A wireless transmission method based on adjustable sparse network coding
By introducing expected budgets and orders between the source and sinks, combining the selection of automatic retransmission requests and optimized Gaussian elimination decoding methods, the problems of transmission delay and unreliable transmission in adjustable sparse network encoding are solved. Achieves wireless transmission with low latency and high reliability.
Patent Information
- Application Number
- CN202310246951.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-14
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2043-03-14
AI Technical Summary
The existing density selection method based on adjustable sparse network encoding has problems such as increasing transmission delay and unreliable transmission when the decoding matrix rank at the sink end is sj+1, and it is impossible to effectively evaluate the degree of reduction in computational complexity.
By introducing expected budgets and orders, the sparseness of encoded data packets is adjusted, combined with the selection of automatic retransmission requests and the optimization of Gaussian elimination decoding method, the calculation complexity is reduced and reliable transmission is ensured.
It realizes that while ensuring reliable transmission, the transmission delay and calculation complexity are reduced, the decoding success rate is improved, and the degree of reduction in calculation complexity is quantitatively evaluated.
Smart Images

Figure CN116318548B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of wireless transmission, and in particular relates to a wireless transmission method based on adjustable sparse network coding. Background Art
[0002] With the development of mobile Internet technology, mankind has entered the 5G era and is about to enter the 6G era. Wireless communication systems are required to have low latency, high reliability and large capacity. The information transmission rate is getting higher and higher. Traditional transmission technologies that ensure reliable transmission, such as automatic request and forward error correction, have problems such as throughput and latency performance degradation and error accumulation in high-speed application scenarios. Network coding has significant advantages in improving network throughput, enhancing network robustness and balancing network traffic, and has become a solution for future wireless communication technology.
[0003] Network Coding (NC) technology is different from the storage-and-forward routing method. It allows the intermediate nodes of the network to encode the received information. In a single-source multicast network, the theoretical upper limit of the network capacity can be reached. However, network coding has a fatal flaw - high computational complexity, which hinders the application of network coding in practical application scenarios. In order to reduce the computational complexity, Danilo et al. proposed sparse network coding (SNC). Since SNC only performs random linear network coding on a subset of the original data packets, a sparse decoding matrix can be obtained. The total number of operations required in the decoding stage will be reduced, thereby reducing the computational complexity. Although SNC reduces the computational complexity, it also increases the transmission delay. The increase in delay is mainly due to its sparsity, which leads to a greatly reduced probability of receiving a linearly independent coded data packet when the rank of the decoding matrix is close to the packet size. In order to solve the above problems, Feizi et al. proposed tunable sparse network coding (TSNC). The main idea of TSNC is to send coded data packets with low sparsity at the beginning when transmitting a packet, and gradually increase the sparsity of the coded data packets as the transmission progresses. This can improve the linear independence of the sent coded data packets and the coded data packets received by the destination, thereby reducing the computational complexity of network coding and the delay of packet transmission.
[0004] Existing technical solution: Chongqing University of Posts and Telecommunications disclosed a density selection method when using adjustable sparse network coding for data transmission in its patent application "A density selection method based on adjustable sparse network coding" (application date: 2021.03.25 application number: 202110322134.5 publication number: CN 113114415A). This method compares the actual budget with the expected budget to select the optimal density value suitable for the current stage, solving the problems of limited bandwidth and insufficient budget caused by unstable link packet loss rate. The implementation steps of this method are: First, the source end first determines the density distribution function, using the staged density distribution function s j = k·(2 j -1) / 2 j Take this as an example; second, the source determines the actual budget T(s) for each stage according to the channel conditions. j ,s j+1 ,k), calculate the expected budget for a certain stage Where w is the sparsity of the coded data packet, and k is the packet size; third, at the source end, for a certain stage, the relationship between the actual budget and the expected budget is determined. If the actual budget is greater than or equal to the expected budget, the sparsity w of the coded data packet can be reduced to reduce the computational complexity until a suitable w value is selected; if the actual budget is less than the expected budget, the w value is increased; fourth, at the source end, the coded data packet is sent with w as the sparsity, and the actual budget of each coded data packet sent at this stage is reduced by 1 until the actual budget of this stage is 0 or the feedback information of the destination end is received, and the coded data packet of the next stage is continued to be sent; fifth, when the rank of the decoding matrix of the destination end is s j+1 , and feed back the rank value of the decoding matrix at this time.
