A Channel Quality Aware Multi-Slot Multi-Channel Audio and Video Narrowband Transmission Method
By employing a channel quality-aware multi-slot, multi-channel audio and video transmission method with dynamic scheduling and reassembly mechanisms, the problems of high latency, severe stuttering, and interruption in audio and video transmission under narrowband and weak network environments are solved, achieving low latency, high availability, and strong fault tolerance, thereby improving the user experience.
Patent Information
- Application Number
- CN202611330870.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-31
- Publication Date
- 2026-11-10
AI Technical Summary
In narrowband weak network environments with limited bandwidth and poor channel quality, traditional single-channel or static multi-channel transmission mechanisms are difficult to balance low latency, high availability and anti-interruption capabilities. Audio and video data transmission is easily affected by interference and loss, non-critical data occupies resources, and the receiving end has difficulty efficiently reassembling out-of-order data, resulting in playback stuttering and image distortion.
By periodically collecting channel parameters, dynamically maintaining the quality index ranking, dividing and encapsulating microslices with global sequence numbers according to priority, implementing time slot-triggered dynamic mapping scheduling, unified buffering and out-of-order reordering at the receiver, and combining cross-channel preemptive retransmission mechanism, channel state awareness and service semantic hierarchical scheduling are achieved.
It achieves low latency, high availability and strong fault tolerance for audio and video transmission under narrowband and weak network conditions, improves user experience quality, and solves the problems of low resource utilization efficiency and data reassembly in traditional methods.
Smart Images

Figure CN123075540A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wireless communication and multimedia transmission technology, and more particularly to a method for multi-slot, multi-channel audio and video narrowband transmission based on channel quality awareness in weak network environments. Background Technology
[0002] With the rapid development of applications such as mobile communication, emergency communication, and remote collaboration, audio and video services are placing higher demands on wireless network transmission. However, in narrowband weak network environments with limited bandwidth (total bandwidth ≤ 512 kbps) and poor channel quality (packet loss rate ≥ 20%, large latency jitter), traditional single-channel or static multi-channel transmission mechanisms struggle to simultaneously achieve low latency, high availability, and anti-interruption capabilities. Existing technologies typically employ fixed channel allocation, uniform priority processing, or simple redundant retransmission strategies, lacking the ability to perceive dynamic channel changes and failing to fully consider the semantic differences and time sensitivity within audio and video data. This results in critical data (such as audio frames and video I-frames) being easily lost due to interference, while non-critical data (such as P / B frames) may consume valuable resources, leading to overall low transmission efficiency and a poor user experience.
[0003] Furthermore, in a multi-channel parallel transmission architecture, without a unified timing identifier and cross-channel scheduling mechanism, the receiver struggles to efficiently reassemble out-of-order data, further exacerbating playback stuttering and image distortion. Therefore, a novel narrowband audio and video transmission method is urgently needed that integrates channel state awareness, service semantic classification, and time-slot-level fine-tuning to achieve stable, smooth, and high-quality real-time communication under extremely limited wireless conditions. Summary of the Invention
[0004] In view of this, the present invention proposes a multi-slot, multi-channel audio and video narrowband transmission method based on channel quality awareness. Through a closed-loop process of "periodic acquisition of channel parameters - dynamic maintenance of quality index sorting - priority segmentation and encapsulation of microslices with global sequence numbers - time slot-triggered dynamic mapping scheduling - unified buffering and out-of-order reordering at the receiving end", the method solves the technical problems of high latency, severe stuttering, and even interruption in audio and video transmission under weak wireless network environments.
[0005] To achieve the above objectives, the present invention provides a method for multi-slot, multi-channel audio and video narrowband transmission based on channel quality awareness, comprising the following steps: S1. The transmitting device periodically collects physical layer parameters of multiple wireless channels, maintains a dynamic state set containing the quality index of each channel, and sorts the available channels in real time according to the quality index. S2. Receive the audio and video data stream input from the application layer, mark the priority according to the data type, divide the data stream into fixed-length micro-slices that are adapted to the physical layer time slot length, and add a global sequence number to each micro-slice; S3. Trigger a scheduling interrupt based on the current time slot, and dynamically map microslices of different priorities to the time slots of the currently available channels for transmission according to the priority of the microslices and the channel sorting results. S4. The receiving device receives microslices from different channels, stores the microslices in a unified buffer pool according to the global sequence number, performs out-of-order rearrangement and frame reassembly, and generates packet loss feedback signaling when a discontinuous sequence number is detected.
[0006] Preferably, in step S1, the calculation and updating of the quality index includes: An exponentially weighted moving average algorithm is used to smooth the currently collected physical layer parameters and historical scores, and the channel score for the current period is calculated. A hysteresis threshold is set, and the channel ranking in the dynamic state set is updated only when the change in the channel score compared to the previous period score exceeds the hysteresis threshold.
[0007] Preferably, in step S2, the encapsulation of the microslices includes: Calculate the maximum payload capacity of the current transmission time slot; The input audio frames, video keyframes, and video reference frames are segmented into multiple consecutive microslices. The global sequence number is encapsulated in the header of each microslice. This global sequence number is shared across all channels and monotonically increases, serving as the sole basis for cross-channel reassembly at the receiver.
[0008] Preferably, in step S3, the scheduling strategy for high-priority audio data includes: The audio microslices are preferentially allocated to the currently ranked best channel slot; The system reads the global packet loss rate in real time. If the global packet loss rate exceeds the preset redundancy threshold, the same audio micro-slice is copied and allocated to the second-ranked suboptimal channel time slot for dual-path concurrent transmission.
