CN111770546B - Delay tolerant network random network coding method based on Q learning - Google Patents
Delay tolerant network random network coding method based on Q learning Download PDFInfo
- Publication number
- CN111770546B CN111770546B CN202010597538.0A CN202010597538A CN111770546B CN 111770546 B CN111770546 B CN 111770546B CN 202010597538 A CN202010597538 A CN 202010597538A CN 111770546 B CN111770546 B CN 111770546B
- Authority
- CN
- China
- Prior art keywords
- node
- coding
- probability
- nodes
- rank
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/22—Communication route or path selection, e.g. power-based or shortest path routing using selective relaying for reaching a BTS [Base Transceiver Station] or an access point
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a Q-learning-based random network coding method for a delay tolerant network, which mainly solves the contradiction that when a Prophet routing algorithm is combined with random linear network coding in the delay tolerant network, the meeting probability of two nodes is overlarge, so that the nodes cannot obtain linearly independent coding packets, and the meeting probability is too small, so that the two nodes cannot meet. The implementation scheme is as follows: carrying out random linear network coding on one generation of a source node, forwarding a coding packet by the source node by using a Prophet routing protocol, and calculating the probability that the relay node receives a linear independent coding packet from other nodes when two relay nodes meet, namely the one-step transition probability of the node coding matrix rank; then, a Q value of the relay node is calculated, in a Q function, a revenue function must be closely related to the one-step transfer probability of the rank and the node encountering probability, and the revenue function is changed into a negative value when the rank is unchanged; and selecting the neighbor node with the maximum Q value to exchange the encoding packet, and updating the Q value of the neighbor node. The invention ensures that the size of the node Q value can reflect the local maximum coding benefit of the network, improves the data packet forwarding throughput and the success rate, and can be used for a delay tolerant network.
Description
Technical Field
The invention belongs to the technical field of communication, and mainly relates to a delay tolerant network random network coding and routing method which can be used for routing decision of a vehicle-mounted ad hoc network, a wild animal monitoring sensor network and a mobile social network.
Background
The delay tolerant network refers to a wireless ad hoc network which can only intermittently communicate among nodes due to the movement of the nodes or energy scheduling and the like under an extreme environment, even is in an interrupted state for a long time, and has the following characteristics compared with the traditional wireless ad hoc network: intermittent connection, due to the reasons of node mobility, node density sparsity and the like, communication links between nodes are frequently split, so that the topological structure of the network is continuously changed, the connection state of the whole network has certain randomness, the connection between the nodes is unpredictable, and the real-time end-to-end routing path cannot be ensured; the time delay is extremely high, the data transmission rate is low, and adjacent nodes can not communicate for a long time, so that the time delay experienced by the message on each hop is greatly increased, and meanwhile, the network has the characteristic of low data transmission rate due to the factors such as limited bandwidth of a link and the like.
In the research of the routing protocol of the delay tolerant network, a Prophet routing algorithm is a typical representative based on a history prediction method, and is proposed for overcoming the blindness of message control in a replication method. Compared with the routing algorithm of infectious diseases, the Prophet routing algorithm has higher average message transmission rate, lower communication load and better average time delay performance.
The specific contents of the Prophet routing algorithm are as follows: when two nodes meet, they exchange respective meeting probabilities with each other, and each node stores meeting probability information of reaching all other nodes, such as the meeting probability P (A, B) indicates the probability that the node A can send information to the node B. The concrete implementation includes the following three steps:
1) updating the probability of an encounter (P) between two nodes A, B meeting init Is an initialization constant) is given by equation (1):
P (A,B) =P (A,B)old +(1-P (A,B)old )×P init
2) if a pair of nodes do not meet for a period of time, the meeting probability will decay. The equation for attenuation is as in equation (2):
P (A,B) =P (A,B)old ×γ T
where γ is a decay factor, T represents the number of time units from the last update, and if two nodes do not meet within T time units, the meeting probability between the two nodes decays. The value of the time unit is set according to the specific network condition.