[0005] A density selection method based on adjustable sparse network coding disclosed in the prior art patent document (CN 113114415A) has the following shortcomings:
[0006] First, when the rank of the decoding matrix at the destination is s j+1 , and at this time, when the actual budget of the jth order is not 0, a certain amount of coded data packets must be sent while waiting for feedback information, which increases the transmission delay of the coded data packets;
[0007] Second, when the actual budget is 0 at the last stage of transmission and the rank of the decoding matrix at the destination is not k, reliable transmission cannot be guaranteed;
[0008] Third, it only determines whether the decoding matrix is full rank, but does not provide an algorithm suitable for decoding sparse decoding matrices, and cannot quantitatively evaluate the degree of reduction in computational complexity. Summary of the invention
[0009] In order to solve the above problems existing in the prior art, the present invention provides a wireless transmission method based on adjustable sparse network coding. The technical problem to be solved by the present invention is achieved by the following technical solutions:
[0010] The present invention provides a wireless transmission method based on adjustable sparse network coding, comprising:
[0011] Step 1, the source end obtains the original data packet to be sent, and determines the encoding information to be encoded for the original data packet;
[0012] Step 2: The source end groups the original data packets to be sent according to the determined grouping size to obtain multiple groups of original data groups, each of which contains multiple original data packets;
[0013] Step 3, the source end encodes each original data packet according to the encoding information to obtain each encoded data packet, and sends each encoded data packet out; the encoded data packet includes multiple encoded data packets;
[0014] Step 4, the destination determines whether the coded data packet is the last coded data packet of a certain order. If it is the last coded data packet, further determines whether the certain order meets the order condition. If so, extracts the coded data packet with the same order as the certain order from the coded data group where the coded data packet is located;
[0015] Step 5, the destination determines whether the rank of the code coefficient matrix of the coded data packet extracted in step 4 meets the receiving condition, and if so, feeds back information to the source, and if not, executes step 6;
[0016] Step 6, the destination extracts all coded data packets in the coded data group where the coded data packets that do not meet the order condition or the receiving condition are located;
[0017] Step 7, the destination determines whether the number of coded data packets extracted in step 6 meets the number condition. If not, it determines whether the rank of the coding coefficient matrix in the coded data packet extracted in step 6 meets the decoding condition. If it meets the decoding condition, it decodes all coded data packets that meet the decoding condition.
[0018] Step 8: If the coded data packet in step 4 is not the last coded data packet of a certain order, or if the number condition is not met in step 7, the destination updates its receiving buffer queue storing the coded data packets to wait until the next coded data packet is received;
[0019] Step 9: The source determines whether it is necessary to retransmit the encoded data packet according to the received feedback information, and if necessary, performs the retransmission process, and the destination performs steps 4 to 8 to complete the transmission of a group of original data packets.
[0020] Beneficial effects of the present invention:
[0021] (1) The present invention sets the maximum number of retransmissions (the destination sends a group of coded data packets of the jth order or a group of coded data packets, and the destination can use r with a probability close to 1). j ≥s j -s j+1 Or r=k (the maximum number of retransmissions required) can meet the user's requirements for different reliability. Specifically, when the maximum number of retransmissions is 0, a decoding success rate of 0.8 can be guaranteed; when the number of retransmissions is 2, reliable transmission can be guaranteed.
[0022] (2) The present invention combines forward error correction with automatic repeat request selection. Specifically, the number of coded data packets to be transmitted for an original data packet at the jth stage is determined as c. j , during transmission, it is first expanded according to the deletion probability ε of the channel to That is, the number of retransmissions is reduced by adding redundancy in advance; when using the automatic repeat request selection, unlike other schemes that feedback retransmissions in units of packets, the present invention not only feedbacks retransmissions in units of packets but also feedbacks retransmissions in units of a certain order of packets. Compared with feedback retransmissions only in units of packets, the feedback method used by the present invention can effectively reduce the delay; when the source end is waiting for the sink end to feedback information, the sink end can also send coded data packets of other orders of the current coded data packet or the next coded data packet. By using these three measures, the transmission delay can be effectively reduced.
[0023] (3) The present invention proposes an optimized Gaussian elimination decoding method, and the decoding matrix obtained by the linear density adjustment method fully meets the requirements of the optimized Gaussian elimination decoding method for the decoding matrix, which can effectively reduce the computational complexity of decoding. The degree of reduction in computational complexity can be quantitatively evaluated by counting the number of addition operations and multiplication operations on the finite field during decoding.
[0024] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 It is a flow chart of a wireless transmission method based on adjustable sparse network coding provided by the present invention;
[0026] Figure 2 It is a schematic diagram of the transmission process provided by the present invention. DETAILED DESCRIPTION
[0027] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.
[0028] Embodiment 1
[0029] like Figure 1 As shown, the present invention provides a wireless transmission method based on adjustable sparse network coding, comprising:
[0030] Step 1, the source end obtains the original data packet to be sent, and determines the encoding information to be encoded for the original data packet;
[0031] Step 2: The source end groups the original data packets to be sent according to the determined grouping size to obtain multiple groups of original data groups, each of which contains multiple original data packets;
[0032] Step 3, the source end encodes each original data packet according to the encoding information to obtain each encoded data packet, and sends each encoded data packet out; the encoded data packet includes multiple encoded data packets;
[0033] Figure 1 The deletion channel in is the sending channel, and the source end sends the encoded data packets from the deletion channel.
[0034] Step 4, the destination determines whether the coded data packet is the last coded data packet of a certain order. If it is the last coded data packet, further determines whether the certain order meets the order condition. If so, extracts the coded data packet with the same order as the certain order from the coded data group where the coded data packet is located;
[0035] Step 5, the destination determines whether the rank of the coding coefficient matrix of the coded data packet extracted in step 4 meets the receiving condition, and if so, feeds back information to the source, and if not, executes step 6;
[0036] Step 6, the destination extracts all coded data packets in the coded data group where the coded data packets that do not meet the order condition or the receiving condition are located;
[0037] Step 7, the destination determines whether the number of coded data packets extracted in step 6 meets the number condition. If not, it determines whether the rank of the coding coefficient matrix in the coded data packet extracted in step 6 meets the decoding condition. If it meets the decoding condition, it decodes all coded data packets that meet the decoding condition.
