A budget-based density adjustment method for sparse network coding

Abstract

The present invention belongs to the technical field of wireless network communication, and specifically relates to a budget-based sparse network coding density adjustment method; the method comprises: obtaining the number of original packets of a source node and a link packet loss rate; setting a sparsity threshold; initializing transmission parameters, the transmission parameters comprising sparsity, an expected budget, a remaining budget and an available budget threshold; performing regional division according to the number of original packets, and obtaining regional division results, including the number of regional divisions and the regenerated packet capacity of each region; if the remaining budget is greater than 0 and the decoding matrix is ​​not full rank, performing coding density adjustment in the current region and updating the remaining budget, switching to the next region and then performing density adjustment of the source node; otherwise, performing coding packet transmission according to the decoding matrix state; the present invention can more accurately and better evaluate the relevant performance of actual network data packet transmission, and reduce transmission overhead.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless network communications, and in particular relates to a budget-based sparse network coding density adjustment method. Background Art

[0002] With the development of human science and technology, from the initial postman communication to wired communication, and then to the current wireless communication technology, as well as the advent of the fifth-generation mobile communication network, people have higher and higher requirements for communication quality, and need high-reliability, low-latency, and high-security network transmission technology to ensure the reliability of information transmission. However, due to the particularity of wireless network transmission media, the transmission channel is easily affected by noise, signal interference, and multipath effects, which reduces the efficiency of wireless communication transmission, increases the transmission delay, and affects the user experience. How to meet the needs of users for diversified and high-quality communication services, achieve higher data transmission efficiency and higher throughput under limited network resources, and achieve greater network performance gains with less network overhead has become one of the important research areas of wireless communication.

[0003] For the above purpose, Ahlswede et al. first proposed the concept of network coding in 2000. The introduction of network coding breaks the "store-and-forward" transmission mode of traditional network intermediate nodes. It allows nodes in the communication network to encode, decode and re-encode the received data packets according to certain rules, and then forward them to downstream nodes. The destination node decodes the received encoded packets according to the corresponding decoding rules. Compared with traditional transmission mechanisms, network coding can carry more data in one transmission, improve data transmission efficiency and bandwidth utilization, and achieve the theoretical maximum network capacity of multicast networks.

[0004] At present, the most widely used in network coding research are random linear network coding and sparse network coding. Since random linear network coding has the problems of high decoding complexity and large header transmission overhead, the concept of sparse network coding is proposed. Its core is to randomly select a part of the original data packet and form a coded packet through linear combination. Since only a part of the data packet participates in the encoding, the proportion of the number of original data packets in the coded packet is reduced, so the transmission overhead of the header of the transmitted coded packet is reduced, and at the same time, the complexity of decoding the coded packet at the destination node is reduced. Although sparse network coding solves the above problems, since it uses a fixed number of original packets to participate in the encoding to form the coded packet when conducting relevant performance research, this increases the transmission overhead in the later stage of transmission and reduces the decoding probability. It is impossible to achieve a balance between the transmission overhead, decoding probability and decoding complexity. In order to solve the problems of sparse network coding, a density-variable sparse network coding scheme is proposed, that is, the number of original packets constituting the coding is dynamic, and the number of original packets increases linearly with the progress of the network transmission stage, which can effectively reduce the transmission overhead in the later stage of transmission.

[0005] At present, there are two main density adjustment schemes for sparse network coding, one is a budget-based density adjustment scheme, and the other is a Markov chain model-based density adjustment scheme. These schemes idealize the network scenarios and only consider density adjustment under ideal channels, but this is not applicable to relevant performance analysis and research in the case of packet loss in actual network channels, and cannot be used to evaluate actual network indicators. At the same time, the linearly independent probability used in the above scheme is only an upper bound expression, which leads to a large density value in the final calculation, resulting in a large problem in decoding complexity. Therefore, there is an urgent need for a density adjustment method in a packet loss environment, which makes the density value more accurate and can be better used to evaluate relevant performance indicators of the actual network, so that it can provide a better balance between reducing transmission overhead, decoding complexity and increasing decoding probability. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a budget-based sparse network coding density adjustment method, which includes:

[0007] S1: Obtain the number of original packets from the source node and the link packet loss rate; set the sparsity threshold;

[0008] S2: Initialize transmission parameters, including sparsity, expected budget, remaining budget and available budget threshold; the initial decoding matrix is ​​empty;

[0009] S3: performing regional division according to the number of original packages, and obtaining regional division results, including the number of regional divisions and the regenerated package capacity of each region;

[0010] S4: If the remaining budget is greater than 0 and the decoding matrix is ​​not full rank, the coding density is adjusted in the current region and the remaining budget is updated, the next region is switched, and the process returns to step S4; otherwise, the coded packet is transmitted according to the decoding matrix state.