3) Transitive updating with encounter probability. For example, if a frequently encounters B, which in turn frequently encounters C, then it can be said that a passes through B and the probability that a can send information to node C is not low. The specific transfer equation is as shown in the formula (3):
P (A,C) =P (A,C)old +(1-P (A,C)old )×P (A,B) ×P (B,C) ×β
where β is the transfer coefficient, reflecting the magnitude of the effect of transfer on the encounter probability. After the three steps of meeting probability updating are carried out, the two nodes start to carry out data packet transmission. The sending method comprises the following steps: if the destination node is D, if P (B,D) >P (A,D) Node a will send a data packet to the encountering node B. The information is forwarded continuously according to the sending method until the data packet reaches the destination node D.
In the case of frequent breakage of the delay tolerant network links, the data transmission capability is a primary goal of the routing algorithm design. The random linear network coding is an information exchange technology integrating coding and routing, and on the basis of a traditional storage and forwarding routing method, the coding information integration is allowed to be carried out on a plurality of received data packets, so that the information quantity of single transmission is increased, and the random linear network coding shows superiority in the aspects of improving the throughput of a delay tolerant network, the routing reliability and the like. In the routing protocol of the delay tolerant network adopting random linear network coding, when two nodes meet, coding packets subjected to coding processing are transmitted to the meeting node, the coding packets are generated by a part of data packets in a node cache through random linear network coding, one coding packet contains information of a plurality of original data packets, and a destination node receives an independent coding packet and can obtain information which is useful for recovering all the original data packets. As long as the destination node receives a sufficient number of encoded packets, the complete information of all the original data packets can be obtained by decoding.
By applying a random linear network coding method in a Prophet routing protocol, throughput and reliability of data transmission can be improved, but if the meeting probability of two nodes is too high, the probability that a coding packet received by the node and an existing coding packet in a cache do not have linear independence is increased, for example: the two nodes meeting each other meet each other again immediately, no new data packet exists between the two nodes, and therefore the encoded packets obtained by the nodes do not have linear independence. How to solve the contradiction between the linear independence of the encounter probability and the received coding packet is one of the problems of improving the performance of the Prophet routing protocol.
Disclosure of Invention
The invention aims to overcome the technical defect of combining a Prophet routing algorithm with random linear network coding, and provides a delay tolerant network random network coding method based on Q learning, wherein the Q learning belongs to a value function estimation method in the field of reinforcement learning, and the Q function is an evaluation function, Q(s) r ,a m ) Is shown in state s r Then, perform action a m The evaluation value obtained thereafter. In the delay tolerant network, when a certain node meets other nodes, only the coding packet of the node with the maximum Q value is received, and the linear independence of the coding packet is ensured to be met under the condition of higher meeting probability.
The Q learning-based random network coding method for the delay tolerant network comprises the following specific steps:
initializing the probability of encounter P of each node with other nodes in a delay tolerant network init The Q value initial value of each node is 0, in the source node, a batch of data packets which are put together for random linear network coding are classified, and one classified original data packet with the number of K is consideredOne linear combination of these K packets is:
whereinIs a finite field of size q, the coefficient α being (α) 1 ,…,α K ) Called code vector, the generated linear combination x is called code packet, if the code vectors of two or more code packets are linearly independent, the code packets are said to be linearly independent, if a certain node carries a set of code packets containing at most r linearly independent code packets x 1 ,…,x r Let us say that the rank of the node is r, and the parallel vector is an r × K matrix of r coding vectors as the coding matrix of the node, and the source node forwards the coded packet by using a Prophet routing protocol;
the relay node meets other nodes once, if a linear independent coding packet can be received, the rank of the coding matrix of the node is added with 1, otherwise, the rank is not changed until the full rank r of the coding matrix is K, the state change of the rank is only related to the current state and is not related to the historical state, therefore, the process that the node receives the linear independent coding packet is a Markov chain, when the full rank is reached, the absorption state is reached, the probability that the node can receive the linear independent coding packet from the meeting node is calculated, and the probability is the one-step transition probability of the rank state;