[0038] Step 8: If the coded data packet in step 4 is not the last coded data packet of a certain order, or if the number condition is met in step 7, the destination updates its receiving buffer queue storing the coded data packets to wait for receiving the next coded data packet;
[0039] Step 9: The source determines whether it is necessary to retransmit the encoded data packet according to the received feedback information, and if necessary, performs the retransmission process, and the destination performs steps 4 to 8 to complete the transmission of a group of original data packets.
[0040] The present invention provides a wireless transmission method based on adjustable sparse network coding to reduce the computational complexity of network coding and ensure reliable transmission. The specific idea of implementing the present invention is: considering the introduction of expected budget and order, the sparsity of the coded data packet can be adjusted by adjusting the values of the expected budget and order, thereby adjusting the computational complexity. The source end first determines the transmission parameters according to the usage scenario and uses the adjustable sparse network coding density adjustment algorithm to determine the rank expected to be achieved per order, the number of coded data packets sent per order, and the sparsity of the coded data packets sent per order, and uses a linear density adjustment method when encoding the original data packet. During transmission, the source end uses the automatic retransmission request protocol to continuously send coded data packets. When the source receives feedback information, it selects whether to retransmit by order or by group according to the relationship between the maximum number of retransmissions and the current number of retransmissions. If retransmission is required, the coded data packet generated according to the feedback information is inserted into the head of the sending queue. Due to transmission in the transmission channel, there is a phenomenon of coded data packet loss in the current coded data packet to the destination end. At the destination, whether feedback is required is determined based on the relationship between the rank of a decoding matrix of a certain order of a certain coded data group and the rank to be achieved by a certain order, and between the rank of a decoding matrix of a certain coded data group and the rank to be achieved by a certain coded data group.
[0041] Embodiment 2
[0042] As an optional embodiment of the present invention, step 1 includes:
[0043] Determine the group size, deletion probability ε, order m and expected budget B according to the usage scenario, and determine the coding information; the coding information includes: the rank s expected to be achieved at the jth (j = 0, 1, 2, ..., m) order j , the number of coded data packets sent at the jth stage n j and the sparsity w of the j-th order coded packet j (w j =0); the process of determining the coded information includes:
[0044] (1a) Calculate s j = <k(2 j -1) / 2 j >(j=0,1,...,m-1),s m = k, where <> means rounding to the nearest integer;
[0045] (1b) Calculate c j = j >(j=1,2,...,m-1),cm = m-1 > Calculate in, Indicates rounding up;
[0046] (1c) For the jth order, calculate the sparsity w of the encoded data packet j , if c j =s j -s j-1 ,c j =c j +1, for w = 1:k, once there is Then j =w.
[0047] The present invention introduces an expected budget and an order, and can adjust the computational complexity and delay of a wireless transmission method based on adjustable sparse network coding by adjusting the values of the expected budget and the order according to different usage scenarios. Specifically, when the packet size is constant, for devices with strong computing power or application scenarios with insufficient budget and limited bandwidth, the expected budget can be adjusted to be slightly larger than the packet size, and the order can be smaller; for devices with weak computing power or application scenarios with sufficient budget and abundant bandwidth resources, the expected budget can be adjusted to be larger than the packet size, and the order can be larger. The specific adjustment method is that when the packet size and the channel deletion probability are constant, the expected budget can be determined first, and then the order can be increased until the computational complexity remains basically unchanged. If the computational complexity still needs to be reduced at this time, the expected budget needs to be increased;
[0048] Embodiment 3
[0049] As an optional embodiment of the present invention, step 2 includes:
[0050] The source end groups P original data packets into groups of k. If the last group is less than k original data packets, it is filled with all-zero data packets. Then the P original data packets can be divided into Groups;
[0051] The value of k ranges from 1 to the number of all original data packets to be sent.
[0052] Embodiment 4
[0053] As an optional embodiment of the present invention, step 3 includes:
[0054] (3a) A set of original data is grouped as P = [p 1 ... p k ], the sparsity of the a-th coded data packet of the j-th (j=1,...,m)th order of a group of original data packets is min{w j-1 +a,w j}, Represents a coded information packet; a=1,2,...,n j , f(i) is randomly selected from the set {1,...,k}, and any i≠i', then from F q \{0} is independently randomly selected and
[0055] (3b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g a,f(i) , get the coding coefficient vector of the a-th coded data packet, where L = k;
[0056] (3c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j of the coding information packet in columns to obtain a coding data packet, and setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order; if yes, the flag is set to 1, otherwise, it is set to 0, and the coding data packet is added to the sending queue of the source end;
[0057] (3d) Repeat (3a)(3b)(3c) until a raw data packet is generated. coded data packets, and composing the coded data packets into a coded data group;
[0058] (3e) The source sends the coded data packets in the sending queue to the wireless channel in sequence, sending one coded data packet in each time slot.
