End-to-end reliable transmission method for supporting drone swarm swarm network

Abstract

The application discloses an end-to-end reliable transmission method supporting a UAV cluster swarm network, and mainly solves the problem of poor transmission performance of a traditional end-to-end reliable transmission control protocol in a high dynamic scene of the UAV cluster swarm network. The application has the following characteristics: 1) the no-rate characteristic of a fountain code is utilized on a poor link, so that the protocol does not need to perform a timeout retransmission after packet loss; 2) a congestion detection and control mechanism is introduced, and a transmission timeout is not used to judge network congestion; 3) a node adjusts a degree distribution mode of the fountain code, a relay node forwarding mode and coding redundancy according to a network congestion degree and a link state; and 4) a current congestion index is used to adjust a pipeline number and a coding window size, so as to perform congestion control. The application can further improve a transmission rate and maintain service continuity in the high dynamic scene.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to an end-to-end reliable transmission method supporting a UAV cluster swarm network. BACKGROUND

[0002] In a UAV cluster swarm network, reliable end-to-end data transmission is a core problem. These networks usually work in environments without fixed infrastructure support, and nodes dynamically establish connections through wireless communication. Due to the mobility of nodes and the instability of wireless channels, the network topology changes frequently, causing traditional transmission protocols to perform poorly in such environments. For example, the TCP protocol was designed primarily for wired networks, and its congestion control mechanism reduces the sending rate of data when encountering packet loss to avoid network congestion. However, in a UAV cluster swarm network, packet loss is not always caused by congestion, but also by routing failure or signal interference caused by node movement. Therefore, it is incorrect to consider packet loss caused by these non-congestion factors as network congestion, which will trigger congestion control frequently, causing reduced throughput and increased transmission delay. In addition, the per-packet end-to-end acknowledgment mechanism is inefficient in multi-hop networks, because each intermediate node needs to wait for the acknowledgment of the downstream node, which will cause serious delay and efficiency problems in networks with rapidly changing network topology. Solving this problem, i.e., how to achieve efficient and reliable end-to-end data transmission in a UAV cluster swarm network, has important practical value for improving network performance and meeting critical communication needs. In particular, in hostile environments or emergency situations, such as military communication and disaster rescue, reliable data transmission is crucial for ensuring the success of tasks and the safety of personnel. Therefore, researching and developing an end-to-end reliable transmission method suitable for the characteristics of a UAV cluster swarm network has important theoretical and application value for improving the practicality and reliability of the network.

[0003] Sreenivasa B.C and G.C.Bhanu Prakash in their published paper "Linkdb-TCP: A Congestion Control Technique For MANET Based On Link Layer Measurements" proposed a congestion control method based on link layer measurements, called Linkdb-TCP. The steps of this method are as follows: (1) the receiver predicts congestion by analyzing network delay, calculates network delay using information provided by the sender, and makes congestion prediction based on this. (2) By collecting the delay history of the last ten packets, the average delay is calculated and compared with the previous delay to predict the congestion in the network. (3) Calculate the Smoothed Round Trip Time (SRTT) and use it as an indicator of network congestion to control transmission parameters. (4) Adjust the congestion window (cwnd) according to the SRTT value, packet size, transmission rate and congestion window value of the previous packet to respond to the congestion in the network. However, due to the frequent changes in the topology of the communication nodes, the dynamic nature and the unreliability of the link layer, it will affect the accuracy of the congestion prediction. At the same time, this method relies on the feedback of the receiver to adjust the behavior of the sender, which may not be timely enough in high delay or high packet loss rate environments. In addition, Linkdb-TCP behaves similarly to TCP when the network conditions are good, but when congestion occurs, it collects link parameters by sending special packets called Linkdb, which may increase the additional overhead and affect the network performance.

[0004] Noor Mast et al. in their published paper "A Cross-Layer Solution for Contention Control to Enhance TCP Performance in Wireless Ad-Hoc Networks" propose a cross-layer solution called CSCC (Cross-layer Solution for Contention Control) aimed at enhancing TCP performance in networks without fixed infrastructure. The steps of this method are as follows: (1) Each node calculates the weighted moving average (WMA) of the number of attempts to access the medium at the MAC layer to estimate channel contention. (2) When the WMA reaches a pre-set threshold CCThresh, the node begins marking data packets to inform the source node of the presence of contention in the network. (3) After receiving the contention notification, the source node adjusts its congestion window (cwnd) size to control the proportion of data packets injected into the network. (4) To ensure fairness for each flow, flows with larger cwnds are punished more. However, since the calculation of WMA relies on the selection of the alpha value, this can lead to insufficient sensitivity or excessive sensitivity to network contention conditions, affecting network performance. In addition, the CSCC mechanism may face challenges in high mobility environments, as the rapid movement of nodes can lead to inaccurate contention estimates, affecting the congestion control effect of TCP.

