2	MS：0.8	MR：0.16
				11	FV：0.2	LS：0	LR：0
12	FV：0.2	VLS：0	VLR：0

To above The reasoning results, get weighted average and finally exported, can get weighted average according to following formula,

$\mathrm{Route}-\mathrm{Stability}=\frac{\mathrm{\Σ}{u}_i{x}_i}{\mathrm{\Σ}{u}_i}$ （7）

Wherein, u _iValue be the substrate (being VLR, LR, MR, HR corresponding constant 0,0.3,0.6,0.9 respectively) of fuzzy set Route-Stability correspondence, x _iBe the corresponding output valve of each bar rule, i.e. degree of membership.Result of calculation as Route-Stability in this example is:

$\mathrm{Routre}-\mathrm{Stability}=\frac{0.12\×0.9+0.48\×0.9+0.04\×0.9+0.16\×0.6}{0.12+0.48+0.04+0.6}=0.84$

To after fuzzy subsystem FC1 and the FC2 defuzzification, can access routing stability and packet loss reason respectively, with to three kinds of packet loss reasons: link congestion, link error code, link switch takes different treatment Countermeasures respectively.

Two, based on fuzzy control and stride the TCP congestion control policy of layer design

Tradition TCP only distinguishes two states: normal condition and congested state of a control, and normal condition refers to traditional startup slowly and congestion avoidance phase here, and congested state of a control refers to quick re-transmission and fast quick-recovery stage among the TCPReno.VF-TCP has increased by two kinds of new states: wait and stop state and " bad " channel status, the triggering of state all realizes by fuzzy control.Cross-layer optimizing by network layer and transport layer improves traditional TCP congestion control policy, based on fuzzy control and the VF-TCP four state transformational relations of striding layer design as shown in Figure 3.

After packet loss took place, the TCP transmitting terminal obtained the packet loss reason by the packet loss differentiating algorithm based on fuzzy control.And determine Link State according to the packet loss reason, take corresponding TCP congestion control policy according to Link State.

If judge that packet loss is former because link disconnects, then trigger to wait and stop state; TCP transmitting terminal replacement ssthresh(starts thresholding slowly), the cwnd(congestion window), the rwnd(receive window) value, and suspend to send data; After treating that route is rebuild successfully, the network layer transmitting terminal rebulids success by striding a layer communication notifications TCP transmitting terminal route, ssthresh, cwnd, rwnd and the SYN(sequence number of record were synchronous before the TCP transmitting terminal recover to be reset) value, and reduce cwnd and send data behind the 4/5cwnd; After retransmitting successfully, recover normal condition.The link error code if packet loss is former, then indicating channel is in " bad " channel status; The TCP transmitting terminal is revised as (4/5) * cwnd with the ssthresh value, and cwnd resets to ssthresh+3, then retransmission of lost packets, when whenever receiving a repeat ACK, ssthresh adds 1 and send 1 message segment, and when the new ACK of the next one arrived, putting cwnd was ssthresh; After retransmitting successfully, recover normal condition.

The link congestion if packet loss is former then enters congested state of a control; Retransmit fast and fast quick-recovery according to TCP Reno congestion control policy; After retransmitting successfully, recover normal condition.

Three, the route switchover policy that reaches based on fuzzy control

In transmission of data packets, routing stability is more high, and the possibility that link disconnects is just more low, and TCP transmission end to end is just more stable.In the DSR Routing Protocol, source node can keep many routes that arrive destination node, does not but have regulation should how to select a route of optimum wherein.Fuzzy control subsystem FC1 of the present invention can estimate route stability, and this just provides useful reference for the Route Selection of DSR.Fuzzy control is carried out in route foundation and process of reconstruction, selected a most stable jumping route, can improve the TCP transmission performance indirectly.Therefore, the present invention improves the routing policy of DSR Routing Protocol, and is corresponding with VF-TCP, and the DSR Routing Protocol after improving is called VF-DSR.The concrete enforcement of VF-DSR Routing Protocol is as follows:

Carry out in route foundation and the process of reconstruction according to the VF-DSR Routing Protocol, source node at first upgrades routing table, obtains all routes that arrive destination node, and selects optimum route based on fuzzy control.Owing among the fuzzy control subsystem FC1, choose the minimum reference of jumping as routing stability of stability in the multi-hop route, to two routes identical with the routing stability grade, think that then the more few route of jumping figure is more excellent.Therefore, select the criterion of optimum route to be according to VF-DSR:

1. selecting routing stability is the route of " highly stable " or " more stable ";

2. not selecting routing stability is the route of " very unstable ";

3. exist many stability all preferably during route, select the minimum route of jumping figure.

After route was rebuild and to be finished, the route layer was by striding a layer communication notifications TCP transmitting terminal, received that the RREP(route should divide into groups) after, the packet of TCP transmitting terminal retransmission of lost also finishes to wait and stops state.

As mentioned above, VF-TCP distinguishes the packet loss reason fast by fuzzy control, and introduce and stride a layer thought, improve TCP congestion control mechanism and route switchover policy based on fuzzy control, so that the different packet loss reason is taked different treatment Countermeasures, strengthen routing stability simultaneously, reduce route and switch the transmission interruption that brings, make throughput of system and average delay improve.