[0059] The present invention proposes a linear density adjustment method, which can effectively reduce the encoding and decoding complexity by combining with a specific decoding algorithm compared with the traditional step-type density adjustment method;
[0060] Embodiment 5
[0061] As an optional embodiment of the present invention, step 4 includes:
[0062] (4a) The destination receives an encoded data packet;
[0063] (4b) judging whether the received coded data packet is the last coded data packet of the jth order by the flag bit, if it is not the last one, executing step 8;
[0064] (4c) If the received coded data packet is the last one, determine whether the order number j is not greater than the order number m. If so, it is considered that the order condition is met, otherwise, the order condition is not met;
[0065] (4d) If the order condition is met, all coded data packets with the same group number and order number j as the received coded data packet are extracted from its own receiving buffer queue.
[0066] Embodiment 6
[0067] As an optional embodiment of the present invention, step 5 includes:
[0068] (5a) The destination determines the rank r of the coding coefficient matrix of all the j-th order coded data packets extracted in step 4. j (j=1,2,...,m),D j =(s j -s j-1 )-r j , if D j If ≤0, it is considered that the acceptance condition is not met, otherwise it is considered that the acceptance condition is met;
[0069] (5b) The destination end feeds back to the source end the group number, order number j and degrees of freedom of the currently received coded data group
[0070] Embodiment 7
[0071] As an optional embodiment of the present invention, step 7 includes:
[0072] (7a) The destination determines whether the total number of coded data packets extracted from the receiving buffer queue in step 6 is less than n+e. If so, it is confirmed that the number condition is met, where If it is not less than, it is confirmed that the number condition is not met;
[0073] (7b) If the number condition is not met, the destination end calculates the rank r of the coding coefficient matrix in all the coded data packets of the current coded data packet extracted in step 6, D m+1 = kr, if D m+1 =0, it is determined that the decoding condition is met, otherwise, it is confirmed that the decoding condition is not met;
[0074] (7c) If the decoding condition is met, decoding all the encoded data packets that meet the decoding condition;
[0075] (7c) Includes:
[0076] (7c1) The destination end extracts the coding coefficient matrix and the coding information packet from all the coded data packets of the current coded data packet extracted;
[0077] (7c2) using the coding coefficient matrix and the coded information packet to perform an optimized Gaussian elimination decoding algorithm to obtain a set of original data packets;
[0078] The input of the optimized Gaussian elimination decoding algorithm is the coding coefficient matrix G, also called the decoding matrix, with a dimension of (n+e)×k. The received coded information packet U=[u 1 u 2... u n+e ] T , output an original data packet P = [p 1 ...p k ] T ,The optimized Gaussian elimination decoding algorithm uses a, z and r, where a=1:k represents the order of the original data packet, and the i-th element in the row vector a is represented by a i , z is a 1×k row vector, representing the column weight of the lower right corner of the decoding matrix, initialized to the column weight of the decoding matrix, and the i-th element in the row vector z is represented by z i , r is a row vector of (n+e)×1, representing the row weight of the decoding matrix, and is initialized to the row weight of the decoding matrix; the specific implementation steps of the optimized Gaussian elimination decoding algorithm are as follows:
[0079] (11a) c←1, repeat (11b)(11c) until c=k, then execute (11d);
[0080] (11b) Find the column whose value is 1 in z, denoted as C;
[0081] (11b1) If C is not empty, for each element in C corresponding to the column y, swap the cth column of G with the yth column, the cth column of a with the yth column, and the cth column of z with the yth column; find the row q containing the non-zero elements from the cth row to the n+eth row in the cth column, swap the cth row of G with the qth row, the cth row of U with the qth row, and the cth row of r with the qth row; update the column weights, on the one hand, z c =0, on the other hand, the column weight of the column corresponding to the non-zero element on the right of the cth row and cth column is reduced by 1, c←c+1; otherwise, execute step (11b2);
[0082] (11b2) If C is empty, find the row with the smallest weight from the cth row to the n+eth row in r. If there are multiple rows, randomly select one and record it as R; swap the cth row with the Rth row of G, the cth row with the Rth row of U, and the cth row with the Rth row of r; find the column where the non-zero elements in the cth row are located, record it as p, swap the cth column with the pth column of G, the cth column with the pth column of a, and the cth column with the pth column of z; perform Gaussian elimination on the cth column and update the corresponding row weight and column weight;
[0083] (11c) Update column weight, c←c+1;
[0084] (11d)p ai ←u i ,i=1,2,...,k.
[0085] (7d) If the decoding condition is not met, the destination end feeds back the group number, order number m+1 and degrees of freedom of the current coded data group to the source end.
[0086] The present invention proposes an optimized Gaussian elimination decoding method. Compared with the traditional Gaussian elimination decoding algorithm, for sparse matrices, this method introduces less density in the process of transforming the coding coefficient matrix into an upper triangle, thereby reducing the computational complexity of the sparse matrix Gaussian elimination.
