A collaborative data retransmission method based on instantly decomposable network coding

Abstract

The present invention belongs to the field of communications technology, and more particularly, relates to a collaborative data retransmission method based on instantly decodable network coding. During data packet broadcasting from a single base station to multiple user terminals, compared to base station data broadcasting methods based on instantly decodable network coding, the base station and D2D terminal data retransmission mechanism based on instantly decodable network coding proposed in the present invention effectively utilizes the parallel transmission capabilities of some user terminals that have successfully received data packets, and empowers the terminal or base station that completes data transmission first to further utilize additional data transmission opportunities, thereby effectively reducing the number of retransmissions of all data packets, average decoding delay, and system completion time.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a collaborative data retransmission method based on instantly decomposable network coding. Background Art

[0002] In a wireless cellular cell consisting of a single base station and M user terminals, all terminals expect to receive all N data packets broadcast by the base station. To effectively overcome the possibility of transmission errors during the data packet broadcast process, the base station can first broadcast all N data packets to all M terminals, then obtain feedback from each terminal on its data packet reception status. Finally, based on the data packet reception and loss information of all M terminals, the base station can repeatedly broadcast appropriate data packets.

[0003] In the traditional store-and-forward mode, the base station needs to retransmit the missing data packets of the M terminals one by one until each terminal reports that it has received all N data packets. Summary of the Invention

[0004] To address the above-mentioned issues, the present invention proposes a collaborative data retransmission method based on instantly resolvable network coding. This method utilizes instantly resolvable network coding and parallel direct transmission of data packets between adjacent user terminals in a cellular network to effectively reduce the number of retransmissions, average decoding delay, and system completion time required for cellular wireless network base stations to broadcast data packets under realistic transmission error environments. This reduces the number of searches performed by the maximum weighted clique search algorithm, thereby effectively shortening the time it takes for all terminals to receive all data packets and improving the throughput of base station broadcast data packets. In particular, during the data packet retransmission phase, since the optimal retransmission coding packet settings for base stations and user terminals typically require solving the maximum clique search problem in graph theory, the present invention incorporates delay-influencing factors such as each terminal's data packet reception status, user terminal pairs that can participate in D2D direct transmission, the network connection topology between different terminals, the base station-to-terminal data transmission rate, and the data transmission rate between different terminals into the construction of the instantly resolvable network coding graph. These factors are then comprehensively considered along with the graph vertex weights to determine the retransmission coding packet and its target terminal.

[0005] To facilitate the explanation and understanding of the technical solution of the present invention, the basic concepts and principles involved in the present invention are briefly introduced first:

[0006] In the D2D network wireless data transmission model based on instantaneous decomposable network coding, the source base station BS has N data packets of the same length p1, p2, ..., p N It needs to be broadcast in half-duplex mode to M single-antenna user terminals numbered 1, 2, ..., M, where the terminal set within the communication coverage of terminal i is K i, the wireless channel transmission rates from the base station to terminals 1, 2, ..., M are r1, r2, ..., r M , the connection status between each terminal and the wireless channel transmission rate constitute the matrix where c i,j Represents the wireless transmission rate from terminal i to terminal j, that is, c i,j =0 means that terminal i cannot directly transmit data to terminal j, and c i,j >0 means terminal i can reach the target at rate c i,j Transmit data directly to terminal j. For any i, j∈{1,2,…M}, the packet error rate (i.e., the probability of data packet transmission error) from the base station to each terminal i and the packet error rate from terminal i to terminal j are ρ s,i and ρ i,j Without loss of generality, it is assumed that all wireless channels have the same independent and identically distributed Rayleigh fading, the transmit power of each terminal is fixed, and the coverage range of each terminal is also constant.

[0007] The base station's broadcast of data packets is usually divided into two transmission phases. The first phase is the initial transmission phase, in which the base station sends N data packets to be transmitted, p1, p2, ..., p N It is broadcasted to M terminals in succession. Afterwards, each terminal will feed back the data packets it has correctly received to the base station. After the first transmission, each terminal i, where i∈[1,M], will maintain two sets of data packets. The first set is the Has set H i , used to record the data packet set owned by terminal i at the current stage, and the other set is the Want set W i , used to record the set of data packets that terminal i has not correctly received in the current phase. The second phase involves data packet retransmission by the base station and data packet exchange between terminals, helping any terminal that has not correctly received all data packets obtain its missing data packets. During this phase, because D2D direct transmission between user terminals utilizes out-of-band transmission based on the Industrial, Scientific, and Manufacturing (ISM) band, simultaneous data packet retransmissions by the base station and any terminal do not interfere with each other. However, D2D direct transmission between different terminals may cause co-channel interference.

[0008] During data packet transmission, it is assumed that both the base station and the terminal use the variable rate MQAM modulation technology described in "Goldsmith AJ, Chua SG. Variable-rate variable-power MQAM for fading channels[J]. IEEE transactions on communications, 1997, 45(10): 1218-1230". Specifically, in the MQAM technology, M=2 is taken, and the bit error rates from the base station to terminal i and from terminal i to terminal j are obtained as follows:

[0009]

[0010]

[0011] Where erfc is the error complementary function, and its expression is

[0012]

[0013] The bit error rate expression can be further deduced that the packet error rates from the base station to terminal i and from terminal i to terminal j are:

[0014] ρ s,i =1-(1-BER i ) B

[0015] ρ i,j =1-(1-BER i,j ) B

[0016] Where B represents the length of the transmitted data packet.

