CN104270790A - Congestion control method based on equitable distribution of communication channel - Google Patents

Abstract

The invention discloses a congestion control method based on equitable distribution of a communication channel. The method comprises the following steps: firstly, the queue increase rate of an upstream neighbor node and the average transmission delay of a data packet are estimated according to the acquired data of the upstream neighbor node; then, the queue length of the upstream neighbor node is calculated according to the estimated queue increase rate, the total queue length of the node and the upstream neighbor node is further calculated, and the congestion detection is carried out in combination with the average transmission delay of the data packet; if congestion occurs, the network congestion degree (namely the congestion degree) is classified, the length of the congestion relief period is determined, and the node is equitably distributed with the communication channel occupation time; finally, a congestion relief algorithm is carried out according to the congestion degree so as to relieve or ease the congestion; if the congestion does not occur, no treatment is carried out. The use equity of the channel is improved, the collision and the packet loss probability are effectively reduced, and the handling capacity is effectively increased.

Description

Based on the jamming control method of channel fair allocat

Technical field

The present invention relates to the communications field, relate to a kind of jamming control method based on channel fair allocat more specifically.

Background technology

In wireless sensor network (Wireless Sensor Network, WSN), the feature such as mutual interference of the limited and wireless link of many-to-one communication mode, wireless bandwidth resource, causes wireless sensor network very easily to occur congested.Network congestion easily causes cache overflow, causes mass data packet loss, increases Network queuing delay, and consumes a large amount of additional energies.Meanwhile, network congestion also can cause access conflict, reduces the utilance of link and the throughput of network.Therefore, efficiently detect in real time and remove the congested hot issue having become guarantee wireless sensor network service quality.

Congestion control can be divided into network resource management and traffic control two kinds of methods.Network resource management refers to the bandwidth resources by managing wireless sensing device network, and namely reasonable distribution bandwidth controls congested.But how rationally and accurately bandwidth is distributed in multihop network, to avoid bandwidth resources under-supply or supply excessive problem, be also difficult at present realize.Traffic control is mainly when network congestion being detected, and node is propagated to prevent congested continuation to downstream node by reducing transmission rate, and upstream nodes sends notice message, to avoid increasing the weight of existing Congestion Level SPCC.Existing traffic congestion control method mainly contains CODA, PCCP, Fusion etc.

CODA (Congestion Detection andAvoidance) agreement detects congested based on buffer storage length and channel utilization, adopts open loop hop-by-hop backpressure mechanism and closed loop multi-source regulation mechanism to carry out congestion control.Open loop hop-by-hop backpressure mechanism is suitable for removing instantaneous congestion, and closed loop multi-source regulation mechanism is suitable for removing persistent congestion.CODA congestion detection mechanism is the congestion condition that the ACK whether receiving sink node feeding back in real time according to source node judges network.If the data volume that source node sends exceedes threshold value, do not receive the ACK of sink node feeding back yet, then to think that network occurs congested.When network congestion being detected, just spread news to source node direction by backpressure mechanism, the node receiving message adopts packet loss or additivity increasing multiplicative to subtract (AIMD) mode to suppress the transmission rate of node, hop-by-hop feeds back to source node, until congested elimination, this mode easily causes the unsteadiness of network throughput, and poor away from the channel allocation fairness of sink node.

The ratio that PCCP (Priority-based Congestion Control Protocol) protocol definition node accepts speed and transmission rate is degree of Congestion, and congested according to degree of Congestion Sampling network.After child node listens to the congestion information of father node, by calculating the degree of Congestion of father node, in conjunction with information such as self priority and son node number, regulating the transmission rate of self: when the degree of Congestion of father node is less, just increasing the transmission rate of self; When degree of Congestion is larger, just reduce the transmission rate of self.PCCP can ensure the transmission rate of node adjustment promptly and accurately self, but have ignored because the adjustment of speed is to the congestion effects of next-hop node.

Fusion is the congestion control solution of cross-layer, merged flow control end to end, source node flow rate limit and have mac-layer protocol 3 kinds of technology of priority.Adopting length of buffer queue and channel to gather two kinds of modes detects congested.When father node detects congested, the congested flag of packet header is just set, to the transmission congestion notification message of child node implicit expression.Child node just stops when receiving congestion notification sending data to father node, gives the transmission priority that congested node is high.The number of child nodes of node listens father node, uses the transmission rate of Token Bucket Policing restriction node self, thus ensure that the identical transmitted traffic of all nodes.Congested in order to remove in time, congested node reduce that Backoff window size is non-congested node 1/4, the data accelerating congested node send, thus give congested node high transmission priority.Fusion can remove congested in time, but lacks the measure guaranteed fairness.