[0087] Embodiment 8
[0088] As an optional embodiment of the present invention, step 9 includes:
[0089] The source receives the feedback information and extracts the group number, order number j and degree of freedom e from it. j , determine whether the number of retransmissions corresponding to the group number and order number j is less than the maximum number of retransmissions. If so, the retransmission process is performed; if not, the source confirms that the transmission of a group of original data packets has been completed. The retransmission process includes:
[0090] (15a) Let the sparsity of a coded data packet be w. When the order number j≤m, for the current original data packet of order j, b(b≤e j ) The sparsity of the retransmitted coded packets w = min{w j-1 +n j +b,w j}; When the order number j = m + 1, the sparsity of all retransmitted coded data packets w = k; a coded information packet Where b = 1, 2, ..., e j , f(i) is randomly selected from the set {1,...,k}, and any i≠i', then f(i)≠f(i'),g b,f(i) , from F q \{0} is independently randomly selected and
[0091] (15b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g b,f(i) , get the coding coefficient vector of the b-th coded data packet, where L = k;
[0092] (15c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j to which the coding information packet belongs in columns to obtain a coding data packet, setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order retransmission, if yes, the flag is set to 1, otherwise, it is set to 0, and adding the coding data packet to the head of the sending queue of the source end;
[0093] (15d) Repeat (15a), (15b), (15c) until e is generated for the jth order of an original data packet. jretransmit the coded data packet, increase the number of retransmissions corresponding to the j-th order of the group by 1, and send the coded data packets in the sending queue to the wireless channel in sequence, with one coded data packet sent in each time slot.
[0094] The present invention combines forward error correction and automatic repeat request selection to reduce the time slot used for successfully transmitting a packet, that is, to reduce the delay of the scheme. The specific combination point is to determine the number of coded data packets to be transmitted in the jth order of a group of original data packets as c. j , during transmission, it is first expanded to That is, the number of retransmissions is reduced by adding redundancy in advance; when using the automatic retransmission request, unlike other schemes that feedback retransmissions in units of packets, the present invention not only feedbacks retransmissions in units of packets but also feedbacks retransmissions in units of a certain order of packets. Compared with only feedback retransmissions in units of packets, the feedback method used in the present invention can effectively reduce the delay.
[0095] The present invention provides a wireless transmission method based on adjustable sparse network coding, the core and key of which include determining the linear density adjustment method in the process of encoding information, feedback from the destination not only in groups but also in groups of a certain order, and optimizing the Gaussian elimination decoding algorithm. The linear density adjustment method can be replaced by the replaced part of (1c) "for w = 1: k, once there is Then j =w. "For the part on the left side of the inequality sign It is the lower bound of the probability that the destination receives a coded data packet with sparsity w and the rank increases by 1 when the decoding matrix rank is i. The calculation of this part can be optimized by using a more accurate probability calculation formula.
[0096] The algorithm implementation process of the present invention is as follows:
[0097] (1) Introducing the order and expected budget, determine the expected rank of each order of a set of coded data packets, the number of coded data packets sent at each order, and the sparsity of the coded data packets sent at each order:
[0098] Determine the group size k, deletion probability ε, order m and expected budget B according to the usage scenario, and determine the expected rank s of the jth (j = 0, 1, 2, ..., m) order j , the number of coded data packets sent at the jth stage n j and the sparsity w of the j-th coded packet j (w j =0). Specifically, the steps performed in this process include:
[0099] (1a) Calculate s j = <k(2 j -1) / 2 j>(j=0,1,...,m-1),s m = k, where <> means rounding to the nearest integer;
[0100] (1b) Calculate c j = j >(j=1,2,...,m-1),c m = m-1 > Calculate in, Indicates rounding up;
[0101] (1c) For the jth order, calculate the sparsity w of the encoded data packet j , if c j =s j -s j-1 ,c j =c j +1, for w = 1:k, once there is Then j =w;
[0102] (2) Group the original data packets to be sent:
[0103] The source end groups P original data packets into groups of k. If the last group is less than k original data packets, it is filled with all-zero data packets. In this way, the P original data packets can be divided into packets, where k ranges from 1 to the number of all original data packets to be sent;
[0104] (3) The source encodes a group of original data packets:
[0105] (3a) A set of original data packets is represented as P = [p 1 ... p k ], the sparsity of the a-th coded data packet of the j-th (j=1,...,m)th order of a group of original data packets is min{w j-1 +a,w j},Right now Represents a coded information packet, where a=1,2,...,n j ,f(i) is randomly selected from the set {1,...,k}, and any i≠i', then f(i)≠f(i'),g a,f(i) From F q \{0} is independently randomly selected and
[0106] (3b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g a,f(i) , get the coding coefficient vector of the a-th coded data packet, where L = k;
[0107] (3c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j of the coding information packet in columns to obtain a coding data packet, setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order, if yes, the flag is set to 1, otherwise, it is set to 0, and adding the coding data packet to the sending queue of the source end;
[0108] (3d) Repeat (3a)(3b)(3c) until a raw data packet is generated. coded data packets, which form a group of coded data packets;
[0109] (3e) The source sends the coded data packets in the sending queue to the wireless channel in sequence, with one coded data packet sent in each time slot;
[0110] (4) The destination determines whether the received coded data packet is the last coded data packet of a certain order:
[0111] After receiving a coded data packet, the destination determines whether the coded data packet is the last coded data packet of the jth order through the flag bit. If not, execute step (13); if yes, execute step (5);
[0112] (5) Determine whether the order number is not greater than the order:
[0113] If the order number j is not greater than the order number m, extract the j-th order coded data packet of the current coded data group, and extract all coded data packets with the same group number and order number j as the received coded data packet from its own receiving buffer queue; otherwise, execute step (8);
[0114] (6) The destination calculates the rank r of the coding coefficient matrix of all the coded data packets of the jth order of the current coded data packet extracted from the receiving buffer queue j (j=1,2,...,m),D j =(s j -s j-1 )-r j , if D j ≤0, execute step (8); otherwise, execute step (7);
[0115] (7) Sending j-th order feedback information of the current coded data packet:
[0116] The destination end feeds back the group number, order number j and degree of freedom of the current coded data group to the source end.