[0005] In their published paper "Fountain Code based Hop-by-Hop Reliable Data Transmission Scheme in Multi-Hop MANETs", Changdong Li et al. combined the digital fountain code technology with the computing resources of relay nodes and proposed a reliable transmission scheme called Fountain code-based Hop-by-Hop Reliable Data Transfer (FHRDT). The steps of this method are as follows: (1) At the source node, the original data is grouped and encoded using fountain codes to form encoded data packets. (2) The encoded data packets are transmitted hop by hop through the multi-hop network to the destination node. (3) At the intermediate nodes, the network layer performs re-encoding operations, while the transport layer attempts to decode the data blocks and generates new re-encoded data packets and sends them to the next hop if not fully decoded. (4) At the destination node, fountain decoding is performed, and after successful decoding, the source node is notified to release the cache through an end-to-end mechanism, and the previous node is notified to start transmitting the next data block. However, the FHRDT scheme may introduce additional computational and storage overhead due to the re-encoding operation at the intermediate nodes, especially in a highly dynamic multi-hop network environment. The rapid movement of nodes may cause the re-encoded data packets to be generated in time or accurately, affecting the reliability and efficiency of transmission. In addition, the one-hop acknowledgement mechanism used in the FHRDT scheme reduces the waiting delay but increases the control overhead, especially when the data packet loss rate is high, which may require more feedback messages, thereby increasing the burden on the network. SUMMARY

[0006] To solve the above problems existing in the prior art, the present application provides an end-to-end reliable transmission method supporting a UAV cluster swarm network. The technical problem to be solved by the present application is solved by the following technical scheme:

[0007] An end-to-end reliable transmission method supporting a UAV cluster swarm network includes:

[0008] S100, in the connection establishment stage, the source node and the destination node establish a communication connection by mutual message sending, and there are multiple relay nodes on the communication link;

[0009] S200, in the transmission stage, the current node periodically cuts off messages from its own message queue according to the encoding window and encodes to obtain encoded symbols in units of transmission rounds; congestion control and pipeline transmission strategy are formulated according to the estimated congestion degree, and the encoded symbols are transmitted to the next node according to the pipeline transmission strategy until the destination node is reached; the destination node recovers the original message from the received encoded symbols through the transmission rounds; the current node is the source node or the relay node;

[0010] S300, in the connection release phase, the source node and the destination node release the communication connection by mutual message sending, and end the transmission.

[0011] Advantages:

[0012] Compared with the prior art, the present application has the following advantages:

[0013] First, the present application proposes a reliable transmission protocol supporting unmanned aerial vehicle cluster swarm network based on fountain code. The source node and the relay node perceive the congestion state of the end-to-end path and the link reliability information, which provides the basis for formulating the pipeline transmission strategy, and can effectively cope with the dynamic topology change and packet loss problem in the communication scene. Compared with the existing protocol, the present application can more reasonably distinguish the packet loss caused by network congestion and non-congestion, avoid unnecessary congestion control triggering caused by misjudgment, and significantly improve the network performance.

[0014] Second, the present application proposes a coding transmission scheme adaptive to congestion and link reliability according to the network perception result. The link reliability information is used to make decisions on the degree distribution mode, relay node forwarding mode and coding redundancy. The congestion information is used to adjust the pipeline number and coding window parameters, further optimizing the continuity of data transmission and the utilization efficiency of network resources, so that the transmission stability and efficiency can be maintained at a high level when facing complex network conditions or frequent node movement.

[0015] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flowchart of an end-to-end reliable transmission method supporting unmanned aerial vehicle cluster swarm network provided by the present application;

[0017] Figure 2 is a total implementation architecture diagram of the present application.

[0018] Figure 3 is a connection establishment stage diagram of the present application;

[0019] Figure 4 is a flowchart of the encoder (source end) in the transmission stage of the present application;

[0020] Figure 5 is a flowchart of the decoder (destination end) in the transmission stage of the present application;

[0021] Figure 6 is a flowchart of the relay node in the transmission stage of the present application;

[0022] Figure 7 is a degree feedback mode transmission diagram of the present application;

[0023] Figure 8 is a schematic diagram of a connection release phase of the present application;

[0024] Figure 9 is a simulation result diagram of the throughput performance of each function of the present application. DETAILED DESCRIPTION

[0025] The present application will be further described in conjunction with specific embodiments, but the embodiments of the present application are not limited thereto.

[0026] The technical idea for achieving the object of the present application is: according to the dynamic topology change and packet loss of the network, the degree of participation of the relay node in coding is adaptively adjusted, an efficient transmission path is constructed, and the performance loss caused by misjudgment of congestion is reduced; by monitoring the network state in real time and dynamically adjusting the transmission strategy, the influence of link failure caused by high mobility of nodes in high dynamic scenarios on network transmission is reduced as much as possible, and the adaptability of the network to high dynamic network scenarios is improved. At the same time, the present application optimizes the continuity of data transmission and the utilization efficiency of network resources through the pipeline transmission mechanism and the priority division scheme, and ensures that efficient and stable data transmission can be maintained under complex network conditions or frequent node movement.

[0027] As shown in Figure 1 , the present application provides an end-to-end reliable transmission method supporting a UAV cluster swarm network, which comprises:

[0028] S100, in a connection establishment phase, a source node and a destination node establish a communication connection by exchanging messages, and there are multiple relay nodes on the communication link;

[0029] S200, in a transmission phase, each current node periodically cuts off messages from its own message queue according to a coding window and codes to obtain coding symbols in units of transmission rounds; congestion control is performed according to the estimated congestion degree, a pipeline transmission strategy is formulated, and the coding symbols are transmitted to the next node according to the pipeline transmission strategy until the destination node is reached; the destination node recovers the original message from the received coding symbols through transmission rounds; the current node is a source node or a relay node;

[0030] Among them, the source node of the present application is the source end, which is provided with a coder, the destination node is the destination end, which is provided with a decoder, and the relay node is provided with both a decoder and a coder.

[0031] S300, in a connection release phase, the source node and the destination node release the communication connection by exchanging messages, and end the transmission.

