Link layer retransmission method for satellite communications

Abstract

The application discloses a link layer retransmission method suitable for satellite communication, and integrates a retransmission control sublayer in a general stream encapsulation layer. The method acquires a network layer protocol data unit, allocates a serial number, and stores and encapsulates the network layer protocol data unit into a baseband frame for transmission; receives and analyzes the baseband frame, performs cross-layer loss detection based on a physical layer decoding state, and generates a feedback report containing a data receiving state; receives the feedback report, acquires a real-time channel state, determines a retransmission granularity strategy based on the feedback report and the real-time channel state, and schedules data to be retransmitted; and acquires physical layer signaling, and dynamically maintains a sending window and a retransmission timer. The application breaks the protocol level barrier, realizes low-delay loss awareness and adaptive scheduling, and improves transmission throughput and bandwidth utilization under a satellite burst fading channel.

Description

Technical Field

[0001] This invention belongs to the field of satellite communication technology, specifically a link-layer retransmission method applicable to satellite communication. Background Technology

[0002] Satellite communication systems are irreplaceable in wide-area coverage and emergency communication networks. Due to the physical characteristics of satellite links, such as long transmission distance, large signal attenuation, and high propagation delay, ensuring data transmission in long-latency, highly dynamic air interface environments can further improve the overall throughput and air interface resource utilization of satellite communication networks.

[0003] Currently, digital video broadcasting satellite standards widely adopt the general stream encapsulation protocol to package network layer data into baseband frames for transmission. Under this architecture, the link layer typically only provides basic fragmentation and reassembly and cyclic redundancy check functions, while the task of recovering lost data packets is handled by the upper transport layer control protocol or end-to-end retransmission mechanism. In actual operation, when the satellite link encounters rain attenuation or shadowing causing sudden bit errors, the failure of the underlying physical layer error correction will result in the entire baseband frame and its payload being directly discarded. Faced with data loss at the lower layer, upper-layer protocols often can only passively detect loss events by relying on long acknowledgment timeouts or coarse-grained sequence number gaps, typically responding with a unified whole-packet retransmission strategy and static backoff algorithms.

[0004] Existing technologies struggle to quickly detect packet loss and cannot perform adaptive scheduling. In satellite burst fading channels, ineffective redundant transmissions are easily generated, resulting in low throughput and bandwidth utilization. Long links also face a high risk of congestion and collapse, and the relevant technological systems still need improvement. Summary of the Invention

[0005] Objective of the Invention: In view of the above-mentioned technical problems to be solved by the present invention, an improved technical solution for a link layer retransmission method suitable for satellite communication is provided.

[0006] In view of this, the present invention adopts the following technical solution: A link-layer retransmission method suitable for satellite communications integrates a retransmission control sublayer within the general stream encapsulation layer of the data link layer, comprising: Obtain network layer protocol data units, assign sequence numbers to network layer protocol data units and cache them, and encapsulate network layer protocol data units into baseband frames for transmission; Receive and parse baseband frames, perform cross-layer loss detection based on physical layer decoding status, and generate a feedback report containing data reception status; Receive feedback reports and obtain real-time channel status; determine retransmission granularity strategy based on feedback reports and real-time channel status; and schedule data to be retransmitted. Obtain physical layer signaling and dynamically maintain the transmission window and retransmission timer based on the physical layer signaling and feedback reports.

[0007] Furthermore, scheduling data to be retransmitted specifically includes: The real-time frame drop rate is compared with the preset frame drop rate threshold, and the system switches between normal retransmission, cautious retransmission, and retransmission suppression states based on the comparison result. In the cautious retransmission mode, the retransmission time interval of the data to be retransmitted is increased, and the frequency of sending feedback reports is reduced. In retransmission suppression mode, the scheduling of data to be retransmitted is paused and the retransmission timer is paused.

[0008] Furthermore, the method also includes: After successfully decoding multiple baseband frames in a row and exiting the retransmission suppression state, a centralized recovery process is executed. The centralized recovery process includes: merging the lost information accumulated during the retransmission suppression state to generate a complete feedback report; compactly filling multiple data to be retransmitted into the same baseband frame for centralized retransmission; and temporarily expanding the transmission window based on the amount of data lost accumulated during the retransmission suppression state.

[0009] Furthermore, dynamically maintaining the sending window also includes: Compare the size of the target sending window with the size of the current sending window; When the target sending window is smaller than the current sending window, the shrinkage rate is determined based on the difference between the current sending window and the target sending window, as well as the network round-trip delay. The current sending window is linearly and progressively shrunk according to the shrinking rate until the current sending window is shrunk to the target sending window.

[0010] Furthermore, based on the amount of data lost cumulatively during the retransmission suppression state, the transmission window is temporarily expanded, specifically including: The upper limit of the expanded sending window is limited to no more than half the total capacity of the available sequence number space to prevent acknowledgment ambiguity caused by sequence number looping during centralized retransmission.

[0011] The beneficial effects of this invention are: it achieves low-latency loss awareness and adaptive scheduling, improving transmission throughput and bandwidth utilization; and it is suitable for scenarios involving satellite burst fading channels. The related technical effects will be described in detail below with reference to specific embodiments. Attached Figure Description

[0012] Figure 1 A flowchart illustrating an example of a link-layer retransmission method for satellite communications provided in this application embodiment.

[0013] Figure 2This is a flowchart illustrating cross-layer loss detection performed during the physical layer decoding state, as provided in an embodiment of this application.

[0014] Figure 3 A flowchart for triggering inference operations for lost data provided in an embodiment of this application.

[0015] Figure 4 This is a flowchart illustrating a feedback report containing data reception status, provided as an embodiment of this application.

