A satellite network-oriented adaptive forwarding method

Abstract

This invention relates to the field of satellite network communication technology, specifically an adaptive forwarding method for satellite networks. This method is applied to satellite forwarding nodes in named data networks. When an interest packet misses its content cache and there are no aggregable entries in the pending interest table, an initial set of candidate interfaces is generated based on a forwarding information database. Available interfaces are supplemented and filtered by combining link switching timetables and detection results, and the risk of reverse path breakage is predicted based on the data return window. Then, interface availability is generated based on the predicted return time, throughput, packet loss rate, and interface queue size in the interface status table, and the forwarding probability and target outgoing interface are determined by combining routing performance values. This invention aims to improve the continuity of interest packet forwarding and data packet return in dynamic satellite networks.

Description

Technical Field

[0001] This invention relates to the field of satellite network communication technology, specifically to an adaptive forwarding method for satellite networks. Background Technology

[0002] Low-Earth orbit (LEO) satellite networks typically achieve cross-regional data transmission through inter-satellite links, satellite-to-ground links, and ground access nodes. Changes in the relative positions of satellite nodes cause the network topology to exhibit significant time-varying characteristics. Named data networks use content names as the forwarding object. They typically use content caching, a pending interest table, and a forwarding information database to handle interest packet forwarding and data packet return. The content cache is used to return data packets from the nearest available source, the pending interest table records unmet interest packets and their incoming interfaces, and the forwarding information database provides the next-hop interface corresponding to the name prefix. Existing forwarding methods mostly rely on routing protocols to update the forwarding information database and, upon receiving an interest packet, match the next-hop interface based on the name to achieve content name-based data request forwarding.

[0003] However, in low-Earth orbit satellite networks, inter-satellite links undergo frequent planned switching due to satellite motion. Space interference, antenna alignment misalignment, and link obstruction can also lead to sudden link anomalies, making it difficult for the forwarding database to reflect interface recovery, interruption, or temporary failure status in a timely manner. When selecting an interface solely based on the forwarding database, interest packets may be forwarded to interfaces that are about to be interrupted or are actually unavailable, resulting in invalid forwarding and request waiting. Simultaneously, data packets rely on the reverse path recorded in the pending interest table for return. If the link on the reverse path within the data return window is broken, even if the interest packet has been successfully forwarded, the data packet may not be able to return along the original path. Therefore, an adaptive forwarding method for satellite networks is needed to solve these problems. Summary of the Invention

[0004] To address the above problems, this invention provides the following technical solution: an adaptive forwarding method for satellite networks, applied to satellite forwarding nodes in named data networks, comprising: S1. When an interest packet is received but the content cache is not hit and there is no aggregateable entry in the pending interest table, a pending interest entry is created, and the forwarding information database is matched according to the interest packet name to obtain an initial candidate interface set. S2. Based on the link switching timetable, determine the interfaces that have been restored to connectivity but have not been synchronized to the forwarding information database, and form the first fast recovery interface set; based on the successful detection results, determine the interfaces that have not been synchronized to the forwarding information database, and form the second fast recovery interface set, and assign initial state attributes and initial routing performance values ​​to the fast recovery interfaces; S3. Based on the link switching timetable and detection results, filter the initial candidate interface set and the first fast recovery interface set, and merge them with the second fast recovery interface set to obtain the final available interface set; S4. Determine the data packet return path based on the incoming interface information in the pending interest entries, and combine the data return window and link switching timetable to perform reverse path breakage risk filtering on the final available interface set to obtain the target available interface set. S5. Read the status attributes of each interface in the target available interface set, obtain the interface availability through normalization and entropy weighting, and generate the forwarding probability by combining the routing performance value of each interface. S6. Determine the target outgoing interface according to the forwarding probability and forward the interest packet. Update the interface status table during the data packet return phase and update the detection results and the second fast recovery interface set during the detection phase.

[0005] Furthermore, the interfaces that have regained connectivity but have not been synchronized to the forwarding information database, determined according to the link switching timetable, form a first fast recovery interface set; the interfaces that have not been synchronized to the forwarding information database, determined according to the successful detection results, form a second fast recovery interface set, including: Based on the link switching timetable, identify the interface that has restored physical connectivity at the current forwarding time but has not been synchronized to the forwarding information database, and add the interface to the first fast recovery interface set; Based on the detection results, identify the interfaces that were successfully detected within the validity period of the results but were not synchronized to the forwarding information database, and add the interfaces to the second fast recovery interface set. Assign initial state attributes and initial routing performance values ​​to the interfaces that are added to the first fast recovery interface set and the second fast recovery interface set.

[0006] Furthermore, the link switching timetable is generated based on satellite ephemeris, relative orbital plane position, and inter-satellite link establishment rules. The link switching timetable includes the connection start time, connection end time, and planned interruption interval for each interface between adjacent satellite nodes.

[0007] Furthermore, the initial candidate interface set and the first fast recovery interface set are filtered based on the link switching timetable and detection results, and then merged with the second fast recovery interface set to obtain the final available interface set, including: According to the link switching timetable, interfaces whose current forwarding time is in the planned interruption interval are removed from the initial candidate interface set; Based on the most recent detection result, interfaces that failed the detection and whose detection results are within the validity period are removed from the initial candidate interface set after removal and the first fast recovery interface set; The final available interface set is obtained by combining the retained initial candidate interface set, the retained first fast recovery interface set, and the second fast recovery interface set.

[0008] Furthermore, the process involves determining the data packet return path based on the incoming interface information in the pending interest entries, and combining this with the data return window and link switching timetable to perform reverse path breakage risk filtering on the final available interface set, thereby obtaining the target available interface set, including: The data packet return path is determined based on the incoming interface information recorded in the pending interest entries; For each interface in the final set of available interfaces, determine the data return window corresponding to that interface; Read the earliest connection termination time of the link between adjacent satellite nodes in the data packet return path from the link switching time table; If the sum of the current forwarding time and the data return window is later than the earliest connection termination time, the interface is removed from the final available interface set. The interfaces that were not removed are identified as the target set of available interfaces.