when two relay nodes (assumed as a node a and a node B) meet each other, it is assumed that the node a has i encoded packets and meets the node B, the node B has j encoded packets, i, j belongs to [1, …, k ], and the node a and the node B both have the probability of the same encoded packet:
when the node A has i code packets, the probability that a linear independent code packet can be received from the node B is the rank state s r To s r+1 The one-step transition probability of (2):
wherein the content of the first and second substances,indicating the probability that the setting of the size of the finite field causes the linear independent coding packet to be received by the node;
the neighbor nodes of the relay nodes are nodes within the communication radius range, when each relay node meets other nodes, the node with the maximum Q value of the neighbor node is selected to exchange the coding packet, if the Q values of the neighbor nodes are the same, the node with more coding packets is selected to exchange the coding packet, and the Q value of the node is updated;
wherein s is r State representing node rank, maxQ(s) r+1 ,a m ) Indicating the next state in memory s r+1 Maximum value of utility value in the act of (a); a is m Indicating an action, a 1 Represents rank plus 1, a 2 Representing rank invariant;in order to learn the rate of speed,μ is a discount factor, μ ∈ (0, 1), ω i As a function of revenue:
wherein, P t (A, B) represents the encounter probability, and the subscript t represents the encounter times of the two nodes;
the target node receives the coded packet of the relay node, when the rank of the target node reaches K (namely full rank), the target node can decode K original data packets through matrix inversion, and the target node replies response information to the source node by adopting a Prophet routing protocol to inform the source node that generation data has been received.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a diagram of a delay tolerant network topology.
Detailed Description
The present invention will be further illustrated below with reference to specific embodiments, which are to be understood as merely illustrative and not limitative of the scope of the present invention.
1) As shown in FIG. 2, the newly-built delay tolerant network has 10 nodes, wherein the node 1 is a source node and the node 10 is a destination node, and the meeting probability P of each node and other nodes is initialized init Each node has a Q value of 0.70, and in the source node, a batch of packets are put together for random linear network coding into one generation. One generation original data packet with the number K being 8A linear combination of these 8 packets is:
whereinIs a finite field with the size q being 256 and the coefficient alpha being (alpha) 1 ,…,α K ) Called the code vector and the resulting linear combination x called the code packet. Two or more code packets are said to be linearly independent if their code vectors are linearly independent. If a node carries a set of code packets containing at most r linearly independent code packets x 1 ,…,x r We say that the rank of this node is r, and refer to r code vectors as r × K order matrix a of the r code vectors as the code matrix of this node. The source node forwards the coded packet by using a Prophet routing protocol;
2) as shown in fig. 2, when the relay node 5 encounters the nodes 3 and 8, it is assumed that the node 5 has 4 code packets and encounters the node 3, the node 3 has 3 code packets, i, j ∈ (1, …, k), and the nodes 5 and 3 both have the probability of the same code packet:
each time node 5 encounters another node, the rank is increased by 1 if a linearly independent coded packet can be received, otherwise, the rank is not changed until the full rank r is 8. The state change of the rank is only related to the current state and is not related to the historical state, and the process that the node 5 receives the linear independent coding packet is a Markov chain and reaches an absorption state when the rank is full. When node 5 has 4 code packets, the probability of receiving a linearly independent code packet from node 3 is state s r To s r+1 The one-step transition probability of (1):
3) as shown in fig. 2, the neighbor nodes of the node 5 are the nodes within the range of the communication radius R, the node 3 and the node 8, and assuming that the Q value of the node 3 is greater than that of the node 8, the node 5 selects the node 3 to exchange the coded packet, and updates the Q value of the node 5:
wherein s is r State representing node rank, maxQ(s) r+1 ,a m ) Indicating the next state in memory s r+1 Maximum value of utility value in the act of (a); a is m Indicating an action, a 1 Represents rank plus 1, a 2 Representing rank invariant; learning rateThe discount factor μ is 0.95, and if node 5 and node 3 meet at 2 nd time, the attenuation factor γ is 0.95, the unit time is 30 seconds, T is 10 before 1 st meeting, and the meeting probability is attenuated as follows:
P (5,3) =P (5,3)old ×0.95 10 =0.70×0.60=0.42
1 st encounter probability update value:
P (5,3) =P (5,3)old +(1-P (5,3)old )×P init =0.42+(1-0.42)×0.70=0.83
the same way can be obtained that the 2 nd encounter probability is updated to 0.80.