[0009] Preferably, in step S3, the scheduling strategy for medium-priority video keyframe data includes: Video keyframe microslices are preferentially allocated to the idle time slots of the currently ranked best channel. If the best channel has no idle time slots, they are allocated to the idle time slots of the second best channel. The forward error correction (FEC) redundancy ratio is dynamically calculated based on the current packet loss rate of the channel, a corresponding number of FEC redundancy slices are generated, and the FEC redundancy slices are filled into the subsequent available time slots.
[0010] Preferably, in step S3, the scheduling strategy for low-priority video reference frame data includes: Video reference frame microslices are transmitted only when there is an idle time slot and the channel score of the channel to which the time slot belongs is higher than a preset threshold. The dwell time of the video reference frame microslice in the transmission queue is detected. If the dwell time exceeds the video frame interval, the microslice is discarded directly.
[0011] Preferably, it further includes a cross-channel preemption and retransmission step, specifically: After receiving a negative acknowledgment signal from the receiver, the sending end parses out the global sequence number of the lost microslice; The priority of the lost microslices is temporarily raised to the highest level; In the next scheduling cycle, the lost microslice will be forcibly retransmitted to the idle channel with the highest current quality index, instead of being forcibly retransmitted on the original transmission channel.
[0012] Preferably, the negative response signaling adopts a compressed bitmap format, specifically: When the receiver detects a global sequence number interruption, it generates compressed signaling containing a reference sequence number and subsequent N microslice reception status bitmaps. The length of the signaling is fixed and less than the control overhead limit of a single time slot.
[0013] Preferably, before step S2, a source rate adaptation step is further included, specifically: Sum the total effective bandwidth of all currently available channels; If the total bandwidth is lower than the current video encoding bitrate, the transmitting processor sends a degradation instruction to the video encoder to reduce the video resolution or frame rate to match the total channel capacity.
[0014] Preferably, in step S4, the specific process of cross-channel reassembly includes: The receiving end establishes a unified jitter buffer across channels; When consecutive global sequence numbers in the buffer constitute a complete frame of data, a decoding interrupt is triggered to deliver the data to the decoder. If a non-continuous global sequence number is detected and the current time is earlier than the difference between the playback timestamp and the round-trip time, it is determined to be a packet loss and the generation of a negative response signal is triggered.
[0015] This application discloses a multi-slot, multi-channel audio and video narrowband transmission method based on channel quality awareness. This method periodically collects physical layer parameters and maintains a dynamic channel quality index set, enabling the system to reflect the real transmission capabilities of each channel in real time. Combined with a hysteresis update mechanism, it effectively suppresses scheduling jitter caused by transient noise, ensuring a stable and reliable resource allocation strategy. By dividing audio and video data into high, medium, and low priorities according to semantic importance and converting them into fixed-length micro-slices, the scheduling engine can accurately match data requirements with channel capabilities within each 10ms-level time slot—high-priority audio prioritizes the use of the optimal channel, critical video frames are protected by FEC, and non-critical frames are discarded as needed, thereby maximizing user experience quality under narrowband constraints. By assigning a globally unique and monotonically increasing sequence number (GSN) to all micro-slices, the receiver can ignore the physical channel origin and only rely on the GSN to complete out-of-order reordering and frame integrity judgment in a unified jitter buffer pool, completely solving the data fragmentation problem caused by concurrent multi-channel transmission. The NACK feedback using compressed bitmap format significantly reduces reverse signaling traffic; combined with the cross-channel preemption retransmission mechanism, lost critical slices can be quickly scheduled to the current best idle channel for retransmission, avoiding repeated failures on the inferior original channel, significantly shortening retransmission latency, and improving end-to-end reliability.
[0016] In summary, this invention, through a three-dimensional collaborative scheduling architecture of "channel-time slot-priority", achieves low latency, high availability and strong fault tolerance in audio and video transmission under conditions of extremely narrow bandwidth and high packet loss and weak network, effectively solving the shortcomings of existing technologies in terms of resource utilization efficiency, service awareness and cross-channel collaboration. Attached Figure Description
[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 The flowchart is provided for a method for multi-slot, multi-channel audio and video narrowband transmission based on channel quality awareness, as an embodiment of the present invention. Detailed Implementation
[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0019] like Figure 1 As shown in some embodiments of this application, this embodiment provides a method for multi-slot multi-channel audio and video narrowband transmission based on channel quality awareness. Specifically, the method includes the following steps: Step S1: The transmitting device periodically collects physical layer parameters of multiple wireless channels, maintains a dynamic state set containing the quality index of each channel, and sorts the available channels in real time according to the quality index.
[0020] Step S1 aims to provide reliable and real-time channel status information for subsequent multi-channel time slot scheduling. The transmitting device (e.g., an audio / video terminal integrating a multi-mode communication module) actively scans or passively listens to all its supported wireless channels (including but not limited to 4G / 5G cellular channels, Wi-Fi channels, dedicated narrowband IoT channels, and ad hoc network channels) at fixed time intervals (e.g., every 10 to 100 milliseconds) to collect physical layer parameters reflecting the current link transmission quality. These parameters typically include, but are not limited to: Received Signal Strength Indicator (RSSI), Signal-to-Noise Ratio (SNR), Signal-to-Interference-Ratio (SINR), Bit Error Rate (BER), Forward Error Correction Margin, Channel Occupancy Rate, and Delay Jitter Estimate.
[0021] Based on the collected raw parameters, the system uses a pre-defined weighted mapping function or machine learning model to fuse and calculate a unified Channel Quality Index (CQI). This CQI quantifies the overall transmission capability of each channel within the current period. The CQIs of all channels are centrally maintained in a dynamic state set, which is updated with each sampling period and serves as the core input for scheduling decisions. Subsequently, the system sorts all available channels in descending order according to their CQI values, forming a real-time updated channel priority queue.
