CN100477639C - A wireless network node buffer data packet processing method - Google Patents

Abstract

The invention discloses a wireless network node buffer data packet processing method, comprising: 1) sorting the input packets in descending order according to the data length, and initializing the value of the water level line t; 2) judging whether the input packet set is empty, if it is empty Then end the operation, otherwise execute the next step; 3) judge whether there is an input packet in the current input packet set, so that the sum of the data lengths of the input packet is just equal to the value of the water level t; if it exists, then execute step 4), otherwise execute Step 5); 4) Splicing the input packet into an output packet with a length of L, after the splicing is successful, storing the spliced output packet, deleting the spliced input packet, and re-executing step 2); 5) Decrease the value of the water level t by one unit, then go to step 2) and execute again. The method of the invention has small code space, fast solution speed and high splicing efficiency, and is very suitable for implementation in wireless network nodes.

Description

A wireless network node buffer data packet processing method

technical field

The invention relates to a wireless network, in particular to a wireless network node buffer data packet processing method.

Background technique

Efficient use of energy and improvement of transmission performance are important issues facing wireless networks. The data packet is the basic operation object of the network, and the data volume, length, and quantity of the data packet directly affect the power consumption of the transmission. By processing data packets and optimizing transmission performance, energy consumption can be significantly reduced. In the usual research on data packet distribution and processing, the research object is a fully connected network topology, and it is assumed that there is at least one complete path from the source to the destination node. As shown in Figure 1, in a clustered wireless network, each cluster has a cluster leader, and the nodes in the cluster send the generated data to the cluster leader. The cluster leader usually does not forward the packet immediately, but puts the packet into the buffer. The buffer collects data packets of different lengths, and when the data accumulates to a certain extent, it is forwarded by the cluster leader. However, in actual use, a wireless network is usually a network with intermittent connection characteristics. The so-called intermittently connected network refers to a network in which nodes cannot always be online due to factors such as energy saving, failure, and interference, but work in an intermittent manner, so the complete path between the source and destination nodes may not exist. As shown in Figure 2, for a network with intermittent connections, since the premise that there must be a complete path between the source and destination nodes in the network is broken, there are several temporarily disconnected clusters in the network, and the transmission path may be disconnected at any time. Therefore, for network routing that can be applied to intermittent connections, it must be able to cache data and tolerate data packets staying at the source node or intermediate node.

In an intermittently connected network, when data is to be sent, the network is just in a state of being isolated into several parts, and the end-to-end application between nodes is destroyed, and data can only be transmitted in an asynchronous manner, allowing greater buffering time and end-of-packet delay. As a result, nodes in intermittently connected networks often buffer a large number of packets. Research has proved that if the data packets are processed before sending, the transmission overhead can be effectively reduced.

There are many ways to process data packets. In terms of data volume, reference 1Shu Lei, S.Y.Lee, Yang Jie "ETRI: A Dynamic Packet Scheduling Algorithm for Wireless SensorNetworks" proposed a two-layer buffer model according to the value and importance of data Differentiate and filter data packets to optimize transmission efficiency; reference 2 "Bhaskar Krishnamachari, Deborah Estrin, Stephen Wicker, "Modeling Data-Centric Routing in Wireless Sensor Networks", proceedings of 6th international workshop on Modeling analysis and simulation of wireless and mobile systems , 2003" and reference 3 "Huseyin Ozgur Tan and Ibrahim Korpeoglu, "PowerEfficient Data Gathering and Aggregation in Wireless Sensor Networks", SIGMOD/PODS volume 32, number 4, December 2003" proposed a data-centric data fusion method, The redundancy of data content is reduced, which significantly reduces the transmission volume.

In the case of a certain amount of data, the choice of data packet length will affect network performance and power consumption. Longer data packets will result in a higher packet loss rate, while small data packets have higher header overhead, so a compromise must be made. In reference 4 "Y.Sankarasubramaniam, I.F.Akyildiz and S.W.McLaughlin, "Energy Efficiency based Packet Size Optimization in Wireless Sensor Networks" in Proc. of the First IEEEInternational Workshop on Sensor Network Protocols and Applications3" and references, 20, Anchorage, A 5 "Eytan Modiano, "An adaptive algorithm for optimizing the packet size used in wireless ARQ protocols" ACM-Baltzer Journal of WirelessNetworks 5, pp279-286, 1999" gives a calculation method for optimizing the packet size, and proposes that there are The most energy efficient packet length.