[0032] In a specific embodiment of the present application, in combination with Figure 2 and Figure 3 , S100 comprises:

[0033] S110, the source node sends CREQ information to the destination node and enters CREQ_SENT state, waits for the destination node to confirm, the CREQ information carries a parameter set, indicating that the source node wants to establish a connection;

[0034] S120, after the destination node receives the CREQ information of the source node, the request is confirmed, and the CREP information is sent to the source node, indicating that the connection is agreed to be established and the initialization of transmission is carried out, and the receiving of the incoming code symbol is prepared; the destination node enters CREQ_RECEIVED state, and waits for the confirmation of the source node;

[0035] S130, after the source node receives the CREP information of the destination node, it sends a CCFM information again to confirm the CREP information of the destination node; the source node enters ESTABLISHED state and starts the transmission phase; after receiving the CCFM information, the destination node also enters ESTABLISHED state, and the connection establishment phase is completed.

[0036] In a specific embodiment of the present application, with reference to Figure 4 , in S200, the current node periodically cuts off messages from the message queue of itself according to the coding window, and encodes to obtain code symbols in the unit of transmission round; according to the estimated congestion degree, the congestion control and the pipeline transmission strategy are made, and the code symbols are transmitted to the next node according to the pipeline transmission strategy until the destination node is reached, including:

[0037] S210a, the source node continuously receives the original message and enters the message queue of itself, and the relay node receives the message transmitted by the previous node and stores it in the message queue of itself;

[0038] S210b, the current node sets the coding window length at each transmission round, and judges whether the length of the message queue of itself is less than the coding window size in a time interval, if yes, the zero byte is filled to make up for it until it is not less than the coding window size; if the message queue is empty, the step of releasing the connection is entered;

[0039] Wherein, if it is the first transmission round, the source node sets the size of the coding window to 1, indicating that 1 original message is selected from the message queue, and the network bandwidth is probed, so that the network gradually adapts to the flow; otherwise, the transmission round number is increased by 1, and the coding window is doubled, until the W cmax is reached, and then the increasing is continued in the form of increasing by 1;

[0040] S210c, the current node takes out the messages with the same size as the coding window from the message queue of itself, and merges them into a byte block, and splits the byte block into a plurality of source symbols according to the size of the source symbol;

[0041] The maximum size of the source symbol should not exceed the merged byte block size MBB (Merge Byte Block). The size of the source symbol is calculated according to the following formula: BDP is the end-to-end delay-bandwidth product, which indicates the maximum amount of data in the network. typ It is related to the message type or packet loss rate. If the packet loss rate is less than 1%, it is set to 1; if the packet loss rate is between 1% and 5%, it is set to 2; if the packet loss rate is higher than 5%, it is set to 4; non-delay sensitive services such as large files can be set to 8. Assuming that the packet loss rate is p loss In a network, when n coded symbols are sent, the probability that at least one of them arrives successfully is not less than a certain threshold P. th ,have From this we can solve: Minimum source symbol size D is the total amount of data to be transmitted.

[0042] S210d, the current node resets the transmission number threshold, and uses the plurality of source symbols to encode to obtain coded symbols; each coded symbol is marked with the current transmission round sequence number and then compressed into a data packet;

[0043] S210e: Congestion control is performed based on the estimated congestion level, and a pipeline transmission strategy is formulated. The data packet is then sent to the next transmission node according to the pipeline transmission strategy until the number of packets sent reaches a threshold, and then the coding interval is increased to continue sending.

[0044] The source node will set the number of coding symbols to be generated according to the established pipeline transmission strategy, continuously generate coding symbols, and mark each coding symbol with the current transmission round number before sending it.

[0045] S210f, the current node monitors the feedback information. If the source node receives an OVER message and the relay node receives a NOVER message, it returns to S210a; if it receives a DFLT message, it returns to S210d.

[0046] In a specific embodiment of the present invention, performing congestion control according to the estimated congestion level in S200 includes:

[0047] S220a: Any current node sets the time interval T as the statistical period and initializes the number of data packets N received within the time period T. p (T) is 0, initialize the maximum sequence number P of the packet received in the nth and n-1th time intervals T n and P n-1 is 0;

[0048] S220b, any current node, whenever a data packet is received within the time interval T, willp (T) is incremented by 1 and P is updated n is the maximum sequence number of the data packets received in the current time interval T;

[0049] S220c, any current node, at the end of each time interval T, uses the formula to calculate the packet loss rate, which is taken as the reliability of the current link, and then updates P n-1 is P n , N is initialized p (T) is 0, for the next update calculation;

[0050] S220d, the queue length threshold Q is set th , the available transmission rate threshold R th , the maximum queuing delay D max , three weight factors w1, w2, w3, the comprehensive congestion index threshold C th , the reference upper limit of the comprehensive congestion index C max , the maximum encoding window size W cmax , the pipeline calculation constant θ;

[0051] S220e, the current node monitors the length of its own sending queue Q, the available transmission rate R, and the maximum queuing delay D

[0052] S220f, the current node calculates the comprehensive congestion index C as an estimate of the congestion degree;

[0053] S220f, if the comprehensive congestion index C satisfies C≥C th , the congestion control mechanism is triggered to reduce the encoding window size and the number of pipelines, and the congestion control mechanism is:

[0054] the encoding window size is set to W

[0055] the new W cmax is set to the reduced window size W c , and is recovered at a speed of doubling every transmission round; the number of pipelines L is set to min(current number of pipelines, ).

[0056] wherein the reference upper limit of the comprehensive congestion index is C max , and the maximum encoding window size is W cmax .