[0022] It is worth noting that “available channels” here refers not only to physically accessible channels, but also to channels that meet the minimum quality of service threshold (such as CQI being higher than a preset threshold), thereby excluding invalid links that are severely degraded or congested.
[0023] For example, suppose in an emergency communication scenario, the transmitting device simultaneously possesses three communication capabilities: 4G public network, local Wi-Fi, and LoRa narrowband private network. The system performs channel probing every 20 milliseconds: on the 4G channel, the measured SINR is 8dB and the BER is 10. -3 On the Wi-Fi channel, interference from surrounding areas resulted in an RSSI of -85dBm and a channel occupancy rate of 70%. On the LoRa private network channel, although the bandwidth was only 50 kbps, the SINR remained stable at 15dB and the BER was below 10. -5 .
[0024] After calculation by the internal scoring model, the corresponding CQIs for the three channels are 65, 40, and 75, respectively. Therefore, the dynamic state set is updated to {Lora:75, 4G:65, Wi-Fi:40}, and a channel ranking is generated accordingly: Lora is the optimal channel, followed by 4G, and Wi-Fi is not currently involved in high-priority data scheduling. This ranking will directly affect whether audio microslices are allocated to the Lora channel in the next scheduling cycle, and whether video keyframes use the 4G channel as a backup path.
[0025] It should be noted that, in specific implementation scenarios, a multi-dimensional parameter fusion strategy can be adopted as a configurable scheme based on the above solutions. This means that the calculation of the channel quality index can support multiple fusion strategies, such as weighted summation based on expert rules, regression models based on historical performance, or lightweight neural network scorers. Users can choose the most suitable evaluation model according to the deployment environment. A diversified channel probing approach can be adopted, meaning that the acquisition of physical layer parameters is not limited to actively sending probe packets. Channel states can also be indirectly derived using ACK / NACK feedback from service data packets, link layer statistics, or cross-layer sensing mechanisms (such as MAC layer retransmission counts), reducing probing overhead. A dynamic state set-supported channel grouping management scheme can be adopted. When the number of channels is large, channels can be logically grouped according to dimensions such as frequency band, operator, and network type, and the order within each subset can be maintained to facilitate load balancing within groups or cross-group redundant scheduling. All of the above optional schemes fall within the scope of protection of this application.
[0026] Step S2: Receive the audio and video data stream input from the application layer, mark the priority according to the data type, divide the data stream into fixed-length microslices that are adapted to the physical layer time slot length, and add a global sequence number to each microslice.
[0027] Step S2 is a crucial preliminary step for implementing service-aware transmission scheduling. The transmitting device first receives the raw data stream from the upper-layer audio / video encoder (such as an H.264 / H.265 video encoder or an Opus audio encoder). This data stream is typically output in frames, containing content of varying semantic importance: for example, audio frames have extremely high real-time requirements, and their loss will directly lead to voice interruptions; video keyframes (I-frames) contain complete image information and are the basis for subsequent P / B frame decoding; while video reference frames (P-frames or B-frames) rely on forward prediction and have relatively high fault tolerance.
[0028] The system automatically assigns a priority tag to each frame of data based on preset rules or encoder metadata (such as NALU type, frame type flag, etc.), typically dividing it into three levels: high priority (such as audio frames, video I-frames), medium priority (such as video P-frames), and low priority (such as video B-frames or non-critical enhancement layers). Subsequently, the system further divides these priority-tagged data frames into fixed-length microslices. The length of each microslice strictly matches the maximum payload bytes that a single time slot can carry (e.g., 128 bytes of payload per time slot), ensuring that each microslice can be completely loaded into a single time slot for transmission, avoiding the scheduling complexity and additional overhead caused by cross-time slot fragmentation.
[0029] Finally, each microslice is assigned a Global Sequence Number (GSN) before encapsulation. This GSN is globally unique, monotonically increasing, and independent of the physical channel used throughout the entire transmission session. Regardless of which channel or time slot the microslice is subsequently transmitted through, its GSN remains unchanged, serving as the sole basis for cross-channel out-of-order reordering and packet loss detection at the receiver.
[0030] For example, suppose the maximum payload of the current physical layer timeslot is 100 bytes. At a certain moment, the application layer inputs a 200-byte audio data frame (high priority) and a 350-byte video I-frame (medium priority). The system first labels the audio frame as "high priority" and the video I-frame as "medium priority". Subsequently: the audio frame is divided into two 100-byte microslices, assigned GSN=1001 and GSN=1002 respectively; the video I-frame is divided into four microslices (the first three are 100 bytes each, and the last one is padded with 50 bytes to reach 100 bytes), assigned GSN=1003 to 1006.
[0031] These microslices, along with their priority identifiers and GSNs, enter the transmission queue, awaiting allocation to specific channel time slots in subsequent steps based on channel sorting results. Even if GSN=1001 is transmitted via a 4G channel and GSN=1002 is transmitted via a Wi-Fi channel, the receiving end can still accurately identify that they belong to the same audio frame using the GSN and reassemble them in order.
[0032] It should be noted that, in specific implementation scenarios, a more refined priority granularity scheme can be adopted based on the above scheme. That is, in addition to the three-level division, it can be expanded to multiple priority levels (such as five levels) according to business needs. For example, it can distinguish between voice activation segments and silent segments, and between video base layers and enhancement layers, to achieve more refined QoS control. A micro-slice length dynamic adaptation scheme can be adopted. Although a fixed-length design is usually used to simplify scheduling, in some embodiments, the micro-slice length can also be dynamically adjusted according to the MTU (Maximum Transmission Unit) or time slot structure of the currently available channels, as long as the slice length within the same frame is consistent and does not exceed the maximum time slot capacity. A diversified global sequence number generation mechanism scheme can be adopted. The GSN can be generated by an independent counter, or it can be combined with a timestamp and session ID to form a composite identifier. As long as it satisfies global uniqueness and monotonically increasing properties, it meets the requirements of this scheme. All of the above optional schemes are within the scope of protection of this application.