Through data fusion and scheduling processing, the amount of data is effectively controlled. On this basis, the choice of packet length becomes an important issue, and it also determines the number of data packets on the network. In the actual network, there are often a large number of fragmented data packets smaller than the optimal length, which affects the efficiency of transmission. These fragmented data packets have to pay routing overhead at each hop, and the startup transient current of the sending circuit during forwarding will also cause additional power consumption. Therefore, reducing the number of fragmented data packets is an effective way to save energy.

Contents of the invention

The object of the present invention is to provide a method for splicing fragmented data packets into a data packet of a specified length in order to reduce the impact of fragmented data packets on network transmission efficiency.

In order to achieve the above object, the present invention provides a wireless network node buffer data packet processing method for splicing input data packets of different lengths into an output data packet of a unique length, including:

1), each input packet in the input packet set is arranged in descending order according to the data length, and the value of the water level t is initialized, and the value of the water level is determined as the length value L of the output packet;

2), judge whether the input packet set is empty, if it is empty, end the operation, otherwise execute the next step;

3), judge whether there is one or more input packets in the current input packet set, so that the sum of the data lengths of these input packets is equal to the value of the water level line t; if there is such an input packet, then perform step 4), otherwise Execute step 5);

4), the input packet is spliced into an output packet with a length of L, if the value of t is less than L, then the insufficient part in the output packet is filled with 0, after the splicing is successful, the output packet obtained by storing the splicing is stored, and the already The spliced input data packet is deleted from the input data packet collection, and step 2 is re-executed);

5) Decrease the value of the water level line t by one unit, then turn to step 2) and execute again.

The present invention also provides a wireless network node buffer data packet processing method for splicing input data packets of different lengths into multiple output data packets of different lengths, including:

a), arrange each input packet in the input packet set in descending order according to length;

b) According to the number of types of output packets with different lengths in the output packet set, set the number of water levels, one water level corresponds to an output packet of one length, and then initialize the value of each water level and the value of the global lower limit. When , the value of the water level is the same as the length of the corresponding output package, and the value of the global lower limit is 100%; finally select a water level and its corresponding output package;

c), judge whether the input package set is empty, if empty, end the operation, otherwise, execute the next step;

d), judging whether there are one or more input packets in the current input packet collection, so that the sum of the data lengths of these input packets is equal to the value of the current water level; if there is such an input packet, then perform step e), otherwise Execute step f);

e) Splice the input data packet into an output packet corresponding to the current water level. If the value of the water level is less than the length of the output packet, fill the insufficient part of the output packet with 0. After the splicing is successful, store the spliced output package, and delete the spliced input data packet from the input data packet collection, and then re-execute step c);

f), judging whether there are output packets of other lengths in the output packet set, so that the ratio of the value of the water level corresponding to the output packet to the length of the output packet is higher than the value of the global lower limit, and if it exists, use the corresponding output packet The water level line replaces the current water level line, and go to step c), if it does not exist, perform step g);

g), reduce the value of the current water level, and compare the value of the lowered water level with the output packet length corresponding to the water level, if their ratio is higher than or equal to the value of the global lower limit, then jump to step c ), otherwise, update the global lower limit to the ratio, and then jump to step c) to continue execution.

In the case that an extra packet header is added to both the input saturation and output packets, the wireless network node buffering data packet processing method is applicable to the case that the output packet and the input packet have an extra packet header.

The advantages of the present invention are:

1. The wireless network node buffer data packet processing method of the present invention has small code space and fast solution speed, and is very suitable for implementation in wireless network nodes.

2. The multi-watermark heuristic search method adopted in the wireless network node buffer data packet processing method of the present invention has made global optimization on the transmission process, and the splicing utilization rate is relatively high.

3. The wireless network node buffering data packet processing method of the present invention performs splicing processing on data packets, which will sacrifice computing resources, response time, etc., but in exchange for the reduction of transmission overhead and routing overhead, it is more suitable for some networks that emphasize saving transmission power. An environment that consumes less energy and has lower real-time requirements.

Description of drawings

Fig. 1 is a schematic diagram of the topology of a fully connected wireless network;

FIG. 2 is a schematic diagram of a topology structure of a wireless network with intermittent connections;

Fig. 3 is a schematic diagram of multi-water line control splicing;

Fig. 4 is the flow chart when the wireless network node buffering data packet processing method of the present invention splices input data packets of different lengths into an output data packet of a unique length;

FIG. 5 is a flow chart of splicing input data packets of different lengths into output data packets of different lengths by the wireless network node buffer data packet processing method of the present invention.