ω i As a function of revenue:
if the Q value before the node 5 meets the node 3 is 0.53 and the maximum effective Q value in the memory is 0.58, the updated Q value of the node 5 is:
Q(s r ,a 1 )=(1-0.1)×0.53+0.1[0.76+0.95×0.58]=0.61
5) the destination node receives the coded packet of the relay node, and when the rank of the destination node reaches r ═ 8 (namely, full rank), the destination node can decode K original data packets by matrix inversion. The destination node replies response information to the source node by adopting a Prophet routing protocol, and informs the source node that generation data is received.
Compared with the prior art, the invention has the following advantages:
1. in the delay tolerant network, a random network coding technology is applied to a Prophet routing protocol, after an original data packet is coded, a coding packet received by an intermediate node contains information of a plurality of original data packets, and the intermediate node also has coding capacity, so that the times of data transmission are reduced, and the throughput of data transmission is improved.
2. In a Prophet routing protocol, a Q-learning delay tolerant network random network coding method is adopted, how to efficiently obtain a linearly independent coding packet by an intermediate node is the key for improving coding benefit, a gain function of Q learning not only considers the meeting probability of two nodes, but also provides judgment of transition probability of node rank increase, so that the linearly independent coding packet is efficiently obtained under the condition of higher meeting probability, and the contradiction that the two nodes cannot meet due to overlarge meeting probability of the two nodes, the node cannot obtain the linearly independent coding packet and undersize meeting probability is solved.
Claims (1)
1. A delay tolerant network random network coding method based on Q learning is characterized in that: the method comprises the following specific steps:
(1) initializing the probability of encounter P of each node with other nodes in a delay tolerant network init The Q value initial value of each node is 0, in the source node, a batch of data packets which are put together for random linear network coding are classified, and one classified original data packet with the number of K is consideredOne linear combination of these K packets is:
whereinIs a finite field of size q, the coefficient α being (α) 1 ,…,α K ) Called code vector, the generated linear combination x is called code packet, if the code vectors of two or more code packets are linearly independent, the code packets are said to be linearly independent, if a certain node carries a set of code packets containing at most r linearly independent code packets x 1 ,…,x r Let us say that the rank of the node is r, and the parallel vector is an r × K matrix of r coding vectors as the coding matrix of the node, and the source node forwards the coded packet by using a Prophet routing protocol;
(2) the relay node meets other nodes once, if a linear independent coding packet can be received, the rank of the coding matrix of the node is added with 1, otherwise, the rank is not changed until the full rank r of the coding matrix is K, the state change of the rank is only related to the current state and is not related to the historical state, therefore, the process that the node receives the linear independent coding packet is a Markov chain, when the full rank is reached, the absorption state is reached, the probability that the node can receive the linear independent coding packet from the meeting node is calculated, and the probability is the one-step transition probability of the rank state;
the probability calculation process of receiving a linear independent coding packet from the encountering node comprises the following steps:
when two relay nodes (assumed as a node a and a node B) meet each other, it is assumed that the node a has i encoded packets and meets the node B, the node B has j encoded packets, i, j belongs to [1, …, k ], and the node a and the node B both have the probability of the same encoded packet:
when the node A has i code packets, the probability that a linear independent code packet can be received from the node B is the rank state s r To s r+1 The one-step transition probability of (1):