[0033] Step S3: Trigger a scheduling interrupt based on the current time slot, and dynamically map microslices of different priorities to the time slots of the currently available channels for transmission according to the priority of the microslices and the channel sorting results.
[0034] Step S3 is the core scheduling step for achieving efficient utilization of multi-channel resources and ensuring service differentiation. The system adopts a time-slot-driven scheduling mechanism: whenever a physical layer transmission time slot begins (e.g., every 10 milliseconds), the hardware or operating system triggers a scheduling interrupt, waking up the scheduler to execute the current round of resource allocation.
[0035] The scheduler first reads the latest channel sorting results maintained in step S1 (i.e., the list of available channels arranged in descending order of channel quality index), as well as the microslices to be transmitted and their priority tags generated and cached in the transmission queue in step S2. Subsequently, the scheduler processes the microslices in each priority queue in sequence according to the preset priority strategy (e.g., high priority > medium priority > low priority), and dynamically maps them to the idle time slots of each channel in the current period.
[0036] Dynamic mapping means that: the same microslice is not bound to a fixed channel, but the optimal available path is selected based on real-time channel quality; multiple microslices can be allocated in parallel to the same time slot on different channels (if the hardware supports multiple transmissions); if a channel is already occupied or unavailable in the current time slot, it is automatically skipped, and the next channel is tried; scheduling decisions are strictly limited to the time window of the current time slot to ensure low-latency response. This process achieves a triple collaboration of "channel quality awareness + service priority driving + time slot-level granularity," ensuring that limited narrowband resources always prioritize serving the most critical data units.
[0037] For example, suppose there are three available channels in the current scheduling period, ordered by CQI as follows: Channel A (best), Channel B (second best), and Channel C (available but of lower quality). Each channel has one idle transmission slot in this time slot (i.e., each supports one microslice).
[0038] At this time, the transmission queue contains: 1 high-priority audio microslice (GSN=2001); 2 medium-priority video I-frame microslices (GSN=2002, 2003); and 3 low-priority video P-frame microslices (GSN=2004~2006).
[0039] The scheduler first processes high-priority data, allocating GSN=2001 to the time slot of channel A; then it processes medium-priority data, allocating GSN=2002 to channel B. GSN=2003 cannot be used because channel A is already occupied, so it checks whether there is still room in channel B (if each channel only supports one microslice per time slot, then GSN=2003 can only wait for the next cycle); low-priority data is not sent temporarily because all channel time slots are currently occupied by high / medium-priority data, and is kept in the queue for subsequent scheduling.
[0040] It should be noted that, in specific implementation scenarios, in addition to the above solutions, a multi-micro-slice concurrent scheduling scheme can be adopted. That is, if the hardware platform supports single-channel multi-stream or MIMO transmission, a single channel can carry multiple micro-slices within a single time slot. The scheduler can thus improve throughput efficiency, especially suitable for batch transmission of low-to-medium priority data. A time slot reservation mechanism can be introduced. To ensure deterministic transmission of high-priority data, some optimal channel time slot resources can be reserved in advance for audio or key frames (e.g., reserving one time slot for audio every 10ms), avoiding congestion of critical data due to sudden traffic. A dynamic scheduling granularity adjustment scheme can be adopted. In scenarios with extremely unstable channels or extremely low load, the scheduling period can be extended from 10ms to 20ms or shortened to 5ms to balance scheduling overhead and response sensitivity, as long as it is still based on "time slot triggering." All of the above optional solutions are within the scope of protection of this application.
[0041] Step S4: The receiving device receives microslices from different channels, stores the microslices in a unified buffer pool according to the global sequence number, performs out-of-order rearrangement and frame reassembly, and generates packet loss feedback signaling when a discontinuous sequence number is detected.
[0042] Step S4 is a crucial step in achieving reliable cross-channel reception and ordered delivery. Because the transmitter employs a multi-channel parallel transmission strategy, different microslices of the same audio / video frame may arrive at the receiver via different physical paths (such as 4G, Wi-Fi, or private networks) with varying delays, leading to out-of-order delivery, duplication, or partial loss. To address this issue, the receiver establishes a unified cross-channel jitter buffer (also known as a unified buffer pool). This buffer pool does not distinguish the source channel of the microslice, using only its Global Sequence Number (GSN) as a unique index.
[0043] When a microslice is received on any channel, the receiver immediately parses its GSN and stores it in the corresponding location in the buffer pool. The system continuously monitors the continuity of GSNs in the buffer pool: once a missing GSN is detected (e.g., GSNs 3005 and 3007 have been received, but 3006 has not been received), it is determined to be a packet loss event. At this time, if the current time has not exceeded the playback deadline corresponding to the microslice (usually the playback timestamp minus the estimated round-trip time), a packet loss feedback signaling (such as a negative acknowledgment NACK) is triggered and sent back to the sender through the reverse control channel to request retransmission.
[0044] Simultaneously, when a consecutive GSN interval in the buffer pool happens to constitute a complete frame of data (e.g., an audio frame contains GSN=3001–3002, and a video I-frame contains 3003–3006), the system performs a frame reassembly operation, piecing together these micro-slices in order to restore the original frame, which is then delivered to the decoder for playback. This mechanism ensures that regardless of how out of order the micro-slices arrive, the final delivered data always maintains semantic integrity and temporal correctness.