Detailed ways

The method of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

The wireless network node buffer data packet processing method of the present invention is to splice a large number of data packets with different lengths into a data packet with an optimized target length. When splicing data packets, several data packets collected by a certain node are called input packets, and the packets generated after splicing are called output packets. Among them, the length of the input packet depends on the length of the collected data of the node, the number of packets of each length is not limited, the length of the output packet can be selected from a limited number of fixed lengths, and the number of packets of each length is not limited. If the spliced input packet is smaller than the output packet length, you need to add 0 to simplify the sending and receiving process. The splicing of input packets into output packets will have additional packet header overhead, which is used for packet decapsulation after reaching the destination node.

Below in conjunction with embodiment, the method of the present invention is described in detail.

Embodiment 1: A splicing method with a fixed output packet length.

When the length of the output packet is fixed, to realize the splicing of data packets is to solve the optimal splicing scheme of splicing all n input packets into an output packet of length L. In the method of the present invention, the utilization rate is used as the criterion for judging the effect of data splicing. The utilization rate is the ratio between the total length of the input packet data and the total length of the output packet data, and the higher the utilization rate, the better the effect of the splicing scheme is obviously.

For this kind of data packet splicing problem, a general greedy algorithm can be used to solve it, that is, each time the current largest input packet is spliced into the output packet, and when the output packet can no longer be accommodated, another new output packet is spliced into it until All input packets are processed. Although this method is simple, it may miss the optimal solution due to the selection of a local optimum, and the total length of the data will increase too much, resulting in a waste of bandwidth.

In order to improve splicing efficiency, in this embodiment, a single water level method may be used. The waterlines indicated represent the scan positions when a stitching attempt is made. Use t to represent the water level line, define the water level line as an integer from the output packet length L to 0, and the water level line t starts to scan in descending order from L. The specific implementation steps are as follows:

Step 11, arrange each input packet in the input packet set in descending order according to the data length, and initialize the value of the water level t, and set the value of the water level as the length value L of the output packet;

Step 12, judge whether the input packet collection is empty, if it is empty, then end the operation, otherwise execute the next step;

Step 13, judging whether there are one or more input packets in the current input packet collection, so that the sum of the data lengths of these input packets is just equal to the value of the water level t; if there is such an input packet, then perform step 14, otherwise Execute step 15;

Step 14, splicing these input data packets into an output data packet with a length of L, if the value of t is less than L, filling the insufficient part in the output data packet with 0, after the splicing is successful, storing the spliced output packet, and Delete the spliced input data packet from the input data packet collection, and re-execute step 12;

Step 15. Decrease the value of the water level t by one unit, and then go to step 12 to execute again.

The algorithm of the above method is implemented as follows:

Use PDisc to represent the input package, the structure of the package is as follows:

struct PDisc{

int len : the length of the packet

int count: the number of packets of this length

PDisc*next: Next package description

}

Let P be a linked list with PDisc as an element, representing input packets of various lengths, and arrange the packets in descending order. R is a linked list with PDisc as an element, representing a splicing scheme. Start solving from the first element p of the P list. t is the water level, cn is the number of times the currently attempted input packet is used, and the initial value is 0. The function to solve a perfect splicing scheme is Perfect(p, t), and the method is as follows:

1. First try to splicing a p-type package into t;

2. If p->len is exactly equal to t, find a perfect splicing, generate an R list element, r->len=t, r->count=1, return;

3. If t is less than p->len, p packets cannot be spliced into t, then solve Perfect(p->next, t) and return;

4. If p->c-cn=1, the p-type package is just used up, then cn is set to 0, a p-type package is spliced into t, and Perfect(p->next, L-t) is solved;

5. If p->count-cn>1, add 1 to cn to solve Perfect(p, L-t);

6. If the plan in 5 contains p-type packages, judge whether the number of times it uses exceeds the available number. If it exceeds, add 1 to the use times of p-type packages in the plan and return; if it exceeds, discard this plan and return ;

7. If the P list is not empty or the scheme R is not found, p=p->next, go to 1;

8. If a plan is found, it means that a p-type packet and the packets in the packet splicing scheme with a length of t-p->len can just be spliced into t, and the p-type packet is added to the head of the scheme list R.

The above method will be described below in conjunction with examples.