[0009] Furthermore, the data return window is determined based on the maximum return time of the interface in the cache hit sample when the interface cache hit probability is not less than the cache hit probability threshold; and based on the maximum round-trip latency of the interface in the producer return sample when the interface cache hit probability is less than the cache hit probability threshold.

[0010] Furthermore, the status attributes include predicted return time, throughput, packet loss rate, and interface queue size. The predicted return time is determined based on the interface cache hit probability, cache hit return latency, and producer round-trip latency. The throughput is determined based on the number of data packets transmitted through the interface within the statistical window. The packet loss rate is determined based on the number of interest packets sent through the interface and the number of returned data packets within the statistical window. The interface queue size is determined based on the data packet queue length of the interface.

[0011] Furthermore, the step of reading the status attributes of each interface in the target available interface set, and obtaining the interface availability after normalization and entropy weighting, includes: Predicted return time, packet loss rate, and interface queue size are normalized as cost-based metrics, while throughput is normalized as a revenue-based metric. Construct an attribute proportion matrix based on the normalized state attributes, and calculate the information entropy of each state attribute based on the attribute proportion matrix. Determine the corresponding entropy weight based on the information entropy of each state attribute; The interface availability is generated by weighting and summing the normalized state attributes according to the entropy weight.

[0012] Furthermore, the step of generating forwarding probabilities by combining the routing performance values ​​of each interface includes: For interfaces originating from the initial candidate interface set, the hop count of the corresponding path in the forwarding information database is converted into a routing performance value; For interfaces originating from the first fast recovery interface set and interfaces originating from the second fast recovery interface set, the initial routing performance value allocated when the interface is added to the corresponding fast recovery interface set is used as the routing performance value. The normalized forwarding probability of each interface is generated by the exponentially weighted product of the interface availability and the routing performance value. The target output interface is selected from the set of available target interfaces according to the roulette wheel selection method.

[0013] Furthermore, updating the interface status table during the data packet return phase and updating the detection results and the second fast recovery interface set during the detection phase includes: When the satellite relay node receives a data packet corresponding to the interest packet from the target output interface, it sends back the data packet and clears the pending interest entries according to the incoming interface information recorded in the pending interest entries, updates the interface status table according to the transmission information of the data packet, and writes the data packet into the content cache. Maintain a probe timer for each interface and send single-hop probe interest packets. Perform two fast reprobes when the probe times out. Mark the probe result of the corresponding interface as failed after three consecutive probe timeouts. Mark the probe result of the corresponding interface as successful when a probe response packet is received. If the interface with successful probe is not synchronized to the forwarding information database, add the interface to the second fast recovery interface set and assign initial state attributes and initial routing performance values.

[0014] Compared with the prior art, the present invention has the following beneficial effects: An adaptive forwarding method for satellite networks generates a set of available interfaces by combining link switching timetables, probe results, and a forwarding information database before forwarding interest packets. Furthermore, it predicts the risk of reverse path breakage based on data return windows, enabling interest packets to avoid interfaces prone to planned interruptions, probe failures, and potential data packet backhaul obstructions, thus reducing forwarding path failures caused by frequent satellite link switching. Simultaneously, an interface status table records the predicted return time, throughput, packet loss rate, and interface queue size for each interface. After normalization and entropy weighting, interface availability is generated and combined with routing performance values ​​to form probabilistic forwarding decisions. This allows forwarding nodes to select the most suitable outgoing interface based on the real-time data plane status when routing information is not yet synchronized or when local link states change abruptly, thereby improving the continuity and reliability of interest packet forwarding and data packet backhaul. Attached Figure Description

[0015] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a system model diagram of the present invention; Figure 3 Flowchart for forwarding interest packages; Figure 4 Here is a flowchart of the data packet return process; Figure 5 This is a comparison chart of the average round-trip latency performance of various forwarding schemes in the scenario of the first group of request nodes and content source nodes; Figure 6 This is a comparison chart of the average round-trip latency performance of various forwarding schemes in the scenario of the second group of request nodes and content source nodes; Figure 7 This is a comparison chart of the average round-trip latency performance of various forwarding schemes in the scenario of the third group of request nodes and content source nodes; Figure 8 A comparison chart of interest satisfaction rates for various forwarding schemes under different combinations of request nodes and content source nodes; Figure 9 A comparison chart of the average throughput of various forwarding schemes under different combinations of request nodes and content source nodes; Figure 10 A comparison chart of the average round-trip time performance of various forwarding schemes under different link failure probabilities; Figure 11 A comparison chart of interest satisfaction rates for various forwarding schemes under different link failure probabilities; Figure 12 A comparison chart of the average throughput of various forwarding schemes under different link failure probabilities; Figure 13 A comparison chart of the average round-trip time of the named data link state routing protocol and the present invention at different runtimes; Figure 14 A comparison chart of throughput and interest satisfaction rate under different protocol combinations. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The following is combined Figures 1 to 5 The embodiments of the present invention will be described below. Figure 1 The overall process of the method of the present invention is shown. Figure 2 The system model of the present invention is shown. Figure 3 The interest packet forwarding process is shown. Figure 4 The data packet return process is shown.

[0019] In this embodiment, an adaptive forwarding method for satellite networks is applied to satellite forwarding nodes in a named data network. The satellite forwarding node includes a content cache, a pending interest table, a forwarding information database, and an interface status table. A link switching timetable, probe result records, a first fast recovery interface set, and a second fast recovery interface set are configured in the forwarding plane. The content cache stores returned data packets, the pending interest table records unmet interest packets and their incoming interface information, the forwarding information database stores next-hop interfaces based on name prefixes, and the interface status table records state attributes such as predicted return time, throughput, packet loss rate, and interface queue size.