[0045] For example, suppose the transmitter divides a video I-frame into four micro-slices, with GSNs of 5001, 5002, 5003, and 5004 respectively. GSNs 5001 and 5003 arrive via the 4G channel at t=105ms; GSN 5004 arrives via the Wi-Fi channel at t=108ms; GSN 5002 is lost due to sudden channel interference and does not arrive.
[0046] The receiving end stores the three received microslices into a unified buffer pool, located at positions 5001, 5003, and 5004 respectively. The system detects that GSN=5002 is missing, and the current time is 110ms, while the playback deadline for this frame is 130ms (considering RTT≈20ms). Therefore, it determines that there is still a retransmission window and generates a NACK signal indicating that GSN=5002 is lost.
[0047] Subsequently, if the retransmitted GSN=5002 is received before 125ms, the buffer pool is filled with 5001–5004, triggering frame reassembly and sending the complete I-frame to the decoder; if it is not received after the timeout, the application policy determines whether to skip the frame or enable the error hiding mechanism.
[0048] It should be noted that, in specific implementation scenarios, a dynamic buffer pool capacity adjustment scheme can be adopted based on the above solutions. This involves dynamically adjusting the maximum depth of the buffer pool according to the current network jitter level or playback latency tolerance to balance jitter resistance and end-to-end latency. A duplicate microslice filtering scheme can be adopted, meaning that if a microslice from the same GSN arrives multiple times due to retransmission or redundant transmission, the receiver can identify and discard duplicate copies based on the GSN, avoiding decoding errors. A frame boundary recognition-assisted reassembly scheme can be adopted, meaning that in addition to the GSN, the microslice header can also carry frame start / end flags to help the receiver quickly determine when a complete frame has been received, improving reassembly efficiency. A packet loss feedback strategy with hierarchical structure can be adopted, meaning that for high-priority microslices (such as audio), NACK can be generated immediately; for low-priority microslices (such as B-frames), feedback can be delayed or batched to reduce control signaling overhead. All of the above optional schemes are within the scope of protection of this application.
[0049] In some embodiments of this application, to avoid scheduling strategy oscillations caused by minor fluctuations and to improve the stability of multi-channel scheduling and the overall system transmission efficiency, the calculation and updating of the quality index in step S1 includes: An exponentially weighted moving average algorithm is used to smooth the currently collected physical layer parameters and historical scores, and the channel score for the current period is calculated. A hysteresis threshold is set, and the channel ranking in the dynamic state set is updated only when the change in the channel score compared to the previous period score exceeds the hysteresis threshold.
[0050] As described above, in step S1, to avoid frequent jumps in channel quality assessment due to instantaneous fluctuations, the system employs an Exponential Weighted Moving Average (EWMA) algorithm to smooth the channel score. Specifically, after periodically collecting the physical layer parameters (such as SINR, RSSI, BER, etc.) of the current wireless channel, they are first mapped to an original channel score for the current period. Subsequently, this original score is weighted and fused with historical cumulative scores according to preset weights to generate a smoothed channel score for the current period. This method assigns higher weight to recent observations while retaining historical trend information, making the channel quality assessment more stable and representative.
[0051] To further suppress frequent adjustments to channel ranking, the system introduces a hysteresis threshold mechanism. That is, in each update cycle, a channel's quality is considered to have changed significantly only if the absolute change between its current smooth score and its score in the previous cycle exceeds a preset hysteresis threshold, thus triggering a recalculation and update of the channel ranking in the dynamic state set; if the change does not exceed the threshold, the original ranking remains unchanged. This mechanism effectively avoids scheduling strategy oscillations caused by minor fluctuations, improving the stability of multi-channel scheduling and the overall system transmission efficiency.
[0052] In some embodiments of this application, in order to achieve out-of-order buffering, integrity judgment, and frame-level reassembly of microslices received across different channels, the encapsulation of the microslices in step S2 includes: Calculate the maximum payload capacity of the current transmission time slot; The input audio frames, video keyframes, and video reference frames are segmented into multiple consecutive microslices. The global sequence number is encapsulated in the header of each microslice. This global sequence number is shared across all channels and monotonically increases, serving as the sole basis for cross-channel reassembly at the receiver.
[0053] As described above, in step S2, the microslice encapsulation process first calculates the maximum number of valid data bytes that the current physical layer transmission time slot can carry, i.e., the maximum payload capacity, based on the structure and overhead of the current time slot (such as preamble, parity bit, control field, etc.). This value determines the upper limit of the data portion length of each microslice, ensuring that the microslice can be completely filled into a single time slot without cross-time slot fragmentation.
[0054] Subsequently, the system distinguishes and processes the audio and video data frames from the application layer according to their type: audio frames, video keyframes (I-frames), and video reference frames (P-frames or B-frames) are respectively divided into one or more consecutive micro-slices. The data length of each micro-slice does not exceed the maximum payload capacity mentioned above. If the last slice is insufficient, it is padded to maintain a fixed-length format, which is convenient for the scheduler to process uniformly.
[0055] After data segmentation, each microslice is appended with a Global Sequence Number (GSN) to its header. This GSN is maintained uniformly by the sender throughout the entire communication session, shared across all channels, and strictly monotonically increasing. Regardless of which channel a microslice is subsequently transmitted through, its GSN uniquely identifies its sequential position within the original data stream. The receiver relies on this GSN to perform out-of-order buffering, integrity checks, and frame-level reassembly of microslices received across different channels. Therefore, the GSN is the core basis for cross-channel cooperative transmission and reliable delivery.
[0056] In some embodiments of this application, to ensure the continuity and low latency of voice communication, the scheduling strategy for high-priority audio data in step S3 includes: The audio microslices are preferentially assigned to the currently ranked best channel slot; The system reads the global packet loss rate in real time. If the global packet loss rate exceeds the preset redundancy threshold, the same audio micro-slice is copied and allocated to the second-ranked suboptimal channel time slot for dual-path concurrent transmission.