Suppose there is a set of input packets arranged in descending order {8, 7, 6, 5, 4, 3, 2, 1, 1}, the numbers in this set represent the data length of each input packet, and the fixed length of the output packet for 10. When using the single waterline method to process data packets, according to step 10, first set the length of the waterline to 10, and then check whether there is one or more data packets whose length is equal to 10 in the input packet collection. Obviously, the input packet The sum of two packets with lengths 3 and 7 is 10, and the two input packets are concatenated to obtain a new output packet, which can be counted as a, and then these two input packets are deleted from the set to obtain the input packet set is {1, 1, 5, 6, 8, 4, 2}. Then check the remaining input packets in the input packet set to see if there are any input packets with a length of 10. Similarly, two input packets of length 8 and 2, and two input packets of length 6 and 4 can be obtained. The two output packets obtained by splicing these data packets are recorded as b and c respectively, and the final remaining input packet set is {1, 1, 5}. The remaining input packets cannot be spliced using the water level line with a length of 10, and the length of the water level line needs to be decreased. Decrease the length of the water level by one unit to 9, and the remaining input packets cannot be spliced. Therefore, continue to decrease until the length of the water level becomes 7. At this time, the length of the input packet set can be 1, 1 The three input packets of 5 and 5 are spliced to obtain d of the output packet, and in the output packet, 0 is added to the insufficient part. Finally, 9 input packets in the input packet set {1, 1, 3, 5, 6, 8, 4, 7, 2} get 4 output packets, which has a total utilization of 92.5%.

Embodiment 2: Splicing method with multiple output packet lengths.

In the method of splicing input packets into output packets, the output packet length may not be unique, but there are multiple optional output packet lengths. When there are multiple output packet lengths to choose from, it is necessary to comprehensively consider possible combinations of splicing input packets into output packets of different lengths, so there are multiple search branches. There are two ways to search: depth-first traversal and breadth-first traversal. Depth-first, after the output packet of one length is completely scanned, it goes to the next one. Breadth-first is to scan in parallel in turn on output packets of various lengths according to certain switching rules. The present invention adopts a multi-waterline heuristic search algorithm. As shown in FIG. 3 , the multi-waterline heuristic search algorithm is a breadth-first approach. The method sets a global lower limit, which is represented by splicing utilization rate of output packets, and the splicing utilization rate is the ratio of the total length of input packets to the length of spliced output packets. The watermark is initially set to the length of the first output packet, and a watermark is recorded for each output packet. Judging whether the current water level can accommodate some input packets, if it exists, it will output; if it does not exist, it will continue to search. The water level is scanned on the corresponding output package. After each search, the water level is reduced by 1, and the utilization rate is recalculated. When the utilization rate is lower than the global lower limit, it is scanned on the next output package, and the current water level is changed to a new output. The water level of the package. In this way, by using the global offline as the basis for water level change control, the water level of each output package changes in the same proportion, which has global unity and can find the global optimal solution. The specific implementation steps are as follows:

Step 21, arrange each input packet in the input packet set in descending order according to length;

Step 22. Set the number of water levels according to the types of output packets of different lengths in the output packet set. One water level corresponds to an output packet of one length, and then initialize the values of each water level and the value of the global lower limit. , set the value of the water level as the length of the corresponding output package, and set the value of the global lower limit as 100%; and select a water level and its corresponding output package;

Step 23, judging whether the input packet set is empty, if empty, end the operation, otherwise, execute the next step;

Step 24, judge whether there is one or more input packets in the current input packet collection, so that the sum of the data lengths of these input packets is just equal to the value of the water level; if there is such an input packet, then perform step 25, otherwise Execute step 26;

Step 25. Splice these input data packets into an output packet corresponding to the current water level. If the value of the water level is less than the length of the output packet, fill the insufficient part of the output packet with 0. After the splicing is successful, store the spliced result. Output the packet, and delete the spliced input data packet from the input data packet collection, and re-execute step 23;

Step 26. Determine whether there is an output packet of other length in the output packet set, so that the ratio of the value of the water level corresponding to the output packet to the length of the output packet is higher than the global lower limit value, and if it exists, use the output packet to correspond replace the current water level, and go to step 23, if it does not exist, go to step 27;

Step 27. Reduce the value of the current water level, and compare the value of the lowered water level with the output packet length corresponding to the water level. If their ratio is higher than or equal to the value of the global lower limit, then jump to step 23 , otherwise, update the global lower limit to the ratio, and then jump to step 23 to continue execution.

The algorithm of the above splicing method with multiple output packet lengths is described as follows:

Set the input package list P and the output package list Q. Both lists are linked lists with PDisc as elements, and the elements in the linked list are arranged in descending order of package length.