[0016] Figure 5 A flowchart illustrating the retransmission granularity strategy based on feedback reports and real-time channel status provided in this application embodiment.

[0017] Figure 6 A flowchart illustrating the dynamic maintenance of the sending window based on physical layer signaling and feedback reports, provided for embodiments of this application.

[0018] Figure 7 The retransmission timer provided in this application embodiment includes a retransmission polling timer and a reordering timer, and a flowchart of dynamically maintaining the retransmission timer according to physical layer signaling. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in sequences other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] To overcome the above shortcomings, combined with Figures 1 to 5 The present application will be further illustrated with reference to specific embodiments.

[0022] The hardware selection, parameter values, and process steps given in this application are merely exemplary embodiments and are not intended to limit the scope of protection of this invention. Based on the core technical logic of this invention, the devices, parameters, and implementation methods can be equivalently replaced and adjusted according to the actual application scenario.

[0023] On the one hand, this application provides a link-layer retransmission method suitable for satellite communication. This method integrates a retransmission control sublayer within the general stream encapsulation layer of the data link layer, specifically including the following steps: Step 101: Obtain network layer protocol data units, assign sequence numbers to network layer protocol data units and cache them, and encapsulate network layer protocol data units into baseband frames for transmission; In the protocol stack architecture of a satellite communication system, the retransmission control sublayer, as a functional entity within the general stream encapsulation layer, is located between the network layer data input and the general stream encapsulation layer packetization operation.

[0024] When obtaining network layer protocol data units, the transmission mode can also be identified in the protocol type field of the general stream encapsulation layer according to the service type to which the network layer protocol data unit belongs.

[0025] The transmission modes specifically include: transparent mode, unacknowledged mode, and acknowledged mode.

[0026] Among them, the transparent mode is applicable to broadcast services and does not perform additional encapsulation processing on the data; Unacknowledged mode is suitable for real-time services. It only adds a sequence number for duplicate detection and reordering at the receiving end, without triggering retransmission. The confirmation mode is suitable for bidirectional link services with high reliability requirements.

[0027] In this scheme, the retransmission control sublayer only performs the operations of allocating sequence numbers and scheduling data to be retransmitted when the transmission mode is acknowledgment mode, thereby achieving compatible transmission of multiple service streams on the same physical link.

[0028] In acknowledgment mode, the sender maintains the next sequence number to be assigned and the earliest sequence number to be acknowledged.

[0029] When a network layer protocol data unit arrives, a sequence number is assigned to it, and it is stored in the retransmission queue for caching.

[0030] Before network layer protocol data units are fragmented and encapsulated into baseband frames, a retransmission extension header can be added before the payload of the network layer protocol data unit.

[0031] For example, the retransmission extension header may optionally adopt a data structure with extension type identifier 0x01, which contains at least the assigned sequence number, the fragment offset when the data is fragmented, and a polling indicator bit for actively requesting feedback reports from the receiving end.

[0032] Based on this, multiple fragments of the same network layer protocol data unit can be tracked independently; After encapsulation, a conventional low-density parity check coding module can be used to perform channel coding on the baseband frame and then transmit it.

[0033] Step 102: Receive and parse the baseband frame, perform cross-layer loss detection based on the physical layer decoding status, and generate a feedback report containing the data reception status; Alternatively, the physical layer at the receiving end processes the baseband frame using conventional decoding algorithms after receiving the signal.

[0034] In this scheme, when the baseband frame is successfully parsed and passed up to the general stream encapsulation layer, the receiver parses the retransmission extension header; To support subsequent fragment-level retransmissions, the receiver's general stream encapsulation layer can optionally be configured to not discard incomplete network layer protocol data unit fragments due to reassembly timer expiration, but instead collect the fragment set along with the fragment receive bits maintained for it. Figure 1 It is then submitted to the retransmission control sublayer, thus pushing the data integrity judgment logic down to this sublayer.

[0035] Furthermore, the receiving end maintains the next expected in-order sequence number variable and the highest received sequence number variable, and performs reordering buffering for out-of-order arriving segments.

[0036] Furthermore, when a loss event is detected or a baseband frame with the polling indicator bit set is received, the receiver generates a feedback report. For example, the feedback report may use a data structure with an extended type identifier of 0x03, which contains a cumulative acknowledgment sequence number to indicate that all data before that sequence number has been received correctly, and a bitmap indicating the specific location of the missing segment.

[0037] The feedback report is sent to the sending end via the reverse link, providing status input for the sending end's retransmission decision.

[0038] Step 103: Receive feedback reports and obtain real-time channel status; determine retransmission granularity strategy based on feedback reports and real-time channel status; and schedule data to be retransmitted. In some embodiments, the real-time channel state can be obtained based on the physical layer decoding state; Data to be retransmitted can be retrieved from the cache.

[0039] In other words, after parsing the feedback report, the sending end uses the cumulative confirmation sequence number to advance the sending window and clear the confirmed data in the retransmission queue.

[0040] For data indicated as lost in the bitmap of the feedback report, the sending end, in conjunction with the real-time channel status reported by the physical layer, decides whether to retransmit only the lost individual segment or to retransmit the entire network layer protocol data unit.

[0041] After determining the retransmission granularity, the sending end prioritizes filling the baseband frame to be sent with the data in the retransmission queue.

[0042] In this scheme, the sending end maintains a retransmission counter for each network layer protocol data unit; This counter increments each time a scheduled retransmission is performed; When the number of retransmissions of a certain data exceeds the preset maximum retransmission threshold, the sending end stops retransmitting the data, triggers the reset process, and sends a reset control command to the receiving end to reinitialize the state variables of both parties, so as to avoid continuous link congestion due to deadlock.