[0011] Preferably, the initial remaining budget is equal to the expected budget, and the initial available budget threshold is equal to half of the initial expected budget.

[0012] Preferably, the formula for performing area division is:

[0013]

[0014] Among them, R j represents the regenerated packet capacity of the jth area, n represents the number of original packets, and m represents the total number of area divisions.

[0015] Preferably, the process of adjusting the coding density and updating the remaining budget includes:

[0016] S41: Determine whether the sparsity is less than the sparsity threshold. If so, update the sparsity and execute step S42; otherwise, the source node transmits the coded packet to the sink node according to the current sparsity until the number of regenerated packets received by the sink node reaches the regenerated packet capacity of the area, and calculate the budget spent in the current area;

[0017] S42: Determine whether the budget spent in the current area is less than the available budget threshold. If so, the source node transmits the coded packet to the sink node according to the updated sparsity until the number of regenerated packets received by the sink node reaches the regenerated packet capacity of the area, and calculates the budget spent in the current area; otherwise, return to step S41;

[0018] S43: Update the remaining budget and available budget thresholds.

[0019] Furthermore, the formula for updating sparsity is:

[0020] w′=w+k

[0021] Wherein, w′ represents the updated sparsity, and k represents the sparsity increase value.

[0022] Furthermore, the process of calculating the budget spent in the current area includes: calculating the coding density according to the sparsity; calculating the linear independence probability according to the coding density and the number of original packets; and calculating the budget spent according to the area division result, the link packet loss rate and the linear independence probability.

[0023] Furthermore, the formula for calculating the budget spent is:

[0024]

[0025] Among them, B(Rj ,R j-1 ,d) represents the budget spent in the jth region, R j represents the regenerated packet capacity of the jth region, p′(i,n) represents the linearly independent probability, and ε represents the link packet loss rate.

[0026] Furthermore, the process of updating the available budget threshold includes: calculating the available budget threshold according to the remaining budget.

[0027] Furthermore, the available budget threshold is half of the remaining budget.

[0028] Preferably, in step S4, the process of transmitting the coded packet according to the decoding matrix state includes: if the decoding matrix is ​​full rank, the destination node decodes the coded packet and completes the information transmission; otherwise, the coded packet is transmitted with the current sparsity until the decoding matrix is ​​full rank and the destination node decodes the coded packet.

[0029] The beneficial effects of the present invention are as follows: the present invention proposes a budget-based sparse network coding density adjustment method applied to wireless networks, the main purpose of which is to solve the problem of increased transmission overhead in the later stage of transmission when using fixed sparsity to study the performance of sparse network coding. The density adjustment method can make the number of original packets in the coding packet dynamically change and increase linearly with the progress of transmission. The present invention comprehensively considers the phenomenon of data packet loss in the actual network transmission channel, and considers the overhead caused by packet loss in addition to the overhead caused by the linear correlation of the data packet itself during the transmission process. At the same time, the method uses an approximate expression of linearly independent probability as the basic formula for calculating the budget, so that the density calculated at the end can be more accurate and can better evaluate the relevant performance of the actual network data packet transmission, thereby reducing the transmission overhead. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the budget-based sparse network coding density adjustment method in the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The present invention proposes a budget-based sparse network coding density adjustment method, such as Figure 1 As shown, the method includes the following contents:

[0033] S1: Obtain the number of original packets from the source node and the link packet loss rate; set the sparsity threshold.

[0034] Get the number of original packets n of the source node and the link packet loss rate ε; set the sparsity threshold, preferably, set the sparsity threshold to

[0035] S2: Initialize the transmission parameters, including sparsity w, expected budget B, and remaining budget B S and available budget thresholds;

[0036] Prioritize, the available budget threshold is set to Initially T=0, indicating that the budget has been spent. Before the source node performs density adjustment, the decoding matrix of the sink node is initialized to be empty.