Initialization: The current scanning object is the first output packet element q of the list Q, and the current water level t is the first output packet length q->len. The global lower limit is the current utilization rate: current water level/current output packet length=100%.

1. When the P list is empty, exit, otherwise, scan the current output package of q, and t is the water level value saved corresponding to the output package, and solve Perfect(p, t);

2. If the scheme is not empty, splice the corresponding input packets into an output packet whose length is q->len, and remove the corresponding input packets from P;

3. If the scheme is empty, and t is less than the sum of the maximum and minimum packet lengths in the input packet list, and t is the largest among all watermarks, select the shortest output packet for the maximum length packet and complete the encapsulation ;

4. The current water level t is decremented by one unit, and the water level of the object is saved.

5. If t/q->len is higher than or equal to the global lower limit, go to 1;

If it is lower than the global lower limit, recalculate the current utilization rate, update the global lower limit, replace the scanning object, make q=q->next, go to 1;

The above description is illustrated with reference to examples.

Suppose there is a set of input packets in descending order {18, 16, 16, 9, 9, 7, 6, 5, 5, 4, 3, 3, 2, 1}, and there are two lengths in the set of output packets The output packets are 10 and 20, respectively. Since the output packets have two different lengths, there are two water level lines, respectively marked as t1 and t2, where t1 corresponds to an output packet with a length of 10, and t2 corresponds to an output packet with a length of 20. Initially, the global lower limit is 100%.

First, select the water level line t2 and its corresponding output packet with a length of 20. Through searching, it can be seen that there are input packets with a length of 18 and 2, 16 and 4, 16, 3 and 1, 9, 6 and 5 in the input packet set It can be spliced into output packets to obtain output packets A, B, C, and D, and the corresponding input packets are deleted from the input packet set, and the remaining input packet set is {7, 3, 9, 5}. Since the current water level cannot find a new splicing scheme, the current water level is changed, and the water level t1 and its corresponding output packet with a length of 10 are selected. The input packets whose lengths are 7 and 3 in the input packet set can be spliced into an output packet to obtain an output packet E, and the remaining input packet set is {9, 5}. Reduce the value of the water level line t1 to 9, 9/10=90%, which is less than the current global lower limit, and lower the global lower limit to 90%. Then lower the value of the water level line t1 to 9. At this time, the input packet with a length of 9 can be sent to the output packet, and the remaining input packets are {5}. Lower the global lower limit again to 50%, and lower the watermark t1 value to 5, sending it into the output packet. When all the input packets in the input packet set are spliced, the entire buffer data processing process is also completed, and the total utilization rate is 94.55%.

In the above example, only the case of two output packet lengths is selected, and the method of the present invention is also applicable to the case of more than two output packet lengths. The following describes the case that there are four output packet lengths.

Suppose there is a set of input packets arranged in descending order {19, 16, 15, 13, 12, 10, 9, 8, 6, 5, 4, 2, 2, 1}, and the length of the output packets is assumed to be 20, 15 , 10, 5. Since there are four types of output packet lengths, there are four different water levels marked as t3, t4, t5, and t6. Initially, the global lower limit is 100%. First select t3, the length of the output packet corresponding to it is 20. Through searching, we know that 19 and 1, 16 and 4, 15 and 5, 12 and 8, 10, 6, 2 and 2 are spliced into an input packet with a length of 20, and the remaining input packets are {13, 9}. Then select t4, the length of the output packet corresponding to it is 15, and there is no input packet that can be spliced. Then select t5, the length of the output packet corresponding to it is 10, and there is no input packet that can be spliced. Finally, t6 is selected, its corresponding output packet length is 5, and there is no input packet that can be spliced. At this point, the input packet set is not empty yet, and the length of the water level needs to be reduced. While reducing the water level, it is necessary to check the global lower limit value. When the length of t5 is reduced to 9, the input packet can be spliced into the output packet, and the global lower limit value at this time is 90%. When the length of t4 is 13, the input packet with a length of 13 can be spliced into the output packet, and the global lower limit value at this time is 86.7%.

In the previous description, the additional calculation overhead and extra headers generated in data splicing were not considered. Adding extra headers will make the algorithm more complicated, which can be solved by using a packet length correction method. Adding the length of the input packet and the output packet to the length of an extra packet header can equivalently transform the case of considering the extra header into the situation of not considering the extra header.

prove:

Criterion 1: When considering the extra packet header, the length is c, for a packet sequence of length K1, K2...Kn, if (K1+c)+(K2+c)...+(Kn-1 +c)+Kn=K, then it is said that this packet sequence can be spliced into a K-length packet when considering the extra packet header.