[0057] In a specific embodiment of the present application, the pipeline transmission strategy in S200 according to the estimated congestion degree comprises:

[0058] S230a, the source node sets the degree distribution decision threshold P th1 and P th2 th1 th2 , the relay node sets the behavior decision threshold p th , C th ;

[0059] S230b, the source node calculates the end-to-end packet loss rate P before each transmission round starts e2e , the calculation method is the product of the packet loss rate of each link P e2e = ΠPLR i ;

[0060] S230c, the relay node reads the packet loss rate PLR of the next hop link and the integrated congestion degree C after receiving the encoded symbol of the previous node next ;

[0061] S230d, the source node judges whether P e2e < P th1 is true, if yes, the pipeline transmission strategy is set as the degree 1 distribution strategy, and the pipeline transmission strategy is set as no need to perform XOR encoding; if P th1 ≤ P e2e < P th2 , the pipeline transmission strategy is set as the robust arc wave distribution; if P e2e ≥ P th2 , the pipeline transmission strategy is set as the degree feedback strategy; the degree feedback strategy includes: the construction stage one, the construction stage two and the completion stage;

[0062] S230e, the relay node judges whether PLR next < pt h and C < C h is true, if true, the relay node sets the pipeline transmission strategy as directly forwarding the encoded symbol; if PLR next ≥ pt h or C ≥ C th , the relay node sets the pipeline transmission strategy as the strategy of forwarding the encoded symbol while decoding;

[0063] S230f, the source node and the relay node both adjust the degree of the pipeline transmission strategy of themselves with a single transmission round as a period.

[0064] In a specific embodiment of the present application, the strategy of forwarding the encoded symbol while decoding includes:

[0065] S230e1, when the source node adopts the degree 1 distribution and the robust arc wave distribution transmission, referring to Figure 5 , the relay node forwards according to the strategy 1, and the strategy 1 is: ​​

[0066] When the relay node receives the encoded packet for the first time, it initializes the matched source node and destination node, and performs the operation of decoding first and encoding later while continuing to forward each encoded packet; the relay node decodes according to the above transmission steps using the newly initialized destination node, and if all source symbols are successfully recovered, it replies with NOVER information and continues to use the source node encoding for the decoded source symbols; the source node receiving the NOVER information suspends sending the encoded symbols, and the source node or relay node receiving the OVER information immediately enters the next transmission round;

[0067] S230e2, when the degree feedback strategy is adopted for transmission, referring to Figure 6 , the relay node forwards according to strategy 2, and the strategy 2 is:

[0068] In the construction phase one, the source node randomly selects input symbols for encoding and sends encoded symbols with a constant degree value of 2 until it receives a BU1 feedback from the next hop, and after the feedback is processed, it does not continue to forward to the source direction, and then enters the construction phase two, and the current node will only receive encoded symbols with a degree value of 2; if the encoded symbol contains only one color of input symbol and is located in a single component, the symbol is ignored and does not continue to forward to the node in the destination direction; if the encoded symbol is composed of two all-white nodes from two different components, it means that the two components can be combined into one, and the encoded symbol should be processed and the single-part graph is updated; continue to judge whether the size of the combined component reaches the maximum connected component, if yes, send BU1 feedback and immediately enter the construction phase two; if the encoded symbol is composed of one black and one white node, the component containing the white node can be recovered and the single-part graph is updated; finally, the encoded symbol is forwarded to the next node; the meaning of component is that the decoder only needs to know the result of the XOR operation of two source symbols, and the two source symbols are connected, all connected source symbols are called components, the size of the component refers to the number of source symbols in the component, and a single unknown source symbol is the smallest component with a size of 1; the maximum connected component refers to the largest component among all components; the component is recovered, which means that once any source symbol in the component is known, all source symbols in the component become known, i.e. the component is recovered; the single-part graph refers to a set composed of all components and all known source symbols, and the known source symbols in the single-part graph are black and the unknown source symbols are white;

[0069] S230e3, when the source node adopts the degree feedback strategy for transmission, referring to Figure 6 , the relay node forwards according to strategy 3, and the strategy 3 is:

[0070] In the second construction phase, the source node randomly selects input symbols for encoding and sends the encoded symbols with a constant degree value of 1 until receiving a BU2 feedback from the next hop; the feedback is processed and no longer forwarded to the source direction, and then enters the completion phase; in the second construction phase, the current node will only receive encoded symbols with a degree value of 1 and continue to forward to the destination direction; while decoding, the monopartite graph should be recovered, and it is checked whether the largest connected component is recovered, if so, a BU2 feedback is sent, and the completion phase is entered; if the decoding is completed, a NOVER feedback is sent to the source node, and the corresponding source node is initialized, and the transmission round of the adjacent node is synchronized;

[0071] S230e4, when the degree feedback strategy is adopted for transmission, reference Figure 6 , the relay node forwards according to strategy 4, and the strategy 4 is:

[0072] If an OVER feedback is received in the second construction phase, the source node stops sending, the feedback is processed and should continue to forward to the source direction, and then enters the next transmission round; if a NOVER is received, the sending is stopped, a suspended state is entered, and the feedback is continued to be forwarded to the source direction;

[0073] S230e5, when the degree feedback strategy is adopted for transmission, reference Figure 6 , the relay node forwards according to strategy 5, and the strategy 5 is:

[0074] In the completion phase, the source node randomly selects input symbols for encoding and sends the encoded symbols according to the calculated optimal degree value until receiving an FDB feedback, and the feedback is no longer forwarded to the source direction, and then the current optimal degree value is updated and the sending is continued, and the completion phase is maintained; in the completion phase, the destination node will receive the encoded symbols of the optimal degree value, if the encoded symbols only contain one color of input symbols and are all located in a single component, the symbols are ignored and no longer forwarded to the nodes in the destination direction; if the encoded symbols are composed of all white nodes from two different components, the encoded symbols should be processed and the monopartite graph is updated; then the decoding progress FDB is fed back, and the symbols are forwarded to the nodes in the destination direction; if the number of white nodes of the encoded symbols is 1, the monopartite graph is updated, the decoding progress FDB is fed back, and the encoded symbols are forwarded to the nodes in the destination direction; if the destination node as a relay node decodes completely, a NOVER feedback is sent to the source direction, the corresponding source node is initialized, and the transmission phase of the source node is set according to the decoding phase of the next hop; if the destination node as a destination node decodes completely, only an OVER feedback is sent to the source direction, and then the next transmission round is entered;

[0075] S230e6, when the degree feedback strategy is adopted for transmission, reference Figure 6The relay node forwards according to policy 6, which is:

[0076] If OVER feedback is received in the completion phase, the source node stops sending, the feedback should continue to forward to the source direction after being processed, and then turn to the next transmission round; if NOVER is received, stop sending, enter the suspended state, and continue to forward the feedback to the source direction.

[0077] In a specific embodiment of the present application, the number of pipelines in the pipeline transmission strategy is dynamically adjusted according to the link utilization rate, and the adjustment process includes:

[0078] Setting the current pipeline level L, the transmission round completion time T cycle , the time consumed by the feedback returning to the source node T ACK , the link utilization rate threshold U th ;

[0079] Defining the number of pipeline levels as the number of coding modules used to complete a certain flow at the same time on a certain node. For example, for the source node, if there are two encoders currently encoding and transmitting different messages belonging to the same flow, it is called two-level pipeline.

[0080] The source node calculates the link utilization rate according to the following formula:

[0081] If U < U th , a new pipeline is added, if U = U th , the number of pipelines is not changed, and if U > U th , a pipeline is deleted.

[0082] In a specific embodiment of the present application, the expected number of encoding symbols is limited in the degree 1 distribution strategy, the robust arc wave distribution strategy and the degree feedback strategy;

[0083] For the degree 1 distribution strategy, the expected number of encoding symbols is: Where k is the total number of source symbols obtained by combining the message blocks from the message queue in the current transmission round; δ is the maximum value of the allowed probability of decoding interruption events; decoding interruption event refers to that before decoding succeeds, the decoder has no encoding symbol with a degree of 1.

[0084] For the robust arc wave distribution strategy, the expected number of encoding symbols is: Refers to a smaller infinitesimal than of higher order when k tends to infinity;

[0085] For the degree feedback strategy, in the construction phase, the expected number of degree 2 encoding symbols is: The number of encoding symbols of degree 1 is: The number of encoding symbols required for recovering k source symbols in the completion phase is The total overhead is: E tr (k) = E(N comp (k)) + E(N1) + E(N2), a is the proportion of the number of source symbols contained in the largest connected component in the single graph at the end of the construction phase to the total number of source symbols k in the current transmission round; c is the average degree of the source symbols, i.e. the number of times each source symbol participates in the formation of an encoding symbol on average before the start of the completion phase, which is obtained according to the formula a + e -ca = 1; P(i) is the probability that a new encoding symbol causes a decoding progress event in the completion phase; the decoding progress event refers to the case that, when the proportion of recovered source symbols reaches , only one or two source symbols among all the source symbols that constitute the new encoding symbol are unknown to the decoder, and the rest of the source symbols that constitute the new encoding symbol are known.

[0086] In a specific embodiment of the present application, with reference to Figure 7 , the recovery of the original message by the destination node from the received encoding symbols in the transmission round in S200 includes:

[0087] The destination node continuously receives the encoding symbols and checks whether the transmission round number in the encoding symbols matches; if not, the OVER / NOVER information is retransmitted and the encoding symbol is discarded; the decoding is performed on the matching encoding symbol, and if all the source symbols are successfully recovered, the OVER / NOVER information is returned to the source node; the destination node assembles the source symbols into byte blocks and re-splits them into original messages according to the length of a single original message.

[0088] In a specific embodiment of the present application, with reference to Figure 8 , S300 includes:

[0089] S310, the destination node starts releasing the connection after the transmission is completed, and the source node completes the data transmission and prepares to release the connection, sends a DREQ information to the destination node, indicating that the source node hopes to disconnect, and enters the DREQ_SENT state, waiting for the confirmation of the destination node;

[0090] S320, after receiving the DREQ information of the source node, the destination node confirms the request and starts preparing to disconnect; the destination node sends a DREP information to the source node, indicating that the connection is agreed to be disconnected, and enters the DREQ_RECEIVED state, waiting for the further action of the source node;

[0091] S330, after the source node receives the DREP information of the destination node, the last step of connection release is performed, a DCFM message segment is sent to the destination node to confirm the DREP information of the destination node, and it is informed that the source node is ready to release all related resources and enters a WAIT_RELEASE state to wait for the final release signal of the destination node;

[0092] S340, after the destination node receives the DCFM information of the source node, it is confirmed that the connection release process is close to completion, a DREL information is sent to the source node to indicate that all resources have been released and the connection has been completely disconnected; the destination node enters a CLOSED state to prepare to accept a new connection request or end communication;

[0093] S350, after the source node receives the DREL information, it also enters the CLOSED state to complete the communication connection release.

[0094] The effect of the application will be further described in combination with a simulation experiment:

[0095] 1. Simulation experiment conditions:

[0096] The application platform of the simulation experiment is: the processor is a 32-core AMD Ryzen Threadripper PRO 5975WX 64-bit CPU, the main frequency is 3.6 GHz, and the memory is 128 GB.

[0097] The software platform of the simulation experiment is: Windows10 operating system.

[0098] The network scene of the simulation experiment is a space of 200kmx100km, in which a plurality of network nodes are randomly distributed, each node can communicate with nodes within the transmission range, the motion model of the node adopts a random waypoint model, and the movement of each node is not related.

[0099] The simulation parameter settings are shown in Table 1

[0100] Table 1 Simulation parameter table

[0101] Maximum power 31 dBm Path loss exponent 2 Receiver sensitivity -89 dBm Signal propagation loss 20 dB Routing protocol DSR MAC protocol TDMA

[0102] 2. Simulation content and result analysis:

[0103] Simulation 1, under the network scene in Table 1, each mechanism in the application is respectively used, and the traditional TCP is used under different node moving speeds, two nodes are randomly selected as the source node and the destination node respectively, the influence of random packet loss on transmission performance is compared, each data point is simulated for 10 times to obtain the average value, and the results are as follows Figure 9 .

[0104] From Figure 9It can be seen that the transmission effect is better when the application is used for reliable transmission in a dynamic scene. TCP has frequent route switching packet loss, so the congestion window is always maintained at a low level, and the performance drops sharply. When the node moving speed reaches about 120 m / s, the effective throughput is close to zero. When only degree 1 symbols are sent, the transmission effect is better when the node moving speed is slow, but as the node moving speed increases, not enough coding gain is introduced, and the throughput drops sharply. When only the robust arc wave distribution is used, since this degree distribution function is more suitable for single-hop erasure channels with random packet loss, it does not perform well in the UAV cluster swarm network, and it is not flexible enough for long-distance transmission in high dynamicity, and the performance is mediocre. When only the degree feedback mode is used for transmission, since this mode divides the fountain code into multiple stages, the interaction of feedback information is more frequent, and it is not worth the loss when the packet loss probability is small, so it is more suitable as a transmission scheme to combat high packet loss rate. When the relay node only participates in forwarding, the transmission resources in the first half will be wasted due to the route failure of the unstable path in the second half. Because the relay node cannot proxy the source node to continue sending, retransmission can only rely on the source node, so the transmission efficiency is low. When only a single pipeline is used, the next round of encoding cannot be performed until the decoding success confirmation feedback is received. Especially in the case of feedback loss or long path transmission, the confirmation information will be delayed, and the sending end will waste time waiting during this period, so the performance is not good. The application selects the appropriate transmission strategy at the right time by detecting the state of the transmission path, thereby improving the transmission performance.

[0105] It should be noted that the terms "first", "second" in the present application are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0106] Although the application is described herein in conjunction with various embodiments, other variations of the disclosed embodiments can be understood and implemented by those skilled in the art by viewing the described figures, disclosure, and appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality.

[0107] The above is further detailed description of the present application in combination with specific preferred embodiments, and cannot be deemed as limitation of the specific implementation of the present application to these descriptions. For those skilled in the art to which the present application belongs, without departing from the concept of the present application, a number of simple deductions or substitutions can be made, and all of them shall be deemed as falling within the protection scope of the present application.

Claims

1. A method for end-to-end reliable transmission supporting drone swarm networks, characterized in that: include: S100, in the connection establishment phase, the source node and the destination node establish a communication connection by exchanging messages, and there are multiple relay nodes on the communication link; S200, in the transmission phase, each current node periodically intercepts messages from its own message queue according to the coding window and encodes them in transmission rounds to obtain coded symbols. Congestion control and pipeline transmission strategies are implemented based on the estimated congestion level, and the coded symbols are transmitted to the next node according to the pipeline transmission strategy until they reach the destination node. The destination node recovers the original message from the received coded symbols through transmission rounds; the current node is a source node or a relay node; S300, in the connection release phase, the source node and the destination node release the communication connection by exchanging messages, thus ending the transmission; The congestion control according to the estimated congestion level in S200 includes: S220a: Any current node sets the time interval T as the statistical period and initializes the number of data packets N received within the time period T. p (T) is 0, initialize the maximum sequence number P of the packet received in the nth and n-1th time intervals T n and P n-1 is 0; S220b, any current node, whenever a data packet is received within the time interval T, will p (T) increments by 1 and updates P n is the maximum sequence number of the data packets received within the current time interval T; S220c, any current node, at the end of each time interval T, uses the formula Calculate the packet loss rate and use it as the reliability of the current link, then update P n-1 P n , initialize N p (T) is 0, used for the next update calculation; S220d, set the queue length threshold Q th , available transmission rate threshold R th , maximum queuing delay D max , three weight factors w1, w2, w3, comprehensive congestion index threshold C th , Comprehensive congestion index reference upper limit C max , maximum value of the coding window W cmax , pipeline calculation constant S220e, the current node monitors its own sending queue length Q, available transmission rate R, and maximum queuing delay in real time S220f, the current node calculates the comprehensive congestion index As an estimator of congestion; S220f, if the comprehensive congestion index C≥C th , then the congestion control mechanism is triggered to reduce the encoding window size and the number of pipelines. The congestion control mechanism is: Set the next transmission round to reduce the encoding window size to The new W cmax Set to the reduced window size W c , and recover at a rate doubling per transmission round; set the number of pipelines 2. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 1 is characterized in that: S100 includes: S110, the source node sends a CREQ message to the destination node and enters the CREQ_SENT state, waiting for the destination node to confirm. The CREQ message carries a parameter set, indicating that the source node wants to establish a connection; S120: After receiving the CREQ information from the source node, the destination node confirms the CREQ information and sends a CREP message to the source node, indicating that it agrees to establish a connection and initialize the transmission, and is ready to receive the incoming coded symbols. The destination node enters the CREQ_RECEIVED state and waits for confirmation from the source node. S130, after receiving the CREP information of the destination node, the source node sends a CCFM message to confirm the CREP information of the destination node; the source node enters the ESTABLISHED state and starts the transmission phase; the destination node also enters the ESTABLISHED state after receiving the CCFM message, completing the connection establishment phase.

3. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 1 is characterized in that: In S200, each current node periodically intercepts messages from its own message queue according to the coding window and encodes them in units of transmission rounds to obtain coded symbols; performs congestion control and formulates a pipeline transmission strategy based on the estimated congestion level, and transmits the coded symbols to the next node according to the pipeline transmission strategy until they reach the destination node, including: S210a, the source node continues to receive the original message and adds it to its own message queue, and the relay node receives the message transmitted by the previous node and stores it in its own message queue; S210b, the current node sets the coding window length in each transmission round and determines whether the length of its message queue is less than the coding window size within a time interval. If so, it fills the message queue with all zero bytes until it is no less than the coding window size. If the message queue is empty, the connection release step is entered. S210c, the current node takes out messages of the same size as the coding window from its own message queue, merges them into byte blocks, and splits the byte blocks into a number of source symbols according to the size of the source symbols; S210d, the current node resets the transmission number threshold, and uses the plurality of source symbols to encode to obtain coded symbols; each coded symbol is marked with the current transmission round sequence number and then compressed into a data packet; S210e: Congestion control is performed based on the estimated congestion level, and a pipeline transmission strategy is formulated. The data packet is then sent to the next transmission node according to the pipeline transmission strategy until the number of packets sent reaches a threshold, and then the coding interval is increased to continue sending. S210f, the current node monitors the feedback information. If the source node receives an OVER message and the relay node receives a NOVER message, it returns to S210a; if it receives a DFLT message, it returns to S210d.

4. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 1 is characterized in that: In S200, the pipeline transmission strategy is formulated based on the estimated congestion level, including: S230a, the source node sets the degree distribution decision threshold P th1 and P th2 , P th1 <P th2 , the relay node sets the behavior decision threshold p th 、C th ; S230b, the source node calculates the end-to-end packet loss rate P before each transmission round begins. e2e , calculated as the product of the packet loss rate of each link P e2e =∏PLR i ; S230c, after receiving the coding symbol from the previous node, the relay node reads the packet loss rate PLR ​​of the next hop link next and the comprehensive congestion level C; S230d, the source node determines P e2e <P th1 Is it true? If so, then formulate the pipeline transmission strategy as degree 1 distribution strategy, formulate the pipeline transmission strategy as no need for XOR encoding; if P th1 ≤P e2e <P th2 , formulate the pipeline transmission strategy as robust arc wave distribution; if P e2e ≥P th2 ,The pipeline transmission strategy is formulated as the degree feedback strategy; the degree feedback strategy includes: construction phase 1, construction phase 2 and completion phase; S230e, the relay node determines the PLR next <p th And C <C th Is it true? If so, the relay node formulates a pipeline transmission strategy to directly forward the coded symbols; if PLR next ≥p th or C≥C th , then the relay node formulates a pipeline transmission strategy of decoding and forwarding coded symbols at the same time; S230f: The source node and the relay node both readjust their pipeline transmission strategies based on a single transmission round.

5. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 4 is characterized in that: The strategy of forwarding coded symbols while decoding includes: S230e1: When the source node adopts degree 1 distribution and robust arc wave distribution transmission, the relay node forwards according to strategy 1, which is: When a relay node receives a coded packet for the first time, it initializes the matching source and destination nodes, and continues to forward each coded packet while performing decoding-first-then-encoding operations. The relay node uses the newly initialized destination node to decode according to the above transmission steps. If all source symbols are successfully recovered, it replies with a NOVER message and continues to use the source node to encode the successfully decoded source symbols. The source node that receives the NOVER message suspends sending coded symbols, and the source node or relay node that receives the OVER message immediately enters the next transmission round. S230e2: When the degree feedback strategy is adopted for transmission, the relay node forwards according to strategy 2, which is: In the first construction phase, the source node randomly selects input symbols with a constant degree of 2, encodes them, and sends the encoded symbols until it receives BU1 feedback from the next hop. After the feedback is processed, it is not forwarded to the source direction. Then it enters the second construction phase. The current node will only receive the encoded symbols with a degree of 2. If the encoded symbol contains only input symbols of one color and all of them are in a single component, it is ignored and not forwarded to the node in the destination direction. If the encoded symbol consists of all-white nodes from two different components, it means that the two components can be merged into one. The encoded symbol should be processed and the monopartite graph should be updated. It is further determined whether the size of the combined component reaches the maximum connected component. If so, BU1 feedback is sent and the second construction phase is immediately entered. If the encoded symbol consists of two nodes, one black and one white, the component containing the white node can be restored and the monopartite graph is updated. Finally, the encoded symbol is forwarded to the next node. S230e3: When the source node adopts the degree feedback strategy for transmission, the relay node follows strategy 3, which is: In the second construction phase, the source node randomly selects input symbols with a constant degree of 1, encodes them, and sends the encoded symbols until it receives BU2 feedback from the next hop. After the feedback is processed, it does not continue to forward it to the source, and then enters the completion phase. In the second construction phase, the current node only receives the encoded symbols with a degree of 1 and continues to forward them to the destination. While decoding, the unipartite graph should be restored, and the maximum connected component should be checked. If so, BU2 feedback is sent, and the completion phase is entered. At this time, if the decoding is complete, NOVER feedback is sent to the source node, and the corresponding source node is initialized, synchronizing the transmission rounds of adjacent nodes. S230e4: When the degree feedback strategy is adopted for transmission, the relay node follows strategy 4, which is: If an OVER feedback is received in the second construction phase, the source node stops sending. After the feedback is processed, it should continue to forward it to the source and then enter the next transmission round. If a NOVER feedback is received, it stops sending, enters the suspended state, and continues to forward the feedback to the source. S230e5: When the degree feedback strategy is adopted for transmission, the relay node follows strategy 5, which is: In the completion phase, the source node randomly selects input symbols according to the calculated optimal degree value, encodes them, and transmits the encoded symbols until it receives FDB feedback, which stops forwarding the feedback to the source. It then updates the current optimal degree value and continues transmitting, remaining in the completion phase. In the completion phase, the destination node receives the encoded symbol with the optimal degree value. If the encoded symbol contains only input symbols of one color, all in a single component, it is ignored and not forwarded to the destination node. If the encoded symbol consists of all-white nodes from two different components, it should be processed and the unipartite graph updated. The decoding progress FDB is then fed back, and the symbol is forwarded to the destination node. If the number of white nodes in the encoded symbol is 1, the unipartite graph is updated, the decoding progress FDB is fed back, and the encoded symbol is forwarded to the destination node. If the destination node, acting as a relay node, completes decoding, it sends NOVER feedback to the source, initializes the corresponding source node, and sets the source node's transmission phase according to the decoding phase of the next hop. If the destination node, acting as a destination node, completes decoding, it simply sends OVER feedback to the source and enters the next transmission round. S230e6: When the degree feedback strategy is adopted for transmission, the relay node forwards according to strategy 6, which is: If OVER feedback is received during the completion phase, the source node stops sending. After the feedback is processed, it should continue to forward it toward the source and then enter the next transmission round. If NOVER is received, it stops sending, enters the suspended state, and continues to forward the feedback toward the source.

6. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 4 is characterized in that: The number of pipelines in the pipeline transmission strategy is dynamically adjusted according to link utilization, and the adjustment process includes: Set the current pipeline level L and the transmission round completion time T cycle , the time it takes for feedback to be transmitted back to the source node is T ACK , link utilization threshold U th ; The source node calculates the link utilization rate according to the following formula If U th , then add a new pipeline, if U=U th The number of pipelines will not be changed if U>U th Delete a pipeline.​ 7. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 4 is characterized in that: The desired number of coding symbols is defined in the degree 1 distribution strategy, the robust arc wave distribution strategy, and the degree feedback strategy; For the degree 1 distribution strategy, the expected number of encoding symbols is: Where k is the total number of source symbols that are split into byte blocks from the message queue in the current transmission round; δ is the maximum allowed probability of a decoding interruption event; a decoding interruption event means that the decoder has no coded symbols with a degree of 1 before decoding is successful; For the robust arc wave distribution strategy, the expected number of coding symbols is: It means that when k tends to infinity, infinitesimals of higher order; For the degree feedback strategy, during the construction phase, the expected number of degree 2 coding symbols required is: The number of coding symbols for degree 1 is: In the completion phase, the number of encoding symbols required to recover k source symbols is The total cost is: E tr (k) = E(N comp (k))+E(N1)+E(N2), α refers to the ratio of the number of source symbols contained in the largest connected component of the unipartite graph at the end of the construction phase to the total number of source symbols k in the current transmission round; c refers to the average degree of the source symbols; P(i) refers to the probability of the event that the new encoded symbols advance the decoding progress in the completion phase.

8. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 1 is characterized in that: In S200, the destination node recovers the original message from the received coded symbols through the transmission rounds, including: The destination node continuously receives the coded symbols and checks whether the transmission round sequence number in the coded symbols matches. If not, the OVER / NOVER information is retransmitted and the coded symbol is discarded. The matching coded symbols are decoded. If all source symbols are successfully recovered, the OVER / NOVER information is replied to the source node. The destination node assembles the source symbols into byte blocks and re-splits them into original messages according to the length of a single original message.

9. The end-to-end reliable transmission method supporting drone cluster swarm network according to claim 1 is characterized in that: S300 includes: S310, the destination node starts to release the connection after completing the transmission. After the source node completes the data transmission, it is ready to release the connection and sends a DREQ message to the destination node, indicating that the source node wishes to disconnect the connection. It then enters the DREQ_SENT state and waits for the destination node's confirmation. S320: After receiving the DREQ information from the source node, the destination node confirms the DREQ information and begins to prepare to disconnect. The destination node sends a DREP message to the source node, indicating that it agrees to disconnect, and enters the DREQ_RECEIVED state, waiting for further action from the source node. S330: After receiving the DREP information from the destination node, the source node performs the final step of connection release and sends a DCFM segment to the destination node to confirm the DREP information of the destination node and inform the source node that it is ready to release all related resources. The source node then enters the WAIT_RELEASE state and waits for the final release signal from the destination node. S340: After receiving the DCFM message from the source node, the destination node confirms that the connection release process is nearly complete and sends a DREL message to the source node, indicating that all resources have been released and the connection has been completely disconnected. The destination node enters the CLOSED state, ready to accept new connection requests or terminate communication. S350: After receiving the DREL information, the source node also enters the CLOSED state and completes the release of the communication connection.

Images