[0037] Before the source node performs density adjustment, the parameter conditions required for completing density adjustment are set, including: setting the initial sparsity w to 0, that is, the coding density d is also 0, and the expected budget B of the system is given; since the source node does not transmit any data packets at the beginning, the number of regenerated packets in the decoding matrix is ​​0, and there is no budget used at this time, so the initial remaining budget is B S = B; preferably, the initial available budget threshold is equal to half of the initial expected budget. When performing density adjustment for the first area, if

[0038] However, when the budget spent is greater than the available budget threshold, in order to prevent the system from stopping transmission, set w 2 If the initial expected budget is insufficient, the source node uses w 2 The value of the encoding package is constructed, that is, w 2 =1, indicating that only one original packet is used to form a coded packet, ensuring that data transmission can proceed normally.

[0039] Among them, the expected budget B represents the number of coded packets that need to be transmitted for the destination node to decode the original packet of the source node; the coding density d represents the ratio of the number of original packets corresponding to the non-zero coefficients in the coded packet to the total number of original packets, and the sparsity w is the number of original packets involved in the coding.

[0040] S3: The source node divides the area according to the number of original packets and obtains the area division result, including the number of area divisions and the regenerated packet capacity of each area.

[0041] The source node uses the rank distribution function to divide the area according to the number of original packets, and obtains the area division result, including the number of area divisions and the regeneration packet capacity of each area; the area refers to a division designed by the present invention on how the sink node should accept the number of regeneration packets, and the regeneration packet capacity refers to the total number of regeneration packets that the sink node should receive in each area. As long as the number of regeneration packets received by the sink node reaches the specified value, feedback information is sent to the source node. The rank distribution function is a function used for area division, and its area division can be expressed as:

[0042]

[0043] Among them, R j represents the regeneration packet capacity of the jth area, m represents the total number of area divisions; j also represents the number of feedbacks, that is, the number of density adjustments, R 0 = 0 indicates the initial condition; the calculated R j When the value is less than 2, take R j The value is 1.

[0044] In some embodiments, the source node has 64 original packets {x 1 ,x 2 ,...,x 64}, the regional division results are shown in Table 1:

[0045] Table 1 Regional division results

[0046] Regional division Recycled pack quantity R0 0 R1 32 R2 48 R3 56 R4 60 R5 62 R6 63 R7 64

[0047] S4: If the remaining budget is greater than 0 and the decoding matrix is ​​not full rank, the coding density is adjusted in the current region and the remaining budget is updated, the next region is switched and the process returns to step S4; otherwise, the coded packet is transmitted according to the decoding matrix state.

[0048] When the remaining budget is greater than 0 and the decoding matrix is ​​not full rank, the process of adjusting the coding density and updating the remaining budget includes:

[0049] S41: Determine whether the sparsity is less than the sparsity threshold. If so, update the sparsity and execute step S42; otherwise, the source node transmits the encoded packet to the destination node according to the current sparsity until the number of regenerated packets received by the destination node reaches the regenerated packet capacity of the area, and calculates the budget spent in the current area.

[0050] Determine whether the sparsity is less than the sparsity threshold; set a local variable w 1 , so that w 1 =w,w 1 Mainly used to temporarily record the change of sparsity, so as to facilitate the subsequent selection of appropriate density and then update the sparsity w 1 .

[0051] Starting from the first area, density adjustment is performed in order of areas; if Update sparsity w 1 And execute step S42, preferably, the sparsity w 1 Linear increase, the formula for updating sparsity is:

[0052] w′=w+k

[0053] Wherein, w′ represents the updated sparsity, k represents the sparsity increase value, and preferably, k=1.

[0054] like No density adjustment is performed at this time, and the previous area R is continued. j-1 The source node transmits the coded packet according to the current sparsity (the sparseness that has not been updated this time). Specifically, the source node constructs the coded packet from the corresponding original packet according to the coding density and sends the coded packet to the sink node. If the coded packet currently sent by the source node is linearly independent of the coded packet previously received by the sink node, the currently sent coded packet is a regenerated packet, otherwise it is a common coded packet. The sink node receives the coded packet until the number of regenerated packets reaches the region R. j The regeneration package capacity is calculated to calculate the budget spent in the current area. The process includes:

[0055] Calculate the encoding density based on sparsity:

[0056] Calculate the linear independence probability p′(i,n) based on the coding density and the number of original packets:

[0057]

[0058]

[0059] Among them, p 0 =1-d, q represents the size of the finite field, preferably, q=2.

