Weak network audio and video transmission system based on edge transcoding and sharing agent

Abstract

The application relates to the weak network audio and video transmission technical field and discloses a weak network audio and video transmission system based on edge transcoding and sharing agents, which comprises the following steps: a network monitoring module acquires network state parameters periodically reported by a plurality of weak network terminals; an edge server performs edge transcoding processing on high-definition audio and video streams; a local strong network agent device determines the comprehensive priority of each narrowband stream, performs dynamic time slot allocation and bandwidth multiplexing processing on the multiple narrowband streams, and generates an aggregated audio and video stream; the bandwidth usage of an uplink is acquired; when the bandwidth is over the limit, the narrowband stream with the lowest priority is discarded; when the bandwidth is still over the limit after being discarded, content integration processing is performed on the audio and video streams of the plurality of weak network terminals to form a single compressed audio and video stream; and a center platform receives the aggregated audio and video stream or the single compressed audio and video stream. The application reduces the connection load of the center platform and the public network bandwidth consumption, and shortens the end-to-end transmission delay.

Description

Technical Field

[0001] This invention relates to the field of weak network audio and video transmission technology, and more specifically, to a weak network audio and video transmission system based on edge transcoding and shared proxy. Background Technology

[0002] In typical weak network application scenarios such as emergency command, field operations, and industrial inspection, audio and video communication is a core component to ensure the security of mission execution and decision-making. These environments often face challenges such as extremely limited uplink bandwidth on the terminal side, poor network quality, high jitter, and uplink resource contention caused by concurrent uploads from multiple terminals.

[0003] Existing patent CN119729062A discloses a weak network audio and video transmission method with end-to-end state awareness. Based on the SFU architecture, it dynamically adjusts retransmission, redundancy strategies, and receiver buffering through uplink and downlink state reports, significantly improving packet loss resistance in single-sender-multiple-receiver scenarios. However, its technical logic fundamentally conflicts with multi-terminal weak network uplink scenarios: under weak network conditions at the terminal side, the retransmission + redundancy strategy requires additional transmission of redundant packets, which will drastically amplify uplink bandwidth pressure when multiple terminals are connected concurrently; it only adjusts packet strategies at the transport layer and does not perform content-level compression on the audio and video streams themselves; the SFU server needs to process media streams independently for each receiver, resulting in command instructions and ordinary video frames being treated the same during transmission, making critical services easily interrupted and lost. CN118101941A discloses a method and system for resisting weak networks in audio and video transmission, in which terminals directly upload high-definition streams to the central platform, edge servers use fixed bitrate transcoding, and each terminal independently occupies the uplink. The core issues are: multiple weak network terminals simultaneously compete for limited uplink resources from base stations, leading to a decrease in overall effective throughput due to channel contention; edge nodes transcode at a preset code rate, making it impossible to dynamically adjust based on real-time terminal bandwidth and packet loss rate; and the lack of priority quantization makes it impossible to guarantee the transmission of critical information when bandwidth is limited. In summary, existing technologies struggle to achieve coordinated optimization across four dimensions: single-stream adaptation accuracy, multi-stream aggregation efficiency, service priority guarantee, and congestion emergency resilience.

[0004] Therefore, it is necessary to design a weak network audio and video transmission system based on edge transcoding and shared proxy to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a weak network audio and video transmission system based on edge transcoding and shared proxy, which aims to solve the problem that existing technologies are unable to achieve coordinated optimization in four dimensions: single-stream adaptation accuracy, multi-stream aggregation efficiency, service priority guarantee, and congestion emergency resilience.

[0006] This invention proposes a weak network audio and video transmission system based on edge transcoding and shared proxy, comprising: The network monitoring module is used to obtain network status parameters periodically reported by several weak network terminals and generate a network status monitoring dataset. An edge server is used to perform edge transcoding on high-definition audio and video streams based on the network status monitoring dataset and bandwidth adaptability mapping relationship, and generate a narrowband stream dataset. A local strong network proxy device is used to add user role identifiers, data type identifiers, and urgency level identifiers to each narrowband stream based on the narrowband stream dataset, generating an identifierd narrowband stream set; based on the identifierd narrowband stream set, determine the comprehensive priority of each narrowband stream, generating a priority ranking dataset; based on the priority ranking dataset, perform dynamic time slot allocation and bandwidth multiplexing processing on multiple narrowband streams to generate aggregated audio and video streams; The local strong network proxy device is also used to obtain the uplink bandwidth usage. When the bandwidth exceeds the limit, the narrowband stream with the lowest priority is discarded according to the priority sorting dataset. If the bandwidth still exceeds the limit after discarding, the audio and video streams of several weak network terminals are integrated, and the remaining narrowband streams are mixed and screen-combined and encoded to form a single compressed audio and video stream. The central platform is used to receive the aggregated audio and video stream or the single-channel compressed audio and video stream.