In summary, existing congestion detection method mainly realizes based on the buffer storage length information of individual node and channel status.When individual node does not compete channel for a long time, easily cause buffer memory sharply to increase, its information shared in the channel can not reflect real-time buffer status really, if do not consider whole channel interior nodes state, may cause congested void inspection and detection leakage phenomenon.

And existing congestion control scheme, mainly by reducing the data transmission rate of source node or the forwarding rate of intermediate node, and increase packet loss carrys out alleviate congestion, and to sacrifice bag transmissibility with increase overhead for cost increases throughput.The channel collision how to be caused due to excessive competition by reduction, is guaranteed that the channel resource allocation of effective and the higher bag transmissibility of maintenance are the key technology difficult problems that alleviate congestion needs solution badly, is not still had suitable solution at present.

Summary of the invention

For the problems referred to above, the object of the invention is to, propose a kind of jamming control method based on channel fair allocat.By estimation node and the buffer memory total length of upstream neighbor node and the average transmission time delay of packet, solve the void inspection and undetected problem that exist in wireless sensor network congestion detection.The fairness that the present invention was used to improve channel by the time of fair allocat node busy channel, thus reduce collision and packet loss, and increase throughput.

The present invention, first estimates the average transmission time delay of its queue rate of rise and packet according to the data volume obtaining upstream neighbor node.Then, calculate the queue length of its upstream neighbor node according to the queue rate of rise of estimation, the total length of further computing node itself and upstream neighbor node queue thereof, and perform congestion detection in conjunction with the average transmission time delay of packet; If occur congested, then by network congestion degree classification (i.e. degree of Congestion), and determine the relieve congestion cycle, to the node fair allocat busy channel time.Finally, according to degree of Congestion, perform relieve congestion algorithm, thus remove or alleviate congestion; If do not occur congested, then do not process.

Concrete steps of the present invention are as follows:

Step one preliminary treatment: calculate the time T successfully sent needed for a packet _s ¹; Calculate in the ideal situation, a node is at a virtual period T _cyclethe interior maximum amount of data D that can transmit _max;

The forward node (being defined as node x) that step 2 has multiple upstream neighbor node carries out following data processing when successfully receiving a packet at every turn: (1) estimation present node and upstream neighbor node queue total length M thereof _max; (2) the average transmission time delay T of estimated data's bag _trans;

Step 3 node x after successfully receiving a packet, according to the average transmission time delay T of packet _transwith queue total length M _maxperform congestion detection algorithm; If it is congested that testing result shows that network occurs, then go to step four; Otherwise, go to step ten;

Step 4 node x carries out classification according to current queue size and upstream neighbor node queue size cases thereof to network congestion degree, is namely divided into different degrees of Congestion;

Step 5 node x, according to degree of Congestion, determines the relieve congestion cycle;

Step 6 node x is according to degree of Congestion, and computing node sends the priority relationship between data and reception data; If send preferential, then go to step eight; Otherwise, go to step seven;

Step 7 node x according to the upstream neighbor node queue length known or estimate and relieve congestion cycle, to each upstream neighbor node fair allocat Channel holding time;

Step 8 performs relieve congestion algorithm;

Step 9 relieve congestion end cycle then goes to step three;

Step 10 terminates.

Compared with the jamming control method of existing MAC layer, the invention has the advantages that:

What 1, the present invention proposed passes through the queue total length predicting node and upstream neighbor node thereof, and the congestion detection method of average transmission time delay in conjunction with packet, the void of network congestion can not only be prevented to examine and undetected problem, comprehensively can also reflect network congestion degree, avoid and existingly whether overflow with the buffer storage length of single node the undetected and empty inspection phenomenon that the congestion detection model for standard exists.

2, the present invention performs relieve congestion algorithm and allocated channel holding time according to degree of Congestion, improves the fairness that channel uses, and effectively reduces collision and packet loss, and increases throughput.