[0043] Step 104: Obtain physical layer signaling, and dynamically maintain the sending window and retransmission timer based on the physical layer signaling and feedback reports.

[0044] In this scheme, the transmitting end continuously listens for physical layer signaling fed back by the physical layer adaptive coding and modulation mechanism; This signaling contains the modulation and coding scheme used by the current link.

[0045] Since the physical layer frame length and payload capacity differ for different modulation and coding modes, the transmitter can use the modulation and coding mode information, combined with the preset network round-trip delay, to update the maximum allowed sequence number range in real time, that is, maintain the size of the transmission window.

[0046] In this scheme, the transmitting end can optionally adjust the timeout threshold of the retransmission polling timer synchronously according to the change of the physical layer frame length, so that in the dynamically changing satellite channel, the flow control and retransmission waiting time always match the underlying physical capacity.

[0047] On the other hand, performing cross-layer loss detection based on the physical layer decoding state can specifically include: Step 201: After the baseband frame is decoded at the physical layer, a decoding state primitive is generated and reported to the retransmission control sublayer. The decoding state primitive includes at least the baseband frame sequence number and the decoding result.

[0048] Among them, the decoding status primitive is reported to the retransmission control sublayer of the data link layer through the cross-layer interface.

[0049] In the decoding state primitive, the baseband frame sequence number is used to identify the timing position of the data frame, and the decoding result indicates the verification status of the current baseband frame, specifically whether it is successful or failed.

[0050] In one alternative implementation, the decoding state primitive also includes an identifier of the modulation and coding mode used by the current baseband frame, and a mapping table describing the generic stream encapsulation segments carried within the baseband frame, included when decoding is successful.

[0051] In other words, the above steps can also be described as: After receiving and using conventional error correction algorithms to complete the channel decoding operation of the baseband frame, the physical layer of the receiver generates a corresponding decoding state primitive for each baseband frame.

[0052] Step 202: When the decoding result is a failure, trigger the inference operation for the lost data based on the decoding state primitive to obtain the inference result of the lost data.

[0053] Accordingly, after receiving the decoding status primitive, the retransmission control sublayer determines the decoding result.

[0054] If the decoding result is successful, extract the data segment and update the receiving status according to the normal data processing procedure; If the decoding result fails, it is determined that the payload carried in the baseband frame has been lost, and the inference process for the damaged data is started without waiting for subsequent data to arrive and generate a sequence number hole.

[0055] According to one aspect of this application, triggering an inference operation for lost data specifically includes: Step 202.1: Take the baseband frame whose decoding result is a failure as the target baseband frame, and obtain the previous correctly received baseband frame and the next correctly received baseband frame that are adjacent in time and correctly received.

[0056] Since the receiver cannot directly obtain the payload structure inside the baseband frame that failed to decode, this method uses expected content inference logic.

[0057] Optionally, based on the baseband frame sequence number, the target baseband frame is located in the receive buffer, and the nearest preceding correctly received baseband frame in terms of timing is obtained by forward search, and the nearest following correctly received baseband frame in terms of timing is obtained by backward search.

[0058] Step 202.2: Extract the boundary information of the last data segment contained in the previous correctly received baseband frame, and the boundary information of the first data segment contained in the next correctly received baseband frame.

[0059] From the previous correctly received baseband frame, the last data segment carried at its tail is parsed out, and the sequence number and segment offset of the data segment are extracted as forward boundary information; similarly, from the next correctly received baseband frame, the first data segment carried at its header is parsed out, and the corresponding sequence number and segment offset are extracted as backward boundary information.

[0060] Step 202.3: Based on boundary information and preset data encapsulation rules, infer the set of missing data segments in the target baseband frame.

[0061] Given that the sending end uses a pre-defined data encapsulation rule of sequential greedy filling, the expected content inference table can be constructed using the acquired forward boundary information and backward boundary information.

[0062] Specifically, the sequence number and the fragment offset are combined into a tuple. By comparing the logical size between the tuples, all data fragments between the forward boundary information and the backward boundary information are calculated and included in the set of lost data fragments.

[0063] The specific inference logic formula is as follows: ; in: Lost represents the set of missing data segments in the inferred target baseband frame; SN represents the sequence number of the data segment to be inferred; SO represents the segment offset of the data segment to be inferred; This indicates the sequence number of the last data segment in the previous correctly received baseband frame; This indicates the segment offset of the last data segment in the previous correctly received baseband frame; This indicates the sequence number of the first data segment in the next correctly received baseband frame; This represents the segment offset of the first data segment in the next correctly received baseband frame.

[0064] The logical size comparison operation rules are defined here as follows: For any two pairs and , The conditions for establishment are , or in equal In the case of .

[0065] According to one aspect of this application, a feedback report including the data reception status is generated, specifically including: Step 1: Based on the inferred set of lost data segments, generate a feedback report containing segment-level loss indications.

[0066] Based on the inference results, a feedback report with an extended type identifier of 0x03 is generated.

[0067] For example, the feedback report includes the cumulative confirmed sequence number, the starting value of the unconfirmed sequence number, and the corresponding reception status bitmap; On the other hand, the feedback report can also include a channel status indication field; The channel state indication field carries the quantized value of the real-time frame loss rate calculated by the receiver based on the sliding window estimator.

[0068] For the inferred set of missing data segments, a segment-level loss indication is added to the feedback report. Specifically, the start and end offsets of the missing segment are added under the corresponding sequence number entry, thereby achieving fine-grained identification of the missing data.

[0069] Step ②: Simultaneously run the reordering timer in the retransmission timer as a collaborative detection mechanism. Compare the inference delay of cross-layer loss detection with the timeout delay of the reordering timer, and use the minimum delay of the two as the condition for triggering the generation of feedback reports.

[0070] In other schemes, inference delay can also be the execution time.