[0060] The budget is calculated based on the regional division results, link packet loss rate and linear independence probability. The calculation formula is:

[0061]

[0062] Among them, B(R j ,R j-1 ,d) represents the budget spent in the jth region, R j represents the regenerated packet capacity of the jth region, p′(i,n) represents the linearly independent probability, and ε represents the link packet loss rate.

[0063] The above calculation formula is different from the previous budget calculation. The previous formula only considered the budget of data packet transmission under ideal conditions. This formula takes into account the budget caused by packet loss during data packet transmission. At the same time, an approximate expression of linearly independent probability is used instead of an upper bound. Therefore, the calculated density value is more accurate and can be used in actual data packet loss scenarios, which can more accurately evaluate network-related performance.

[0064] S42: Determine whether the budget spent in the current area is less than the available budget threshold. If so, the source node transmits the coded packet to the destination node according to the updated sparsity until the number of regenerated packets received by the destination node reaches the regenerated packet capacity of the area, and calculates the budget spent in the current area; otherwise, return to step S41.

[0065] If the budget spent in the current area is less than the available budget threshold, the source node transmits the coded packet to the sink node according to the updated sparsity until the number of regenerated packets received by the sink node reaches the regenerated packet capacity of the area and updates the available budget threshold, calculates the budget spent in the current area, and updates the remaining budget B S , switch to the next area and return to step S4.

[0066] S43: Update the remaining budget and available budget thresholds.

[0067] The remaining budget is calculated based on the expected budget, the current area, and the budgets spent in all previous areas; the available budget threshold is updated based on the remaining budget. Preferably, the available budget threshold is half of the remaining budget, which is expressed as: T represents the budget consumed during the data packet transmission process, and the initial T=0.

[0068] After adjusting the coding density in the current area and updating the remaining budget, the next area is switched and the process returns to step S4.

[0069] When the remaining budget is less than 0 or the decoding matrix is ​​full rank:

[0070] The coded packet is transmitted according to the state of the decoding matrix, and the process includes: if the decoding matrix is ​​full rank, the sink node decodes the coded packet and completes the information transmission; otherwise, the coded packet is transmitted with the current sparsity until the decoding matrix is ​​full rank, the sink node decodes the coded packet and completes the information transmission.

[0071] In some preferred embodiments of the present invention:

[0072] The source node has 64 original packets {x 1 ,x 2 ,...,x 64}, the link packet loss rate is 0.05.

[0073] Initialize the transmission parameters, including: initial sparsity w = 0, expected budget B = 70, spent budget T = 0, w 1 =0,w 2 =1, i=0,R 0 =0, where i represents the number of regenerated packets currently possessed by the decoding matrix.

[0074] According to the rank distribution function and the number of original packets 64, the number of regenerated packets that each area should reach is calculated, that is, each area {R 0 ,R 1 ,R 2 ,R 3 ,R 4 ,R 5 ,R 6 ,R 7}The corresponding numbers of regenerated packets are {0,32,48,56,60,62,63,64} respectively.

[0075] If the remaining budget is greater than 0 and the decoding matrix rank is 0, the process of adjusting the coding density in the current region is as follows:

[0076] (1) but According to the regional division results shown in Table 1, calculate the 0 to R 1 The required budget is R 1 The budget to be spent is:

[0077] Its budget is not less than 35, and we continue to execute step 2 to update the sparsity, that is, adjust the encoding density, so that w 1 Linear increase.

[0078] (2) but Calculate from R 0 to R 1 The budget required is:

[0079] Its budget is not less than 35, and we continue to execute step 3 to update the sparsity, that is, adjust the encoding density, so that w 1 Linear increase.

[0080] (3) but Calculate from R 0 to R 1 The budget required is: Its budget is less than 35, and the new density is determined as R 1 The density corresponding to the area is At the same time, R 1The budget spent in the region is counted in T, that is, T = 34.6911.

[0081] B S =BT=70-34.6911=35.3089.