[0007] Furthermore, when the edge server generates the narrowband stream dataset, it includes: Extract the measured available bandwidth of each weak network terminal from the network status monitoring dataset; Based on the measured available bandwidth and the bandwidth adaptability mapping relationship, determine the target output bit rate, resolution, and frame rate that match the current network status of each weak network terminal; Based on the target output bitrate, resolution, and frame rate, edge transcoding is performed on the high-definition audio and video streams corresponding to each weak network terminal to generate narrowband streams; the narrowband streams are then integrated into a narrowband stream dataset according to the terminal identifier.

[0008] Furthermore, when the local strong network proxy device generates the priority ranking dataset, it includes: Based on the user role identifier, data type identifier, and urgency level identifier, weight coefficients are assigned to determine the overall priority of each narrowband stream; based on the overall priority, the narrowband streams are sorted from high to low to generate a priority ranking dataset.

[0009] Furthermore, when the local strong network proxy device generates the aggregated audio and video stream, it includes: Based on the priority sorting dataset, continuous transmission time slots are dynamically allocated to high-priority narrowband streams; low-priority narrowband streams are filled into the idle time slots between the continuous transmission time slots; and the multiple narrowband streams after time slot allocation are bandwidth multiplexed according to the time sequence to generate aggregated audio and video streams.

[0010] Furthermore, when the local strong network proxy device generates the aggregated audio and video stream, it also includes: Based on the priority sorting dataset, narrowband streams with a comprehensive priority greater than the priority threshold are identified as high-priority narrowband streams, and narrowband streams with a comprehensive priority less than or equal to the priority threshold are identified as low-priority narrowband streams.

[0011] Furthermore, when the local strong network proxy device forms a single compressed audio / video stream, it includes: The audio data in the remaining narrowband stream is mixed to generate a synthesized audio stream; the video data in the remaining narrowband stream is combined to generate a synthesized video stream; the synthesized audio stream and the synthesized video stream are encoded and compressed to form a single-channel compressed audio and video stream.

[0012] Furthermore, when the local strong network proxy device encodes and compresses the synthesized audio stream and the synthesized video stream, it includes: The synthesized audio stream and synthesized video stream are time-stamp aligned and synchronized. The synchronized audio and video data are then jointly encoded and compressed using preset encoding parameters to form a single-channel compressed audio and video stream.

[0013] Furthermore, when the network monitoring module generates the network status monitoring dataset, it includes: Receive network status parameters sent by each weak network terminal within the reporting period; perform timestamp alignment and validity verification on the network status parameters, remove abnormal data, and generate qualified network status parameters; classify and integrate the qualified network status parameters according to the terminal identifier to generate a network status monitoring dataset.

[0014] Furthermore, the network status parameters include measured available bandwidth, packet loss rate, and jitter value.

[0015] Furthermore, the local strong network proxy device obtains the current uplink bandwidth usage in real time; compares the current bandwidth usage with the uplink bandwidth threshold; and determines that the bandwidth is exceeded when the current bandwidth usage is greater than the uplink bandwidth threshold.

[0016] Compared with existing technologies, the advantages of this invention are as follows: the edge server dynamically generates narrowband streams based on the real-time network status of the terminal, avoiding stuttering or bandwidth waste caused by fixed bitrates and improving the stability of single-stream transmission; the local strong network proxy device quantifies stream priority through triple identification of user role, data type, and urgency level, combined with dynamic time slot allocation and bandwidth reuse technology, ensuring the transmission of high-value services (such as command instructions and emergency images) under limited bandwidth, thus strengthening the reliability of critical information transmission; when the uplink exceeds the limit, low-priority streams are discarded first to quickly release bandwidth; if the limit is still exceeded, the remaining streams are mixed, screen merged, and jointly encoded to compress multiple information streams into a single high-information-density stream, avoiding complete transmission interruption and maximizing the preservation of core business content, thus improving the system robustness and availability in extreme weak network scenarios; multiple narrowband streams are aggregated and transmitted on the local proxy side, reducing the connection load of the central platform and the consumption of public network bandwidth; edge transcoding and local integration processing shift the computational pressure forward, reducing the processing burden of the central platform, while shortening end-to-end transmission latency and improving the system's real-time response capability. 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 This is a functional block diagram of a weak network audio and video transmission system based on edge transcoding and shared proxy provided in 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] In some embodiments of this application, see Figure 1 As shown, a weak network audio and video transmission system based on edge transcoding and shared proxy is proposed, including: The network monitoring module is used to obtain network status parameters periodically reported by several weak network terminals and generate a network status monitoring dataset. Edge servers are used to perform edge transcoding on high-definition audio and video streams based on network status monitoring datasets and bandwidth adaptability mapping relationships, generating narrowband stream datasets; The local strong network proxy device is used to add user role identifiers, data type identifiers, and urgency level identifiers to each narrowband stream based on the narrowband stream dataset, generating an identifierd narrowband stream set; based on the identifierd narrowband stream set, it determines the comprehensive priority of each narrowband stream, generating a priority ranking dataset; based on the priority ranking dataset, it performs dynamic time slot allocation and bandwidth multiplexing processing on multiple narrowband streams to generate aggregated audio and video streams; The local strong network proxy device is also used to obtain the uplink bandwidth usage. When the bandwidth exceeds the limit, the dataset is sorted according to priority and the narrowband stream with the lowest priority is discarded. If the bandwidth still exceeds the limit after discarding, the audio and video streams of several weak network terminals are integrated and processed. The remaining narrowband streams are mixed and screen merged and encoded and compressed to form a single compressed audio and video stream. The central platform is used to receive aggregated audio and video streams or single-channel compressed audio and video streams.