Use the VanetMobisim instrument to build the traffic simulation scene, use NS2 software to carry out network simulation.According to the IEEE802.11p standard, physical layer and MAC layer parameter are carried out detailed configuration, select the 5.9GHz frequency range, bandwidth is the 10MHz that stipulates in the standard, and the node communication scope is 250m, and the monitoring scope is 550m, critical value S is made as 200m, when euclidean distance between node pair during greater than 200m, with probability P ₀=0.1 the error code packet loss takes place, and continuous FTP transmission is arranged between reference node.VF-TCP and TCP Veno are compared, and wherein the route layer of TCP Veno is selected the DSR agreement for use, and the route layer of VF-TCP will be selected DSR and VF-DSR agreement respectively for use.Node moves with 6 kinds of average speeds respectively: 0m/s(is static), 5m/s, 10m/s, 15m/s, 20m/s, 25m/s, simulation time 500s.Carry out traffic scene emulation for each average speed, and the trace file that will at every turn generate is used VF-TCP respectively and two kinds of agreements of TCP Veno are carried out network simulation, try to achieve throughput and the time delay mean value of VF-TCP and TCP Veno simulation result under each average speed respectively, simulation result such as Fig. 4, shown in Figure 5.As seen under the environment that, change in topology very fast at node speed is frequent, channel status is not good, VF-TCP﹠amp; VF-DSR can effectively improve average throughput, and reduces average end-to-end time delay.

Claims (8)

1. In the vehicle-carrying communication net based on the TCP jamming control method of fuzzy control, it is characterized in that the first fuzzy control subsystem FC1 carries out fuzzy control according to car-mounted terminal relative velocity and distance, obtains the fuzzy set about routing stability; Routing stability fuzzy set, relative round-trip delay rate of change, short-term throughput rate of change are imported the second fuzzy control subsystem FC2; FC2 obtains the packet loss reason according to fuzzy rule base through fuzzy reasoning, and determines Link State according to the packet loss reason, takes corresponding TCP congestion control policy according to Link State; Carry out in route foundation and the reconstruction according to the VF-DSR Routing Protocol, source node upgrades routing table, obtains all routes that arrive destination node, and selects optimum route based on fuzzy control.
2. 2. TCP jamming control method according to claim 1, it is characterized in that, the determining of described routing stability is specially, get the minimum stability of jumping as the whole piece route of stability in the multi-hop route, its domain is [0,1], be divided into four fuzzy sets: { very unstable VLR, more unstable LR generally stablizes MR, highly stable HR}.
3. 3. TCP jamming control method according to claim 1 is characterized in that, relatively the round-trip delay rate of change determine be specially: the value set to the speed variable is divided into three fuzzy sets: SV(is at a slow speed), the MV(middling speed), FV(is quick) }; Will be apart from being divided into four fuzzy sets: the SS(short distance), among the MS(apart from), LS(long apart from), VLS(very long apart from), according to formula:

   

   Calculate relative round-trip delay rate of change dRTT _n, wherein, SRTT _n=(1-α) SRTT _N-1+ α RTT _n, RTT _nBe the RTT sampled value of current message segment, SRTT _nBe level and smooth round-trip delay, α is the smoothing factor between 0 and 1, and subscript n represents to report code.
4. 4. according to one of them described TCP jamming control method of claim 1-3, it is characterized in that the determining of short-term throughput rate of change is specially, according to formula: ${\mathrm{STT}}_n=\frac{\mathrm{Packet}\_\mathrm{num}\×\mathrm{Packet}\_\mathrm{size}}{T}$ Calculate STT _nAccording to formula: ${\mathrm{dSTT}}_n=\frac{{\mathrm{STT}}_n-{\mathrm{STT}}_{n-1}}{{\mathrm{STT}}_n}$ Calculate short-term throughput rate of change dSTT _nWherein, T is measurement period, and Packet_num is data packets for transmission number in the statistic period T, and Packet_size is the packet size, and subscript n represents to report code.
5. 5. according to one of them described TCP jamming control method of claim 1-3, it is characterized in that link disconnects if packet loss is former, then trigger to wait and stop state, the TCP transmitting terminal is reset and is started the value of thresholding ssthresh, congestion window cwnd, receive window rwnd slowly, and suspends the transmission data; After treating that route is rebuild successfully, the network layer transmitting terminal rebulids success by striding a layer communication notifications TCP transmitting terminal route, the synchronous SYN value of ssthresh, cwnd, rwnd and sequence number of record before the TCP transmitting terminal recover to be reset, and reduce cwnd and send data behind the 4/5cwnd; After retransmitting successfully, recover normal condition.
6. 6. according to one of them described TCP jamming control method of claim 1-3, it is characterized in that, the link error code if packet loss is former, then indicating channel is in " bad " channel status; The TCP transmitting terminal is revised as (4/5) * cwnd with the ssthresh value, and cwnd resets to ssthresh+3, then retransmission of lost packets, when whenever receiving a repeat ACK, ssthresh adds 1 and send 1 message segment, and when the new ACK of the next one arrived, putting cwnd was ssthresh; After retransmitting successfully, recover normal condition.
7. 7. according to one of them described TCP jamming control method of claim 1-3, it is characterized in that the link congestion if packet loss is former then enters congested state of a control; Retransmit fast and fast quick-recovery according to the TCPReno congestion control policy; After retransmitting successfully, recover normal condition.
8. 8. according to one of them described TCP jamming control method of claim 1-3, it is characterized in that, describedly select optimum route to be specially based on fuzzy control: select routing stability to be the route of " highly stable " or " more stable "; Not selecting routing stability is the route of " very unstable "; When having the route of many " highly stable " or " more stable ", select the minimum route of jumping figure; After route was rebuild and to be finished, the route layer was by striding a layer communication notifications TCP transmitting terminal, received that route should divide into groups behind the RREP, and the packet of TCP transmitting terminal retransmission of lost also finishes to wait and stops state.

Images