Accompanying drawing explanation

Fig. 1 is the flow chart realizing congestion control of the present invention.

Specific implementation method

The present invention devises the jamming control method of channel fair allocat, composition graphs 1, and the specific implementation method of congestion control is as follows:

Step one, preliminary treatment: calculate the time T successfully sent needed for a packet _s ¹and at perfect condition virtual period T _cyclethe interior maximum amount of data D that can transmit _max, concrete steps are as follows:

1) defining queue maximum is Q _lim, transmission rate is DR, bandwidth is BW, transmission delay is T _delay, request bag RTS length is L _rts, control bag CTS and ACK length be respectively L _ctsand L _ack, data packet length is L _data, keep out of the way that window is CW, time slot is T _stot, the SIFS duration is T _sIFS, the DIFS duration is T _dIFS, then successfully send the time T of a packets need _s ¹for:

${T_s}^1=\mathrm{CW}\×T_{\mathrm{slot}}+\frac{L_{\mathrm{rts}}+L_{\mathrm{cts}}+L_{\mathrm{data}}+L_{\mathrm{ack}}}{\mathrm{DR}}+3\×T_{\mathrm{SIFS}}+4\×T_{\mathrm{delay}}---\left(1\right)$

2) the data required time transmitting full load queue size is:

Q _lim×T _s ¹ (2)

3) arranging virtual period is T _cycle, then the virtual period time is not more than the data required time of transmission full load queue size:

0＜T _cycle≤Q _lim×T _s ¹ (3)

4) the maximum amount of data D that can transmit in virtual period in the ideal situation _maxfor:

$D_{\max }=\frac{T_{\mathrm{cycle}}}{{T_s}^1}---\left(4\right)$

The queue total length M of step 2, estimation node x and upstream neighbor node thereof _maxand the average transmission time delay T of calculated data bag _trans, concrete steps are as follows:

1) suppose that the set of the upstream neighbor node of node x is set A, l is arbitrary node in set A, namely at T _ithe data volume that moment node x successfully receives node l is node l remaining data amount is at T _jmoment detects that the data volume that now node x receives node l is node l remaining data amount is according to the time of upstream neighbor node l data obtained and the size of respective queue, the approximate queue rate of rise R of node l can be estimated _lfor:

$R_l=\frac{N_l^j-N_l^i+M_l^j-M_l^i}{T_j-T_i}---\left(5\right)$

2) if the last data receiving certain upstream neighbor node l of node x are at T _jmoment, then T _k(T _k> T _j) queue length of moment node l for:

If the time that node x obtains node l data exceedes the maximum time T of node l theory competition to channel _max, that is: T _k-T _j> T _max, then can estimate that the queue real time length of node l is by prediction mode:

$M_l^k=(T_k-T_j)\×R_l+M_l^j---\left(6\right)$

Otherwise the real-time queue size of node l is the queue size got the last time:

$M_l^k=M_l^j---\left(7\right)$

Wherein, theoretical competition is to the maximum time T of channel _max=(upstream neighbor nodes+present node number) × (T _s ¹+ T _dIFS);

3) according to the mode of above-mentioned estimation upstream neighbor node queue size, prediction T _ithe queue total length M of moment node x and upstream neighbor node thereof _maxfor:

$M_{\max }=M_x^i+\underset{l\∈A}{\mathrm{\Σ}}{M}_l^i---\left(8\right)$

4) suppose that node x is at T _ithe data volume that moment successfully sends is and there is T ₀=0, the average transmission time delay T of computing node x packet _transfor:

If do not occur congested in network, then:

$T_{\mathrm{trans}}=\frac{T_i-T_0}{(S_x^i-S_x^0+\underset{l\∈A}{\mathrm{\Σ}}(N_l^i-N_l^0))}---\left(9\right)$

Otherwise, suppose that the time of last relieve congestion is T _j(T _i> T _j), then

$T_{\mathrm{trans}}=\frac{T_i-T_j}{(S_x^i-S_x^j+\underset{l\∈A}{\mathrm{\Σ}}(N_l^i-N_l^j))}---\left(10\right)$

Step 3, node x after successfully receiving data, according to the average transmission time delay T of its packet _transwith queue total length M _maxperform congestion detection algorithm, concrete steps are as follows:

1) node x is after successfully receiving a packet, when (γ is a ratio parameter, 0 < γ < 1) and time, represent that average transmission packet institute's time spent is less than time needed for theory, now network may occur congested; Otherwise network does not occur that congested or congested intangibility removes;

2) M is worked as _max> D _maxtime × η (η is an error parameter, 0 < η < 1), represent that the data volume of node x and upstream neighbor node thereof can not be transmitted in a virtual period, now network occurs congested; Otherwise, do not need to carry out relieve congestion.