[0020] Specifically, the network monitoring module periodically receives network status parameters reported by each weak network terminal through a communication protocol (such as RTCP-XR). These network status parameters include measured available bandwidth, packet loss rate, and jitter value. The module performs timestamp alignment and validity verification (removing abrupt outliers) on the reported data, categorizes and integrates it according to terminal identifiers, and generates a structured network status monitoring dataset.

[0021] The edge server has a built-in bandwidth adaptability mapping table, which predefines the target output bitrate, resolution, and frame rate for different network state intervals. The server parses the network state monitoring dataset, extracts the measured available bandwidth of each terminal, queries the mapping relationship to determine the transcoding parameters, and calls a lightweight transcoding engine (such as integrating FFmpeg and a lightweight AI super-resolution module) to transcode the high-definition audio and video streams in real time, generating narrowband streams, and integrating them into a narrowband stream dataset according to the terminal identifier.

[0022] After receiving narrowband streams, the local strong network proxy device adds three service identifiers to each stream: a user role identifier generated based on terminal registration configuration information, a data type identifier automatically identified by parsing media stream header information (e.g., voice stream marker - voice, video I-frame marker - keyframe), and an urgency identifier generated by terminal-side event trigger signals or the proxy device's built-in keyword recognition module. Based on preset weight coefficients (user role, data type, and urgency are assigned weights respectively), the device calculates the comprehensive priority value of each narrowband stream and sorts them from high to low to generate a priority ranking dataset. Dynamic time slot allocation is then performed based on this dataset: continuous transmission time slots are allocated to high-priority narrowband streams to ensure low latency. The system intelligently fills low-priority narrowband streams into time slots and reuses them in time sequence to form aggregated audio and video streams. At the same time, the proxy device monitors the uplink bandwidth usage in real time through the network interface. When the usage exceeds a preset threshold (such as 85% of the nominal bandwidth of the link), it is determined to be a bandwidth overrun. The narrowband stream with the lowest overall priority is discarded first according to the priority sorting of the dataset. If the bandwidth is still overrun after discarding, the content integration process is initiated: the audio data of the remaining narrowband streams is mixed and synthesized to generate a synthesized audio stream, and the video data is gridded and screen-combined according to priority to generate a synthesized video stream. After time stamp alignment and audio-video synchronization, joint encoding compression is performed using preset encoding parameters to form a single compressed audio and video stream.

[0023] The central platform receives aggregated audio and video streams or single-channel compressed audio and video streams through a standard network interface, and completes decoding, distribution and storage for use by upper-layer business systems.

[0024] Understandably, by using dynamic transcoding at the edge to achieve precise matching between single-stream bandwidth and terminal network status, the uplink transmission pressure on weak network terminals is reduced; the local strong network proxy device constructs a priority-driven time slot multiplexing mechanism based on service semantic identifiers, improving uplink resource utilization efficiency and ensuring priority and reliable transmission of critical information such as command instructions; the two-level emergency strategy of hierarchical discarding and content integration compresses multiple core information streams into a single high information density stream when the link is extremely congested, maintaining communication continuity; the overall architecture moves the computation and scheduling pressure forward to the edge and proxy layers, avoiding channel competition caused by direct transmission from multiple terminals and reducing the load on the central platform and public network bandwidth consumption.

[0025] In some embodiments of this application, when the edge server generates a narrowband stream dataset, it includes: Extract the measured available bandwidth of each weak network terminal from the network status monitoring dataset; Based on the measured available bandwidth and bandwidth adaptability mapping relationship, determine the target output bit rate, resolution and frame rate that match the current network status of each weak network terminal; Based on the target output bitrate, resolution, and frame rate, edge transcoding is performed on the high-definition audio and video streams corresponding to each weak network terminal to generate narrowband streams; the narrowband streams are then integrated into a narrowband stream dataset according to the terminal identifier.

[0026] Specifically, the edge server first reads the measured available bandwidth values ​​corresponding to each weak network terminal from the network status monitoring dataset according to the terminal identifier index. These values ​​are valid data verified by the network monitoring module. Then, it calls the bandwidth adaptability mapping relationship pre-installed in the server storage unit (this mapping relationship is stored in the form of a structured data table or parameterized algorithm, clearly defining the correspondence logic between different measured available bandwidth ranges and the target output bitrate, resolution, and frame rate). Based on the measured available bandwidth of each terminal, it accurately queries and determines the target transcoding parameters dynamically matched to the current network status. Subsequently, the edge server calls the built-in lightweight transcoding... The processing unit (supporting encoding standards such as H.264 / AVC and H.265 / HEVC, and integrating adaptive quantization and GOP structure optimization modules) strictly performs real-time transcoding processing on the high-definition audio and video streams uploaded by each terminal according to the target parameters. Under the limited bitrate, it prioritizes the preservation of key visual information such as faces and text, generating narrowband streams that meet the requirements of narrowband transmission. After transcoding is completed, the edge server indexes and encapsulates each narrowband stream according to the unique identifier of the terminal, adds timestamps, transcoding parameters and verification identifiers, integrates them into a narrowband stream dataset with a clear structure and complete metadata, and pushes it to the local strong network proxy device through the internal communication interface.

[0027] Understandably, by using a bandwidth-adaptive mapping relationship to achieve precise dynamic matching between transcoding parameters and the real-time network status of the terminal, the bandwidth waste or transmission stuttering problems caused by fixed bitrate transcoding are avoided; lightweight transcoding is completed centrally at the edge, reducing the computing load and energy consumption of terminals in weak networks, while ensuring a high degree of adaptability between narrowband stream quality and network conditions.

[0028] In some embodiments of this application, when the local strong network proxy device generates the priority ranking dataset, it includes: Based on user role identifier, data type identifier, and urgency level identifier, weight coefficients are assigned to determine the overall priority of each narrowband stream; based on the overall priority, the narrowband streams are sorted from high to low to generate a priority ranking dataset.

[0029] Specifically, the proxy device first parses the service identification information attached to each narrowband stream. The user role identifier is determined based on the terminal registration configuration information (e.g., commander, medic, or ordinary team member, corresponding to a preset role level value). The data type identifier is automatically identified by parsing the media stream encapsulation header information (e.g., audio streams are marked as "-audio," video I-frames as "-keyframe," and P / B frames as "-ordinary frames," and mapped to a data type level value). The urgency level identifier is generated by the terminal-side event trigger signal or the proxy device's built-in lightweight keyword recognition module (e.g., if voice keywords such as "urgent" or "evacuation" are detected, it is marked as "high"; otherwise, it is set to "medium" or "low" according to preset rules and converted to an urgency level value). The proxy device then calls the weight coefficient table (user role weight coefficient α, data type weight coefficient β, and urgency level weight coefficient γ, the sum of which is 1, and can be dynamically adjusted according to business scenarios through the management interface) pre-stored in the configuration storage unit to perform a weighted calculation on each narrowband stream. After the calculation is completed, the proxy device sorts all narrowband streams in descending order of their overall priority values, generating a structured priority sorting dataset. This dataset contains the terminal identifier, overall priority value, sorting number, and original identifier information for each stream.

[0030] Understandably, by quantitatively integrating multi-dimensional business identifiers with configurable weight coefficients, the system achieves refined and objective evaluation of audio and video stream priorities, avoiding the subjectivity and bias of manual settings or single-dimensional judgments. The dynamic configurability of weight coefficients enables the system to flexibly adapt to the business support needs of different scenarios such as emergency command, telemedicine, and industrial inspection. The generated structured priority ranking dataset provides accurate decision-making basis for dynamic time slot allocation, bandwidth reuse, and over-limit emergency handling, ensuring that high-value information such as command instructions and emergency voices receive priority transmission when bandwidth is limited.

[0031] In some embodiments of this application, when a local strong network proxy device generates aggregated audio and video streams, it includes: Based on the priority sorting dataset, continuous transmission time slots are dynamically allocated to high-priority narrowband streams; low-priority narrowband streams are filled into the idle time slots between continuous transmission time slots; and the multiple narrowband streams after time slot allocation are bandwidth multiplexed according to the time sequence to generate aggregated audio and video streams.

[0032] In some embodiments of this application, when the local strong network proxy device generates aggregated audio and video streams, it further includes: Based on the priority sorting dataset, narrowband streams with a comprehensive priority greater than the priority threshold are identified as high-priority narrowband streams, while narrowband streams with a comprehensive priority less than or equal to the priority threshold are identified as low-priority narrowband streams.

[0033] Specifically, when the local strong network proxy device generates aggregated audio and video streams, it first sorts the dataset by priority and defines narrowband streams with a comprehensive priority value greater than a preset priority threshold as high-priority narrowband streams (such as key business streams like commander voices and video keyframes), and defines narrowband streams with a comprehensive priority value less than or equal to the threshold as low-priority narrowband streams (such as auxiliary information streams like non-critical video frames of ordinary team members). The priority threshold is a configurable parameter that can be dynamically adjusted according to business scenarios through the management interface (e.g., set to 7.0 for emergency command scenarios and 8.5 for routine inspection scenarios). Subsequently, continuous transmission time slots are dynamically allocated to each high-priority narrowband stream. The time slot length is calculated in real time based on the stream bit rate, frame interval, and service delay constraints. After allocation, the system accurately identifies the idle intervals between adjacent high-priority time slots, fills the idle time slots with low-priority narrowband streams in descending order of priority, and dynamically extracts data units using an adaptive fragmentation strategy, attaching fragmentation sequence numbers and boundary markers. Finally, all time slot data is strictly aligned according to the global time axis and seamlessly spliced ​​into a single data stream through a time-division multiplexing mechanism, embedding a time slot mapping table and synchronization timestamps to generate a structured and time-accurate aggregated audio and video stream.

[0034] Understandably, the system achieves refined resource scheduling through a two-layer mechanism of threshold definition and dynamic time slot allocation: the configurability of priority thresholds allows the system to flexibly adapt to the guarantee granularity requirements of different business scenarios, and can dynamically adjust the range of high-priority flows without modifying the core algorithm, thus improving the convenience of operation and maintenance and the adaptability of scenarios; continuous time slots ensure low jitter transmission of critical services, and the idle time slot filling strategy fully taps the link utilization rate and avoids bandwidth idleness.

[0035] In some embodiments of this application, when a local strong network proxy device forms a single compressed audio / video stream, it includes: The audio data in the remaining narrowband stream is mixed to generate a synthesized audio stream; the video data in the remaining narrowband stream is combined to generate a synthesized video stream; the synthesized audio stream and the synthesized video stream are encoded and compressed to form a single-channel compressed audio and video stream.

[0036] In some embodiments of this application, when the local strong network proxy device encodes and compresses the synthesized audio stream and the synthesized video stream, it includes: The synthesized audio stream and synthesized video stream are time-stamp aligned and synchronized. The synchronized audio and video data are then jointly encoded and compressed using preset encoding parameters to form a single-channel compressed audio and video stream.

[0037] Specifically, for the remaining narrowband streams after bandwidth over-limit processing, the audio data is first mixed—the mixing weights are dynamically allocated based on the overall priority of each narrowband stream (higher priority means greater weight), a weighted superposition algorithm is used to fuse multiple audio signals, and dynamic range compression and peak limiting techniques are used to suppress signal distortion, generating a clear and distinguishable synthesized audio stream; simultaneously, the video data is combined—based on the priority-sorted dataset, the video frames are arranged in a grid according to their priority (high-priority frames are placed in the center and appropriately enlarged, while low-priority frames are arranged in order around the perimeter), and each frame... The system performs adaptive scaling, boundary padding, and color consistency correction to generate a well-structured composite video stream. Then, it performs timestamp alignment and audio-video synchronization processing on the composite audio and video streams: using the time reference of the high-priority stream as a reference to correct the original timestamp deviation, and using interpolation compensation algorithms to eliminate audio-visual asynchrony, ensuring lip-sync and smooth video playback. Finally, it calls a pre-set encoding parameter library (including encoding standards, target bitrate, GOP structure, and quantization parameters) to jointly encode and compress the synchronized audio and video data, generating a single-channel compressed audio and video stream that is time-accurate, structurally complete, and conforms to transmission protocol requirements.

[0038] Understandably, by integrating mixing, screen merging, synchronization, and compression, multiple core audio and video information streams are efficiently condensed into a single high-information-density stream when the link is extremely congested, thus avoiding complete service interruption. The mixing strategy highlights the recognizability of key voice commands, the screen merging layout ensures the visual prominence and information integrity of high-priority images, and the timestamp alignment and synchronization mechanism ensures smooth playback without stuttering on the terminal. Joint encoding and compression maximize information carrying efficiency under the premise of strictly limiting the output bitrate, thereby improving the communication continuity of the system in extreme weak network scenarios such as satellite link interruption and sudden base station congestion.

[0039] In some embodiments of this application, when the network monitoring module generates a network status monitoring dataset, it includes: Receive network status parameters sent by each weak network terminal within the reporting period; perform timestamp alignment and validity verification on the network status parameters, remove abnormal data, and generate qualified network status parameters; classify and integrate the qualified network status parameters according to the terminal identifier to generate a network status monitoring dataset.

[0040] In some embodiments of this application, network status parameters include measured available bandwidth, packet loss rate, and jitter value.

[0041] Specifically, the module receives network status parameters actively reported by each weak network terminal at a preset reporting period (e.g., 2 seconds) via standard communication protocols (such as RTCP-XR or custom UDP packets). These parameters explicitly include the measured available bandwidth (unit: kbps), packet loss rate (percentage), and jitter value (unit: ms). Upon receiving the data, the module first performs timestamp alignment: using the module's local high-precision clock as a reference, it uniformly maps the timestamps of the data reported by each terminal to a standard time series. Data with timestamp deviations exceeding a preset tolerance (e.g., ±50 milliseconds) is compensated using linear interpolation to ensure strict time alignment of data from multiple terminals. Next, it performs validity verification: for the measured available bandwidth, it discards... Excluding outliers that exceed the theoretical range of the physical link (such as negative values ​​or exceeding 150% of the terminal's nominal bandwidth); for packet loss rate, invalid data less than 0% or greater than 100% is removed; for jitter values, a sliding window (window size of 10 periods) standard deviation detection method is used to remove abrupt changes that deviate from the window mean by more than 3 times the standard deviation; data that passes the verification is marked as qualified network status parameters; finally, the module classifies and integrates qualified parameters based on the terminal's unique identifier (such as device ID or MAC address) to generate a structured network status monitoring dataset. This dataset uses the terminal identifier as the index key and contains the valid parameters of each terminal in the current period, the verification timestamp, and the data quality identifier, which are available for real-time access by the edge server.

[0042] Understandably, by using timestamp alignment and multi-dimensional validity verification mechanisms, the interference of abnormal data caused by terminal reporting timing deviations and network jitter is eliminated, thereby improving the timing consistency and reliability of monitoring data. The structured classification and integration method provides edge servers with accurate and reliable terminal network status input, ensuring the accuracy of dynamic matching of transcoding parameters.

[0043] In some embodiments of this application, it also includes: The local strong network proxy device obtains the current uplink bandwidth usage in real time; compares the current bandwidth usage with the uplink bandwidth threshold; when the current bandwidth usage exceeds the uplink bandwidth threshold, it is determined that the bandwidth is exceeded.

[0044] Specifically, the proxy device continuously acquires real-time traffic data of the uplink physical interface through the operating system kernel-level network monitoring interface (such as the / proc / net / dev file or eBPF traffic statistics module in Linux systems) with a sampling period of 100 milliseconds. It performs sliding window averaging filtering on the instantaneous bandwidth values ​​of five consecutive sampling points to eliminate instantaneous spike interference caused by network jitter and calculates the smoothed current bandwidth usage (unit: kbps). The uplink bandwidth threshold is pre-stored in the device configuration storage unit and set to 85% of the nominal bandwidth of the link (for example, the threshold is set to 1275kbps when the nominal bandwidth of the satellite link is 1.5Mbps). It can also be dynamically adjusted by the management platform according to the link type (4G / 5G / satellite) or service scenario. The proxy device compares the smoothed current bandwidth usage with the threshold in real time. When the usage continuously exceeds the threshold to the preset confirmation window (500 milliseconds), it is determined that the effective bandwidth has exceeded the limit, and an over-limit event signal is immediately generated and the subsequent emergency handling process is triggered.

[0045] Understandably, the triple mechanism of high-frequency sampling, sliding filtering, and continuous confirmation identifies the true state of bandwidth exceeding limits, avoiding false triggers caused by instantaneous traffic fluctuations and improving the reliability of the judgment. The threshold is set as a percentage of the nominal bandwidth and supports dynamic configuration, taking into account the adaptability to different link characteristics and engineering operability. The seamless connection between the over-limit judgment and emergency handling provides timely and reliable triggering basis for strategies such as graded discarding and content integration, avoiding business interruption or resource waste caused by judgment delays or misjudgments.

[0046] The following examples illustrate this in detail: In one incident, the terminal uploaded 1080p audio and video streams via a 4G network (measured uplink bandwidth 30–110kbps, packet loss rate 18%–32%, jitter 40–120ms); the edge server was deployed on the base station side of the emergency communication vehicle; the command cabin had a built-in local strong network proxy device, which connected to the provincial command center platform via a Ka-band satellite link (nominal bandwidth 1.2Mbps, threshold set to 1020kbps).

[0047] Step 1: Edge-Terminal Collaborative Transcoding The terminal reports every 2 seconds via RTCP-XR: {device_id, avg_bandwidth=85kbps, plr=25%, jitter=65ms} (taking the commander's terminal as an example); The edge server uses a pre-defined bitrate mapping table: <50kbps→32kbps / 320×240 / 8fps 50–100kbps→64kbps / 480×360 / 12fps >100kbps→128kbps / 720p / 15fps The commander's stream was transcoded to a 64kbps narrowband stream, the medical staff stream (70kbps bandwidth) was transcoded in the same way, and the search and rescue team member stream (45kbps bandwidth) was transcoded to 32kbps. The transcoding process utilizes a lightweight ResNet super-resolution module to enhance the edges of key content such as face regions and map text, maintaining recognizability at low bitrates and avoiding image blurring and abrupt changes caused by traditional transcoding.

[0048] Step 2: Aggregate Shared Proxy Resource Pool The proxy equipment is for narrowband intelligent marking: Commander's voice stream: user_role=commander (weight 0.5), data_type=voice (weight 0.3), urgency_flag=high (detected keywords "landslide" and "urgently need support", weight 0.2) → Overall priority = 0.5×9 + 0.3×10 + 0.2×10 = 9.5 Medical staff video keyframes: Priority = 7.8 Search and rescue team P-frame: Priority = 3.2 Set a priority threshold of 7.5 to define the first two categories as high-priority flows; Dynamic time slot allocation: Continuous time slots are allocated for command voice (ensuring low latency of <150ms), followed by medical keyframes, and search and rescue team members fill the gaps; the total bandwidth after aggregation is 980kbps, and it is stably uploaded to the central platform.

[0049] Step 3: Two-tier emergency response for extreme congestion Emergency situation: Aftershocks caused instantaneous degradation of the satellite link, and the proxy device detected bandwidth usage of 1080kbps (exceeding the threshold of 1020kbps). Level 1 response: Based on priority, 8 non-critical video streams from search and rescue team members (lowest priority 3.2) are automatically discarded, bandwidth is reduced to 890kbps, and the link is restored to stability; Level 2 Response (Simulated Link Deterioration): If the limit is still exceeded after dropping (e.g., remaining flow bandwidth 1050kbps), immediately initiate content consolidation: Audio: The voices of the commander and three medics were weighted and mixed according to priority (0.55 for the commander and 0.15 for each medic), and the dynamic range was dynamically compressed to ensure that the instructions were clear and indistinguishable; Video: The video adopts a "1+3" grid layout—the commander's image is centered and enlarged (occupying 55% of the screen), and the images of the three medical personnel are arranged around it (each occupying 15%). The composite video stream is generated after color correction and edge smoothing. Synchronous compression: Audio and video synchronization is corrected based on the high-priority stream timestamp, and H.264 Main Profile encoding is used to compress it into a single 480kbps stream to continuously upload core live information.

[0050] The central platform decodes in real time: under normal conditions, it displays 12 split-screen images; in emergency situations, it seamlessly switches to an integrated screen, and the command screen simultaneously displays the dynamics of key personnel and voice commands.

[0051] In summary, the edge server dynamically generates narrowband streams based on the real-time network status of the terminal, avoiding stuttering or bandwidth waste caused by fixed bitrates and improving the stability of single-stream transmission. The local strong network proxy device quantifies stream priority through triple identification of user role, data type, and urgency level. Combined with dynamic time slot allocation and bandwidth reuse technology, it ensures the transmission of high-value services (such as command instructions and emergency images) under limited bandwidth, enhancing the reliability of critical information transmission. When the uplink exceeds the limit, low-priority streams are dropped first to quickly release bandwidth. If the limit is still exceeded, the remaining streams are mixed, screen merged, and jointly encoded to compress multiple information streams into a single high-information-density stream. This avoids complete transmission interruption and maximizes the preservation of core business content, improving the system robustness and availability in extreme weak network scenarios. Multiple narrowband streams are aggregated and transmitted on the local proxy side, reducing the connection load of the central platform and the consumption of public network bandwidth. Edge transcoding and local integration processing shift the computational pressure forward, reducing the processing burden of the central platform, while shortening end-to-end transmission latency and improving the system's real-time response capability.

[0052] 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 weak network audio and video transmission system based on edge transcoding and shared proxy, characterized in that, include: The network monitoring module is used to obtain network status parameters periodically reported by several weak network terminals and generate a network status monitoring dataset. An edge server is used to perform edge transcoding on high-definition audio and video streams based on the network status monitoring dataset and bandwidth adaptability mapping relationship, and generate a narrowband stream dataset. A local strong network proxy device is used to add user role identifiers, data type identifiers, and urgency level identifiers to each narrowband stream based on the narrowband stream dataset, thereby generating an identifierd narrowband stream set; Based on the identified narrowband stream set, the overall priority of each narrowband stream is determined, and a priority ranking dataset is generated; Based on the priority sorting dataset, dynamic time slot allocation and bandwidth multiplexing are performed on multiple narrowband streams to generate aggregated audio and video streams. The local strong network proxy device is also used to obtain the uplink bandwidth usage. When the bandwidth exceeds the limit, the narrowband stream with the lowest priority is discarded according to the priority sorting dataset. If the bandwidth still exceeds the limit after discarding, the audio and video streams of several weak network terminals are integrated, and the remaining narrowband streams are mixed and screen-combined and encoded to form a single compressed audio and video stream. The central platform is used to receive the aggregated audio and video stream or the single-channel compressed audio and video stream.

2. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 1, characterized in that, When the edge server generates a narrowband stream dataset, it includes: Extract the measured available bandwidth of each weak network terminal from the network status monitoring dataset; Based on the measured available bandwidth and the bandwidth adaptability mapping relationship, determine the target output bit rate, resolution, and frame rate that match the current network status of each weak network terminal; Based on the target output bitrate, resolution, and frame rate, edge transcoding is performed on the high-definition audio and video streams corresponding to each weak network terminal to generate narrowband streams; the narrowband streams are then integrated into a narrowband stream dataset according to the terminal identifier.

3. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 1, characterized in that, When the local strong network proxy device generates the priority sorting dataset, it includes: Based on the user role identifier, data type identifier, and urgency level identifier, weight coefficients are assigned to determine the overall priority of each narrowband stream; based on the overall priority, the narrowband streams are sorted from high to low to generate a priority ranking dataset.

4. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 3, characterized in that, When the local strong network proxy device generates aggregated audio and video streams, it includes: Based on the priority sorting dataset, continuous transmission time slots are dynamically allocated to high-priority narrowband streams; low-priority narrowband streams are filled into the idle time slots between the continuous transmission time slots; and the multiple narrowband streams after time slot allocation are bandwidth multiplexed according to the time sequence to generate aggregated audio and video streams.

5. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 4, characterized in that, When the local strong network proxy device generates aggregated audio and video streams, it also includes: Based on the priority sorting dataset, narrowband streams with a comprehensive priority greater than the priority threshold are identified as high-priority narrowband streams, and narrowband streams with a comprehensive priority less than or equal to the priority threshold are identified as low-priority narrowband streams.

6. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 1, characterized in that, When the local strong network proxy device forms a single compressed audio and video stream, it includes: The audio data in the remaining narrowband stream is mixed to generate a synthesized audio stream; the video data in the remaining narrowband stream is combined to generate a synthesized video stream; the synthesized audio stream and the synthesized video stream are encoded and compressed to form a single-channel compressed audio and video stream.

7. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 6, characterized in that, When the local strong network proxy device encodes and compresses the synthesized audio stream and synthesized video stream, it includes: The synthesized audio stream and synthesized video stream are time-stamp aligned and synchronized. The synchronized audio and video data are then jointly encoded and compressed using preset encoding parameters to form a single-channel compressed audio and video stream.

8. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 1, characterized in that, When the network monitoring module generates the network status monitoring dataset, it includes: Receive network status parameters sent by each weak network terminal within the reporting period; perform timestamp alignment and validity verification on the network status parameters, remove abnormal data, and generate qualified network status parameters; classify and integrate the qualified network status parameters according to the terminal identifier to generate a network status monitoring dataset.

9. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 8, characterized in that, The network status parameters include measured available bandwidth, packet loss rate, and jitter value.

10. The weak network audio and video transmission system based on edge transcoding and shared proxy according to claim 1, characterized in that, Also includes: The local strong network proxy device obtains the current bandwidth usage of the uplink in real time; Compare the current bandwidth usage with the uplink bandwidth threshold; When the current bandwidth usage exceeds the uplink bandwidth threshold, it is determined that the bandwidth has exceeded the limit.

Images