[0117] (8) Extract all the coded data packets of the current coded data group:
[0118] The sink extracts the coded data packet with the same group number as the received coded packet from its own receiving buffer queue;
[0119] (9) The destination determines whether the total number of coded data packets extracted from the receive buffer queue is less than n+e, where If yes, execute step (13); otherwise, execute step (10);
[0120] (10) The destination end calculates the rank r of the coding coefficient matrix of all the coded data packets of the current coded data packet extracted from the receiving buffer queue, D m+1 = kr, if D m+1 =0, then execute step (11); otherwise, execute step (12);
[0121] (11) Decoding:
[0122] The destination extracts the coding coefficient matrix and the coding information packet from all the coded data packets of the current coded data packet extracted, and uses the coding coefficient matrix and the coded information packet to execute the optimized Gaussian elimination decoding algorithm to obtain a set of original data packets. The input of the optimized Gaussian elimination decoding algorithm is the coding coefficient matrix G, also called the decoding matrix, with a dimension of (n+e)×k. The received coded information packet U=[u 1 u 2 ... u n+e ] T , output an original data packet P = [p 1 ... p k ] T , a, z and r are used in the specific implementation steps of the optimized Gaussian elimination decoding algorithm, where a=1:k represents the order of the original data packet, and the i-th element in the row vector a is represented by a i , z is a 1×k row vector, representing the column weight of the lower right corner of the decoding matrix, initialized to the column weight of the decoding matrix, and the i-th element in the row vector z is represented by z i , r is a column vector of (n+e)×1, representing the row weight of the decoding matrix, and is initialized to the row weight of the decoding matrix. The specific implementation steps of the optimized Gaussian elimination decoding algorithm are as follows:
[0123] (11a) c←1, repeat (11b)(11c) until c=k, then execute (11d);
[0124] (11b) Find the column whose value is 1 in z, denoted as C;
[0125] (11b1) If C is not empty, for each element in C corresponding to the column y, swap the cth column of G with the yth column, the cth column of a with the yth column, and the cth column of z with the yth column; find the row q containing the non-zero elements from the cth row to the n+eth row in the cth column, swap the cth row of G with the qth row, the cth row of U with the qth row, and the cth row of r with the qth row; update the column weights, on the one hand, z c =0, on the other hand, the column corresponding to the non-zero element on the right of the cth row and cth column is decremented by 1, c←c+1; otherwise, execute step (11b2);
[0126] (11b2) If C is empty, find the row with the smallest weight from the cth row to the n+eth row in r. If there are multiple rows, randomly select one and record it as R. Swap the cth row with the Rth row of G, the cth row with the Rth row of U, and the cth row with the Rth row of r; find the column where the non-zero elements in the cth row are located, record it as p, swap the cth column with the pth column of G, the cth column with the pth column of a, and the cth column with the pth column of z; perform Gaussian elimination on the cth column and update the corresponding row and column weights;
[0127] (11c) Update column weight, c←c+1;
[0128] (11d)p ai ←u i ,i=1,2,...,k
[0129] (12) Sending feedback information of current coded data group:
[0130] The destination end feeds back the group number, order number m+1 and degree of freedom of the current coded data group to the source end. c
[0131] (13) Update the receive buffer queue:
[0132] The destination adds the received coded data packet to its own receiving buffer queue and waits for receiving the next coded data packet;
[0133] (14) The source receives the feedback information and extracts the group number, order number j and degree of freedom e j , determine whether the number of retransmissions corresponding to the group number and the order number j is less than the maximum number of retransmissions, if so, execute step (15), otherwise, execute step (16);
[0134] (15) Retransmit the encoded data packet:
[0135] (15a) Let the sparsity of a coded data packet be w. When the order number j≤m, for the current original data packet of order j, b(b≤e j ) The sparsity of the retransmitted coded packets w = min{w j-1 +nj +b,w j}; When the order number j = m + 1, the sparsity of all retransmitted coded packets is w = k. A coded information packet Where b = 1, 2, ..., e j , f(i) is randomly selected from the set {1,...,k}, and any i≠i', then f(i)≠f(i'),g b,f(i) , from F q \{0} is independently randomly selected and
[0136] (15b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g b,f(i) , get the coding coefficient vector of the b-th coded data packet, where L = k;
[0137] (15c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j to which the coding information packet belongs in columns to obtain a coding data packet, setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order retransmission, if yes, the flag is set to 1, otherwise, it is set to 0, and adding the coding data packet to the head of the sending queue of the source end;
[0138] (15d) Repeat (15a), (15b), (15c) until e is generated for the jth order of an original data packet. j retransmitted coded data packets, add 1 to the number of retransmissions corresponding to the j-th order of the group, and then execute step (3e);
[0139] (16) Complete the transmission of a set of original data packets.