Step 4, when network occur congested time, node x carries out classification according to current queue size and upstream neighbor node queue size cases thereof to network congestion degree, and be namely divided into different degrees of Congestion, tool step body is as follows:

1) when foot is covered with in node x queue and w is when being greater than 0.7, setting degree of Congestion is 3;

2) when node x current queue size is not less than the mean value of maximum message total amount, namely setting degree of Congestion is 2;

3) when node x current queue size is less than the mean value of maximum message total amount, namely (n is upstream neighbor nodes), now setting degree of Congestion is 1.

Step 5, node x are according to degree of Congestion, and determine the relieve congestion cycle, concrete steps are as follows:

1) when degree of Congestion is 3, relieve congestion cycle time 0.5 times, i.e. 0.5 × T _cycle;

2) when degree of Congestion is 1, relieve congestion cycle time 0.4 times, i.e. 0.6 × T _cycle;

3) other situations, the relieve congestion cycle is constant.

Step 6, node x are according to degree of Congestion, and computing node sends the priority relationship between data and reception data, and concrete steps are as follows:

1) when degree of Congestion is 1, intermediate node x is congested comparatively light, and the priority now receiving data is greater than the priority sending data, namely in the channel allocation period, node x preferential receipt data, if node x do not have data receiver time just send data;

2) when degree of Congestion is 2, intermediate node x load is comparatively large, and now receive identical with the priority sending data, namely in the channel allocation period, node x transmits and receive data according to normal channel competition;

3) when degree of Congestion is 3, intermediate node x load is serious, and the priority now sending data is greater than the priority receiving data, and namely in the channel allocation period, node x preferentially sends data, if just receive data when node x does not have data to send.

Step 7, node x are according to the upstream neighbor node queue length known or estimate and relieve congestion cycle, and to each node fair allocat Channel holding time, concrete steps are as follows:

1) when degree of Congestion is 3, the Channel holding time that node x distributes to upstream neighbor node l is 0;

2) when degree of Congestion is 1, the Channel holding time that node x distributes to upstream neighbor node l is:

$T=\frac{M_l^i}{\underset{l\&Element;A}{\mathrm{\&Sigma;}}{M}_l^i}\&times;0.6\&times;T_{\mathrm{cycle}}---\left(11\right)$

3) other situations, the Channel holding time that node x distributes to upstream neighbor node l is:

$T=\frac{M_l^i}{\underset{l\&Element;A}{\mathrm{\&Sigma;}}{M}_l^i}\&times;T_{\mathrm{cycle}}---\left(12\right)$

Step 8, execution relieve congestion algorithm, concrete steps are as follows:

1) when degree of Congestion is 3, intermediate node x load is serious, forbids receiving data, and upstream node waits for that intermediate node sends decision-making competitive channel mechanism again after congestion control information next time;

2) when degree of Congestion is 2, intermediate node x load is comparatively large, no longer preferential receipt data, receives identical with transmission data priority;

3) when degree of Congestion is 1, intermediate node x is congested comparatively light, and preferential receipt data, when data receiver completes, send data.

Step 9, relieve congestion end cycle then go to step three.

Claims (9)

1. based on a jamming control method for channel fair allocat, it is characterized in that, in wireless sensor network, predict node real-time queue length, the average transmission time delay in conjunction with packet detects congested; Degree of Congestion is divided to congestion condition, determine the relieve congestion cycle, and according to the upstream neighbor node queue size fair allocat channel control cycle known or estimate, carry out relieve congestion or alleviation, described method at least comprises the following steps:

Step one preliminary treatment: calculate the time T successfully sent needed for a packet _s ¹; Calculate in the ideal situation, a node is at a virtual period T _cyelethe interior maximum amount of data D that can transmit _max;