[0071] In this scheme, while performing cross-layer loss inference, the receiving end maintains a reordering timer.

[0072] The timer is started when a non-contiguous sequence number is detected.

[0073] To prevent retransmitted data from being skipped by the advanced window during transmission, the timeout threshold setting of the reordering timer must meet the following constraints:

[0074] in: Indicates the timeout threshold for the reordering timer; RTT stands for network round-trip time; This indicates the delay margin for the generation and processing of feedback reports; This indicates the retransmission scheduling delay margin.

[0075] In actual operation, the final latency of the detection mechanism depends on the smaller value between cross-layer inference and timer timeout. The calculation logic is as follows: ; in: This indicates the actual detection delay that triggered the generation of the feedback report; min() represents the minimum value function; This indicates the execution latency of baseband frame decoding and cross-layer inference; This indicates the timeout delay of the reordering timer.

[0076] In the above, when a non-greedy encapsulation strategy is adopted at the transmitting end, or when the inference interval is too large due to the decoding failure of multiple consecutive baseband frames, the reordering timer is used as a fallback mechanism to trigger feedback.

[0077] This embodiment also provides an optional implementation for triggering the generation of feedback reports. By establishing a unified feedback trigger determination node at the receiving end, it is specifically divided into the following parallel branches: Branch 1, cross-layer real-time inference trigger, i.e.: That is, after inferring the lost set based on the baseband frame decoding failure, a feedback report is generated and sent under the condition that the state prohibition timer is not running.

[0078] Branch 2, triggered by polling instructions, i.e.: When the receiving end parses the polling indicator bit in the received retransmission extension header and finds that it is set to a preset value, it temporarily suspends the triggering action and generates a feedback report after the status prohibition timer expires.

[0079] In this embodiment, the preset value can be configured to 1. Branch 3, triggered by a periodic timer, i.e.: As an alternative method, the receiver runs a periodic status timer to generate a feedback report containing the current status of the entire receiving window at set time intervals, in order to counteract the risk of the feedback report itself being lost in the backhaul link.

[0080] In this invention, the underlying decoding state primitives are extracted and combined with the expected content inference mechanism, which alleviates the problem of passive waiting due to the sequence number gaps in traditional protocols, reduces the loss detection delay, and realizes frame-level loss delimitation.

[0081] On the other hand, the specific retransmission scheduling strategy for the three-stage adaptive implementation process of sudden frame loss awareness is described. This strategy introduces a companion state machine to avoid invalid retransmissions during sudden fading of the satellite channel, and its specific execution steps are as follows: Specifically, obtaining the real-time channel status includes: Obtain the decoding results of continuously received baseband frames, use a sliding window of preset length to count the proportion of baseband frames that failed to be decoded within the sliding window, and use the calculated proportion as the real-time frame drop rate.

[0082] Optionally, the real-time frame drop rate can be included in the feedback report and transmitted to the sender.

[0083] In some embodiments, the receiver maintains a sliding window channel state estimator based on the physical layer decoded state primitive sequence.

[0084] During this process, conventional data sliding window iteration logic can be used to record the frame state within the time window.

[0085] Furthermore, the ratio of the number of baseband frames that failed to be decoded within the sliding window to the total number of received baseband frames within that window is calculated to obtain the real-time frame loss rate.

[0086] The specific calculation formula is as follows: ; in: Indicates in Real-time frame drop rate calculated at each step; This indicates the duration of the sliding window or the span of baseband frames. Indicates from arrive Within the time window, the decoding result is the cumulative number of failed baseband frames; Indicates from arrive The total number of baseband frames received by the receiver within the time window.

[0087] Among them, obtained through calculation It is a dimensionless probabilistic characteristic quantity used to characterize the transient error trend of the current satellite physical link.

[0088] In this scheme, the real-time frame loss rate is used as a channel state indicator and is appended to the feedback report by the receiver when generating the feedback report.

[0089] Specifically, the extended information field of the feedback report includes a channel state indication field, which carries a quantized value of the real-time frame loss rate.

[0090] The transmitting end obtains the real-time frame loss rate by parsing the channel state indication field in the feedback report.

[0091] In one alternative implementation, the receiver and the transmitter each maintain their own state machines. The receiver adjusts its feedback behavior autonomously based on the real-time frame loss rate calculated locally, while the transmitter executes scheduling decisions based on the real-time frame loss rate obtained from the feedback report.

[0092] In this embodiment, scheduling the data to be retransmitted specifically includes: The real-time frame drop rate is compared with the preset frame drop rate threshold, and the system switches between normal retransmission, cautious retransmission, and retransmission suppression states based on the comparison result. In the cautious retransmission state, the retransmission interval of the data to be retransmitted is increased, and the receiving end can be instructed by signaling to reduce the frequency of sending feedback reports; in the retransmission suppression state, the scheduling of the data to be retransmitted is suspended and the retransmission timer is paused.

[0093] After obtaining the real-time frame loss rate, the sending end executes a unified scheduling status judgment logic.

[0094] On the other hand, the sending end and the receiving end can each independently maintain a scheduling state machine with the same logic.

[0095] Based on the locally calculated real-time frame loss rate, and the same first and second thresholds as the sender, the receiver independently determines and executes state switching operations related to its own feedback behavior.

[0096] Based on the real-time frame loss rate obtained from the feedback report, the sending end independently determines and executes state switching operations related to retransmission scheduling.

[0097] Because both sides use the same threshold configuration and similar channel observation data, their state transitions remain synchronized. The following describes the state behaviors of the transmitter and receiver respectively: In this embodiment, a first threshold and a second threshold can be set as state switching boundaries, and the following actions are performed based on the comparison results: In some scenarios, when the real-time frame loss rate is less than the first threshold, the channel is determined to be in the normal range, and the state machine maintains or switches to the normal retransmission state.

[0098] In this state, the scheduler responds to the received feedback report, schedules the data to be retransmitted according to the normal priority, and the retransmission timer operates normally.

[0099] In some scenarios, when the real-time frame loss rate is greater than or equal to the first threshold and less than the second threshold, the channel is determined to be in the deterioration range, and the state machine switches to the cautious retransmission state.

[0100] In this state, a waiting interval based on the increasing number of retransmissions is inserted between two consecutive retransmissions of the same network layer protocol data unit, thereby increasing the retransmission time interval of the data to be retransmitted. The timeout threshold of the receiver feedback disable timer is increased several times to reduce the frequency of feedback report transmission and reduce bandwidth usage under degraded channel conditions.

[0101] In some scenarios, when the real-time frame loss rate is greater than or equal to the second threshold, the channel is determined to be in a sudden interruption interval (e.g., rain attenuation or shadow occlusion), and the state machine switches to the retransmission suppression state.

[0102] In this state, all resource allocation and scheduling operations for data to be retransmitted are suspended, and the data to be retransmitted is retained in the retransmission buffer queue. The countdown of the retransmission polling timer on the sending end is paused, and the receiving end pauses the generation and transmission of feedback reports, maintaining only a cumulative record of lost segments locally.

[0103] In this scheme, after multiple baseband frames are successfully decoded and the retransmission suppression state is exited, a centralized recovery process is executed. The centralized recovery process includes: The lost information maintained and accumulated locally at the receiving end during the retransmission suppression state is merged to generate a complete feedback report; Multiple data to be retransmitted are compactly filled into the same baseband frame for centralized retransmission; Additionally, the transmission window is temporarily expanded based on the amount of data lost during the retransmission suppression period.

[0104] For example, when the channel quality recovers, the receiver monitors the number of baseband frames that are successfully decoded consecutively.

[0105] For example, when the number of consecutive successful baseband frames reaches a preset exit threshold, the sudden fading is determined to have ended, triggering the state machine to exit the retransmission suppression state and enter the centralized recovery process.

[0106] In the centralized recovery process, the receiving end extracts the accumulated records maintained locally, merges and encapsulates them into a feedback report covering the entire period of the incident, and sends it out.

[0107] In this scheme, after receiving the feedback report, the sending end schedules all the data that is backed up in the queue and is waiting to be retransmitted at once.

[0108] In the encapsulation stage, a compact padding strategy is adopted to splice multiple retransmission segments belonging to different network layer protocol data units into the same or a few baseband frames to improve padding efficiency.

[0109] In this scheme, the sending end resumes the previously paused retransmission polling timer and performs a temporary expansion of the sending window to allow more data to enter the in-transit state and accelerate the clearing of the backlog queue.

[0110] Based on the amount of data lost during the retransmission suppression period, the sending window is temporarily expanded, specifically including: In some embodiments, the upper limit of the expanded transmission window is limited to no more than half the total capacity of the available sequence number space to prevent acknowledgment ambiguity caused by sequence number looping during centralized retransmission.

[0111] When performing a temporary expansion of the sending window, an anti-ambiguity windowing correction algorithm can be optionally executed.

[0112] For example, the earliest sequence number currently awaiting confirmation, the basic sending window size, and the amount of data lost during the burst are obtained, and the sum of the three is calculated to obtain the upper limit of the initial recovery window; Obtain the earliest serial number and add it to half of the total available serial number space to get the upper limit of the anti-ambiguity window; Compare the initial recovery window upper limit with the anti-ambiguity window upper limit, and extract the smaller value of the two as the final expanded sending window upper limit value.

[0113] The specific calculation formula is as follows:

[0114] in: This indicates the upper limit of the transmission window after the temporary expansion during centralized recovery, which represents the maximum sequence number that is allowed to be transmitted; This indicates the earliest sequence number currently awaiting confirmation, which is the lower bound of the sliding window; W represents the base transmit window size, which is determined by configuration or physical layer parameters; This indicates the cumulative number of network layer protocol data units lost during the retransmission suppression state; This indicates the total available sequence number space under the current protocol configuration; for example, the space corresponding to a 12-bit sequence number is 4096. This represents the maximum theoretical window span to prevent serial number wraparound ambiguity.

[0115] By implementing the above constraints, the receiving end can be prevented from misinterpreting newly transmitted data packets as old retransmission packets due to the cyclic reuse of sequence numbers.

[0116] One example is determining the retransmission granularity strategy based on feedback reports and real-time channel conditions, specifically including: Step 401: Determine the number of lost segments and the total number of segments of the network layer protocol data unit to be retransmitted, and obtain the real-time frame loss rate in the real-time channel status.

[0117] In this embodiment, after receiving a feedback report containing a fragment-level loss indication, the sending end parses the feedback report to obtain the number of missing segments in the network layer protocol data unit to be retransmitted, and determines it as the number of lost segments.

[0118] Based on this, query the encapsulation record of the network layer protocol data unit in the transmission cache to obtain the total number of segments it has been divided into.

[0119] Furthermore, the real-time frame loss rate is obtained through a sliding window channel state estimator, which serves as the data input for subsequent decision-making.

[0120] According to one aspect of this application, in scenarios where system computing power is limited or channel state estimation is not enabled, the transmitter can adopt decision logic based on static thresholds, namely: Calculate the ratio of the number of lost segments to the total number of segments. When the ratio is greater than a preset ratio threshold, select to retransmit the entire network layer protocol data unit. When the ratio is not greater than the preset ratio threshold, select fragment retransmission.

[0121] In this scheme, the static threshold can be set to 50%.

[0122] Step 402: Based on the number of lost segments, the preset basic transmission cost of segments, and the preset protocol header overhead cost, calculate the basic cost of segment retransmission. Use the real-time frame loss rate to perform a secondary loss probability amplification correction on the basic cost of segment retransmission to obtain the expected cost of segment retransmission.

[0123] In one expected cost model, for the fragmented retransmission method, the sum of the basic transmission cost of the fragment and the protocol header overhead cost is calculated, and then multiplied by the number of lost fragments to obtain the basic cost of fragmented retransmission.

[0124] Next, the difference between 1 and the real-time frame loss rate is calculated to obtain the probability that no frames are lost in a single segment transmission.

[0125] Furthermore, the ratio of the base cost of fragment retransmission to the probability of no frame loss in a single fragment transmission is calculated to achieve secondary loss probability amplification correction and obtain the expected cost of fragment retransmission.

[0126] The specific calculation formula is as follows: ; in: This represents the expected cost of fragmented retransmission; Indicates the number of lost segments of the network layer protocol data unit to be retransmitted; This represents the preset basic transmission cost of a data segment, and indicates the number of bits of baseband frame capacity occupied by a single data segment; This represents the preset protocol header overhead cost, and indicates the number of bits of capacity occupied by the extended header added during retransmission. This represents the real-time frame loss rate, which characterizes the probability of data being lost twice under the current channel. This indicates the probability that no frame loss occurs during a single segment transmission; This indicates that the basic cost of fragment retransmission is not considered under the condition of secondary loss.

[0127] Step 403: Based on the total number of segments, the basic transmission cost of segments, and the overhead cost of protocol headers, calculate the basic cost of whole packet retransmission. Using the real-time frame loss rate and the total number of segments, perform multi-segment joint loss probability amplification correction on the basic cost of whole packet retransmission to obtain the expected cost of whole packet retransmission.

[0128] For the whole packet retransmission method, calculate the product of the total number of segments and the basic transmission cost of the segments, and add the product to the protocol header overhead cost to obtain the basic cost of whole packet retransmission.

[0129] Next, the probability of no frame loss in a single segment transmission is used as the base, and the total number of segments is used as the exponent to perform an exponentiation operation to obtain the success rate of multi-segment joint transmission.

[0130] Accordingly, the ratio of the base cost of whole packet retransmission to the success rate of multi-fragment joint transmission is calculated to achieve multi-fragment joint loss probability amplification correction, thus obtaining the expected cost of whole packet retransmission; that is: ; in: This represents the expected cost of retransmitting the entire packet. This indicates the total number of network layer protocol data units to be retransmitted. This indicates the preset basic transmission cost of the segment; This indicates the default protocol header overhead cost; Indicates the real-time frame drop rate; This indicates the probability that no frame loss occurs during a single segment transmission; This represents the joint probability that all segments are successfully received in the same retransmission. This indicates that the basic cost of retransmitting the entire packet under the condition of secondary loss is not considered.

[0131] Step 404: Compare the expected cost of retransmitting the whole packet with the expected cost of retransmitting the fragments, and select the retransmission method with the smaller expected cost as the retransmission granularity strategy.

[0132] Accordingly, the sending end obtains two expected cost values ​​and performs a value comparison operation.

[0133] If the expected cost of fragment retransmission is less than the expected cost of full packet retransmission, a fragment retransmission instruction is generated, and the scheduler only extracts the lost data fragments and fills them into subsequent baseband frames. Conversely, if the expected cost of retransmitting the entire packet is less than or equal to the expected cost of retransmitting fragments, a retransmission instruction for the entire packet is generated, and the scheduler reorders all fragments of the network layer protocol data unit for transmission.

[0134] The cost model shows a non-linear positive correlation between the expected cost of whole packet retransmission and the real-time frame loss rate. As the real-time frame loss rate increases, the denominator decreases due to the exponential effect.

[0135] At the physical level, this manifests as a decrease in the probability that all segments of a data packet containing multiple fragments will successfully retransmit simultaneously when channel conditions deteriorate. To illustrate this trend, a normalized abstract example is provided: Assuming that after normalization, we obtain the basic transmission cost of the fragment, the protocol header overhead cost, the total number of fragments, and the number of lost fragments; When the real-time frame loss rate approaches 0, the expected cost of fragment retransmission and the expected cost of whole packet retransmission are obtained, and the model tends to retransmit fragments. As the real-time frame loss rate increases, the expected cost of fragmented retransmission rises, while the expected cost of whole-packet retransmission is amplified due to the combined success rate. In this situation, the model adaptively strengthens its selection of fragmented retransmission, thereby reducing overall network bandwidth consumption.

[0136] In this application, on the one hand, a three-stage adaptive scheduling state machine is driven by sliding window channel estimation to actively suppress invalid retransmissions during sudden fading and to perform compact centralized compensation after the channel is restored, thereby curbing bandwidth waste under adverse operating conditions. On the other hand, by introducing a secondary loss probability to construct an expected cost model, the system can adaptively tilt towards small-granularity fragment retransmission when the channel deteriorates, thereby reducing the overall retransmission cost through optimized decision-making.

[0137] Another example is dynamically maintaining the transmission window based on physical layer signaling, including: Step 501: Parse the physical layer signaling to obtain the current modulation and coding mode.

[0138] In this step, the physical layer control channel is monitored through the cross-layer interface to extract the adaptive coding and modulation signaling currently used in the forward link, and the currently effective modulation and coding mode identifier is parsed from it.

[0139] Step 502: Determine the corresponding physical layer frame length and baseband frame information bit length according to the modulation and coding mode, and calculate the current effective data rate in combination with the fixed overhead of the protocol header.

[0140] In this step, a preset physical layer parameter mapping table is queried to obtain the baseband frame information bit length and physical layer frame length corresponding to the modulation and coding mode; The difference between the baseband frame information bit length and the fixed overhead of the protocol header is calculated to obtain the number of effective payload bits per frame; The ratio of the number of effective payload bits per frame to the physical layer frame length is calculated to obtain the current effective data rate.

[0141] For example: ; in: This represents the current effective data rate under modulation and coding mode m; This indicates the length of the baseband frame information bits corresponding to the modulation and coding mode m; The fixed overhead bits for the protocol headers of the General Stream Encapsulation Header and the Retransmission Extension Header; This indicates the transmission time span of the next physical layer frame in modulation and coding mode m.

[0142] Step 503: Calculate and determine the target transmission window based on the current effective data rate and the preset network round-trip delay.

[0143] Calculate the product of the current effective data rate, network round-trip delay, and preset bandwidth utilization coefficient to obtain the bandwidth-delay product data volume; Calculate the ratio of the bandwidth-delay product data volume to the target transmission window value, and then round up this ratio.

[0144] The single packet length is the product of the number of bytes in the protocol data unit and 8.

[0145] For example: ; in: This represents the calculated target transmission window, expressed in units of dimensionless protocol data units. This indicates the preset network round-trip time; This represents the preset bandwidth utilization coefficient, which is greater than 1 to provide redundancy margin. This indicates the average byte size of a network layer protocol data unit; 8 is used to convert bytes to bits.

[0146] In this step, there is a conditional decision branch, specifically: Determine whether the calculated target sending window exceeds the current configured sequence number space capacity limit.

[0147] If the limit is not exceeded, the system will operate in the original mode; if the limit is exceeded, for example, if the calculated value is greater than the capacity limit of the twelve-digit serial number space, the system will be triggered to switch to extended serial number mode to prevent window overflow from causing serial number ambiguity.

[0148] In some scenarios, an extended serial number mode, such as a sixteen-bit serial number space, can be used.

[0149] According to an embodiment of this application, dynamically maintaining the sending window further includes: Compare the size of the target sending window with the size of the current sending window; When the target sending window is smaller than the current sending window, the shrinkage rate is determined based on the difference between the current sending window and the target sending window, as well as the network round-trip delay. The current sending window is linearly and progressively shrunk according to the shrinking rate until the current sending window is shrunk to the target sending window.

[0150] In actual link operation, dynamic switching of modulation and coding modes will cause changes in the target transmission window. This step uses branch logic to handle window updates, as follows: In some cases, when the target transmission window is greater than or equal to the current transmission window, it indicates that the channel quality has improved, and the current transmission window is directly updated to the target transmission window.

[0151] In some cases, when the target transmission window is smaller than the current transmission window, it indicates that the channel order has been reduced. If the window is shrunk suddenly, a large amount of data in transit will violate the window rule.

[0152] Accordingly, the difference between the current sending window and the target sending window is calculated, and the difference is divided by the network round-trip delay to obtain the shrinkage rate; Accordingly, based on the contraction rate and the time difference between the current time and the switching time, the asymptotic window value is calculated, and the larger of the target sending window and the asymptotic window value is taken as the current sending window after the actual update.

[0153] For example: ; ; in: This indicates the contraction rate, measured in units per second. Indicates the current sending window before the handover occurred; This represents the calculated target sending window; Indicates in The current sending window is updated in real time; This indicates the initial moment when the modulation and coding mode switch occurs.

[0154] The above avoids misjudgments of timeout retransmissions caused by violent window closing.

[0155] According to the embodiments of this application, the retransmission timer includes a retransmission polling timer and a reordering timer. The retransmission timer is dynamically maintained according to the physical layer signaling, specifically as follows: Parse the physical layer signaling to obtain the current modulation and coding mode and determine the corresponding physical layer frame length; The timeout threshold of the retransmission polling timer is dynamically calculated and updated based on the physical layer frame length, the preset network round-trip delay, and the average number of fragments of the network layer protocol data unit. The timeout threshold of the reordering timer is dynamically calculated and updated based on the physical layer frame length, network round-trip latency, and the maximum number of fragments of network layer protocol data units.

[0156] In this scheme, after obtaining the current modulation and coding mode and physical layer frame length, the product of the physical layer frame length and the average number of fragments is calculated to obtain the estimated time for single packet transmission for the retransmission polling timer. The network round-trip delay is doubled and the estimated time for single packet transmission is doubled to obtain the updated retransmission polling timer timeout threshold.

[0157] For the reordering timer, the worst-case single-packet cross-frame time is obtained by multiplying the physical layer frame length by the maximum number of fragments. The network round-trip delay, the cross-frame time, and the preset protection margin are added together to obtain the updated reordering timer timeout threshold.

[0158] For example, ; ; in: Indicating in modulation and coding mode Timeout threshold for the retransmission polling timer; This indicates the average number of fragments in a network layer protocol data unit; Indicating in modulation and coding mode The timeout threshold for the reordering timer; Indicates the maximum number of fragments for a network layer protocol data unit; This indicates the preset system processing protection margin.

[0159] As can be seen from the above linkage calculation logic, the timer threshold is positively correlated with the physical layer frame length. When the modulation and coding mode decreases, leading to an increase in the physical layer frame length, the timer threshold is adaptively extended. This helps to ensure sufficient waiting time under low-rate links and prevents the timer from prematurely timeing out and triggering redundant actions.

[0160] In this invention, by monitoring the physical layer signaling and dynamically linking the sending window and timer threshold, and supplementing it with a progressive shrinking algorithm, the risk of queue overflow and timeout misjudgment caused by the mismatch between the physical layer dynamic modulation and the upper layer static parameters is reduced. While ensuring smooth traffic on long-delay links, the overall throughput of the system is improved.

[0161] In this application, the receiving end updates the local reordering timer threshold by parsing the modulation and coding mode identifier carried in the received baseband frame header and independently executing the same dynamic calculation logic. Alternatively, the sending end can notify the receiving end of the calculated threshold via signaling.

[0162] Those skilled in the art will understand that the specific values ​​of the first and second thresholds can be adaptively adjusted based on the engineering experience of the satellite link, the latency tolerance of the service, and the retransmission capacity of the system, and appropriate values ​​can be determined without creative effort.

[0163] The optional embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solution of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A link-layer retransmission method suitable for satellite communication, characterized in that, A retransmission control sublayer is integrated within the general stream encapsulation layer of the data link layer, including: Obtain network layer protocol data units, assign sequence numbers to network layer protocol data units and cache them, and encapsulate network layer protocol data units into baseband frames for transmission; Receive and parse baseband frames, perform cross-layer loss detection based on physical layer decoding status, and generate a feedback report containing data reception status; Receive feedback reports and obtain real-time channel status; determine retransmission granularity strategy based on feedback reports and real-time channel status; and schedule data to be retransmitted. Obtain physical layer signaling and dynamically maintain the transmission window and retransmission timer based on the physical layer signaling and feedback reports.

2. The method according to claim 1, characterized in that, Cross-layer loss detection is performed based on the physical layer decoding state, specifically including: After the baseband frame is decoded at the physical layer, a decoding state primitive is generated and reported to the retransmission control sublayer. The decoding state primitive contains at least the baseband frame sequence number and the decoding result. When the decoding result fails, an inference operation for the lost data is triggered.

3. The method according to claim 2, characterized in that, Trigger inference operations for lost data, specifically including: The baseband frame whose decoding result is a failure is taken as the target baseband frame, and the previous and next correctly received baseband frames that are adjacent in time and correctly received are obtained. Extract the boundary information of the last data segment contained in the previous correctly received baseband frame, and the boundary information of the first data segment contained in the next correctly received baseband frame. Based on boundary information and preset data encapsulation rules, the set of missing data segments in the target baseband frame is inferred.

4. The method according to any one of claims 1 or 3, characterized in that, Generate a feedback report that includes the data reception status, specifically including: Based on the inferred set of lost data fragments, a feedback report containing fragment-level loss indications is generated; The reordering timer is used as a collaborative detection mechanism. The inference delay of cross-layer loss detection is compared with the timeout delay of the reordering timer, and the minimum delay between the two is used as the condition for triggering the generation of feedback reports.

5. The method according to claim 1, characterized in that, Obtaining real-time channel status includes: Obtain the decoding results of continuously received baseband frames, use a sliding window of preset length to count the proportion of baseband frames that failed to be decoded within the sliding window, and use the calculated proportion as the real-time frame drop rate.

6. The method according to claim 1, characterized in that, The retransmission granularity strategy is determined based on feedback reports and real-time channel conditions, specifically including: Determine the number of lost segments and the total number of segments of the network layer protocol data unit to be retransmitted, and obtain the real-time frame loss rate in the real-time channel state; Based on the number of lost segments, the basic transmission cost of segments, and the overhead cost of protocol headers, the basic cost of segment retransmission is calculated. The basic cost of segment retransmission is then amplified and corrected by a second loss probability using the real-time frame loss rate to obtain the expected cost of segment retransmission. Based on the total number of segments, the basic transmission cost of segments, and the overhead cost of protocol headers, the basic cost of retransmitting the entire packet is calculated. Using the real-time frame loss rate and the total number of segments, the basic cost of retransmitting the entire packet is corrected by multi-segment joint loss probability amplification to obtain the expected cost of retransmitting the entire packet. By comparing its expected cost with that of fragmented retransmission, the retransmission method with the lower expected cost is selected as the retransmission granularity strategy.

7. The method according to claim 1, characterized in that, The transmission window is dynamically maintained based on physical layer signaling and feedback reports, including: Parse physical layer signaling to obtain the current modulation and coding mode; Based on the modulation and coding mode, determine the corresponding physical layer frame length and baseband frame information bit length, and calculate the current effective data rate in combination with the fixed overhead of the protocol header; The target transmission window is calculated and determined based on the current effective data rate and the preset network round-trip delay.

8. The method according to claim 1, characterized in that, The retransmission timer includes a retransmission polling timer and a reordering timer. The retransmission timer is dynamically maintained based on physical layer signaling, including: Parse the physical layer signaling to obtain the current modulation and coding mode and determine the corresponding physical layer frame length; The timeout threshold of the retransmission polling timer is dynamically calculated and updated based on the physical layer frame length, the preset network round-trip delay, and the average number of fragments of the network layer protocol data unit. The timeout threshold of the reordering timer is dynamically calculated and updated based on the physical layer frame length, network round-trip latency, and the maximum number of fragments of network layer protocol data units.

9. The method according to claim 1, characterized in that, When acquiring network layer protocol data units, the following is also included: Based on the service type to which the network layer protocol data unit belongs, the transmission mode is identified in the protocol type field of the general stream encapsulation layer; The transmission modes include transparent mode, unacknowledged mode, and acknowledged mode. The operations of allocating sequence numbers and scheduling data to be retransmitted are performed only when the transmission mode is acknowledged.

10. The method according to claim 1, characterized in that, Encapsulating network layer protocol data units into baseband frames for transmission specifically includes: A retransmission extension header is added before the payload of the network layer protocol data unit. The retransmission extension header includes at least the assigned sequence number, the fragment offset when the data is fragmented, and a polling indicator bit for actively requesting feedback reports from the receiver.

Images