[0057] As described above, when the scheduler processes high-priority audio microslices, it first assigns them to the idle time slot corresponding to the best channel with the highest quality index in the current channel ranking, so as to achieve reliable transmission by utilizing the best link conditions.
[0058] Meanwhile, the system continuously monitors the global packet loss rate obtained from feedback from the receiver or local statistics. If the packet loss rate exceeds a pre-set redundancy threshold (e.g., 15%), it indicates that the current network environment has deteriorated, and single-path transmission can no longer meet the reliability requirements of audio services. In this case, the scheduler will copy the same audio microslice and, within the current scheduling cycle, in addition to allocating it to the optimal channel, allocate the copy to the available time slot of the second-best channel, thus enabling concurrent transmission of the same audio microslice on two different channels.
[0059] This dual-path redundancy mechanism does not rely on retransmission waiting and can significantly increase the arrival probability of critical audio data within a single transmission opportunity, effectively combating sudden channel interruptions or high packet loss scenarios, thereby maintaining acceptable voice call quality under narrowband and weak network conditions.
[0060] In some embodiments of this application, in order to balance transmission reliability and bandwidth efficiency, the scheduling strategy for medium-priority video keyframe data in step S3 includes: Video keyframe microslices are preferentially allocated to the idle time slots of the currently ranked best channel. If the best channel has no idle time slots, they are allocated to the idle time slots of the second best channel. The forward error correction (FEC) redundancy ratio is dynamically calculated based on the current packet loss rate of the channel, a corresponding number of FEC redundancy slices are generated, and the FEC redundancy slices are filled into the subsequent available time slots.
[0061] As mentioned above, video keyframes (such as I-frames) play a fundamental role in video decoding. Their loss will prevent subsequent frames from being correctly reconstructed, thus requiring high protection. The scheduler first attempts to allocate the micro-slice corresponding to the keyframe to the idle time slot of the channel with the best quality in the current channel ranking (i.e., the channel ranked first). If this best channel has no available time slots in the current scheduling period, it is then allocated to the idle time slot of the second-best ranked channel, ensuring that keyframe data can still be sent in a timely manner and avoiding excessive encoding or playback delays caused by waiting for the best channel.
[0062] Meanwhile, the system dynamically calculates the forward error correction (FEC) redundancy ratio applicable to the key frame based on the historical or real-time packet loss rate of the currently selected channel. For example, the higher the packet loss rate, the greater the FEC redundancy ratio. Based on this ratio, the system generates a corresponding number of FEC redundant slices. These redundant slices are correlated with the original microslices and can be used to recover data when some original slices are lost. The generated FEC redundant slices are treated as low-priority data and filled into available time slots not occupied by high- or medium-priority data in subsequent scheduling cycles. This improves the packet loss resistance of the key frame and avoids the waste of resources caused by reserving fixed bandwidth for redundant data.
[0063] In some embodiments of this application, in order to achieve resource efficiency and timeliness as the core principles, the scheduling strategy for low-priority video reference frame data in step S3 includes: Video reference frame microslices are transmitted only when there is an idle time slot and the channel score of the channel to which the time slot belongs is higher than a preset threshold. The dwell time of the video reference frame microslice in the transmission queue is detected. If the dwell time exceeds the video frame interval, the microslice is discarded directly.
[0064] As mentioned above, since reference frames rely on keyframes for decoding, and their loss usually only affects local image quality rather than causing decoding interruption, the system adopts a non-preemptive, on-demand transmission method for their transmission. Specifically, the scheduler only allocates video reference frame microslices in the queue to a time slot for transmission if there is an idle time slot in the current scheduling period that is not occupied by high-priority or medium-priority microslices, and the channel quality score of the channel to which the idle time slot belongs is higher than a preset minimum available threshold; if the channel quality is too low or there are no idle resources, the scheduling is skipped and transmission is not forced.
[0065] Furthermore, the system continuously monitors the dwell time of each video reference frame microslice in the transmission queue. Once this dwell time exceeds the frame interval duration under the current video encoding configuration (e.g., 30 milliseconds corresponds to 30 fps), it indicates that even if successfully transmitted, the frame has missed its decoding or playback opportunity by the time it arrives at the receiver and cannot effectively contribute to video reconstruction. In this case, the system directly discards the microslice, freeing up queue space and avoiding unnecessary bandwidth consumption, thereby concentrating limited narrowband resources on time-sensitive critical data transmission.
[0066] In some embodiments of this application, in order to efficiently recover lost critical data, a cross-channel preemption retransmission step is also included, specifically: After receiving a negative acknowledgment signal from the receiver, the sending end parses out the global sequence number of the lost microslice; The priority of the lost microslices is temporarily raised to the highest level; In the next scheduling cycle, the lost microslice will be forcibly retransmitted to the idle channel with the highest current quality index, instead of being forcibly retransmitted on the original transmission channel.
[0067] As described above, when the transmitting end receives the negative acknowledgment (NACK) signaling from the receiving end, it first parses the global sequence number (GSN) of the lost microslice contained in the signaling, thereby accurately locating the data unit that needs to be retransmitted.
[0068] Subsequently, the system temporarily elevates the original priority of the lost microslice to the highest level (equivalent to the scheduling level of high-priority audio data), enabling it to preempt in subsequent scheduling. In the next scheduling cycle, the scheduler no longer considers which channel the microslice was originally transmitted through, but instead forcibly assigns it to the channel with the highest quality index in the current dynamic channel ranking and available idle time slots for retransmission.