Criterion 2: When the extra packet header is not considered, for a packet sequence of length K1, K2...Kn, if K1+K2...+Kn-1+Kn=K, it is said that the packet sequence is not It can be spliced into K-length packets when considering additional headers.

When considering the extra packet header, for the packet sequence of input packet lengths L1, L2...Ln, to judge whether it can be spliced into an output packet of length L, criterion 1 needs to be applied. Through packet length adjustment: adjust the input packet to L1+c, L2+c...Ln-1+c, Ln+c, and adjust the output length to L+c, which are respectively recorded as: A1, A2...An, and the output length A. When the extra header is not considered, apply criterion 2 to the above adjusted length, if satisfied, then A1+A2...+An=A, namely (L1+c)+(L2+c)...(Ln- 1+c)+Ln+c=L+c, namely (L1+c)+(L2+c)...(Ln-1+c)+Ln=L, indicating that L1, L2...Ln also satisfy Guidelines 1. And this process is reversible. So there are:

Guideline 3:

For a packet sequence with length K1, K2...Kn and output packet length K, if the packet length is adjusted and the extra header is not considered, criterion 2 is satisfied, which is equivalent to criterion 1 when the extra header is considered. vice versa. Therefore, through the adjustment of the packet length, the algorithm operation result without considering the additional packet header is equivalent to the algorithm operation result when the additional packet header is considered. Therefore, according to the above-mentioned adjusted packet length, the method of the present invention when the extra header is not considered is also applicable to the situation of considering the extra header.

Claims (3)

1, a kind of wireless network node buffering data package processing method is used for the input packet of different length is spliced into the dateout bag of same length, comprising:

1), each input bag in the input bag set is done descending according to data length, and the value of initialization waterline t, be integer with the value defined of waterline from the length value L to 0 of output packet, waterline begins to successively decrease from L and scans;

2), judge whether the set of input bag is empty, if sky end operation then otherwise is carried out next step;

3), judge whether current input bag has more than one input bag in gathering, and the data length sum of described input bag equals the value of waterline t; If there is such input bag, then execution in step 4), otherwise execution in step 5);

4), will import the bag dateout bag that to be spliced into a length be L, if the value of t is less than L, then the insufficient section in the dateout bag is filled 0, after splicing successfully, the output packet that the storage splicing obtains, and the input packet that will splice deletion from input packet set, and execution in step 2 again);

5), with the value of the waterline t unit that descends, forward step 2 then to) re-execute.

2, a kind of wireless network node buffering data package processing method is used for the input packet of different length is spliced into the dateout bag of different length, comprising:

A), each input bag in the input bag set is done descending by length, with the value defined of waterline be integer from the length value L to 0 of output packet;

B), according to the species number of different length output packet in the output packet set, the number of designated water level line,, the output packet of the corresponding a kind of length of waterline, the value of each bar waterline of initialization and the value of overall lower limit then, when initialization, the value of waterline is identical with the length of its corresponding output packet, and the value of overall lower limit is 100%; Choose a waterline and corresponding output packet thereof at last;

C), judge whether the set of input bag is empty, if empty, end operation, otherwise, carry out next step;

D), judge whether current input bag has more than one input bag in gathering, and the data length sum of described input bag equals the value of current waterline; If have such input bag, then execution in step e), otherwise execution in step f);

E), will import packet and be spliced into an output packet corresponding with current waterline, if the value of waterline is less than the length of output packet, then the insufficient section in the output packet is filled 0, after splicing successfully, the output packet that the storage splicing obtains, and the input packet that will splice deletion from the set of input packet, re-execute step c) then;

F), judge the output packet that whether has other length in the output packet set, make the ratio of the value of the pairing waterline of this output packet and this output packet length be higher than the value of overall lower limit, if exist, then the waterline with this output packet correspondence replaces current waterline, and forward step c) to, if there is not execution in step g);

G), reduce the value of current waterline, and the value of the waterline after will the reducing output packet length corresponding with this waterline is made comparisons, if their ratio is greater than or equal to the value of overall lower limit, then jump to step c), otherwise, overall lower limit is updated to this ratio, jumps to then and continue in the step c) to carry out.

3, wireless network node buffering data package processing method according to claim 1 and 2, it is characterized in that, all increase under the situation of an extra packet joint at described input bag and output packet, described wireless network node buffering data package processing method is applicable to that output packet and input bag have the situation in extra packet header.

Images