[0082] Therefore, the source node constructs the coded packet with w=3 until the number of regenerated packets at the destination node reaches R 1 After the value (32) corresponding to the area is reached, the sink sends feedback information to the source, and the source stops sending the coded packet and enters the next area, selecting a suitable coding density for the next area based on the remaining budget. The density adjustment parameters for the first area are shown in Table 2:

[0083] Table 2 Example of density adjustment in the first area

[0084]

[0085] So far, T=34.6911,B=70, R 1 =32, i=32, repeat the above steps and continue to adjust the density for the second area R 2 Select the appropriate density; specifically: assign the sparsity w of the first region to the temporary variable w 1 , that is, w 1 =3, the density adjustment process of the second area is:

[0086] but Calculate from R 1 to R 2 The budget required is: Its budget is less than 17.65445, and the new density is determined to be R 2 The density corresponding to the area is At the same time, the used budget is counted into T, that is, T = 34.6911 + 16.8084 = 51.4995, B S = BT = 70-51.4995 = 18.5005. At this time, the source node constructs the coded packet with w = 4 and sends the coded packet until the number of regenerated packets received by the destination node reaches R 2 The value specified by the area, that is, when the number of regenerated packets is equal to 48, the sink sends feedback to the source, the source stops sending coded packets and enters the next area, and selects the density for the next area based on the remaining budget. The density adjustment parameters for the second area are shown in Table 3:

[0087] Table 3 Example of density adjustment in the second area

[0088]

[0089] At this point, w = 4, T=51.4995,B=70, R 2 =48, i=48, repeat the above steps, continue to adjust the density, select appropriate density for subsequent areas, until the final remaining budget is less than 0 or the decoding matrix is ​​full rank, and stop adjusting the density.

[0090] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A budget-based sparse network coding density adjustment method, characterized in that: include: S1: Obtain the number of original packets from the source node and the link packet loss rate; Set the sparsity threshold; S2: Initialize transmission parameters, including sparsity, expected budget, remaining budget, and available budget threshold; The initial decoding matrix is ​​empty; S3: The source node divides the area according to the number of original packets, and obtains the area division result, including the number of area divisions and the regenerated packet capacity of each area; the formula for area division is: Among them, R j represents the capacity of the regenerated packet in the jth region, n represents the number of original packets, and m represents the total number of regional divisions; S4: If the remaining budget is greater than 0 and the decoding matrix is ​​not full rank, the coding density is adjusted in the current region and the remaining budget is updated, the next region is switched and the step S4 is returned; otherwise, the coding packet is transmitted according to the decoding matrix state; the process of adjusting the coding density and updating the remaining budget includes: S41: Determine whether the sparsity is less than the sparsity threshold. If so, update the sparsity and execute step S42; otherwise, the source node transmits the coded packet to the sink node according to the current sparsity until the number of regenerated packets received by the sink node reaches the regenerated packet capacity of the area, and calculate the budget spent in the current area; The formula for updating sparsity is: w′=w+k Among them, w′ represents the updated sparsity, and k represents the sparsity increase value; S42: Determine whether the budget spent in the current area is less than the available budget threshold. If so, the source node transmits the coded packet to the sink node according to the updated sparsity until the number of regenerated packets received by the sink node reaches the regenerated packet capacity of the area, and calculates the budget spent in the current area; otherwise, return to step S41; The process of calculating the budget spent in the current area includes: calculating the coding density according to the sparsity; calculating the linear independence probability according to the coding density and the number of original packets; calculating the budget spent according to the area division result, the link packet loss rate and the linear independence probability; the formula for calculating the budget spent is: Among them, B(R j ,R j-1 ,d) represents the budget spent in the jth region, R j represents the regeneration packet capacity of the jth region, p′(i,n) represents the linearly independent probability, and ε represents the link packet loss rate; S43: Update the remaining budget and available budget thresholds.

2. The budget-based sparse network coding density adjustment method according to claim 1, characterized in that: Initialize the remaining budget to be equal to the desired budget, and initialize the available budget threshold to be equal to half of the initial desired budget.

3. The budget-based sparse network coding density adjustment method according to claim 1, characterized in that: The process of updating the available budget threshold includes calculating the available budget threshold based on the remaining budget.

4. The budget-based sparse network coding density adjustment method according to claim 1, characterized in that: The available budget threshold is half of the remaining budget.

5. The budget-based sparse network coding density adjustment method according to claim 1, characterized in that: In step S4, the process of transmitting the coded packet according to the decoding matrix state includes: if the decoding matrix is ​​full rank, the sink node decodes the coded packet and completes the information transmission; otherwise, the coded packet is transmitted with the current sparsity until the decoding matrix is ​​full rank and the sink node decodes the coded packet.

Images