[0020] In low-Earth orbit satellite networks, as satellites move along their orbits, inter-satellite links between orbital systems may experience planned interruptions due to entering polar regions, increased relative speed, or changes in antenna pointing conditions. When an interest packet has reached a content source or cache node along a certain path, and the data packet needs to return along the reverse path recorded in the pending interest table, if the adjacent satellite links on that reverse path are broken within the data packet return window, the data packet cannot return along the original path, resulting in a reverse path partitioning phenomenon. This embodiment avoids interfaces that may cause return failures by performing link prediction, interface status evaluation, and data return window determination before interest packet forwarding.

[0021] like Figure 2 As shown, the system model in this embodiment includes an interface extraction module, an interface availability determination and supplementation module, a link status prediction module, a detection module, a reverse path breakage risk prediction module, and a multi-attribute probabilistic decision-making module. The interface extraction module extracts initial candidate interfaces from the forwarding information database based on the packet of interest name; the interface availability determination and supplementation module supplements interfaces for rapid recovery based on the link switching timetable and detection results; the link status prediction module determines the link connectivity time based on satellite ephemeris and inter-satellite link establishment rules; the detection module detects sudden failures or sudden recovery interfaces; the reverse path breakage risk prediction module eliminates interfaces with backhaul risks based on the data packet return path and data return window; and the multi-attribute probabilistic decision-making module determines the final outgoing interface based on interface status attributes and routing performance values.

[0022] S1. When an interest packet is received but the content cache is not hit and there is no aggregateable entry in the pending interest table, a pending interest entry is created, and the forwarding information database is matched according to the interest packet name to obtain an initial candidate interface set. Specifically, such as Figure 3 As shown, when a satellite relay node receives an interest packet at the current relay time, it first queries the content cache using the content name carried by the interest packet as an index. If a data packet corresponding to the content name exists in the content cache, the data packet is returned directly without executing the subsequent interface selection process. If the content cache is not found, the pending interest table is queried. If an entry with the same content name or an aggregable name prefix already exists in the pending interest table, the incoming interface of the interest packet is appended to the existing pending interest entry, enabling subsequent data packets to be returned to multiple request directions simultaneously. If no aggregable entry exists in the pending interest table, a new pending interest entry is created, and information such as the interest packet name, incoming interface, reception time, and lifecycle are recorded.

[0023] Specifically, after creating pending interest entries, the satellite forwarding node performs longest prefix matching in the forwarding information database based on the interest packet name, extracting one or more potential next-hop interfaces corresponding to the matching item to form an initial candidate interface set. The initial candidate interface set reflects the known forwardable directions of the current routing information. However, due to the rapid changes in satellite link status, the initial candidate interface set may omit interfaces that have been restored but have not yet been synchronized to the forwarding information database, or it may include interfaces that are about to be interrupted or whose probes have failed. Therefore, it is necessary to supplement and filter the set in subsequent steps by combining the link switching timetable and probe results.

[0024] S2. Based on the link switching timetable, determine the interfaces that have been restored to connectivity but have not been synchronized to the forwarding information database, and form the first fast recovery interface set; based on the successful detection results, determine the interfaces that have not been synchronized to the forwarding information database, and form the second fast recovery interface set, and assign initial state attributes and initial routing performance values ​​to the fast recovery interfaces; Specifically, the link switching timetable is generated from satellite ephemeris, relative orbital plane positions, and inter-satellite link establishment rules. It describes the start and end times of connectivity between interfaces between adjacent satellite nodes, as well as planned interruption intervals. Satellite relay nodes query the link switching timetable at the current relay time. If an interface has restored physical connectivity, but the relay information database has not yet updated it as an available next hop, then that interface is added to the first fast-recovery interface set. This process allows the data plane to temporarily utilize restored links before routing information synchronization is complete, reducing the window period caused by routing convergence lag after link recovery.

[0025] Furthermore, in one embodiment provided by this application, the step of determining interfaces that have restored connectivity but have not been synchronized to the forwarding information database according to the link switching timetable to form a first fast recovery interface set; and determining interfaces that have not been synchronized to the forwarding information database according to the valid detection success results to form a second fast recovery interface set, includes: determining interfaces that have restored physical connectivity at the current forwarding time but have not been synchronized to the forwarding information database according to the link switching timetable, and adding such interfaces to the first fast recovery interface set; determining interfaces that were successfully detected within the validity period of the detection results but have not been synchronized to the forwarding information database according to the detection results, and adding such interfaces to the second fast recovery interface set; and assigning initial state attributes and initial routing performance values ​​to the interfaces added to the first fast recovery interface set and the second fast recovery interface set.

[0026] Specifically, the second fast recovery interface set is used to supplement unplanned recovery interfaces. The probe module saves the most recent probe result and the validity period of the result for each interface. When the most recent probe result of an interface is successful, and the successful probe result has not exceeded the validity period, but the interface has not yet been synchronized to the forwarding information database, the interface is added to the second fast recovery interface set. Interfaces added to the first or second fast recovery interface set may lack complete historical state, so initial state attributes and initial routing performance values ​​are assigned to them. The initial state attributes can be the average value of interfaces of the same type, the adjacent historical window value, or a preset conservative value; the initial routing performance value can be determined based on the topological distance of the corresponding inter-satellite link or the default hop count estimate. In this way, fast recovery interfaces can participate in subsequent probabilistic forwarding calculations without being excluded due to a lack of historical data.

[0027] Furthermore, in one embodiment provided in this application, the link switching timetable is generated based on satellite ephemeris, relative position of orbital plane and inter-satellite link establishment rules, and the link switching timetable includes the connection start time, connection end time and planned interruption interval of each interface between adjacent satellite nodes.