[0069] This mechanism avoids repeated failures when the original transmission channel may have deteriorated. At the same time, it utilizes the optimal available channel and priority preemption strategy to significantly shorten retransmission delay and improve retransmission success rate. It is especially suitable for the rapid recovery of critical audio and video slices in narrowband weak network environments.
[0070] In some embodiments of this application, to reduce the transmission overhead of feedback signaling and adapt to the strict resource constraints of narrowband time slots, the negative acknowledgment signaling adopts a compressed bitmap format, specifically: When the receiver detects a global sequence number interruption, it generates compressed signaling containing a reference sequence number and subsequent N microslice reception status bitmaps. The length of the signaling is fixed and less than the control overhead limit of a single time slot.
[0071] As described above, when the receiver detects an interruption in the Global Sequence Number (GSN) (i.e., there are unreceived microslices), it does not report the lost sequence numbers one by one. Instead, it selects the most recently correctly received microslice sequence number as the reference sequence number and generates a binary bitmap of length N starting from this reference.
[0072] Each bit in this bitmap corresponds sequentially to the reception status of the 1st to Nth microslice following the reference sequence number: for example, a bit value of "0" indicates that the corresponding GSN microslice has been successfully received, and a bit value of "1" indicates that the microslice has been lost. The entire NACK signaling is composed of the reference sequence number and this N-bit status bitmap, and its total length is designed to be a fixed value, strictly less than the maximum control channel overhead allowed by a single physical layer time slot (e.g., no more than 20 bytes), thereby ensuring that the feedback signaling can be reliably transmitted within a control time slot and will not be fragmented or discarded due to excessive signaling size.
[0073] This compression mechanism efficiently represents multiple packet loss events under limited bandwidth, significantly reduces control traffic on the reverse link, and maintains accurate guidance for retransmission scheduling at the sending end.
[0074] In some embodiments of this application, in order to achieve dynamic matching between the sending end service bit rate and the current network transmission capacity, a source rate adaptation step is included before step S2, specifically: Sum the total effective bandwidth of all currently available channels; If the total bandwidth is lower than the current video encoding bitrate, the transmitting processor sends a degradation instruction to the video encoder to reduce the video resolution or frame rate to match the total channel capacity.
[0075] As described above, the transmitter first summarizes the total effective bandwidth of all available wireless channels in the current period. This effective bandwidth is estimated based on the quality index of each channel, historical throughput, or time slot resources that the scheduler can allocate, reflecting the net data rate that can actually be used for audio and video transmission.
[0076] The system then compares this total bandwidth with the current video bitrate output by the video encoder. If the total available bandwidth is detected to be lower than the current video encoding bitrate, it indicates that the existing channel capacity is insufficient to support the video stream at the current image quality and frame rate. Maintaining the original encoding parameters will lead to continuous packet loss or buffer overflow. At this point, the transmitting processor sends a degradation instruction to the video encoder, requesting it to actively reduce encoding parameters, such as reducing the video resolution (e.g., from 720p to 480p), reducing the frame rate (e.g., from 30fps to 15fps), or adjusting the quantization parameters to reduce the bitrate.
[0077] Through this mechanism, the video source output rate is dynamically adjusted to a level not exceeding the current multi-channel aggregation bandwidth, thereby avoiding transmission degradation caused by overload. This maximizes the utilization of limited narrowband resources while ensuring the overall continuity of audio and video. This adaptive process is independent of subsequent micro-slice scheduling and is a source control measure before transmission, forming a synergistic optimization with channel awareness and priority scheduling in the overall scheme.
[0078] In some embodiments of this application, to address the out-of-order and fragmentation problems caused by multi-channel parallel transmission, the specific process of cross-channel reassembly in step S4 includes: The receiving end establishes a unified jitter buffer across channels; When consecutive global sequence numbers in the buffer constitute a complete frame of data, a decoding interrupt is triggered to deliver the data to the decoder. If a non-continuous global sequence number is detected and the current time is earlier than the difference between the playback timestamp and the round-trip time, it is determined to be a packet loss and the generation of a negative response signal is triggered.
[0079] As mentioned above, the receiver first establishes a unified jitter buffer across channels. This buffer does not distinguish the source channel of the microslice, but uses the Global Sequence Number (GSN) as an index to centrally store all received microslices.
[0080] When there is a continuous and complete GSN sequence in the buffer, and this sequence exactly covers all the microslices corresponding to a complete audio and video frame, the system determines that the frame has been completely received, and then triggers a decoding interrupt, delivering the reconstructed complete frame data to the upper-layer decoder for processing and playback.
[0081] Meanwhile, the system continuously monitors the continuity of the GSN in the buffer. If an interruption is detected in a GSN segment (i.e., there is an unreceived microslice), it further determines whether the current system time is earlier than the playback timestamp of the frame corresponding to the lost microslice minus the estimated round-trip time (RTT). If this condition is met, it means there is still enough time to recover the lost data through retransmission, and it is determined to be a recoverable packet loss event. A negative acknowledgment (NACK) signaling is immediately generated to notify the sender to retransmit. If the current time is close to or exceeds this time limit, retransmission is not requested to avoid invalid feedback and decoding delays. This mechanism ensures the accuracy of reconstruction while balancing real-time performance and feedback efficiency.