[0017] The transmission rate from the base station to terminal i and the transmission rate from terminal i to terminal j are defined as:

[0018]

[0019]

[0020] Where W represents the bandwidth transmitted from the base station to the target user, and W i,j Represents the frequency band width of the wireless channel between the transmitting terminal i and the receiving terminal j.

[0021] Definition 1. State Feedback Matrix (SFM): During the data packet retransmission process, the state of each terminal i in terms of the data packet can be represented by the matrix F = [f i,j ] M×N Indicates that if p j ∈Hi , then there is f i,j =0 holds; otherwise, if p j ∈W i , then there is f i,j =1 holds true.

[0022] Each coded packet retransmitted by the base station or terminal can be represented as The symbols represents the bitwise XOR operation between binary packets, and each data packet p i Each contains B bits.

[0023] Definition 2. Completion Time CT i It is the length of time that a single terminal i takes to receive all data packets broadcast by the base station.

[0024] Definition 3. System Completion Time (SCT) is the time it takes for all M terminals to receive all data packets broadcast by the base station. The present invention uses SCT to represent it, i.e.

[0025] SCT=max i∈D {CT i}

[0026] Definition 4. During the tth data packet retransmission, if terminal i receives the immediately decodable packet P * , then its decoding delay will remain unchanged; otherwise, the decoding delay of terminal i will increase by the time length T of the tth data packet retransmission. t .

[0027] Definition 5. Average decoding delay is the average decoding delay of all terminal i.

[0028] Definition 6. If the wireless channel transmission rate matrix C between M user terminals = [c i,j ] M×M The number of elements greater than 0 in is count, so the network connectivity of the D2D network is

[0029] Definition 7. For a vertex set T in a simple undirected graph G = (V, E), where and A set of vertices T is called a clique of an undirected graph G if there is an edge connecting any two vertices in T.

[0030] Definition 8. For a clique T of an undirected graph G = (V, E), if adding any vertex from the set V\T to T does not form a new clique, then clique T is a maximal clique of graph G.

[0031] Definition 9. Among all the maximal cliques in an undirected graph G, the maximal clique containing the largest number of vertices is called the maximum clique of graph G.

[0032] In order to generate appropriate coded packets based on data packet retransmission, it is necessary to construct the IDNC graph for the base station and D2D network using the following method:

[0033] (1) Design of IDNC graph vertices

[0034] For a base station with a channel transmission rate of r i Data packet p sent to terminal i j , where i∈[1,M], j∈[1,N] and r i ∈R (where R is the transmission rate vector from the base station to the terminal), the base station can generate a vertex of the IDNC graph On the other hand, for the data packet p sent by terminal m to terminal n at the sending rate c[m,n] q , the D2D network can generate a vertex v in the IDNC graph m,n,q,c[m,n] , where m,n∈[1,M], q∈[1,N] and c[m,n]∈C (where C is the terminal-to-terminal transmission rate matrix).

[0035] (2) Construction of IDNC graph edges

[0036] IDNC diagram G generated at the base station BS-IDNC Middle, Vertex With vertex An edge must meet one of the following two conditions:

[0037] Condition 1: i≠m, p n ∈W i and p j ∈W m ;

[0038] Condition 2: i≠m and j=n.

[0039] The IDNC graph G generated in the D2D network D2D-IDNC In the middle, vertex v m,n,q,c[m,n] With vertex v i,j,k,c[i,j] An edge must meet one of the following conditions:

[0040] Condition 3: m = i, n ≠ j, and q = k;

[0041] Condition 4: m = i, n ≠ j, p k ∈W n and p q ∈W j ;

[0042] Condition 5: m≠i, n≠j, and

[0043] (3) Design of vertex weights in IDNC graph

[0044] Finding the largest group in the IDNC graph during the retransmission phase can provide more transmission opportunities in the current time slot, which can make the system decoding delay as low as possible. Considering the need to minimize the completion time, it is necessary to find the largest weighted group in the IDNC graph and make the weight design of the vertices more conducive to the completion time. When designing the vertices of the IDNC graph, the base station vertex information should include the impact of factors such as the receiving terminal, lost data packets, and transmission rate on the completion time. Therefore, the vertices of the base station IDNC graph The weight can be determined by the following three factors.

[0045] a) Set the vertex The inverse of the corresponding transmission completion time is ω1, that is,

[0046]

[0047] The parameters B and r i Represents the packet length and the transmission rate from the base station to terminal i, respectively. This parameter setting takes into account the base station's transmission rate. When the encoded packet length is the same, a faster transmission rate consumes less time, thereby having the opportunity to reduce the overall completion time of the network.

[0048] b) Set the vertex The corresponding link transmission success rate is ω2, that is,

[0049]

[0050] The parameter ρ s,i Represents the packet error rate from the base station to terminal i. This parameter setting takes into account the link's transmission success rate. A higher link transmission success rate, that is, a lower packet error rate, reduces the number of retransmissions.