[0028] Specifically, satellite ephemeris provides the satellite's spatial position at different times, and the relative position on the orbital plane is used to determine whether the geometric conditions for establishing an inter-satellite link are met between adjacent orbital satellites. Inter-satellite link establishment rules are used to define the conditions for link establishment, maintenance, and interruption. Based on this information, connected and interrupted intervals can be formed for each interface in time-ordered sequence. When forwarding interest packets, forwarding nodes only need to read the link status within the current time and prediction window to determine whether the interface is suitable for forwarding interest packets and reverse transmitting data packets.

[0029] Specifically, the link switching timetable is generated from satellite ephemeris, relative orbital plane positions, and inter-satellite link establishment rules. It describes the start and end times of connectivity between interfaces between adjacent satellite nodes, as well as planned interruption intervals. Satellite relay nodes query the link switching timetable at the current relay time. If an interface has restored physical connectivity, but the relay information database has not yet updated it as an available next hop, then that interface is added to the first fast-recovery interface set. This process allows the data plane to temporarily utilize restored links before routing information synchronization is complete, reducing the window period caused by routing convergence lag after link recovery.

[0030] S3. Based on the link switching timetable and detection results, filter the initial candidate interface set and the first fast recovery interface set, and merge them with the second fast recovery interface set to obtain the final available interface set; Specifically, the satellite relay node first performs planned interruption filtering on the initial candidate interface set. For any interface in the initial candidate interface set, it reads the status of that interface at the current relay time from the link switching time table; if the current relay time is within the planned interruption interval of that interface, it means that the link reachability cannot be guaranteed after the interest packet is relayed through that interface, and the interface is removed from the initial candidate interface set; if the current relay time is within the planned connectivity interval, the interface is temporarily retained.

[0031] Furthermore, in one embodiment provided by this application, the step of filtering the initial candidate interface set and the first fast recovery interface set according to the link switching timetable and the detection results, and merging them with the second fast recovery interface set to obtain the final available interface set includes: removing interfaces from the initial candidate interface set whose current forwarding time is within the planned interruption interval according to the link switching timetable; removing interfaces whose detection failed and whose detection results are within the validity period from the removed initial candidate interface set and the first fast recovery interface set according to the most recent detection results; and merging the retained initial candidate interface set, the retained first fast recovery interface set, and the second fast recovery interface set to obtain the final available interface set.

[0032] Specifically, after planned outage filtering, interfaces experiencing sudden failures need to be excluded based on the most recent probe results. For interfaces in the filtered initial candidate interface set and the first fast recovery interface set, if the most recent probe result is a failure, and that failure result is still valid, it indicates that although the interface may be located in the planned connectivity section, the actual link is unavailable due to hardware anomalies, transient obstructions, or channel interference. Therefore, this interface is further removed. For interfaces in the second fast recovery interface set, whose source is itself a valid successful probe result, they directly participate in the merging process. Finally, the retained initial candidate interface set, the retained first fast recovery interface set, and the second fast recovery interface set form the final available interface set. This set includes both available interfaces already in the forwarding information base and recovery interfaces discovered in advance by the data plane, enabling forwarding decisions to simultaneously possess routing foundations and data plane fast recovery capabilities.

[0033] S4. Determine the data packet return path based on the incoming interface information in the pending interest entries, and combine the data return window and link switching timetable to perform reverse path breakage risk filtering on the final available interface set to obtain the target available interface set. Specifically, data packets in named data networks typically return along the reverse path of the interest packet. Therefore, the incoming interface information recorded in the pending interest table limits the reverse path that the data packet must traverse during return. For each interface in the final set of available interfaces, the satellite relay node not only determines whether the interest packet can be sent, but also whether the corresponding data packet can return along the reverse path recorded in the pending interest table within the expected return time. If only the forward link is determined to be available, while ignoring the data packet return path, then in Figure 2 In the reverse path partitioning scenario shown, interest packets may be successfully forwarded, but the packets are blocked by the broken inter-satellite link on their return journey.

[0034] Furthermore, in one embodiment provided in this application, the step of determining the data packet return path based on the incoming interface information in the pending interest entries, and combining the data return window and link switching timetable to perform reverse path breakage risk filtering on the final available interface set to obtain the target available interface set includes: determining the data packet return path based on the incoming interface information recorded in the pending interest entries; determining the data return window corresponding to each interface in the final available interface set; reading the earliest connection end time of the link between adjacent satellite nodes in the data packet return path from the link switching timetable; removing the interface from the final available interface set when the sum of the current forwarding time and the data return window is later than the earliest connection end time; and determining the interfaces that have not been removed as the target available interface set.

[0035] Specifically, let the current forwarding time be... ,interface The corresponding data return window is The earliest connection termination time of the link between adjacent satellite nodes in the data packet return path is When the following conditions are met: When the following condition is met: This indicates that at least one segment of the link on the reverse path has been broken before the expected return of the data packet, the interface poses a risk of reverse path breakage and should be removed from the final set of available interfaces; If the data packet can return while the reverse path remains connected, the interface can be retained in the target set of available interfaces. This decision ensures that interface selection considers not only the forward transmission of interest packets but also the security of data packet return.

[0036] Furthermore, in one embodiment provided in this application, the data return window is determined based on the maximum return time of the interface in the cache hit sample when the interface cache hit probability is not less than the cache hit probability threshold; and based on the maximum round-trip latency of the interface in the producer return sample when the interface cache hit probability is less than the cache hit probability threshold.