[0140] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0141] Although the present application is described herein in conjunction with various embodiments, in the process of implementing the claimed application, those skilled in the art may understand and implement other variations of the disclosed embodiments by reviewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality of components or steps.
[0142] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.
Claims
1. A wireless transmission method based on adjustable sparse network coding, characterized in that: include: Step 1, the source end obtains the original data packet to be sent, and determines the encoding information to be encoded for the original data packet; Step 2: The source end groups the original data packets to be sent according to the determined grouping size to obtain multiple groups of original data groups, each of which contains multiple original data packets; Step 3, the source end encodes each original data packet according to the encoding information to obtain each encoded data packet, and sends each encoded data packet out; the encoded data packet includes multiple encoded data packets; Step 4, the destination determines whether the coded data packet is the last coded data packet of a certain order. If it is the last coded data packet, further determines whether the certain order meets the order condition. If so, extracts the coded data packet with the same order as the certain order from the coded data group where the coded data packet is located; Step 5, the destination determines whether the rank of the coding coefficient matrix of the coded data packet extracted in step 4 meets the receiving condition, and if so, feeds back information to the source, and if not, executes step 6; Step 6, the destination extracts all coded data packets in the coded data group where the coded data packets that do not meet the order condition or the receiving condition are located; Step 7, the destination determines whether the number of coded data packets extracted in step 6 meets the number condition. If not, it determines whether the rank of the coding coefficient matrix in the coded data packet extracted in step 6 meets the decoding condition. If it meets the decoding condition, it decodes all coded data packets that meet the decoding condition. Step 8: If the coded data packet in step 4 is not the last coded data packet of a certain order, or if the number condition is not met in step 7, the destination updates its receiving buffer queue storing the coded data packets to wait until the next coded data packet is received; Step 9, the source determines whether it is necessary to retransmit the encoded data packet according to the received feedback information, and if necessary, performs the retransmission process, and the destination performs steps 4 to 8 to complete the transmission of a group of original data packets; Step 1 includes: Determine the group size k, deletion probability ε, order m and expected budget B according to the usage scenario, and determine the coding information; the coding information includes: the rank s expected to be achieved at the jth (j = 0, 1, 2, ..., m) order j (s0=0), the number of coded data packets sent at the jth stage n j and the sparsity w of the j-th order coded packet j (w0=0); The process of determining the coded information includes: (1a) Calculate s j = <k(2 j -1) / 2 j >(j=0,1,...,m-1),s m = k, where <> means rounding to the nearest integer; (1b) Calculate c j = j >(j=1,2,...,m-1),c m = m-1 > Calculate in, Indicates rounding up; (1c) For the jth order, calculate the sparsity w of the encoded data packet j , if c j =s j -s j-1 ,c j =c j +1, for w = 1:k, once there is Then j =w; Step 3 includes: (3a) A set of original data groups is represented as P = [p1...p k ], the sparsity of the a-th coded data packet of the j-th (j=1,...,m)th order of a group of original data packets is min{w j-1 +a,w j }, Represents a coded information packet; a=1,2,...,n j , f(i) is randomly selected from the set {1,...,k}, and any i≠i', then f(i)≠f(i'), g a,f(i) From F q \{0} is independently randomly selected and (3b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g a,f(i) , get the coding coefficient vector of the a-th coded data packet, where L = k; (3c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j of the coding information packet in columns to obtain a coding data packet, and setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order; if yes, the flag is set to 1, otherwise, it is set to 0, and the coding data packet is added to the sending queue of the source end; (3d) Repeat (3a)(3b)(3c) until a raw data packet is generated. coded data packets, and composing the coded data packets into a coded data group; (3e) The source sends the coded data packets in the sending queue to the wireless channel in sequence, with one coded data packet sent in each time slot; Step 4 includes: (4a) The destination receives an encoded data packet; (4b) judging whether the received coded data packet is the last coded data packet of the jth order by the flag bit, if it is not the last one, executing step 8; (4c) If the received coded data packet is the last one, determine whether the order number j is not greater than the order number m. If so, it is considered that the order condition is met, otherwise, the order condition is not met; (4d) If the order condition is met, extract all coded data packets with the same group number and order number j as the received coded data packet from its own receiving buffer queue; Step 5 includes: (5a) The destination determines the rank r of the coding coefficient matrix of all the j-th order coded data packets extracted in step 4. j (j=1,2,...,m),D j =(s j -s j-1 )-r j , if D j If ≤0, it is considered that the acceptance condition is not met, otherwise it is considered that the acceptance condition is met; (5b) The destination end feeds back to the source end the group number, order number j and degrees of freedom of the currently received coded data group Step 7 includes: (7a) The destination determines whether the total number of coded data packets extracted from the receiving buffer queue in step 6 is less than n+e. If so, it is confirmed that the number condition is met, where If it is not less than, it is confirmed that the number condition is not met; (7b) If the number condition is not met, the destination end calculates the rank r of the coding coefficient matrix in all the coded data packets of the current coded data packet extracted in step 6, D m+1 = kr, if D m+1 =0, it is determined that the decoding condition is met, otherwise, it is confirmed that the decoding condition is not met; (7c) If the decoding condition is met, decoding all the encoded data packets that meet the decoding condition; (7d) If the decoding condition is not met, the destination end feeds back the group number, order number m+1 and degrees of freedom of the current coded data group to the source end. Step 9 includes: The source receives the feedback information and extracts the group number, order number j and degree of freedom e from it. j , determine whether the number of retransmissions corresponding to the group number and order number j is less than the maximum number of retransmissions. If so, the retransmission process is performed; if not, the source end confirms the completion of the transmission of a group of original data packets.
2. A wireless transmission method based on adjustable sparse network coding according to claim 1, characterized in that: Step 2 includes: The source end groups P original data packets into groups of k. If the last group is less than k original data packets, it is filled with all-zero data packets. Then the P original data packets can be divided into Groups; The value of k ranges from 1 to the number of all original data packets to be sent.
3. The wireless transmission method based on adjustable sparse network coding according to claim 1, characterized in that: (7c) Includes: (7c1) The destination end extracts the coding coefficient matrix and the coding information packet from all the coded data packets of the current coded data packet extracted; (7c2) using the coding coefficient matrix and the coded information packet to perform an optimized Gaussian elimination decoding algorithm to obtain a set of original data packets; The input of the optimized Gaussian elimination decoding algorithm is the coding coefficient matrix G, also called the decoding matrix, with a dimension of (n+e)×k. The received coded information packet U=[u1 u2...u n+e ] T , output an original data packet P = [p1...p k ] T ,The optimized Gaussian elimination decoding algorithm uses a, z and r, where a=1:k represents the order of the original data packet, and the i-th element in the row vector a is represented by a i , z is a 1×k row vector, representing the column weight of the lower right corner of the decoding matrix, initialized to the column weight of the decoding matrix, and the i-th element in the row vector z is represented by z i , r is a row vector of (n+e)×1, representing the row weight of the decoding matrix, and is initialized to the row weight of the decoding matrix; the specific implementation steps of the optimized Gaussian elimination decoding algorithm are as follows: (11a) c←1, repeat (11b)(11c) until c=k, then execute (11d); (11b) Find the column whose value is 1 in z, denoted as C; (11b1) If C is not empty, for each element in C corresponding to the column y, swap the cth column of G with the yth column, the cth column of a with the yth column, and the cth column of z with the yth column; find the row q containing the non-zero elements from the cth row to the n+eth row in the cth column, swap the cth row of G with the qth row, the cth row of U with the qth row, and the cth row of r with the qth row; update the column weights, on the one hand, z c =0, on the other hand, the column weight of the column corresponding to the non-zero element on the right of the cth row and cth column is reduced by 1, c←c+1; otherwise, execute step (11b2); (11b2) If C is empty, find the row with the smallest weight from the cth row to the n+eth row in r. If there are multiple rows, randomly select one and record it as R; swap the cth row with the Rth row of G, the cth row with the Rth row of U, and the cth row with the Rth row of r; find the column where the non-zero elements in the cth row are located, record it as p, swap the cth column with the pth column of G, the cth column with the pth column of a, and the cth column with the pth column of z; perform Gaussian elimination on the cth column and update the corresponding row weight and column weight; (11c) Update column weight, c←c+1; (11d)p ai ←u i ,i=1,2,...,k。 4. The wireless transmission method based on adjustable sparse network coding according to claim 1, characterized in that: The retransmission process includes: (15a) Let the sparsity of a coded data packet be w. When the order number j≤m, for the current original data packet of order j, b(b≤e j ) The sparsity of the retransmitted coded packets w = min{w j-1 +n j +b,w j }; When the order number j = m + 1, the sparsity of all retransmitted coded data packets w = k; a coded information packet Where b = 1, 2, ..., e j , f(i) is randomly selected from the set {1,...,k}, and any i≠i', then f(i)≠f(i'),g b,f(i) , from F q \{0} is independently randomly selected and (15b) Generate a 1-row L-column all-zero row vector and set the value of the f(i)th element to g b,f(i) , get the coding coefficient vector of the b-th coded data packet, where L = k; (15c) splicing a coding information packet, a coding coefficient vector, and the group number and order number j to which the coding information packet belongs in columns to obtain a coding data packet, setting a flag in the coding data packet to mark whether the coding data packet is the last coding data packet of a certain order retransmission, if yes, the flag is set to 1, otherwise, it is set to 0, and adding the coding data packet to the head of the sending queue of the source end; (15d) Repeat (15a), (15b), (15c) until e is generated for the jth order of an original data packet. j retransmit the coded data packet, increase the number of retransmissions corresponding to the j-th order of the group by 1, and send the coded data packets in the sending queue to the wireless channel in sequence, with one coded data packet sent in each time slot.