[0051] c) Set the vertex The corresponding rate r i The same packet p is lost with terminal i j The number of terminals is ω3, that is,

[0052]

[0053] The parameter f n,j ∈F represents whether terminal n loses data packet p in the state feedback matrix j This parameter setting takes into account the loss of data packets p jThe more terminals there are, the higher the priority of transmitting the data packet, thus effectively reducing the number of retransmissions and shortening the completion time. and , first find a r x can maximize their product. , then the weight value of the vertex is expressed as follows:

[0054]

[0055] For any edge e in the base station IDNC graph, if the edge connects two vertices and , then the weight of edge e is

[0056] Similarly, the weight values of the D2D network vertices are expressed as follows:

[0057] ω(v m,n,q,c[m,n] )=ω2(v m,n,q,c[m,n] )×max[ω1(v m,n,q,c[m,n] )×ω3(v m,n,q,c[m,n] )]

[0058] Among them, ω1(v m,n,q,c[m,n] ) represents the vertex v m,n,q,c[m,n] The corresponding inverse of the transmission completion time, ω2(v m,n,q,c[m,n] ) represents the link transmission success rate, ω3(v m,n,q,c[m,n] ) represents the vertex v m,n,q,c[m,n] Correspondingly, within the coverage of terminal m and at rate c[m,n], the same packet p is lost with terminal n. q The number of terminals.

[0059] For any edge e in the D2D network IDNC graph, if the edge connects two vertices v m,n,q,c[m,n] and v k,l,j,c[k,l] , then the weight value of edge e is ω(v m,n,q,c[m,n] )+ω(v k,l,j,c[k,l] ).

[0060] Based on the IDNC diagram of the above base station and D2D network, the technical solution to be adopted by the present invention is:

[0061] For a simple undirected IDNC graph with weights, a maximum weight clique can be searched based on the following algorithm.

[0062] Algorithm 1: Maximum weight clique search algorithm for IDNC graph G = (V, E) based on edge weights

[0063] a. Initialize the maximum clique U max is an empty set, and generates the adjacency matrix M of the IDNC graph G;

[0064] b. Search for the edge e with the largest weight in the edge set E. max ;

[0065] c. Move edge e max Connect two vertices v i and v k Join the largest group U max Delete vertex v from vertex set V i and v k , and delete e from the edge set E max ;

[0066] d. Delete all points from the point set V that are related to v. i Not adjacent to v k Non-adjacent vertices and delete all edges connected to these vertices from the edge set E;

[0067] e. Delete all points from the point set V that are related to v. i and v k Vertices that are not adjacent at the same time, and all edges connected to these vertices are deleted from the edge set E;

[0068] f. Find all the items with v i and v k At the same time, adjacent vertices are deleted from the edge set E and the vertices are connected to v i The edges connected to them and their relationship with v k connected edges;

[0069] g. If the edge set E is an empty set, then add the vertex with the largest weight in the vertex set V to the maximum clique U max In the process, stop the algorithm execution and output the largest group U max ; Otherwise, return b;

[0070] For a D2D network with one base station, M terminals, and N data packets to be transmitted, a transmission scheme can be set for the base station and terminals based on the following algorithm, which includes the set of target terminals, the coded packets, and the transmission rate from the sender to the target terminals.

[0071] Algorithm 2: Data retransmission algorithm between base stations and D2D terminals based on IDNC

[0072] S1. The base station generates a state feedback matrix SFM based on user feedback information and initializes the data retransmission round to s = 1;

[0073] S2. At the beginning of the sth round of data retransmission, the base station generates a vertex set V and a base station IDNC graph G based on SFM and the rate from the base station to each terminal. BS-IDNC , and is the graph G BS-IDNCCalculate the weight of each vertex in the base station vertex set V and the base station IDNC graph G BS-IDNC Bring it into Algorithm 1 to obtain the maximum group if Then it means that all lost data packets have been recovered and the algorithm execution ends; otherwise, in the vertex set V and graph G BS-IDNC Delete the vertex set U from max , change the terminal x i ,x i+1 ,…,x i+m Set as the target user for base station transmission, XOR to get the encoded packet And set the sending rate to The length of time for base station data transmission in the sth round of data retransmission is T BS =B / r BS , where B is the length of the coded packet;

[0074] S3. The base station updates its state feedback matrix SFM according to the latest vertex set V. Based on the updated state feedback matrix SFM, the base station generates the vertex set V of the D2D network D2D In the vertex set V D2D Delete all the items that satisfy x j ∈{x i ,x i+1 ,…,x i+m} or y j ∈{x i ,x i+1 ,…,x i+m}vertices And generate IDNC graph G for subsequent D2D network transmission D2D-IDNC ;

[0075] S4, the transmission vertex set V of the D2D network D2D and IDNC Figure G D2D-IDNC Bring it into Algorithm 1 to obtain the maximum group if Then there is no D2D transmission opportunity in the sth round. The base station updates the state feedback matrix SFM according to the information fed back by all terminals, updates s=s+1, and returns S2; otherwise, from the vertex set V D2D Delete U D2D-max , and the set U D2D-max The l transmitting terminals participating in D2D transmission are divided into l different subsets, namely as well as

[0076] S5, terminal x1 sends at a rate Broadcast packets Give terminal y 11 ,y 12 ,…,y 1a , terminal x2 sends at a rate Broadcast packets Give terminal y 21 ,y 22 ,…,y 2b ,…, terminal x l At the sending rate Sending Data Packets Give terminal y l1 ,y l2 ,…,y lh .

[0077] S6. The base station sets the D2D transmission time of the sth round of data retransmission to And the state feedback matrix SFM is updated according to the information fed back by all terminals;

[0078] S7, according to the idle time |T D2D -T BS |, continue searching for additional transmission opportunities in the sth round of data retransmission:

[0079] S71. For each terminal x i , if in the vertex set V D2D There is at least one vertex in satisfy Then the terminal x i will be at a rate c[x i ,y] transmits data packet z to terminal y;

[0080] S72: The base station updates the state feedback matrix SFM according to the information fed back by all terminals, and searches for additional transmission opportunities of the base station in the following manner:

[0081] S721, if T D2D ≤T BS , then the base station has no additional transmission opportunity, updates s=s+1, and returns to S2; otherwise, jumps to S722;

[0082] S722. Delete the vertex set V that meets the condition B / r. g >T D2D -T BS All vertices of If the vertex set Then the base station has no additional transmission opportunity, updates s=s+1, and returns to S2; otherwise, the base station needs to bring the vertex set V and the base station IDNC graph into Algorithm 1 to obtain the maximum clique Data packets XOR to get the encoded packet and at a rate Send the coded packet to terminal m i ,m i+1 ,…,m i+k , so that the base station can D2D -T BS Complete the transfer within.

[0083] S723: After the additional data transmission of the base station is completed, each target terminal feeds back the coded packet reception status to the base station, so that the base station can update its state feedback matrix SFM.

[0084] S724. Update s=s+1 and return to S2.

[0085] The beneficial effects of the present invention are:

[0086] In the process of broadcasting data packets from a single base station to multiple user terminals, compared with the base station data broadcasting method based on instantly decodable network coding, the base station and D2D terminal data retransmission mechanism based on instantly decodable network coding proposed in the present invention effectively utilizes the parallel transmission capabilities of some user terminals that have successfully received data packets, and empowers the terminal or base station that completes data transmission first to further utilize additional data transmission opportunities, thereby effectively reducing the number of retransmissions of all data packets, the average decoding delay and the system completion time.

[0087] Compared with the weighted maximum clique search algorithm based on point weights in “Wang Meng. Research on retransmission scheme based on immediately decomposable network coding in D2D networks [D]. Chongqing University of Posts and Telecommunications, 2018”, the weighted maximum clique search algorithm based on edge weights proposed in this invention effectively improves the efficiency of vertex weight search on the basis of assigning appropriate weights to each edge of the IDNC graph of the base station and the D2D network. In particular, when the maximum number of vertices in an IDNC graph is m, the number of searches required by the weighted maximum clique search algorithm based on point weights does not exceed m, while the number of searches required by the weighted maximum clique search algorithm based on edge weights does not exceed m. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a state feedback matrix SFM after a base station of a D2D network with 5 terminals sends 4 data packets.

[0089] Figure 2 The present invention is based on Figure 1The state feedback matrix shown is generated by the base station IDNC diagram.

[0090] Figure 3 (a) Base station IDNC graph G generated by data retransmission based on Algorithm 2 of the present invention BS-IDNC and (b) Figure G BS-IDNC Example of the largest weighted clique in .

[0091] Figure 4 (a) D2D direct transmission IDNC graph G generated by data retransmission based on Algorithm 2 of the present invention D2D-IDNC Figure and (b) Figure G D2D-IDNC Example of the largest weighted clique in .

[0092] Figure 5 This is a schematic diagram showing how the number of searches for the edge-weight-based maximum weight clique search algorithm proposed in the present invention and the point-weight-based maximum weight clique search algorithm proposed in the literature [2] changes with the number of vertices in the IDNC graph.

[0093] Figure 6 Yes, when the number of terminals M∈[5,50], the number of data packets N=10, the data packet length B=1Mbits, and its network connectivity is 0.6, a comparative diagram of the base station and D2D terminal data retransmission method based on edge weight search, the base station and D2D terminal data retransmission method based on point weight search, the base station data retransmission method based on edge weight search, and the base station data retransmission method based on point weight search as the number of terminals changes.

[0094] Figure 7 Yes, when the number of terminals M∈[5,50], the number of data packets N=10, the data packet length B=1Mbits, and its network connectivity is 0.6, a comparative diagram of the average decoding delay of the base station and D2D terminal data retransmission method based on edge weight search, the base station and D2D terminal data retransmission method based on point weight search, the base station data retransmission method based on edge weight search, and the base station data retransmission method based on point weight search as the number of terminals changes.

[0095] Figure 8 Yes, when the number of terminals M∈[5,50], the number of data packets N=10, the data packet length B=1Mbits, and its network connectivity is 0.6, a comparative diagram of the system completion time of the base station and D2D terminal data retransmission method based on edge weight search, the base station and D2D terminal data retransmission method based on point weight search, the base station data retransmission method based on edge weight search, and the base station data retransmission method based on point weight search as the number of terminals changes. DETAILED DESCRIPTION

[0096] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0097] Example 1

[0098] In this embodiment, consider a base station of a D2D network including 5 terminals sending 4 data packets, such as Figure 1 The state feedback matrix SFM of this example is shown. When the channel transmission rates from the base station to terminals 1, 2, 3, 4 and 5 are 4kbps, 2kbps, 1kbps, 5kbps and 4kbps respectively, it can be based on Figure 1 And vertex weights are generated as Figure 2 The IDNC graph G=(V,E) shown in the figure, where V={v 1,1,4 ,v 5,1,4 ,v 1,3,4 ,v 3,2,1 ,v 4,1,5 ,v 5,4,4} and E={(v 1,1,4 ,v 3,2,1 ),(v 1,1,4 ,v 4,1,5 ),(v 5,1,4 ,v 3,2,1 ),(v 5,1,4 ,v 4,1,5 ),(v 3,2,1 ,v 4,1,5 ),(v 1,3,4 ,v 5,4,4 ),(v 1,3,4 ,v 3,2,1 ),(v 5,4,4 ,v 4,1,5 ),(v 5,4,4 ,v 3,2,1 ),(v 1,1,4 ,v 5,1,4 )}.