[0037] Specifically, the interface cache hit probability is used to distinguish whether a data packet might be returned by a neighboring cache or by the content source node. Let the interface... The cache hit probability is The cache hit probability threshold is ,interface The maximum return time in the cache hit samples is ,interface The maximum round-trip time in the producer-returned sample is The data return window will then be:

[0038] in, The number of packets returned by the interface that were cached and hit within the statistics window, and the number of interest packets sent, are determined as follows:

[0039] in, Represents the statistics window Inner Canon Interface The number of packets that returned as cache hits when the sent interest packets are obtained. Represents the statistics window Inner Canon Interface The number of interest packets sent. If the cache hit probability is high, the maximum return time in the cache hit sample is used as a shorter data return window that closely reflects the cache scenario; if the cache hit probability is low, the maximum round-trip latency in the producer return sample is used as a more conservative data return window to avoid underestimating the data packet return time.

[0040] S5. Read the status attributes of each interface in the target available interface set, obtain the interface availability through normalization and entropy weighting, and generate the forwarding probability by combining the routing performance value of each interface. Specifically, after obtaining the set of available target interfaces, the satellite forwarding node reads the status attributes of each interface from the interface status table. These status attributes describe the actual forwarding capability of the interface under the current network conditions, including not only static routing information but also dynamic indicators obtained from real-time observations of the data plane. Figure 2 In the multi-attribute probabilistic decision-making module, the satellite forwarding node normalizes the status attributes of each interface and performs entropy weight processing to obtain the interface availability, and then combines the routing performance value to generate the forwarding probability.

[0041] Furthermore, in one embodiment provided in this application, the state attributes include predicted return time, throughput, packet loss rate, and interface queue size. The predicted return time is determined based on the interface cache hit probability, cache hit return latency, and producer round-trip latency. The throughput is determined based on the number of data packets transmitted through the interface within the statistical window. The packet loss rate is determined based on the number of interest packets sent through the interface and the number of returned data packets within the statistical window. The interface queue size is determined based on the data packet queue length of the interface.

[0042] Specifically, the predicted return time reflects the time that an interest packet might wait for after being forwarded through the interface before the data packet is returned. Since the data packet may be returned from either a nearby cache or the content source node, the predicted return time can be determined jointly by the cache hit probability, cache hit return latency, and producer round-trip latency.

[0043] in, Interface Predicted return time, Interface Latest cache hit return latency Interface Round-trip latency for forwarding to the content source node and receiving data packets.

[0044] Throughput is used to reflect the data transmission capacity of an interface. Let the statistical window length be... ,interface The number of data packets transmitted within the statistics window is Then the interface throughput is:

[0045] in, Interface In the statistics window The throughput within the window. Packet loss rate reflects the proportion of interfaces that fail to receive a returned data packet after sending an interest packet. Let the statistical window contain the throughput of the interfaces. The number of interest packets sent is The number of returned data packets is The packet loss rate is:

[0046] The size of the interface queue is determined by the length of the interface's data packet queue. The longer the queue, the greater the current forwarding pressure on the interface, and the longer the queuing time for subsequent data packets to return or be forwarded may be.

[0047] Furthermore, in one embodiment provided in this application, the step of reading the state attributes of each interface in the target available interface set and obtaining the interface availability through normalization and entropy weighting includes: normalizing the predicted return time, packet loss rate, and interface queue size as cost indicators, and normalizing the throughput as a revenue indicator; constructing an attribute proportion matrix based on the normalized state attributes, and calculating the information entropy of each state attribute based on the attribute proportion matrix; determining the corresponding entropy weight based on the information entropy of each state attribute; and performing a weighted summation of the normalized state attributes based on the entropy weight to generate the interface availability.

[0048] Specifically, suppose the target has a total of [number] available interfaces. Select one interface Each state attribute participates in the interface evaluation, constructing a decision matrix:

[0049] in, Indicates the first The first interface Each status attribute value. For revenue-related metrics like throughput, a higher value indicates stronger interface capabilities, which is normalized to:

[0050] For cost-related metrics such as predicted return time, packet loss rate, and API queue size, smaller values ​​indicate better API performance, which can be normalized as follows:

[0051] in, and They represent the first The maximum and minimum values ​​of each status attribute across all available target interfaces. If the maximum and minimum values ​​are equal, it indicates that the attribute cannot distinguish the differences between the current interfaces. In this case, the normalization result of the attribute can be set to the same constant to avoid division by zero and maintain the feasibility of the evaluation process.

[0052] After normalization, calculate the first... The state attribute in the th ... Percentage of attributes in each interface:

[0053] Calculate the next Information entropy of each state attribute:

[0054] when When it is zero, let Set to zero. Determine the first [unit] based on information entropy. Entropy weights of state attributes:

[0055] The interface is obtained by weighting and summing the normalized state attributes according to the entropy weight. Interface availability:

[0056] in, Interface Overall availability within the current set of available target interfaces. Indicates the first Entropy weight of each state attribute, Indicates the first The interface in the first Normalized values ​​for each state attribute. Entropy weighting can automatically assign weights based on the degree of difference between each indicator under the current network state. When a certain indicator differs significantly between different interfaces, its role in distinguishing the quality of interfaces becomes more obvious, and its weight is increased accordingly.

[0057] Furthermore, in one embodiment provided in this application, the step of generating forwarding probabilities by combining the routing performance values ​​of each interface includes: for interfaces originating from the initial candidate interface set, converting the hop count of the corresponding path in the forwarding information database into routing performance values; for interfaces originating from the first fast recovery interface set and interfaces originating from the second fast recovery interface set, using the initial routing performance value allocated when the interface is added to the corresponding fast recovery interface set as the routing performance value; generating a normalized forwarding probability for each interface based on the exponentially weighted product of the interface availability and the routing performance value; and selecting a target outgoing interface from the target available interface set according to a roulette wheel selection method.

[0058] Specifically, interfaces from the initial candidate interface set have corresponding paths in the forwarding information database. The hop count of these paths can be directly read and converted into routing performance values. Since a lower hop count generally indicates a lower path cost, the interfaces can be... The routing performance value is set to:

[0059] in, Interface The number of hops for the corresponding path. Interface The routing performance value. For interfaces originating from the first or second fast recovery interface set, since they have not yet been synchronized to the forwarding information database, the initial routing performance value assigned when joining the fast recovery interface set is used in the calculation.

[0060] Let the exponential weight of interface availability be... The exponential weight of the routing performance value is Then the interface The forwarding probability is:

[0061] in, Interface The probability of being selected as the target outgoing interface. Interfaces with higher availability and better routing performance will receive a higher forwarding probability. At the same time, the roulette wheel selection method will not always select a single interface with the highest score, but will allow other interfaces with a certain level of availability to receive a small number of forwarding opportunities. This avoids all interest packets being concentrated on the same link, causing local congestion, and also allows forwarding nodes to continuously observe the status changes of alternative paths.

[0062] S6. Determine the target outgoing interface according to the forwarding probability and forward the interest packet. Update the interface status table during the data packet return phase and update the detection results and the second fast recovery interface set during the detection phase.

[0063] Specifically, the satellite relay node constructs a probability interval based on the relay probability of each interface, generates random numbers that fall within the corresponding probability interval, and thus determines the target outgoing interface. After determining the target outgoing interface, the relay node sends the interest packet through the target outgoing interface and retains the incoming interface and relay status of the interest packet in the pending interest entry, so that subsequent returned data packets can be processed accordingly. Figure 4 The process shown is returned.

[0064] Furthermore, in one embodiment provided by this application, updating the interface status table during the data packet return phase and updating the detection results and the second fast recovery interface set during the detection phase includes: when the satellite forwarding node receives a data packet corresponding to the interest packet from the target outgoing interface, returning the data packet according to the incoming interface information recorded in the pending interest entries and clearing the pending interest entries, updating the interface status table according to the transmission information of the data packet, and writing the data packet into the content cache; maintaining a detection timer for each interface and sending a single-hop detection interest packet, performing two fast re-detections when the detection times out, marking the detection result of the corresponding interface as failed after three consecutive detection timeouts, marking the detection result of the corresponding interface as successful when a detection response packet is received, and adding the interface to the second fast recovery interface set and assigning initial state attributes and initial routing performance values ​​when the successfully detected interface is not synchronized to the forwarding information database.

[0065] Specifically, such as Figure 4 As shown, when a satellite relay node receives a data packet corresponding to a packet of interest from the target outgoing interface, it first sends the data packet back to the requester along the reverse path according to the incoming interface information recorded in the pending interest entry. If a pending interest entry records multiple incoming interfaces, the data packet is forwarded to the corresponding incoming interface respectively. After the return is completed, the pending interest entries are cleared to prevent satisfied interest packets from continuing to occupy table resources. The return latency, return quantity, byte count, and corresponding interface queue changes of the data packet are used to update the predicted return time, throughput, packet loss rate, and interface queue size in the interface status table, so that the next interest packet forwarding can use the updated interface status. The data packet is also written to the content cache so that subsequent identical or matching interest packets can be directly satisfied by the cache.

[0066] Specifically, for sudden link failures or sudden recovery events where the link switching schedule cannot be predicted in advance, the detection module maintains a detection timer for each interface and sends single-hop detection interest packets. These single-hop detection interest packets are only used to confirm the local availability of the link corresponding to the adjacent interface and do not propagate to multi-hop paths, thus reducing detection overhead. If a detection does not receive a response within a predetermined time, two fast re-probes are immediately executed; if three consecutive detections time out, the detection result for that interface is marked as failed and made available for S3 filtering within its validity period. When a detection response packet is received, the detection result for that interface is marked as successful; if the successfully detected interface has not yet been synchronized to the forwarding information database, it is added to the second fast recovery interface set and assigned initial state attributes and initial routing performance values. These updates enable the interface state table, detection results, and the second fast recovery interface set to be directly invoked during the next interest packet forwarding, forming a closed loop between packet feedback, interface state updates, and subsequent forwarding decisions.

[0067] Combination Figures 5 to 14The experimental results of this invention are explained below. In the experiments, the scheme of this invention was compared with the native optimal routing scheme and the real-time link learning forwarding scheme. The technical effectiveness of this invention was verified by measuring average round-trip latency, interest satisfaction rate, average throughput, and forwarding performance under different link failure conditions.

[0068] like Figure 5 As shown, in the scenario of the first group of request nodes and content source nodes, this group of nodes experiences the process of the inter-track link changing from interruption to recovery. This invention can proactively include new recovery interfaces in the candidate range when the link recovery window arrives using a link switching timetable, and select low-latency paths by combining the interface status table. In the experiment, within the critical switching window, the average round-trip latency of this invention only showed a brief increase, then quickly converged to a low level; compared to the higher latency before link recovery, this invention can also discover new paths with lower latency. The native optimal routing scheme, due to its reliance on a forwarding information database that has not been updated in a timely manner, is prone to forwarding stagnation after the old path fails; while the real-time link learning forwarding scheme can gradually learn the link recovery status, it relies on historical feedback and suffers from response lag. This invention, by supplementing available interfaces in advance through a first fast recovery interface set and a second fast recovery interface set, is superior in terms of proactive new link identification capability and low-latency path selection.

[0069] like Figure 6 As shown, in the scenario of the second group of request nodes and content source nodes, this group of nodes experiences a change in the link from connectivity to planned interruption. Before the physical disconnection of the link, this invention identifies the risk of reverse path failure through a link switching timetable and uses a data return window to identify and eliminate interfaces that may cause data packet return failures, allowing interest packets to switch to a more stable interface before the planned interruption occurs. Experimental results show that the average round-trip latency of this invention remains at a low level overall, with only brief fluctuations near the link switching point, followed by a rapid drop. In contrast, the real-time link learning and forwarding scheme requires going through link failure, packet loss, feedback learning, and reselection processes, making it prone to communication interruptions during switching; the native optimal routing scheme lacks rapid recovery capabilities after link interruption, making it difficult to maintain effective communication.

[0070] like Figure 7As shown, in the scenario of the third group of request nodes and content source nodes, this group of nodes remains in a relatively stable link state region for a long period. This invention, the native optimal routing scheme, and the real-time link learning forwarding scheme all maintain relatively stable average round-trip latency, indicating that in scenarios without significant drastic topology changes, this invention will not disrupt the normal forwarding of stable paths due to the introduction of link prediction, detection, and probabilistic selection. This invention exhibits slight latency fluctuations at individual moments because the roulette wheel selection mechanism attempts other potential interfaces with a low probability, thereby continuously observing alternative paths and avoiding the concentration of all interest packets on a single link. This fluctuation is small and does not affect the overall steady-state forwarding performance.

[0071] like Figure 8 As shown, under different combinations of request nodes and content source nodes, the interest satisfaction rate of this invention is generally better than or no less than that of the comparative scheme. In the first scenario, this invention can quickly discover and utilize the new recovery path after the link is restored, thus significantly improving the interest satisfaction rate compared to the native optimal routing scheme; compared to the real-time link learning and forwarding scheme, it also has a slight advantage because it does not need to wait for a long feedback learning process. In the second scenario, this invention, through reverse path breakage risk prediction, actively avoids high-risk interfaces before planned link disconnection, avoiding the problem that interest packets are successfully forwarded but data packets cannot be returned, thus improving the interest satisfaction rate even more significantly. In the third stable scenario, the interest satisfaction rate of each scheme remains at a high level, indicating that this invention has an enhancing effect in dynamic scenarios and does not reduce normal data delivery capabilities in stable scenarios.

[0072] like Figure 9 As shown, under different combinations of request nodes and content source nodes, the average throughput of this invention is consistent with the trend of interest satisfaction rate. In the first and second dynamic link scenarios, this invention can reduce the forwarding of invalid interest packets caused by failed interfaces, interfaces about to be interrupted, and interfaces with backhaul risks, enabling more interest packets to obtain corresponding data packet returns, thus increasing the amount of data successfully returned per unit time. Especially in the second planned interruption scenario, the prediction of reverse path breakage risk avoids a large number of data packet backhaul failures, resulting in a more significant increase in throughput. In the third stable scenario, the throughput of this invention is close to that of the comparative scheme, indicating that multi-attribute probabilistic selection does not introduce significant additional overhead in stable links.

[0073] like Figures 10 to 12As shown, under different link failure probabilities, the performance of the three forwarding schemes all decreases with increasing failure probability, but the present invention consistently maintains superior performance. When the link failure probability is low, the present invention avoids occasional low-quality interfaces in a timely manner through detection results and interface status tables, keeping the average round-trip time or invalid forwarding overhead low, and the interest satisfaction rate and average throughput high. When the link failure probability is high, the native optimal routing scheme tends to continuously forward interest packets to failed paths, while the real-time link learning forwarding scheme may be affected by the lag in historical link status and choose outdated paths. The present invention, relying on the most recent detection results, interface status tables, and probabilistic forwarding decisions, can switch to effective interfaces more quickly after a sudden failure, reducing the additional overhead caused by waiting, retransmission, and link failure.

[0074] Specifically, in both low and high failure probability scenarios, this invention improves interest satisfaction rate and average throughput compared to the native optimal routing scheme. Compared to the real-time link learning and forwarding scheme, the difference is small in low failure probability scenarios, while the advantages of this invention expand in high failure probability scenarios. This indicates that this invention is applicable not only to predictable orbital link switching scenarios but also to unplanned link failure scenarios caused by equipment malfunctions, obstructions, or space interference. The detection results are used to identify sudden failures, and the link switching timetable is used to identify predictable interruptions. The combination of the two can cover both planned and unplanned changes in satellite network link status.

[0075] like Figure 13 As shown, in a hybrid fault scenario incorporating a named data link state routing protocol, while relying solely on a dynamic routing protocol can restore communication through route convergence after a long-term fault, the convergence process takes time, leaving a service interruption window between the fault occurrence and route convergence completion. Static routing schemes lack dynamic adjustment capabilities after long-term faults, potentially leading to continuous communication interruptions. When this invention collaborates with a named data link state routing protocol, short-term faults are quickly handled by the forwarding plane using probes and interface state awareness, while long-term faults are handled by the routing plane through route convergence, thus enabling a division of labor between the control plane and the forwarding plane. This invention generates only a small fluctuation at the moment of a fault, allowing for rapid switching to an available path and maintaining the continuity of the data retrieval process.

[0076] like Figure 14As shown, the comparison of throughput and interest satisfaction rate under different protocol combinations further illustrates that this invention has a better synergistic effect when combined with the Named Data Link State Routing Protocol. Compared with using the Named Data Link State Routing Protocol alone in conjunction with the native optimal routing scheme, this invention can temporarily select an available interface by the forwarding plane before the routing protocol has converged, and continue to use the updated forwarding information database for decision-making after the routing protocol has been updated. Therefore, this invention does not replace the routing protocol, but rather compensates for the routing protocol when routing convergence is delayed or there are sudden changes in local links. Experimental results show that this invention can improve the average throughput and interest satisfaction rate, indicating that it can enhance the continuity of interest packet forwarding and data packet backhaul in dynamic low-Earth orbit satellite networks.

[0077] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0078] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An adaptive forwarding method for satellite networks, applied to satellite forwarding nodes in named data networks, characterized in that, include: S1. When an interest packet is received but the content cache is not hit and there is no aggregateable entry in the pending interest table, a pending interest entry is created, and the forwarding information database is matched according to the interest packet name to obtain an initial candidate interface set. S2. Based on the link switching timetable, determine the interfaces that have been restored to connectivity but have not been synchronized to the forwarding information database, and form the first fast recovery interface set; based on the successful detection results, determine the interfaces that have not been synchronized to the forwarding information database, and form the second fast recovery interface set, and assign initial state attributes and initial routing performance values ​​to the fast recovery interfaces; S3. Based on the link switching timetable and detection results, filter the initial candidate interface set and the first fast recovery interface set, and merge them with the second fast recovery interface set to obtain the final available interface set; S4. Determine the data packet return path based on the incoming interface information in the pending interest entries, and combine the data return window and link switching timetable to perform reverse path breakage risk filtering on the final available interface set to obtain the target available interface set. S5. Read the status attributes of each interface in the target available interface set, obtain the interface availability through normalization and entropy weighting, and generate the forwarding probability by combining the routing performance value of each interface. S6. Determine the target outgoing interface according to the forwarding probability and forward the interest packet. Update the interface status table during the data packet return phase and update the detection results and the second fast recovery interface set during the detection phase.

2. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The interfaces that have been restored to connectivity but have not been synchronized to the forwarding information database, determined according to the link switching timetable, form a first fast recovery interface set; the interfaces that have not been synchronized to the forwarding information database, determined according to the successful detection results, form a second fast recovery interface set, including: Based on the link switching timetable, identify the interface that has restored physical connectivity at the current forwarding time but has not been synchronized to the forwarding information database, and add the interface to the first fast recovery interface set; Based on the detection results, identify the interfaces that were successfully detected within the validity period of the results but were not synchronized to the forwarding information database, and add the interfaces to the second fast recovery interface set. Assign initial state attributes and initial routing performance values ​​to the interfaces that are added to the first fast recovery interface set and the second fast recovery interface set.

3. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The link switching timetable is generated based on satellite ephemeris, relative orbital plane position and inter-satellite link establishment rules. The link switching timetable includes the connection start time, connection end time and planned interruption interval of each interface between adjacent satellite nodes.

4. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The step involves filtering the initial candidate interface set and the first fast recovery interface set based on the link switching timetable and detection results, and then merging them with the second fast recovery interface set to obtain the final available interface set, including: According to the link switching timetable, interfaces whose current forwarding time is in the planned interruption interval are removed from the initial candidate interface set; Based on the most recent detection result, interfaces that failed the detection and whose detection results are within the validity period are removed from the initial candidate interface set after removal and the first fast recovery interface set; The final available interface set is obtained by combining the retained initial candidate interface set, the retained first fast recovery interface set, and the second fast recovery interface set.

5. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The process involves determining the data packet return path based on the incoming interface information in the pending interest entries, and then filtering the final available interface set for reverse path breakage risk using a combination of the data return window and link switching timetable, to obtain the target available interface set, including: The data packet return path is determined based on the incoming interface information recorded in the pending interest entries; For each interface in the final set of available interfaces, determine the data return window corresponding to that interface; Read the earliest connection termination time of the link between adjacent satellite nodes in the data packet return path from the link switching time table; If the sum of the current forwarding time and the data return window is later than the earliest connection termination time, the interface is removed from the final available interface set. The interfaces that were not removed are identified as the target set of available interfaces.

6. The adaptive forwarding method for satellite networks according to claim 5, characterized in that, The data return window is determined based on the maximum return time of the interface in the cache hit sample when the interface cache hit probability is not less than the cache hit probability threshold; and based on the maximum round-trip latency of the interface in the producer return sample when the interface cache hit probability is less than the cache hit probability threshold.

7. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The status attributes include predicted return time, throughput, packet loss rate, and interface queue size. The predicted return time is determined based on the interface cache hit probability, cache hit return latency, and producer round-trip latency. The throughput is determined based on the number of data packets transmitted through the interface within the statistical window. The packet loss rate is determined based on the number of interest packets sent through the interface and the number of returned data packets within the statistical window. The interface queue size is determined based on the data packet queue length of the interface.

8. The adaptive forwarding method for satellite networks according to claim 7, characterized in that, The process of reading the status attributes of each interface in the target available interface set, and obtaining the interface availability after normalization and entropy weighting, includes: Predicted return time, packet loss rate, and interface queue size are normalized as cost-based metrics, while throughput is normalized as a revenue-based metric. Construct an attribute proportion matrix based on the normalized state attributes, and calculate the information entropy of each state attribute based on the attribute proportion matrix. Determine the corresponding entropy weight based on the information entropy of each state attribute; The interface availability is generated by weighting and summing the normalized state attributes according to the entropy weight.

9. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The process of generating forwarding probabilities by combining the routing performance values ​​of each interface includes: For interfaces originating from the initial candidate interface set, the hop count of the corresponding path in the forwarding information database is converted into a routing performance value; For interfaces originating from the first fast recovery interface set and interfaces originating from the second fast recovery interface set, the initial routing performance value allocated when the interface is added to the corresponding fast recovery interface set is used as the routing performance value. The normalized forwarding probability of each interface is generated by the exponentially weighted product of the interface availability and the routing performance value. The target output interface is selected from the set of available target interfaces according to the roulette wheel selection method.

10. The adaptive forwarding method for satellite networks according to claim 1, characterized in that, The steps of updating the interface status table during the data packet return phase and updating the probe results and the second fast recovery interface set during the probe phase include: When the satellite relay node receives a data packet corresponding to the interest packet from the target output interface, it sends back the data packet and clears the pending interest entries according to the incoming interface information recorded in the pending interest entries, updates the interface status table according to the transmission information of the data packet, and writes the data packet into the content cache. Maintain a probe timer for each interface and send single-hop probe interest packets. Perform two fast reprobes when the probe times out. Mark the probe result of the corresponding interface as failed after three consecutive probe timeouts. Mark the probe result of the corresponding interface as successful when a probe response packet is received. If the interface with successful probe is not synchronized to the forwarding information database, add the interface to the second fast recovery interface set and assign initial state attributes and initial routing performance values.

Images