wherein the content of the first and second substances,indicating the probability that the setting of the size of the finite field causes the linear independent coding packet to be received by the node;
(3) the neighbor nodes of the relay nodes are nodes within the communication radius range, when each relay node meets other nodes, the node with the maximum Q value of the neighbor node is selected to exchange the coding packet, if the Q values of the neighbor nodes are the same, the node with more coding packets is selected to exchange the coding packet, and the Q value of the node is updated;
and calculating the Q value:
wherein s is r State representing the node rank, max Q(s) r+1 ,a m ) Indicating the next state in memory s r+1 Maximum value of utility value in the act of (a); a is m Indicating an action, a 1 Represents rank plus 1, a 2 Representing rank invariant;in order to learn the rate of speed,μ is a discount factor, μ ∈ (0, 1), ω i As a function of gain:
wherein, P t (A, B) represents the encounter probability, and the subscript t represents the encounter times of the two nodes;
(4) the target node receives the coded packet of the relay node, when the rank of the target node reaches K (namely full rank), the target node can decode K original data packets through matrix inversion, and the target node replies response information to the source node by adopting a Prophet routing protocol to inform the source node that generation data has been received.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010597538.0A CN111770546B (en) | 2020-06-28 | 2020-06-28 | Delay tolerant network random network coding method based on Q learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010597538.0A CN111770546B (en) | 2020-06-28 | 2020-06-28 | Delay tolerant network random network coding method based on Q learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111770546A CN111770546A (en) | 2020-10-13 |
CN111770546B true CN111770546B (en) | 2022-09-16 |
Family
ID=72722133
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010597538.0A Active CN111770546B (en) | 2020-06-28 | 2020-06-28 | Delay tolerant network random network coding method based on Q learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111770546B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112564712B (en) * | 2020-11-26 | 2023-10-10 | 中国科学院计算技术研究所 | Intelligent network coding method and equipment based on deep reinforcement learning |
CN114374613B (en) * | 2022-01-11 | 2023-09-15 | 江西理工大学 | Vehicle-mounted delay tolerant network coding maximum stream setting method based on soft interval support vector machine |
CN114980185A (en) * | 2022-05-12 | 2022-08-30 | 重庆邮电大学 | Vehicle-mounted self-organizing network routing method based on topological evolution |
CN115134374B (en) * | 2022-06-29 | 2023-08-01 | 中国人民解放军国防科技大学 | Directory data synchronization method and device for opportunistic network |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103346858A (en) * | 2013-06-03 | 2013-10-09 | 哈尔滨工业大学深圳研究生院 | System LT code compiling method based on superposition degree |
CN106953803A (en) * | 2017-02-28 | 2017-07-14 | 北京科技大学 | A kind of improvement probabilistic routing method based on deep space delay-tolerant network |
CN107995114A (en) * | 2017-12-08 | 2018-05-04 | 江西理工大学 | Delay Tolerant Network method for routing based on Density Clustering |
CN109121105A (en) * | 2018-09-17 | 2019-01-01 | 河海大学 | Operator's competition slice intensified learning method based on Markov Game |
CN110083064A (en) * | 2019-04-29 | 2019-08-02 | 辽宁石油化工大学 | A kind of network optimal track control method based on non-strategy Q- study |
CN110138432A (en) * | 2019-05-16 | 2019-08-16 | 哈尔滨工业大学(深圳) | DTN data transmission method based on network code and relaying caching auxiliary |
WO2019231289A1 (en) * | 2018-06-01 | 2019-12-05 | Samsung Electronics Co., Ltd. | Method and apparatus for machine learning based wide beam optimization in cellular network |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2469496A1 (en) * | 2010-12-23 | 2012-06-27 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Concept for encoding data defining coded positions representing a trajectory of an object |
WO2017087755A1 (en) * | 2015-11-20 | 2017-05-26 | Veniam, Inc. | Systems and methods for network coded mesh networking in a network of moving things |
-
2020