[0082] Compared with existing technologies, this application discloses a multi-slot, multi-channel audio and video narrowband transmission method based on channel quality awareness. This method periodically collects physical layer parameters and maintains a dynamic channel quality index set, enabling the system to reflect the real transmission capabilities of each channel in real time. Combined with a hysteresis update mechanism, it effectively suppresses scheduling jitter caused by transient noise, ensuring a stable and reliable resource allocation strategy. By dividing audio and video data into high, medium, and low priorities according to semantic importance and converting them into fixed-length micro-slices, the scheduling engine can accurately match data requirements with channel capabilities within each 10ms-level time slot—high-priority audio prioritizes the use of the optimal channel, critical video frames are protected by FEC, and non-critical frames are discarded as needed, thereby maximizing user experience quality under narrowband constraints. By assigning a globally unique and monotonically increasing sequence number (GSN) to all micro-slices, the receiver can ignore the physical channel origin and only rely on the GSN to complete out-of-order reordering and frame integrity judgment in a unified jitter buffer pool, completely solving the data fragmentation problem caused by concurrent multi-channel transmission. The NACK feedback using compressed bitmap format significantly reduces reverse signaling traffic; combined with the cross-channel preemption retransmission mechanism, lost critical slices can be quickly scheduled to the current best idle channel for retransmission, avoiding repeated failures on the inferior original channel, significantly shortening retransmission latency, and improving end-to-end reliability.
[0083] In summary, this invention, through a three-dimensional collaborative scheduling architecture of "channel-time slot-priority", achieves low latency, high availability and strong fault tolerance in audio and video transmission under conditions of extremely narrow bandwidth and high packet loss and weak network, effectively solving the shortcomings of existing technologies in terms of resource utilization efficiency, service awareness and cross-channel collaboration.
[0084] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0086] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0087] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A method for multi-slot, multi-channel audio and video narrowband transmission based on channel quality awareness, characterized in that, Includes the following steps: S1. The transmitting device periodically collects physical layer parameters of multiple wireless channels, maintains a dynamic state set containing the quality index of each channel, and sorts the available channels in real time according to the quality index. S2. Receive the audio and video data stream input from the application layer, mark the priority according to the data type, divide the data stream into fixed-length micro-slices that are adapted to the physical layer time slot length, and add a global sequence number to each micro-slice; S3. Trigger a scheduling interrupt based on the current time slot, and dynamically map microslices of different priorities to the time slots of the currently available channels for transmission according to the priority of the microslices and the channel sorting results. S4. The receiving device receives microslices from different channels, stores the microslices in a unified buffer pool according to the global sequence number, performs out-of-order rearrangement and frame reassembly, and generates packet loss feedback signaling when a discontinuous sequence number is detected.
2. The method as described in claim 1, characterized in that, In step S1, the calculation and updating of the quality index includes: An exponentially weighted moving average algorithm is used to smooth the currently collected physical layer parameters and historical scores, and the channel score for the current period is calculated. A hysteresis threshold is set, and the channel ranking in the dynamic state set is updated only when the change in the channel score compared to the previous period score exceeds the hysteresis threshold.
3. The method as described in claim 1, characterized in that, In step S2, the encapsulation of the microslices includes: Calculate the maximum payload capacity of the current transmission time slot; The input audio frames, video keyframes, and video reference frames are segmented into multiple consecutive micro-slices. The global sequence number is encapsulated in the header of each microslice. This global sequence number is shared across all channels and monotonically increases, serving as the sole basis for cross-channel reassembly at the receiver.
4. The method as described in claim 1, characterized in that, In step S3, the scheduling strategy for high-priority audio data includes: The audio microslices are preferentially assigned to the currently ranked best channel slot; The system reads the global packet loss rate in real time. If the global packet loss rate exceeds the preset redundancy threshold, the same audio micro-slice is copied and allocated to the second-ranked suboptimal channel time slot for dual-path concurrent transmission.
5. The method as described in claim 1, characterized in that, In step S3, the scheduling strategy for medium-priority video keyframe data includes: Video keyframe microslices are preferentially allocated to the idle time slots of the currently ranked best channel. If the best channel has no idle time slots, they are allocated to the idle time slots of the second best channel. The forward error correction (FEC) redundancy ratio is dynamically calculated based on the current packet loss rate of the channel, a corresponding number of FEC redundancy slices are generated, and the FEC redundancy slices are filled into the subsequent available time slots.
6. The method as described in claim 1, characterized in that, In step S3, the scheduling strategy for low-priority video reference frame data includes: Video reference frame microslices are transmitted only when there is an idle time slot and the channel score of the channel to which the time slot belongs is higher than a preset threshold. The dwell time of the video reference frame microslice in the transmission queue is detected. If the dwell time exceeds the video frame interval, the microslice is discarded directly.
7. The method as described in claim 1, characterized in that, It also includes a cross-channel preemption and retransmission step, specifically: After receiving a negative acknowledgment signal from the receiver, the sending end parses out the global sequence number of the lost microslice; The priority of the lost microslices is temporarily raised to the highest level; In the next scheduling cycle, the lost microslice will be forcibly retransmitted to the idle channel with the highest current quality index, instead of being forcibly retransmitted on the original transmission channel.
8. The method as described in claim 7, characterized in that, The negative response signaling uses a compressed bitmap format, specifically: When the receiver detects a global sequence number interruption, it generates compressed signaling containing a reference sequence number and subsequent N microslice reception status bitmaps. The length of the signaling is fixed and less than the control overhead limit of a single time slot.
9. The method as described in claim 1, characterized in that, Before step S2, a source rate adaptation step is also included, specifically: Sum the total effective bandwidth of all currently available channels; If the total bandwidth is lower than the current video encoding bitrate, the transmitting processor sends a degradation instruction to the video encoder to reduce the video resolution or frame rate to match the total channel capacity.
10. The method as described in claim 1, characterized in that, In step S4, the specific process of cross-channel reassembly includes: The receiving end establishes a unified jitter buffer across channels; When consecutive global sequence numbers in the buffer constitute a complete frame of data, a decoding interrupt is triggered to deliver the data to the decoder. If a non-continuous global sequence number is detected and the current time is earlier than the difference between the playback timestamp and the round-trip time, it is determined to be a packet loss and the generation of a negative response signal is triggered.