[0099] against Figure 2 The steps for obtaining the maximum weighted cluster using the method of the present invention are as follows:

[0100] S1, the largest group U max Initialize it to an empty set and generate the following adjacency matrix M based on the IDNC graph, where the 1st, 2nd, ..., 6th rows of the matrix M correspond to the vertices v of the IDNC graph respectively. 1,1,4 ,v 1,3,4 ,v 3,2,1 ,v 4,1,5 ,v 5,1,4 ,v 5,4,4 , and the same is true for the 1st, 2nd,…, 6th columns of the matrix M.

[0101]

[0102] S2. Since the edge set E is not empty, the edge with the largest weight in E is selected, that is, e max =(v 1,1,4 ,v 5,1,4 );

[0103] S3, e max Connect two vertices v 1,1,4 and v 5,1,4 Add to the maximum clique U max and update U max ={v 1,1,4 ,v 5,1,4},V=V\{v 1,1,4 ,v 5,1,4}={v 1,3,4 ,v 3,2,1 ,v 4,1,5 ,v 5,4,4}, and E=E\{e max}={(v 1,1,4 ,v 3,2,1 ),(v 1,1,4 ,v 4,1,5 ),(v 5,1,4 ,v 3,2,1 ),(v 5,1,4 ,v 4,1,5 ),(v 3,2,1 ,v 4,1,5 ),(v 1,3,4 ,v 5,4,4 ),(v 1,3,4 ,v 3,2,1 ),(v 5,4,4 ,v 4,1,5 ),(v 5,4,4 ,v 3,2,1 )};

[0104] S4. Delete all points in the point set V that are related to v. 1,1,4 Not adjacent to v 5,1,4 Non-adjacent vertices, i.e. v 1,3,4 and v 5,4,4 , and delete all edges connected to these vertices in the edge set E. Thus, update the vertex set V = V\{v 1,3,4 ,v 5,4,4}={v 3,2,1 ,v 4,1,5} and edge set E=E\{(v 1,3,4 ,v 5,4,4 ),(v 1,3,4 ,v 3,2,1 ),(v 5,4,4 ,v 4,1,5 ),(v 5,4,4 ,v 3,2,1 )}={(v 1,1,4,v 3,2,1 ),(v 1,1,4 ,v 4,1,5 ),(v 5,1,4 ,v 3,2,1 ),(v 5,1,4 ,v 4,1,5 ),(v 3,2,1 ,v 4,1,5 )};

[0105] S5. Delete all points in the point set V that are related to v. 1,1,4 and v 5,1,4 Vertices that are not simultaneously adjacent, and all edges connected to these vertices are deleted from the edge set E. Since there are no vertices and edges that meet these conditions at this time, the sets V and E remain unchanged;

[0106] S6. Find all the items with v 1,1,4 and v 5,1,4 At the same time, the adjacent vertex is v 3,2,1 and v 4,1,5 , and delete v from the edge set E 3,2,1 or v 4,1,5 With v 1,1,4 or v 5,1,4 All connected edges, that is, (v 3,2,1 ,v 1,1,4 ), (v 4,1,5 ,v 1,1,4 ), (v 3,2,1 ,v 5,1,4 ), (v 4,1,5 ,v 5,1,4 ). Therefore, the updated set E is {(v 3,2,1 ,v 4,1,5 )};

[0107] S7. Since the edge set E and the vertex set V are not empty at this time, return to S2.

[0108] S2. Since the edge set E is not empty, take out the edge with the largest weight in the set E, that is, e max ={(v 3,2,1 ,v 4,1,5 )};

[0109] S3, e max Connect two vertices v 3,2,1 and v 4,1,5 Add to the maximum clique U max and update U max ={v 1,1,4 ,v 5,1,4 ,v 3,2,1 ,v 4,1,5}, as well as

[0110] S4. At this time, both the vertex set V and the edge set E are empty and do not need to be deleted;

[0111] S5. At this time, both the vertex set V and the edge set E are empty and do not need to be deleted;

[0112] S6. At this time, both the vertex set V and the edge set E are empty and do not need to be deleted;

[0113] S7: At this point, both the vertex set V and the edge set E are empty, so the algorithm stops executing and the maximum clique U is output. max ={v 1,1,4 ,v 5,1,4 ,v 3,2,1 ,v 4,1,5}.

[0114] Example 2

[0115] In this example, a D2D network with 9 terminals sends 4 data packets. The state feedback matrix SFM and the channel transmission rate matrix C between the 9 terminals in the D2D network are shown below.

[0116]

[0117] When the channel transmission rates from the base station to terminals 1, 2, ..., 9 are 4 kbps, 2 kbps, 1 kbps, 5 kbps, 1 kbps, 5 kbps, 8 kbps, 1 kbps, and 7 kbps respectively, the specific steps of the base station and D2D network retransmission strategy generated by Algorithm 2 of the present invention are as follows:

[0118] S1. The base station initializes the state feedback matrix SFM according to the information fed back by the user, and initializes the data retransmission round to s=1.

[0119] S2. At the beginning of the first round of data retransmission, the base station generates a vertex set V = {v 1,1,4 ,v 3,2,1 ,v 4,3,5 ,v 5,4,1 ,v 6,1,5 ,v 7,1,8 ,v 7,2,8 ,v 8,3,1 ,v 8,4,1} and Figure 3 Base station IDNC diagram G shown in (a) BS-IDNC . The vertex sets V and G BS-IDNC Substitute into Algorithm 1 to obtain Figure 3 The largest clique U shown in (b) max ={v 1,1,4 ,v 4,3,5 ,v 5,4,1,v 6,1,5 ,v 7,1,8}.

[0120] because So from the vertex set V and graph G BS-IDNC Delete the vertex set U from max , set terminals 1, 4, 5, 6 and 7 as the target users of the base station transmission, perform XOR on data packets p1, p3 and p4 to obtain the coded packet And set the sending rate to r BS =min{r1,r4,r5,r6,r7}=1kbps. Under this setting, the duration of this round of base station data transmission is T BS =B / r BS = 0.1s, where B = 100 bits;

[0121] S3. The base station updates its state feedback matrix SFM according to the latest vertex set V as follows.

[0122]

[0123] According to the updated state feedback matrix SFM and the channel transmission rate matrix C between the 9 terminals, the base station generates the vertex set V of the D2D network IDNC graph in the following way D2D For example, in SFM, f 3,2 =1 means that terminal 3 still misses data packet p2, then in the 3rd column of the channel transmission rate matrix C, find the rows with transmission rate greater than 0, i.e. rows 2 and 7. 7,3 =1, so terminal 7 cannot recover data packet p2 for terminal 3, while terminal 2 can transmit data packet p2 to terminal 3 at a rate of 2kbps, so a V D2D The vertex v 2,3,2,2 . Using a similar method, V D2D The remaining vertices are v 1,7,2,1 ,v 2,7,2,1 ,v 1,8,3,2 ,v 9,7,2,5 ,v 9,8,3,3 ,v 9,8,4,3 Therefore, the initial V D2D For {v 2,3,2,2 ,v 1,7,2,1 ,v 2,7,2,1 ,v 1,8,3,2 ,v 9,7,2,5 ,v 9,8,3,3 ,v 9,8,4,3}.

[0124] In V D2D Delete all the items that satisfy xj ∈{1,4,5,6,7} or y j Vertices ∈{1,4,5,6,7} That is v 1,8,3,2 ,v 1,7,2,1 ,v 2,7,2,1 ,v 9,7,2,5 , thereby updating V D2D ={v 2,3,2,2 ,v 9,8,3,3 ,v 9,8,4,3}, and generate Figure 4 The D2D network IDNC diagram G shown in (a) D2D-IDNC ;

[0125] S4, in the vertex set V D2D and IDNC Figure G D2D-IDNC Based on Algorithm 1, the following is generated: Figure 4 The largest clique U shown in (b) D2D-max ={v 2,3,2,2 ,v 9,8,3,3}. From the vertex set V D2D Delete U D2D-max , and U D2D-max All vertices in are classified according to the sending terminal numbers participating in D2D transmission, and are divided into two categories: 2,3,2,2} and {v 9,8,3,3}.

[0126] S5. According to the above classification, terminal 2 sends data packet p2 to terminal 3 at a sending rate of c[2,3]=2kbps, and terminal 9 sends data packet p3 to terminal 8 at a sending rate of c[9,8]=3kbps.

[0127] S6: Set the transmission time of the D2D terminal to T D2D =B / min{c[2,3],c[9,8]}=0.05s. Terminals 3 and 8 provide feedback to the base station on whether they have received the required data packets correctly. The base station updates the state feedback matrix SFM based on the feedback information.

[0128] S7, according to the idle time |T D2D -T BS | = 0.05s, continue searching for additional data retransmission opportunities:

[0129] S7.1. D2D terminal transmission opportunity search: traverse U D2D-max For example, for terminal 2, due to its transmission time t 2,3 =B / c[2,3]=0.05s, and in the vertex set V D2D There is no terminal y that satisfies B / c[2,y]≤max(TD2D ,T BS )-t 2,3 = 0.05s, so terminal 2 has no chance to continue transmitting. For terminal 9, since its transmission time t 9,8 =B / c[9,8]=0.033s, and in the vertex set V D2D There is a terminal 8 that satisfies B / c[9,8]=0.033s≤max(T D2D ,T BS )-t 9,8 =0.067s, so terminal 9 has the opportunity to continue transmitting data, that is, terminal 9 transmits data packet p4 to terminal 8 at rate c[9,8]=3kbps;

[0130] S7.2, the base station updates the state feedback matrix SFM according to the information fed back by all terminals, and searches for additional transmission opportunities of the base station based on the following method: D2D ≤T BS Therefore, the base station has no chance to continue retransmitting, updates s=s+1=2, and jumps to S2.

[0131] Example 3

[0132] In this example, randomly generated An adjacency matrix is used to represent the specific edge connections of the IDNC graph with n vertices. At the same time, the weight ω of each vertex is uniformly distributed in the interval [0.1, 2]. Figure 5 A diagram shows how the search times for both vertex-weighted and edge-weighted maximum-weight clique search algorithms vary with the number of vertices in the IDNC graph. The diagram shows that the search times for both algorithms increase as the number of vertices in the IDNC graph increases, while the edge-weighted maximum-weight clique search algorithm performs fewer searches than the vertex-weighted maximum-weight clique search algorithm. This is because the vertex-weighted search algorithm searches only one vertex in the IDNC graph during a single search, while the edge-weighted search algorithm searches two vertices connected by an edge during each search, resulting in a lower average search time than the vertex-weighted search algorithm.

[0133] In the D2D network data packet retransmission mode based on Instantaneous Decipherable Network Coding (IDNC), the base station can select several missing data packets and perform binary XOR-based encoding to generate coded packets to be sent. It can also help terminals with missing data packets recover their missing data packets by performing binary XOR decoding with the data packets they have correctly received, by retransmitting the coded packets themselves or specifying certain terminals to retransmit the coded packets.

[0134] Compared to traditional store-and-forward methods, the benefits of D2D network packet retransmission based on instantly resolvable network coding include helping more terminals successfully decode their missing packets with fewer coded packet transmissions, effectively shortening the time required for the base station to successfully broadcast a packet. Furthermore, allowing the base station and some terminals to simultaneously send coded packets to terminals with missing packets fully utilizes the frequency and spatial resources of the cellular cell, further shortening the time required for all terminals to receive all N packets. The reduction in packet transmission latency also helps improve the overall data transmission throughput of the wireless cellular cell. Furthermore, compared to the matrix inversion and multiplication operations required for packet retransmission decoding based on linear or random network coding, the D2D network packet retransmission based on instantly resolvable network coding involves only simple decoding calculations based on binary XOR, significantly reducing the computational complexity required for the receiving terminal to recover the packet and improving the practicality of the packet retransmission scheme.

Claims

1. A collaborative data retransmission method based on instantly decodable network coding is proposed. The D2D network is assumed to include one base station, M terminals, and N data packets to be transmitted. The length of the data packet is B. The set of terminals within the communication coverage of terminal i is defined as K. i , the wireless channel transmission rates from the base station to terminals 1, 2, ..., M are r1, r2, ..., r M , the connection status between each terminal and the wireless channel transmission rate form a matrix where c i,j Represents the wireless transmission rate from terminal i to terminal j, that is, c i,j =0 means that terminal i cannot directly transmit data to terminal j, and c i,j >0 means terminal i can reach the target at rate c i,j Transmit data directly to terminal j; for any i,j∈{1,2,…M}, the packet error rate from the base station to each terminal i and the packet error rate from terminal i to terminal j are ρ s,i and ρ i,j ; The base station's broadcast of data packets is divided into two transmission phases. The first phase is the initial transmission phase, in which the base station sends N data packets to be transmitted, p1, p2, ..., p N It is broadcasted to M terminals in succession, and then each terminal feeds back the data packets it has correctly received to the base station; after the first transmission, each terminal i will maintain two sets of data packets, i∈[1,M], the first set is the Has set H i , used to record the data packet set owned by terminal i in the current stage, and the second set is the Want set W i , used to record the set of data packets that terminal i has not correctly received in the current stage; characterized in that, The data retransmission method comprises the following steps: S1, the base station generates a state feedback matrix SFM based on the user feedback information, and initializes the data retransmission round to s = 1; the definition of the state feedback matrix SFM is: during the data packet retransmission process, the data packet status of each terminal i is expressed by the matrix F = [f i,j ] M×N Indicates that if p j ∈H i , then f i,j =0; if p j ∈W i , then f i,j =1; S2. At the beginning of the sth round of data retransmission, the base station generates a vertex set V and a base station IDNC graph G based on SFM and the rate from the base station to each terminal. BS-IDNC , specifically: for the base station with a channel transmission rate of r i Data packet p sent to terminal i j , the base station generates a vertex of the IDNC graph where i∈[1,M], j∈[1,N] and r i ∈RR is the transmission rate vector from the base station to the terminal, and the IDNC graph G generated by the base station BS-IDNC Middle, Vertex With vertex Edges must satisfy the conditions i≠m, p n ∈W i and p j ∈W m Or i≠m and j=n; For graph G BS-IDNC The weight of each vertex is calculated as follows: The weight of edge e in the base station IDNC graph is Using the base station vertex set V and base station IDNC graph G BS-IDNC Calculate the maximum clique The maximum clique is defined as a vertex set T in a simple undirected graph G = (V, E) where and If there is an edge connecting any two vertices in T, then the vertex set T is called a clique of the undirected graph G. If adding any vertex from the set V\T to T does not form a new clique, then the clique T is a maximal clique of the graph G. Among all the maximal cliques in the undirected graph G, the maximal clique with the largest number of vertices is called the maximum clique of the graph G. If All lost data packets have been recovered and the data retransmission process ends; otherwise, in the vertex set V and graph G BS-IDNC Delete the vertex set U from max , change the terminal x i ,x i+1 ,…,x i+m Set as the target user for base station transmission, XOR to get the encoded packet Set the sending rate to The length of time for base station data transmission in the sth round of data retransmission is T BS =B / r BS ; Maximum group U max The calculation method is: a. Initialize the maximum clique U max is an empty set, and generates the adjacency matrix M of the IDNC graph G; b. Search for the edge e with the largest weight in the edge set E. max ; c. Move edge e max Connect two vertices v i and v k Join the largest group U max Delete vertex v from vertex set V i and v k , and delete e from the edge set E max ; d. Delete all points from the point set V that are related to v. i Not adjacent to v k Non-adjacent vertices and delete all edges connected to these vertices from the edge set E; e. Delete all points from the point set V that are related to v. i and v k Vertices that are not adjacent at the same time, and all edges connected to these vertices are deleted from the edge set E; f. Find all the items with v i and v k At the same time, adjacent vertices are deleted from the edge set E and the vertices are connected to v i The edges connected to them and their relationship with v k connected edges; g. If the edge set E is an empty set, then add the vertex with the largest weight in the vertex set V to the maximum clique U max In the process, stop the algorithm execution and output the largest group U max ; Otherwise, return b; S3. The base station updates its state feedback matrix SFM according to the latest vertex set V. Based on the updated state feedback matrix SFM, the base station generates the vertex set V of the D2D network D2D , in the vertex set V D2D Delete all the items that satisfy x j ∈{x i ,x i+1 ,…,x i+m } or y j ∈{x i ,x i+1 ,…,x i+m }vertices And generate IDNC graph G for subsequent D2D network transmission D2D-IDNC Specifically, the data packet p sent by terminal m to terminal n at the sending rate c[m,n] q , the D2D network generates a vertex v in the IDNC graph m,n,q,c[m,n] , where m,n∈[1,M], q∈[1,N] and c[m,n]∈C, C is the terminal-to-terminal transmission rate matrix, and the IDNC graph G generated in the D2D network D2D-IDNC In the middle, vertex v m,n,q,c[m,n] With vertex v i,j,k,c[i,j] Edges must satisfy the conditions m=i, n≠j and q=k or m=i, n≠j, p k ∈W n and p q ∈W j or m≠i, n≠j, and Figure G D2D-IDNC The weight values of the vertices are: ω(v m,n,q,c[m,n] )=ω2(v m,n,q,c[m,n] )×max[ω1(v m,n,q,c[m,n] )×ω3(v m,n,q,c[m,n] )] Among them, ω1(v m,n,q,c[m,n] ) represents the vertex v m,n,q,c[m,n] The corresponding inverse of the transmission completion time, ω2(v m,n,q,c[m,n] ) represents the link transmission success rate, ω3(v m,n,q,c[m,n] ) represents the vertex v m,n,q,c[m,n] Correspondingly, within the coverage of terminal m and at rate c[m,n], the same packet p is lost with terminal n. q The number of terminals, the weight of edge e is ω(v m,n,q,c[m,n] )+ω(v k,l,j,c[k,l] ); S4, use the largest group U in S2 max The calculation method uses the transmission vertex set V of the D2D network D2D and IDNC Figure G D2D-IDNC Get the largest group if Then there is no D2D transmission opportunity in the sth round. The base station updates the state feedback matrix SFM according to the information fed back by all terminals, updates s=s+1, and returns S2; otherwise, from the vertex set V D2D Delete U D2D-max , and the set U D2D-max The l transmitting terminals participating in D2D transmission are divided into l different subsets, namely as well as S5, terminal x1 sends at a rate Broadcast packets Give terminal y 11 ,y 12 ,…,y 1a , terminal x2 sends at a rate Broadcast packets Give terminal y 21 ,y 22 ,…,y 2b ,…, terminal x l At the sending rate Sending Data Packets Give terminal y l1 ,y l2 ,…,y lh ; S6. The base station sets the D2D transmission time of the sth round of data retransmission to And the state feedback matrix SFM is updated according to the information fed back by all terminals; S7, according to the idle time |T D2D -T BS |, continue searching for additional transmission opportunities in the sth round of data retransmission: S71. For each terminal x i , if in the vertex set V D2D There is at least one vertex in satisfy Terminal x i At the rate c[x i ,y] transmits data packet z to terminal y; S72: The base station updates the state feedback matrix SFM according to the information fed back by all terminals, and searches for additional transmission opportunities of the base station in the following manner: S721, if T D2D ≤T BS , the base station has no additional transmission opportunity, updates s=s+1, and returns to S2; otherwise, jumps to S722; S722. Delete the vertex set V that meets the condition B / r. g >T D2D -T BS All vertices of If the vertex set If the base station has no additional transmission opportunities, it updates s=s+1 and returns to S2; otherwise, the base station brings the vertex set V and the base station IDNC graph into the maximum clique U. max The calculation method of the maximum cluster is obtained Data packets XOR to get the encoded packet and at a rate Send the coded packet to terminal m i ,m i+1 ,…,m i+k , so that the base station at time T D2D -T BS Complete the transmission within S723. After the additional data transmission of the base station is completed, each target terminal feeds back the coded packet reception status to the base station, so that the base station updates its state feedback matrix SFM. S724. Update s=s+1 and return to S2.

Images