The forward node (being defined as node x) that step 2 has multiple upstream neighbor node carries out following data processing when successfully receiving a packet at every turn: (1) estimation present node and upstream neighbor node queue total length M thereof _max; (2) the average transmission time delay T of estimated data's bag _trans;

Step 3 node x after successfully receiving a packet, according to the average transmission time delay T of packet _transwith queue total length M _maxperform congestion detection algorithm; If it is congested that testing result shows that network occurs, then go to step four; Otherwise, go to step ten;

Step 4 node x carries out classification according to current queue size and upstream neighbor node queue size cases thereof to network congestion degree, is namely divided into different degrees of Congestion;

Step 5 node x, according to degree of Congestion, determines the relieve congestion cycle;

Step 6 node x is according to degree of Congestion, and computing node sends the priority relationship between data and reception data; If send preferential, then go to step eight; Otherwise, go to step seven;

Step 7 node x according to the upstream neighbor node queue length known or estimate and relieve congestion cycle, to each upstream neighbor node fair allocat Channel holding time;

Step 8 performs relieve congestion algorithm;

Step 9 relieve congestion end cycle then goes to step three;

Step 10 terminates.

2. the method for claim 1, is characterized in that described preliminary treatment, at least also comprises:

1) defining queue maximum is Q _lim, transmission rate is DR, bandwidth is BW, transmission delay is T _delay, request bag RTS length is L _rts, control bag CTS and ACK length be respectively L _ctsand L _ack, data packet length is L _data, keep out of the way that window is CW, time slot is T _slot, the SIFS duration is T _sIFS, the DIFS duration is T _dIFS, then successfully send the time T of a packets need _s ¹for:

${T_s}^1=\mathrm{CW}{\&times;T}_{\mathrm{slot}}+\frac{L_{\mathrm{rts}}+L_{\mathrm{cts}}+L_{\mathrm{data}}+L_{\mathrm{ack}}}{\mathrm{DR}}+3\&times;T_{\mathrm{SIFS}}+4\&times;T_{\mathrm{delay}}---\left(1\right)$

2) the data required time transmitting full load queue size is:

Q _lim×T _s ¹ (2)

3) arranging virtual period is T _cycle, then the virtual period time is not more than the data required time of transmission full load queue size:

0＜T _cycle≤Q _lim×T _s ¹ (3)

4) the maximum amount of data D that can transmit in virtual period in the ideal situation _maxfor:

$D_{\max }=\frac{T_{\mathrm{cycle}}}{{T_s}^1}---\left(4\right)$

3. the method for claim 1, is characterized in that the queue total length M of described estimation node x and upstream neighbor node thereof _maxand the average transmission time delay T of packet _trans, at least also comprise:

1) suppose that the set of the upstream neighbor node of node x is set A, l is arbitrary node in set A, namely at T _ithe data volume that moment node x successfully receives node l is node l remaining data amount is at T _jmoment detects that the data volume that now node x receives node l is node l remaining data amount is according to the time of upstream neighbor node l data obtained and the size of respective queue, the approximate queue rate of rise R of node l can be estimated _lfor:

$R_l=\frac{N_l^j-N_l^i+M_l^j-M_l^i}{T_j-T_i}---\left(5\right)$

2) if the last data receiving certain upstream neighbor node l of node x are at T _jmoment, then T _k(T _k> T _j) queue length of moment node l for:

If the time that node x obtains node l data exceedes the maximum time T of node l theory competition to channel _max, that is: T _k-T _j> T _max, then can estimate that the queue real time length of node l is by prediction mode:

$M_l^k=(T_k-T_j)\&times;R_l+M_l^j---\left(6\right)$

Otherwise the real-time queue size of node l is the queue size got the last time:

$M_i^k=M_l^j---\left(7\right)$

Wherein, theoretical competition is to the maximum time T of channel _max=(upstream neighbor nodes+present node number) × (T _s ¹+ T _dIFS);

3) according to the mode of above-mentioned estimation upstream neighbor node queue size, prediction T _ithe queue total length M of moment node x and upstream neighbor node thereof _maxfor:

$M_{\max }=M_x^i+\underset{l\&Element;A}{\mathrm{\&Sigma;}}{M}_l^i---\left(8\right)$

4) suppose that node x is at T _ithe data volume that moment successfully sends is and there is T ₀=0, the average transmission time delay T of computing node x packet _transfor:

If do not occur congested in network, then:

$T_{\mathrm{trans}}=\frac{T_i-T_0}{(S_x^i-S_x^0)+\underset{l\&Element;A}{\mathrm{\&Sigma;}}(N_l^i-N_i^0))}---\left(9\right)$

Otherwise, suppose that the time of last relieve congestion is T _j(T _i> T _j), then:

$T_{\mathrm{trans}}=\frac{T_i-T_j}{(S_x^i-S_x^j)+\underset{l\&Element;A}{\mathrm{\&Sigma;}}(N_l^i-N_l^j)}---\left(10\right)$

4. the method for claim 1, is characterized in that the described average transmission time delay T according to node x packet _transwith queue total length M _maxperform congestion detection algorithm, at least also comprise:

1) node x is after successfully receiving a packet, works as T _trans< T _s ¹× γ (γ is a ratio parameter, 0 < γ < 1) and time, represent that average transmission packet institute's time spent is less than time needed for theory, now network may occur congested; Otherwise network does not occur that congested or congested intangibility removes;

2) M is worked as _max> D _maxtime × η (η is an error parameter, 0 < η < 1), represent that the data volume of node x and upstream neighbor node thereof can not be transmitted in a virtual period, now network occurs congested; Otherwise, do not need to carry out relieve congestion.

5. the method for claim 1, is characterized in that described degree of Congestion divides, at least also comprises:

1) when foot is covered with in node x queue and w is when being greater than 0.7, setting degree of Congestion is 3;

2) when node x current queue size is not less than the mean value of maximum message total amount, namely setting degree of Congestion is 2;

3) when node x current queue size is less than the mean value of maximum message total amount, namely (n is upstream neighbor nodes), now setting degree of Congestion is 1.

6. the method for claim 1, is characterized in that described according to the degree of Congestion determination relieve congestion cycle, at least also comprises:

1) when degree of Congestion is 3, relieve congestion cycle time 0.5 times, i.e. 0.5 × T _cycle;

2) when degree of Congestion is 1, relieve congestion cycle time 0.4 times, i.e. 0.6 × T _cycle;

3) other situations, the relieve congestion cycle is constant.

7. the method for claim 1, is characterized in that the described priority relationship sending data and reception data according to degree of Congestion determination node x, at least also comprises:

1) when degree of Congestion is 1, intermediate node x is congested comparatively light, and the priority now receiving data is greater than the priority sending data, and namely in the channel allocation period, node x preferential receipt data, if node x does not have just to send data during data receiver;

2) when degree of Congestion is 2, intermediate node x load is comparatively large, and now receive identical with the priority sending data, namely in the channel allocation period, node x transmits and receive data according to normal channel competition;

3) when degree of Congestion is 3, intermediate node x load is serious, and the priority now sending data is greater than the priority receiving data, and namely in the channel allocation period, node x preferentially sends data, if just receive data when node x does not have data to send.

8. the method for claim 1, is characterized in that the upstream neighbor node queue length according to knowing or estimate and relieve congestion cycle, to each node fair allocat Channel holding time, at least also comprising:

1) when degree of Congestion is 3, the Channel holding time that node x distributes to upstream neighbor node l is 0;

2) when degree of Congestion is 1, the Channel holding time that node x distributes to upstream neighbor node l is:

$T=\frac{M_l^i}{\underset{l\&Element;A}{\mathrm{\&Sigma;}}{M}_l^i}\&times;0.6\&times;T_{\mathrm{cycle}}---\left(11\right)$

3) other situations, the Channel holding time that node x distributes to upstream neighbor node l is:

$T=\frac{M_l^i}{\underset{l\&Element;A}{\mathrm{\&Sigma;}}{M}_l^i}\&times;T_{\mathrm{cycle}}---\left(12\right)$

9. the method for claim 1, is characterized in that describedly performing different relieve congestion algorithms according to different degrees of Congestion, at least also comprises:

1) when degree of Congestion is 3, intermediate node x load is serious, forbids receiving data, and upstream node waits for that intermediate node sends decision-making competitive channel mechanism again after congestion control information next time;

2) when degree of Congestion is 2, intermediate node x load is comparatively large, no longer preferential receipt data, receives identical with transmission data priority;

3) when degree of Congestion is 1, intermediate node x is congested comparatively light, and preferential receipt data, when data receiver completes, send data.