- 2020-06-28 CN CN202010597538.0A patent/CN111770546B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103346858A (en) * | 2013-06-03 | 2013-10-09 | 哈尔滨工业大学深圳研究生院 | System LT code compiling method based on superposition degree |
CN106953803A (en) * | 2017-02-28 | 2017-07-14 | 北京科技大学 | A kind of improvement probabilistic routing method based on deep space delay-tolerant network |
CN107995114A (en) * | 2017-12-08 | 2018-05-04 | 江西理工大学 | Delay Tolerant Network method for routing based on Density Clustering |
WO2019231289A1 (en) * | 2018-06-01 | 2019-12-05 | Samsung Electronics Co., Ltd. | Method and apparatus for machine learning based wide beam optimization in cellular network |
CN109121105A (en) * | 2018-09-17 | 2019-01-01 | 河海大学 | Operator's competition slice intensified learning method based on Markov Game |
CN110083064A (en) * | 2019-04-29 | 2019-08-02 | 辽宁石油化工大学 | A kind of network optimal track control method based on non-strategy Q- study |
CN110138432A (en) * | 2019-05-16 | 2019-08-16 | 哈尔滨工业大学(深圳) | DTN data transmission method based on network code and relaying caching auxiliary |
Non-Patent Citations (5)
Title |
---|
Dynamical coding method for intermittent connectivity wireless network;P. Yang, D. Liu and Y. Zhang;《The 2014 2nd International Conference on Systems and Informatics (ICSAI 2014)》;20150115;全文 * |
Enhancement of PRoPHET routing;B. B. Bista and D. B. Rawat;《2016 IEEE Region 10 Conference (TENCON)》;20170209;全文 * |
容迟网络编码节点状态感知的数据转发策略;王汝言等;《重庆邮电大学学报(自然科学版)》;20130415;全文 * |
延迟容忍网络中基于流间网络编码的传输策略研究;彭雷;《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》;20180215;全文 * |
自由空间光通信中基于Q学习算法的Raptor10码译码策略;敖珺,梁积卫等;《中国激光》;20170519;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN111770546A (en) | 2020-10-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111770546B (en) | Delay tolerant network random network coding method based on Q learning | |
CN107547436B (en) | Coding and decoding method of transmission control protocol based on network coding | |
CN109347604B (en) | Multi-hop network communication method and system based on batched sparse codes | |
CN109041010B (en) | Vehicle-mounted delay tolerant network data forwarding method based on half-naive Bayes classifier | |
CN102006237A (en) | Routing decision method for delay tolerant network | |
CN106953803B (en) | Improved probability routing method based on deep space delay tolerant network | |
CN111917512A (en) | High-efficiency coding design method based on BATS code in multi-hop transmission system | |
CN101667885A (en) | Method for reducing redundancy message by using network coding technique on DTN or ICN network | |
CN107994971B (en) | Coding transmission method and coding communication system for limited buffer area relay link | |
Darabkh | Queuing Analysis and Simulation of Wireless Access and End Point Systems using Fano Decoding. | |
Han et al. | NCAnt: A network coding-based multipath data transmission scheme for multi-UAV formation flying networks | |
CN113490239B (en) | Heterogeneous wireless link concurrent transmission control method based on adaptive network coding | |
CN111970731B (en) | Spray and Wait mobile opportunistic network routing method based on node reputation | |
CN110430011B (en) | BATS code coding method based on regular variable node degree distribution | |
CN107995114A (en) | Delay Tolerant Network method for routing based on Density Clustering | |
CN110536309B (en) | Mobile social network routing method based on node activity and energy factors | |
CN115843083B (en) | Underwater wireless sensor network routing method based on multi-agent reinforcement learning | |
CN116113008A (en) | Multi-agent routing algorithm for unmanned aerial vehicle self-organizing network | |
Budyal et al. | Intelligent agent based delay aware QoS unicast routing in mobile ad hoc networks | |
CN103427951B (en) | With the data forwarding method that coding redundancy controls | |
CN111865488B (en) | Code selection method for multi-hop short packet communication | |
CN114172619A (en) | Network communication method based on distributed batch sparse codes | |
CN108429683B (en) | Network data routing method, system and device | |
CN113852569B (en) | Data packet coding and real-time transmission method using hysteresis feedback information | |
Facenda et al. | Deep Reinforcement Learning for Latency-Sensitive Communication With Adaptive Redundant Retransmissions |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |