A multi-functional integrated digital network broadcasting system based on intelligent algorithm

Abstract

The application relates to the technical field of digital network broadcasting, in particular to a multifunctional integrated digital network broadcasting system based on an intelligent algorithm.The system comprises a multi-modal data acquisition module, a broadcasting content generation module, a broadcasting scheduling decision module, an abnormal interference suppression module and a terminal cooperative feedback module; the multi-modal data acquisition module collects multi-source information in a covered area in real time; the broadcasting content generation module performs intelligent clustering analysis based on the information to generate structured content; the broadcasting scheduling decision module adopts a graph attention scheduling algorithm to optimize a scheduling strategy; the abnormal interference suppression module identifies an interference source by using a residual convolution network and repairs a disturbed signal by using a recurrent neural network; and the terminal cooperative feedback module executes playing and feedback data and dynamically adjusts receiving parameters and a pushing strategy by means of federal reinforcement learning.The application significantly improves the intelligent level, adaptability, user satisfaction and overall service efficiency of the digital network broadcasting system.

Description

Technical Field

[0001] This invention relates to the field of digital network broadcasting technology, and more particularly to a multifunctional integrated digital network broadcasting system based on intelligent algorithms. Background Technology

[0002] With the rapid development of information and communication technologies, digital network broadcasting systems are widely used in various scenarios such as public safety early warning, emergency command transmission, rail transit guidance, and smart campus information dissemination. Traditional broadcasting systems mainly rely on a central server to push preset content on a scheduled basis. Their core mechanism usually adopts a static scheduling strategy, a fixed playlist, and a one-way transmission channel, making it difficult to adjust in real time according to the dynamic changes in the actual application environment and user needs. This rigid mechanism not only affects the timeliness and accuracy of information delivery, but also has obvious limitations in the context of network fluctuations, terminal diversity, and content diversity. Currently, the following problems still exist: Existing systems mainly rely on manual settings or static templates for the generation and distribution of broadcast content, making it difficult to dynamically adjust the content structure according to the actual environment, and lacking comprehensive perception and intelligent analysis of multi-source heterogeneous data; Traditional broadcast scheduling decision-making methods are mostly based on preset rules or simple priority ranking, and cannot achieve dynamic adaptive resource allocation and task optimization based on terminal location, regional congestion status, and user historical response behavior; In complex electromagnetic environments, existing systems have limited ability to identify and suppress abnormal factors such as channel conflicts, signal attenuation, and malicious interference, and lack efficient anomaly detection and signal repair mechanisms, which can easily lead to broadcast signal distortion, interruption, or even loss, affecting the continuity and reliability of broadcast services. Summary of the Invention

[0003] To address the aforementioned issues, this invention provides a multi-functional integrated digital network broadcasting system based on intelligent algorithms. This system overcomes the limitations of traditional digital broadcasting systems in areas such as intelligent content generation, adaptive scheduling optimization, and abnormal interference suppression. It enables precise distribution of broadcast content and intelligent full-process management under the fusion of multi-source heterogeneous data, thereby improving the intelligence level, adaptability, user satisfaction, and overall service efficiency of the digital network broadcasting system.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A multi-functional integrated digital network broadcasting system based on intelligent algorithms includes a multi-modal data acquisition module, a broadcast content generation module, a broadcast scheduling decision module, an abnormal interference suppression module, and a terminal collaborative feedback module that are connected in sequence.

[0006] The multimodal data acquisition module is used to collect multi-source information within the broadcast coverage area in real time;

[0007] The broadcast content generation module is used to perform cluster analysis on the semantic features of the content using the sparse tensor semantic clustering algorithm based on the multi-source information, and to construct a content-context tag mapping relationship by combining historical playback data to generate structured broadcast content.

[0008] The broadcast scheduling decision module is used to construct a broadcast task flow graph with time window constraints based on the structured broadcast content, combined with terminal location, regional congestion status and historical broadcast response data, and to jointly optimize the scheduling strategies of broadcast time window, content priority and terminal scheduling order by adopting a graph attention scheduling algorithm based on asynchronous advantage update, and generate broadcast scheduling decision instructions.

[0009] The abnormal interference suppression module is used to identify potential interference sources during broadcast signal transmission using a residual convolutional neural network algorithm, and to reconstruct the disturbed broadcast segment through a signal repair mechanism driven by a recurrent neural network; the potential interference sources include channel conflict, signal attenuation, and malicious interference;

[0010] The terminal collaborative feedback module is used to receive the broadcast signal controlled by the broadcast scheduling command, perform playback operations on the terminal, collect terminal playback status data, user interaction feedback data and environmental response data, and dynamically adjust the terminal receiving parameters and content push strategy through a federated reinforcement learning algorithm.

[0011] Furthermore, the multi-source information includes environmental sound signals, real-time voice signals collected by microphones, video surveillance image data, regional meteorological parameters, electromagnetic interference signal detection data, terminal device geographical location and motion status data, user interaction behavior data, and network link status parameters.

[0012] Furthermore, the operation of the broadcast content generation module includes the following steps:

[0013] The multi-source information is standardized and preprocessed, and a multi-order feature tensor containing semantic, temporal, and modal dimensions is constructed.

[0014] Based on the multi-order feature tensor, the sparse tensor semantic clustering algorithm is used to compress the semantic units and cluster them into topics. The semantic core units in multi-source content are identified by low-rank representation and semantic sparsity constraints to form a set of candidate broadcast semantic segments.

[0015] Based on the candidate broadcast semantic fragment set, combined with historical broadcast playback data and terminal response records, a content-context tag mapping relationship is constructed using a tag migration modeling method, and a content-context matching matrix is ​​generated using a multi-scenario adaptation model based on tag weight fitting.

[0016] Based on the candidate broadcast semantic fragment set and content-context matching matrix, a broadcast structure generation mechanism based on constraint satisfaction problem solving is adopted to reconstruct the order and optimize the format of the candidate content, generating structured broadcast content including time series logic, context adaptability and multimodal fusion features.

[0017] Furthermore, the formula for the sparse tensor semantic clustering algorithm is as follows:

[0018]

[0019] in, Let S represent the set of optimal semantic clustering results; S represents the set of all clustering partitioning schemes. Represents multi-order feature tensor elements; This represents the content-context label weight matrix; i represents the semantic unit number; j represents the time window number; k represents the modality number; This represents the set of data indices within the c-th cluster; This represents the number of multi-source data entries in the c-th cluster; This represents the cluster indicator vector for the c-th cluster; The c-th topic cluster contains the number of feature entries; C represents the final number of topic categories. This represents the sparsity adjustment parameter.

[0020] Furthermore, the operation of the broadcast scheduling decision module includes the following steps:

[0021] Based on the structured broadcast content, combined with the terminal spatial distribution, congestion status within the region, and historical broadcast response data, a broadcast task flow graph with time window constraints is constructed, where nodes represent specific broadcast content, edges represent the temporal or logical dependencies between content, and time windows are used to limit the executable time range of each task node.

[0022] Collect information on the link accessibility, reception capability, interaction behavior characteristics and content preference of each terminal, construct a terminal state tensor, and fuse it with the broadcast task graph to generate a state-enhanced scheduling graph.

[0023] Based on the state-enhanced scheduling graph, a graph attention scheduling algorithm based on an asynchronous advantage update mechanism is adopted to dynamically calculate the scheduling value function of each broadcast task node. Based on the structural correlation between nodes, content priority and terminal feedback potential, attention weights are allocated, and the broadcast time window allocation, content delivery priority ranking and target terminal scheduling order are jointly optimized through asynchronous strategy.

[0024] Based on the optimization results, broadcast scheduling decision instructions are generated, including content identifiers, target terminal ID lists, predetermined time windows, and frequency resource allocation information. These instructions are then encapsulated and issued through the heterogeneous link control module.

[0025] Furthermore, the asynchronous advantage update mechanism specifically includes the following steps:

[0026] Based on the state-enhanced scheduling graph, for each broadcast task node, the node state transition sequence is asynchronously sampled based on the node's historical scheduling state, current environmental characteristics, and terminal feedback data, and the scheduling value function of the task node is dynamically updated.

[0027] By performing weighted difference on the scheduling value function at different time steps, the advantage estimation function is calculated, and combined with the context attention weight of each node in the graph structure, local sensitivity adjustment of the scheduling priority of broadcast content and the selection of target terminals is achieved.

[0028] In scenarios involving multi-task concurrency and resource contention, a distributed parallel computing approach is adopted to independently optimize the scheduling parameters of each node and update the global scheduling strategy parameters in an asynchronous manner.

[0029] The global scheduling strategy parameter update based on the asynchronous advantage update mechanism integrates the latest scheduling priority and resource allocation suggestions of the nodes into the final broadcast scheduling decision output, including content distribution order, target terminal grouping and time slot resource allocation information, and sends scheduling instructions to the heterogeneous link control module.

[0030] Furthermore, the formula for the graph attention scheduling algorithm is as follows:

[0031]

[0032] in, This represents the final scheduling priority score of broadcast task node i; This represents the content importance score corresponding to the i-th task node in the structured broadcast content; This represents the user activity index within the target area of ​​task i; This represents the historical feedback score for content i; This indicates the current link bandwidth utilization rate of the geographical area where task i is located; Indicates the maximum link bandwidth capacity; This represents the congestion risk coefficient of task i in the state-enhanced scheduling graph; Indicates the remaining schedulable time window for task i; , and This represents the weighting factor for the score; The exponential amplification factor representing the response feedback score; and These represent the slope of the Sigmoid function and the center position parameter in the time window urgency score, respectively.

[0033] Furthermore, the operation of the abnormal interference suppression module includes the following steps:

[0034] The system performs real-time acquisition of multi-channel signal streams in the broadcast signal transmission link and uses a multi-scale decomposition method to extract multi-dimensional characteristic parameters such as signal energy spectrum distribution, instantaneous frequency characteristics, channel utilization, and interference index.

[0035] Based on the residual convolutional neural network algorithm, the above feature parameters are modeled in time and fused in space. Combined with channel history data and statistical thresholds, the system can automatically detect abnormal signal segments and classify the interference type.

[0036] For the detected abnormal segments, an adaptive signal repair mechanism driven by a recurrent neural network is adopted to reconstruct the content of the disturbed signal based on the temporal pattern and amplitude characteristics of historical reference segments.

[0037] By comparing the signals before and after reconstruction, and combining the current link dynamics with the target terminal status, an interference suppression control vector is generated, including strategy parameters at the levels of forwarding delay adjustment, link remapping, frequency band avoidance, and content reconstruction. The abnormal suppression processing results are then synchronously fed back to the terminal collaborative feedback module.

[0038] Furthermore, the terminal collaborative feedback module constructs a policy adaptation mechanism based on a federated reinforcement learning architecture. It periodically optimizes the receiving policy using playback status data, interaction feedback data, and environmental awareness parameters collected locally on the terminal, and achieves cross-terminal policy updates through global policy aggregation.

[0039] The beneficial effects of this invention are as follows:

[0040] This invention utilizes a multimodal data acquisition module to acquire real-time heterogeneous information from multiple sources within the broadcast area, including audio, video, environmental data, and user behavior. Combined with a sparse tensor semantic clustering algorithm, it extracts core content features at the semantic level, constructing a semantic mapping model between content and context. This makes broadcast content more aligned with audience needs and environmental context, significantly improving the effectiveness and acceptance rate of information dissemination. The broadcast scheduling decision module, considering terminal location distribution, regional congestion status, and historical response data, constructs a task flow graph with time window constraints. It identifies key nodes through a graph attention mechanism and dynamically optimizes scheduling using an asynchronous advantage update strategy, effectively avoiding conflicts and duplication of broadcast content, improving the timeliness and accuracy of scheduling, and ensuring that high-priority content is delivered first at key nodes. The abnormal interference suppression module integrates a residual convolutional neural network algorithm to identify potential anomalies such as channel conflicts and malicious interference. Leveraging a recurrent neural network-driven temporal feature modeling mechanism, it achieves real-time repair and reconstruction of disturbed signal segments, significantly improving the stability and reliability of the broadcast link in complex transmission environments and reducing the risk of content interruption. The terminal collaborative feedback module not only realizes the real-time collection of terminal playback status and user behavior data, but also makes personalized dynamic adjustments to terminal receiving parameters and pushed content through a federated reinforcement learning mechanism. This ensures user privacy and security while improving the receiving experience of individual users, enabling the system to have a closed-loop optimization capability of "perception-feedback-control". Attached Figure Description

[0041] Figure 1 This is a schematic diagram of a multifunctional integrated digital network broadcasting system based on intelligent algorithms according to the present invention.

[0042] Figure 2 This is a flowchart illustrating the operation of a broadcast content generation module according to an embodiment of the present invention.

[0043] Figure 3 This is a flowchart illustrating the operation of the broadcast scheduling decision module provided in an embodiment of the present invention. Detailed Implementation

[0044] Please see Figure 1-3 As shown, the present invention relates to a multifunctional integrated digital network broadcasting system based on intelligent algorithms.

[0045] Example

[0046] A multi-functional integrated digital network broadcasting system based on intelligent algorithms includes a multi-modal data acquisition module, a broadcast content generation module, a broadcast scheduling decision module, an abnormal interference suppression module, and a terminal collaborative feedback module that are connected in sequence.

[0047] The multimodal data acquisition module is used to collect multi-source information within the broadcast coverage area in real time; the multi-source information includes environmental sound signals, real-time voice signals collected by microphones, video surveillance image data, regional meteorological parameters, electromagnetic interference signal detection data, terminal device geographical location and motion status data, user interaction behavior data, and network link status parameters.

[0048] Specifically, this module integrates multiple types of sensors and data acquisition terminals, which are deployed at key nodes within the broadcast coverage area (such as teaching buildings, subway stations, business districts, community entrances and exits, etc.).

[0049] Ambient sound collection: High-sensitivity digital ambient sound sensors are installed at key locations in the broadcast coverage area (such as subway passages, shopping malls, campuses, etc.). A hybrid layout of directional and omnidirectional sensors is adopted to achieve full coverage and collect background noise and abnormal sound sources (such as alarms, noise, vehicle horns, etc.) in real time 24 hours a day.

[0050] Real-time microphone voice acquisition: Install high-sensitivity array microphones in densely populated areas to acquire real-time voices and sudden verbal commands from on-site personnel, providing voice input for event response and emergency broadcasting.

[0051] Video surveillance image data: Integrates high-definition cameras to collect video information such as the distribution of people, abnormal behavior, and environmental changes in the area, and supports content recognition and event-driven processing.

[0052] Regional meteorological parameter collection: Deploy multi-parameter sensors such as temperature, humidity, air pressure, wind speed, and rainfall to dynamically collect environmental meteorological data, providing a basis for content delivery and scheduling optimization under extreme weather conditions.

[0053] Electromagnetic interference signal detection: Establish electromagnetic environment monitoring points, deploy dedicated radio frequency detectors and broadband signal acquisition devices, scan key operating frequency bands and spare frequency bands, and monitor and record signal strength, noise levels and interference events within specific frequency bands.

[0054] Geographic and motion data of terminal devices: By using the built-in GPS / BeiDou module and accelerometer of the broadcast terminal, the location of the terminal and its motion trajectory are collected in real time to achieve spatial precision in content distribution.

[0055] User interaction behavior data: The terminal side integrates touch, button, and voice interaction interfaces to collect user feedback, commands, and interaction data on broadcast content.

[0056] Network link status parameters: Periodically detect the bandwidth, latency, packet loss rate, etc. of wireless and wired network links within the area to evaluate network quality and provide real-time support for scheduling and push strategies.

[0057] All collected data undergoes initial aggregation and preprocessing (such as data cleaning, anomaly removal, and feature compression) at local edge nodes before being securely uploaded to the system data center.

[0058] The broadcast content generation module is used to perform cluster analysis on the semantic features of the content using the sparse tensor semantic clustering algorithm based on the multi-source information, and to construct a content-context tag mapping relationship by combining historical playback data to generate structured broadcast content.

[0059] The operation of the broadcast content generation module includes the following steps:

[0060] The multi-source information is standardized and preprocessed, and a multi-order feature tensor containing semantic, temporal, and modal dimensions is constructed.

[0061] Specifically, data from different sources (such as text, audio, video, and environmental parameters) undergo format standardization, missing value completion, anomaly detection, and filtering. For example, audio signals are first subjected to noise suppression and normalization, video frames are standardized in resolution, text data is standardized in character encoding, and environmental parameters are subjected to anomaly removal and interpolation completion.

[0062] Align all modal features to form a three-dimensional or higher-dimensional tensor, with the axes as follows:

[0063] Semantic dimension (different semantic units, such as keywords, topic tags, etc.)

[0064] The time dimension (the temporal information of data fragments, ensuring the order and context of events)

[0065] Modal dimension (data source type, such as text, audio, video, environment, etc.)

[0066] Sparsity detection and indexing of tensor data structures facilitate efficient clustering and topic discovery in the future.

[0067] Based on the multi-order feature tensor, the sparse tensor semantic clustering algorithm is used to compress the semantic units and cluster them into topics. The semantic core units in multi-source content are identified by low-rank representation and semantic sparsity constraints to form a set of candidate broadcast semantic segments.

[0068] Specifically, on the data center server, for the constructed multi-level feature tensors, a tensor structure integrity check is first performed to remove extremely sparse dimensions caused by collection anomalies, ensuring the accuracy of subsequent clustering. Using efficient sparse tensor analysis toolkits (such as Tensorly), the tensor data is progressively decomposed, automatically selecting the most representative topic feature components with high signal-to-noise ratios, achieving high-density aggregation and noise reduction of semantic information.

[0069] Based on the topic feature space obtained from tensor decomposition, the system automatically distinguishes between global topics (such as emergency alarms, traffic announcements, and routine notifications) and local scene topics (such as temporary events at specific locations or time periods). It employs various distribution-independent clustering strategies, such as spectral clustering and density clustering, to ensure that even a small number of important peripheral events or niche scene content can be identified and extracted, preventing mainstream data topics from "drowning" long-tail information.

[0070] The system traces the source of each clustering result, automatically locating high-weight segments in the original multi-source data (such as a voice recording that elicits a high response, a video clip containing abnormal behavior, or a parameter change range corresponding to a sudden weather event). It generates a complete metadata structure for each candidate segment, including the source modality, start and end times, key semantic tags, and relevant contextual descriptions; some segments support multimodal cross-validation (such as the same event being simultaneously tagged by audio, video, and environmental sensor data). All candidate segments are pushed to a review interface for manual review or automated filtering, ultimately selecting a "semantic segment whitelist" that can participate in content generation.

[0071] Based on the candidate broadcast semantic fragment set, combined with historical broadcast playback data and terminal response records, a content-context tag mapping relationship is constructed using a tag migration modeling method, and a content-context matching matrix is ​​generated using a multi-scenario adaptation model based on tag weight fitting.

[0072] Specifically, the system automatically captures all recent historical broadcast content and its corresponding terminal feedback data (such as user click-through rate, playback completion rate, interaction frequency, complaints, or likes). Using "content fragment - contextual tag - feedback performance" as the primary key, a high-dimensional data table is built in the knowledge base, forming a response mapping between content and actual application scenarios. External data sources (such as holiday schedules, weather service APIs, and city event calendars) are automatically imported to enrich the semantic breadth and adaptability of the contextual tag library.

[0073] For newly generated candidate semantic segments, the system uses methods such as semantic embedding similarity analysis and contextual behavior feature comparison to migrate effective contextual tags from historical content to the new segments. Each segment can obtain multiple candidate contextual tags and their adaptation reliability, and the system automatically adjusts the tag weights based on historical data. For example, if content has a high playback response during "rainy morning rush hour," the corresponding contextual tag weight will be increased. For cold-start content that cannot be automatically adapted, the system can use a weak tag recommendation mechanism to prioritize the allocation of contextual tags with the highest similarity and simultaneously collect terminal responses after the first push to quickly supplement the tag weights.

[0074] The system establishes a real-time, updatable, multi-dimensional adaptation matrix, with content fragments as rows and context tags as columns. This matrix not only includes historical statistical adaptation scores but also incorporates dynamic weighted adjustments based on real-time terminal responses and external event factors, ensuring that matrix weights continuously optimize as the environment and user needs change. All content-context matching matrices support fast retrieval and reverse indexing, facilitating real-time push and distribution based on scenario and content adaptation by the subsequent scheduling module.

[0075] Based on the candidate broadcast semantic fragment set and content-context matching matrix, a broadcast structure generation mechanism based on constraint satisfaction problem solving is adopted to reconstruct the order and optimize the format of the candidate content, generating structured broadcast content including time series logic, context adaptability and multimodal fusion features.

[0076] Specifically, the system defines multiple constraints for each broadcast content generation task, such as total duration (must not exceed limits), topic coverage (must include key events), redundancy (the same topic is not allowed to be broadcast consecutively), and content exclusivity (different types of warnings cannot be broadcast simultaneously). User groups, scenario types, and special regional needs serve as additional constraint inputs to guide content assembly and arrangement.

[0077] Based on constraints, a priority queue and multi-round filtering algorithm are used to optimally sort the candidate semantic fragment set, ensuring that emergency-related and high-priority content is placed first, regular content is reasonably interspersed, and less-interested content is automatically delayed or merged for push. Intelligent transitional phrases (such as "The following is a weather warning, please be careful"), time prompts, and related sound effects are inserted to enhance the overall logical coherence and auditory / visual experience. Special adaptations are made for content combinations in special scenarios (such as emergency broadcasts and nighttime silent mode), such as prioritizing text content, adjusting speech rate, and automatic noise reduction.

[0078] Based on the type and capabilities of the push terminal (such as smart speakers, mobile apps, LED screens, etc.), assign appropriate multimodal output formats to each piece of content. Implement joint encapsulation of text, voice, images, and video (such as text with voice playback, video with ambient sound, image carousel, etc.), supporting simultaneous display or differentiated playback on multiple terminals. Embed detailed metadata within each structured content package, such as content ID, topic, priority, push time period, target terminal, adaptation context, and redundancy checksum, facilitating downstream system scheduling, distribution, monitoring, and feedback loop.

[0079] All generated structured broadcast content is archived in the content library. The system automatically adds a "recommended" tag to highly relevant content, enabling automatic reuse and continuous iterative optimization. Archived content synchronously supports tag updates and context-adaptive historical review, facilitating subsequent system evaluation of the effectiveness of push strategies and achieving self-evolution and precise service.

[0080] Furthermore, the formula for the sparse tensor semantic clustering algorithm is as follows:

[0081]

[0082] in, Let S represent the set of optimal semantic clustering results; S represents the set of all clustering partitioning schemes. Represents multi-order feature tensor elements; This represents the content-context label weight matrix; a represents the semantic unit number; e represents the time window number; d represents the modality number. This represents the set of data indices within the c-th cluster; This represents the number of multi-source data entries in the c-th cluster; This represents the cluster indicator vector for the c-th cluster; The c-th topic cluster contains the number of feature entries; C represents the final number of topic categories. This represents the sparsity adjustment parameter.

[0083] The broadcast scheduling decision module is used to construct a broadcast task flow graph with time window constraints based on the structured broadcast content, combined with terminal location, regional congestion status and historical broadcast response data, and to jointly optimize the scheduling strategies of broadcast time window, content priority and terminal scheduling order by adopting a graph attention scheduling algorithm based on asynchronous advantage update, and generate broadcast scheduling decision instructions.

[0084] The operation of the broadcast scheduling decision module includes the following steps:

[0085] Based on the structured broadcast content, combined with the terminal spatial distribution, congestion status within the region, and historical broadcast response data, a broadcast task flow graph with time window constraints is constructed, where nodes represent specific broadcast content, edges represent the temporal or logical dependencies between content, and time windows are used to limit the executable time range of each task node.

[0086] It's important to note that the system first automatically parses the attributes of each structured broadcast content, including the content topic, target audience, priority, expected duration, recommended playback scenarios, compatible terminal types, and emergency or regular tags. Each broadcast content is treated as an independent task node, embedding all scheduling-related metadata and assigning a globally unique identifier. For example, the "Emergency Weather Alert" node is marked as high priority and required to be pushed within 10 minutes. Content with duplicate topics or overlapping times is deduplicated and merged to reduce redundant nodes in the task flow graph and optimize subsequent scheduling efficiency.

[0087] The system collects the spatial location (GPS, WiFi, Bluetooth positioning, etc.) of all terminals in real time and their corresponding physical areas (such as buildings, subway stations, shopping malls, etc.). It collects data on the number of online terminals, the proportion of currently active terminals, wireless link channel utilization, and historical broadcast peak data for each area to assess local resource pressure. Areas with dense terminal concentrations, severe channel congestion, or high mobility are automatically marked; these areas will be prioritized for scheduling tasks.

[0088] For content that needs to be played sequentially (such as previews, event summaries, and emergency response procedures), the system automatically identifies logical dependencies between content through semantic analysis and historical playback sequences, establishing directed edge relationships. Constraints such as "only one emergency broadcast is allowed at a time" are added to prevent content mixing and audience interference. Based on the urgency of the content, historical response time distribution, and user sleep patterns, the system sets the earliest and latest push times for each task node, creating a task execution window. For example, traffic information should be pushed during morning rush hour, while silent / text notifications should be pushed at night.

[0089] The system retrieves terminal response data from historical broadcast tasks, including the actual completion time of the task, the playback success rate in the audience area, and user feedback (such as likes / complaints). Based on areas or time periods with slow historical responses, the time window for relevant task nodes is moved forward or extended; for tasks with excellent historical feedback, their time window is appropriately shortened to improve the overall system timeliness and satisfaction. The system supports real-time visualization of the task flow graph, making it easy for dispatchers to view critical task links, bottleneck nodes, and high-risk areas.

[0090] Collect information on the link accessibility, reception capability, interaction behavior characteristics and content preference of each terminal, construct a terminal state tensor, and fuse it with the broadcast task graph to generate a state-enhanced scheduling graph.

[0091] Specifically, the system periodically monitors the terminal's signal strength, current bandwidth, available network type (e.g., 5G, WiFi, wired), latency, and packet loss rate to identify terminals with poor or limited network connectivity. It collects data on terminal CPU, memory, and storage utilization, battery level, and temperature rise, automatically downgrading the priority of resource-constrained or potentially faulty devices to avoid pushing high-load content. Terminals periodically upload user interaction behaviors (e.g., actively playing, pausing, skipping, repeating, rejecting / closing content), and the system statistically analyzes terminal activity, response latency, and preferred content types to automatically classify the terminal's responsiveness to broadcasts. It continuously tracks the terminal's historical content response curves (e.g., preferences for news, music, emergency information, and advertisements) to assign preference tags and weighted recommendation parameters to the terminal.

[0092] After standardizing the state characteristics (network, device, interaction, preferences, etc.) of each terminal, they are mapped to state vectors. All terminal state vectors are stacked according to the terminal ID primary key, forming a high-dimensional tensor with the structure [terminal ID, feature category, time slice]. This allows for tracking historical state changes and periodic behaviors, facilitating anomaly detection and trend analysis. Terminals exhibiting extreme anomalies in the tensor (such as prolonged disconnections or persistent abnormal loads) are automatically flagged, and subsequently excluded from scheduling or have their task push priority reduced.

[0093] For each content task node, a set of the most matching terminals is dynamically selected based on its push target, compatible terminal type, and historical terminal response records, and a content-terminal mapping relationship is established. Each task node is associated with the real-time status characteristics of its target terminal. Graph node attributes are no longer simple content tags, but embed multi-dimensional contextual information such as the current terminal's actual reachability, load capacity, and preference matching degree. The entire scheduling graph supports visual display; the system can refresh the attributes of each node and edge in real time, such as node color changes indicating status changes and edge thickness reflecting terminal availability or congestion levels. For terminals that may fail or are prone to errors, the system can automatically designate backup terminals, achieving fault tolerance and redundancy in scheduling tasks, improving broadcast arrival rate and system robustness.

[0094] Based on the state-enhanced scheduling graph, a graph attention scheduling algorithm based on an asynchronous advantage update mechanism is adopted to dynamically calculate the scheduling value function of each broadcast task node. Based on the structural correlation between nodes, content priority and terminal feedback potential, attention weights are allocated, and the broadcast time window allocation, content delivery priority ranking and target terminal scheduling order are jointly optimized through asynchronous strategy.

[0095] Specifically, in the state-enhanced scheduling graph, the system assigns initial attention weights to each content node, terminal node, and the edges between them, reflecting the structural relationships between nodes, content priority, historical feedback potential, etc. For example, if a content node is strongly connected to a highly active, high-feedback terminal, its scheduling priority is increased.

[0096] The system adopts a multi-threaded / distributed architecture, independently sampling the scheduling status of each content node (including historical scheduling status, current terminal status, regional congestion changes, etc.) and asynchronously updating the node's scheduling value function. For each task node, its state transition sequence at different time points and under different target terminals is asynchronously sampled, accumulating scheduling feedback information brought about by scheduling history and environmental changes.

[0097] By applying a weighted differential to the scheduling value function (e.g., "new strategy value - average historical value"), a superiority evaluation function is calculated. The calculation results are then combined with the contextual attention weights of nodes to dynamically adjust the content push order, time window allocation, and target terminal scheduling order, achieving adaptive and sensitive optimization for sudden events, trending content, and special regions. For example, if a terminal has been particularly responsive to similar content recently, its scheduling weight is increased, and its allocated content time window and priority are simultaneously optimized.

[0098] In scenarios involving regional hotspots and high content concurrency, the system optimizes the scheduling parameters for each content node and its corresponding terminal in parallel. All optimization results are written to the global scheduling parameter library asynchronously, without waiting for synchronization from all nodes, thus improving the system's real-time performance and scalability under large-scale conditions. After each global parameter update, node priorities, content terminal allocation, and resource allocation suggestions are re-integrated to prepare for the generation of scheduling instructions.

[0099] Based on the optimization results, broadcast scheduling decision instructions are generated, including content identifiers, target terminal ID lists, predetermined time windows, and frequency resource allocation information. These instructions are then encapsulated and issued through the heterogeneous link control module.

[0100] Specifically, the system generates a broadcast scheduling instruction set based on the optimization results, including: a unique identifier for each piece of content, a list of corresponding target terminal IDs, a specific time window for allocation, and resource configuration information such as frequency / bandwidth. The instruction set can be grouped according to multi-dimensional conditions such as region, terminal type, and content urgency.

[0101] The scheduling commands are encapsulated using a heterogeneous link control module to ensure compatibility with various terminals (such as 4G / 5G, WiFi, wired Ethernet, LoRa, and other links) and different manufacturers' equipment protocols. During encapsulation, the command format is automatically adjusted according to the target terminal's capabilities and real-time network conditions. For example, if some terminals only support text content, the system automatically transcodes and simplifies unnecessary content.

[0102] After encapsulation, the instructions are sent to the target terminal through the scheduling gateway and link distribution mechanism, pushed in batches, and support breakpoint resumption and retransmission fault tolerance. The system monitors the instruction issuance and terminal execution status in real time, collects execution logs and feedback information, and automatically alarms or resends in case of abnormalities, forming a complete closed loop.

[0103] The asynchronous advantage update mechanism specifically includes the following steps:

[0104] Based on the state-enhanced scheduling graph, for each broadcast task node, the node state transition sequence is asynchronously sampled based on the node's historical scheduling state, current environmental characteristics, and terminal feedback data, and the scheduling value function of the task node is dynamically updated.

[0105] Specifically, each broadcast task node (such as a piece of content to be broadcast) is handled by an independent thread or scheduling agent, and can initiate state sampling independently without waiting for all nodes to synchronize. Nodes periodically access the scheduling graph, terminal state tensor, and the latest environmental parameters to autonomously collect their own relevant scheduling status, such as the time of the most recent push, the reception success rate of associated terminals, the congestion situation in the area, and the urgency of the content. If the content belonging to a node experienced failure or response delay in the last push, the node can autonomously mark its current status as "to be optimized," triggering a self-correction strategy.

[0106] After collecting data, nodes compare their current scheduling status with historical sampling sequences (such as performance in each scheduling cycle within a week) to identify recent trends in environmental and terminal response, such as improved terminal response, optimized regional networks, or changes in content requirements. Through comparative analysis, nodes can promptly detect fluctuations in their own scheduling value (such as push delivery success rate, timeliness, and user feedback), providing data support for the dynamic adjustment of the subsequent value function.

[0107] By performing weighted difference on the scheduling value function at different time steps, the advantage estimation function is calculated, and combined with the context attention weight of each node in the graph structure, local sensitivity adjustment of the scheduling priority of broadcast content and the selection of target terminals is achieved.

[0108] Specifically, each broadcast task node has a local state cache that automatically records the scheduling performance of the most recent scheduling cycles (e.g., 10, 30), including multi-dimensional data such as push time period, number of assigned terminals, actual coverage, user feedback score, and terminal response speed. After a new scheduling cycle, the scheduling node automatically compares its performance with historical averages and historical maximum / minimum performance. For example, if the user feedback for the current content push is significantly better than the historical average, it is recorded as a "positive advantage" event. For new content, cold start nodes, or nodes whose performance changes drastically due to sudden environmental changes, the system allows automatic adjustment of the comparison window length to improve the speed of adaptation to sudden events.

[0109] Nodes dynamically calculate the "weighted difference" between the current scheduling value and the historical average. This not only focuses on absolute improvement but also adjusts the difference's influence based on the content's actual weight, time sensitivity, and regional specificities. For example, if a piece of content receives better feedback during the morning peak hours than similar content in the past, the difference's weight increases more significantly, and the scheduling system will prioritize pushing that content during subsequent peak hours. For performance spikes followed by long-term declines, the system uses moving averages, sliding windows, and threshold mechanisms to prevent occasional data noise from affecting the overall scheduling strategy.

[0110] During node scheduling, the system proactively reads the scheduling status and attention weights of adjacent nodes in the task flow graph (e.g., logical dependencies, topic associations, temporal proximity, etc.) in the most recent period. If some preceding nodes have recently increased their priority, causing a delay in the push of this node, the system will automatically identify this and fine-tune the scheduling priority of this node in the context to prevent critical content from being delayed indefinitely. For nodes with hot topics, emergency content, or high-frequency user feedback, the attention weights of their related nodes will also be dynamically increased, forming a multi-dimensional scheduling association based on semantics, logic, and time.

[0111] By combining weighted differential and contextual attention, the system dynamically updates node priorities, target terminal combinations, and content distribution windows in each round of scheduling decisions. If a terminal is detected to have repeatedly provided positive feedback on similar content recently, the system will automatically increase its distribution weight for that content; conversely, it will decrease its weight, ensuring that scheduling resources are tilted towards high-potential targets. All fine-tuning and strategy changes are recorded with detailed reasons and supporting data, facilitating subsequent system evaluation and strategy optimization.

[0112] In scenarios involving multi-task concurrency and resource contention, a distributed parallel computing approach is adopted to independently optimize the scheduling parameters of each node and update the global scheduling strategy parameters in an asynchronous manner.

[0113] Specifically, each broadcast task node runs independently in a distributed architecture and can be deployed on different servers, edge nodes, or cloud platforms, achieving true physical parallelism and fault-tolerant isolation. Each node has its own local optimization process, responsible for collecting the necessary environmental data, terminal status, and scheduling feedback in real time, and autonomously adjusting parameters (such as push time, terminal allocation, content order, and resource pre-allocation). Nodes only need to asynchronously upload optimization results at a set frequency (such as every 5 seconds or every minute) or under certain conditions (such as changes in key indicators), without waiting for global synchronization.

[0114] The latest optimization results for all nodes, including scheduling priorities, resource allocation suggestions, and attention weights, are transmitted back to the global scheduling center in real time via a secure, low-latency message queue or API interface. The scheduling center automatically maintains parameter version numbers and timestamps for each node. When multiple nodes contend for the same terminal resource, the allocation request with the highest priority or the most recent one is prioritized, and the conflict resolution result is synchronously fed back to the relevant nodes. For large-scale scenarios (such as hundreds of broadcast task nodes and thousands of terminals), the system supports batch asynchronous transmission and rapid updates of local parameters, greatly improving the overall scheduling response speed and concurrency capabilities.

[0115] If a node fails to transmit the latest parameters for a short period due to network failure, abnormal crash, or other reasons, the scheduling center will temporarily use its most recently valid transmitted parameters, or use locally optimal historical parameters as a temporary scheduling basis to ensure global scheduling continuity and uninterrupted service. After recovery, the node automatically transmits missing parameters and scheduling logs. The scheduling center re-evaluates the parameter validity based on the actual results to ensure the long-term optimality of the global strategy. All anomalies and compensation actions are automatically recorded in the logging system for easy post-event traceability and system health assessment.

[0116] The scheduling center periodically aggregates, resolves conflicts, and maintains consistency of all returned optimization parameters, employing rules such as maximum priority, timing consistency, and resource mutual exclusion. The aggregated global scheduling policy is used to generate broadcast scheduling instructions for the next cycle, covering all key parameters such as content push order, terminal grouping, time slot allocation, and resource reservation. After the global policy changes, the system immediately pushes the changes to all nodes and terminals, ensuring that the scheduling policy is dynamically controllable and highly adaptive.

[0117] The global scheduling strategy parameter update based on the asynchronous advantage update mechanism integrates the latest scheduling priority and resource allocation suggestions of the nodes into the final broadcast scheduling decision output, including content distribution order, target terminal grouping and time slot resource allocation information, and sends scheduling instructions to the heterogeneous link control module.

[0118] Specifically, the global scheduling center periodically or as needed summarizes the latest scheduling parameters of all nodes, and comprehensively sorts and uniformly plans the scheduling priorities, target terminal groups, and time slot resource allocation of all nodes. The system combines the resource competition between nodes and the current environmental constraints (such as frequency, bandwidth, total task duration, etc.) to reasonably coordinate the resource requests of each node, preventing resource conflicts and scheduling bottlenecks.

[0119] The scheduling center generates the final scheduling instruction set based on the latest global parameters, including detailed information such as content distribution order, push groups for each terminal, specific playback time windows, and link allocation strategies. All instructions are encapsulated by the heterogeneous link control module and then sent to each terminal, while the system monitors the sending and execution results. Execution feedback (such as successful push, terminal non-response, content loss, etc.) is collected and analyzed in real time to provide basic data for the next round of asynchronous advantage updates.

[0120] The system continuously tracks scheduling feedback, terminal behavior, and network environment changes, and through continuous asynchronous updates, it achieves self-learning and adaptive evolution of scheduling strategies. In the event of sudden events, large traffic fluctuations, or the addition / removal of terminals, the scheduling system can quickly adjust itself to ensure that broadcast tasks always efficiently and reliably cover the target terminal group.

[0121] Furthermore, the formula for the graph attention scheduling algorithm is as follows:

[0122]

[0123] in, This represents the final scheduling priority score of broadcast task node i; This represents the content importance score corresponding to the i-th task node in the structured broadcast content; This represents the user activity index within the target area of ​​task i; This represents the historical feedback score for content i; This indicates the current link bandwidth utilization rate of the geographical area where task i is located; Indicates the maximum link bandwidth capacity; This represents the congestion risk coefficient of task i in the state-enhanced scheduling graph; Indicates the remaining schedulable time window for task i; , and This represents the weighting factor for the score; The exponential amplification factor representing the response feedback score; and These represent the slope of the Sigmoid function and the center position parameter in the time window urgency score, respectively.

[0124] The calculation formula is as follows:

[0125]

[0126] in, This represents the set of context labels corresponding to broadcast content i; This indicates the semantic matching strength between content i and tag k; This represents the historical response frequency index corresponding to label k; The weighting coefficient (0~1) represents the semantic similarity and historical click frequency, used to control the weight ratio of semantic structure and behavioral feedback.

[0127] The calculation formula is as follows:

[0128]

[0129] in, This indicates the dependency path depth of task i in the broadcast task flow graph; This represents the local task density of task i, and the number of its neighboring nodes (upstream and downstream tasks), used to evaluate the logical coupling between tasks. This represents the number of broadcast tasks that occur concurrently with task i within the current time period; and This represents the adjustment weight parameter that depends on depth and local density.

[0130] The abnormal interference suppression module is used to identify potential interference sources during broadcast signal transmission using a residual convolutional neural network algorithm, and to reconstruct the disturbed broadcast segment through a signal repair mechanism driven by a recurrent neural network; the potential interference sources include channel conflict, signal attenuation, and malicious interference;

[0131] The operation of the abnormal interference suppression module includes the following steps:

[0132] The system performs real-time acquisition of multi-channel signal streams in the broadcast signal transmission link and uses a multi-scale decomposition method to extract multi-dimensional characteristic parameters such as signal energy spectrum distribution, instantaneous frequency characteristics, channel utilization, and interference index.

[0133] Specifically, multi-channel high-sensitivity acquisition units are configured at key nodes in the broadcast signal transmission link (such as base stations, relays, and terminal front-ends) to support the simultaneous acquisition of raw signal streams from the main channel, backup channel, and environmental channel. Each signal acquisition includes basic parameters such as timestamp, frequency band identifier, signal amplitude, and phase. The system can adaptively adjust the sampling rate and automatically increase the sampling density for abnormal or high-risk periods.

[0134] For the raw signals acquired from each channel, multi-scale analysis methods such as wavelet decomposition and empirical mode decomposition (EMD) are employed to decompose the signal energy at different time scales, capturing sudden anomalies and long-term trends. The energy spectrum distribution, signal dominant frequency, and instantaneous frequency drift characteristics of each decomposition layer are calculated, with a focus on extracting anomaly indicators related to interference, such as local amplitude spikes and frequency spurious emissions. Combined with link status data, channel utilization (e.g., spectrum occupancy, signal passband utilization) and interference indices (e.g., SINR, BER statistics) are extracted simultaneously to form a complete set of multi-dimensional feature parameters, which serve as input for subsequent intelligent identification.

[0135] Based on the residual convolutional neural network algorithm, the above feature parameters are modeled in time and fused in space. Combined with channel history data and statistical thresholds, the system can automatically detect abnormal signal segments and classify the interference type.

[0136] Specifically, multidimensional feature parameters are encoded according to time series and spatial channels respectively, forming a three-dimensional tensor of [time, channel, feature category]. A multi-layer convolutional neural network (ResNet-CNN) with residual structure is used to extract temporal patterns (such as anomalous waveforms and signal abrupt changes) and spatial co-features (such as coherence between multiple channels and noise coupling between different links in the same frequency band) in parallel. The residual structure effectively avoids the degradation of deep models and improves detection robustness under complex interference environments.

[0137] The neural network model is calibrated using historical channel data during the training phase (such as labeled typical samples of channel collisions, signal attenuation, and malicious interference). During runtime, the model discriminates against the real-time input feature tensors, outputting anomaly probabilities and classification results for each time segment in the signal stream. Once the anomaly probability exceeds a dynamic threshold, the system automatically locks the corresponding signal segment and labels the interference type (such as channel collisions, signal attenuation, malicious injection, etc.) based on the model's discrimination results.

[0138] For each detected abnormal event, the system automatically records the start and end time, associated channels, main abnormal characteristics, and preliminary classification results, and archives them in the abnormal detection log for subsequent source tracing and statistical analysis.

[0139] For the detected abnormal segments, an adaptive signal repair mechanism driven by a recurrent neural network is adopted to reconstruct the content of the disturbed signal based on the temporal pattern and amplitude characteristics of historical reference segments.

[0140] Specifically, for each signal segment marked as anomalous, the system retrieves recent historical "healthy" signal segments from that link, prioritizing samples with amplitude and frequency patterns similar to the current task. These historical reference segments are used to assist signal reconstruction, improving the similarity and coherence of the repair.

[0141] Recurrent neural networks such as LSTM or GRU are used to input the contextual data of historical reference segments and current anomalous segments into the model. Based on temporal correlation and amplitude dynamics, the model generates or corrects the signal content of anomalous segments frame by frame, including amplitude recovery, missing data completion, and smoothing transitions. The repair process supports collaborative modeling of multi-channel signals, ensuring the consistency and synchronization of signals across all links.

[0142] The repair strategy is dynamically adjusted based on the type and scope of the anomaly: for minor noise interference, interpolation smoothing can be used; for severe signal loss, deep generative repair or switching to backup channel data is employed. The repaired signal is then compared with the original reference signal to assess differences, triggering secondary repair or marking it as a severe anomaly requiring manual intervention if necessary.

[0143] By comparing the signals before and after reconstruction, and combining the current link dynamics with the target terminal status, an interference suppression control vector is generated, including strategy parameters at the levels of forwarding delay adjustment, link remapping, frequency band avoidance, and content reconstruction. The abnormal suppression processing results are then synchronously fed back to the terminal collaborative feedback module.

[0144] Specifically, the system automatically compares the main indicators of the signal before and after repair (such as energy spectrum, signal integrity, bit error rate, etc.), and combines the current link dynamics (such as remaining channel bandwidth, latency changes) with the target terminal status (such as playback failure, feedback anomalies). It then evaluates the effectiveness of the signal repair and determines whether subsequent link strategies need dynamic adjustment.

[0145] Based on the comprehensive evaluation results, an interference suppression control vector is automatically generated, containing multi-level policy parameters, such as:

[0146] Forwarding delay adjustment: Extend or shorten the link waiting time for broadcast tasks to avoid high-interference time slots.

[0147] Link remapping: temporarily switching to backup links, redundant channels, or multiple links for concurrent forwarding to improve content delivery rate.

[0148] Frequency band avoidance: Dynamically hop frequencies to available frequency bands with lower interference to reduce the probability of channel collisions.

[0149] Content Reconstruction Level: Set the depth of content reconstruction (such as full segment repair, partial smoothing, and downgraded push), balancing content integrity with system resource consumption.

[0150] All control parameters are encapsulated into standard control vectors and sent to relevant broadcast nodes and link scheduling modules.

[0151] The anomaly suppression results are synchronized to the terminal collaborative feedback module in real time, supporting the collection of anomaly recovery status and user feedback on the terminal side. The system periodically collects terminal feedback to evaluate the effectiveness of the suppression strategy, providing a data loop for subsequent model training and strategy fine-tuning.

[0152] The terminal collaborative feedback module is used to receive the broadcast signal controlled by the broadcast scheduling command, perform playback operations on the terminal, and simultaneously collect terminal playback status data, user interaction feedback data, and environmental response data. It also dynamically adjusts the terminal receiving parameters and content push strategy through a federated reinforcement learning algorithm. The terminal collaborative feedback module constructs a policy adaptive mechanism based on the federated reinforcement learning architecture, periodically optimizes the receiving strategy using the playback status data, interaction feedback data, and environmental awareness parameters collected locally on the terminal, and achieves cross-terminal policy updates through global policy aggregation.

[0153] Specifically, after receiving the system broadcast scheduling command, the terminal device automatically parses the pushed content and calls the corresponding broadcast control components according to the content type (text, audio, video, mixed text and images, etc.). It supports various broadcast strategies such as timed playback, insertion, and looping to ensure that the content is played on time according to priority. It collects real-time data on the terminal's current playback status (playback progress, content type, volume settings, playback anomalies), device health status (power supply, battery, storage space, CPU / memory utilization), and network connection quality (bandwidth, signal strength, latency, packet loss, etc.). It integrates environmental acoustic sensors, temperature and humidity sensors, etc., to collect real-time acoustic environment data (such as background noise, user conversations, emergency alarm sounds, etc.), temperature, humidity, and light intensity in the space where the terminal is located, providing an environmental baseline for subsequent content adaptation and interference detection.

[0154] The terminal features multiple interaction channels, including a touchscreen, physical buttons, and voice input, allowing users to provide feedback on broadcast content (such as content confirmation, skipping, repeating playback, complaining, liking, rating, and tagging). Through algorithms such as speech recognition and semantic analysis, it captures key intentions and emotional nuances in user feedback (such as satisfaction, dissatisfaction, urgency, and confusion). The video terminal can also be equipped with a camera for facial expression recognition to assess user attention and engagement (subject to compliance). For special scenarios such as emergency broadcasts and disaster warnings, it collects users' emergency response actions (such as confirming escape routes, reporting location, and making SOS calls) and feedback, enhancing the broadcast system's emergency response capabilities.

[0155] Each terminal, based on a federated reinforcement learning architecture, utilizes locally collected playback data, interaction feedback, and environmental parameters to construct a user preference model and content push strategy. Through real-time feedback signals (such as user-initiated playback, positive feedback, playback interruption, etc.), it adaptively adjusts parameters such as content category, playback time, volume control, and push frequency to continuously improve the user experience. It dynamically adjusts parameters and push strategies based on different environments (such as noisy environments, quiet spaces, outdoor environments, etc.), different user groups (such as the elderly, students, merchants, etc.), and different terminal forms (mobile devices, wall-mounted terminals, public large screens, etc.). It periodically detects playback anomalies (such as stuttering, frame drops, silence, playback failure, etc.) and network anomalies (disconnection, low speed, switching, etc.), automatically switching locally cached content, adjusting playback schemes, attempting to resume playback, and promptly reporting anomaly information.

[0156] The terminal only uploads local model parameters or policy gradients, not raw user data. The central server aggregates model updates from multiple terminals, and after completing global model aggregation, the optimization results are synchronously distributed to each terminal, achieving system-level collaborative adaptation. When significant optimizations occur in the global policy or specific content policies require urgent adjustments, the system supports real-time distribution of policy parameters to the target terminals, ensuring the timeliness and effectiveness of push notifications for emergencies and dynamic trending content. Terminals are grouped based on factors such as their region, user profile, and historical performance, enabling personalized and differentiated management of content push notifications, improving the overall efficiency and satisfaction of the system's push notifications.

[0157] For abnormal states and system security risks detected locally (such as forged commands, terminal tampering, unauthorized access, etc.), security logs are automatically generated and reported to the backend, supporting rapid system response and security self-healing. Backend administrators can view the status of each terminal in real time through the system platform and remotely issue control commands, such as terminal restart, content update, policy switching, blacklist processing, etc., to achieve controllable and robust system operation.

[0158] In summary, this invention, through a multimodal data acquisition module, integrates multi-source information such as environmental audio, real-time voice, video images, meteorological parameters, electromagnetic interference signals, user behavior, and terminal status to achieve comprehensive dynamic perception of the broadcast coverage area.

[0159] This invention introduces a sparse tensor semantic clustering algorithm, constructing a semantic-temporal-modal third-order tensor for multimodal features to mine representative candidate broadcast segments. A label migration modeling method is then used to achieve accurate content-scene matching, thereby generating structured, context-adaptive broadcast content. The scheduling module employs a graph attention scheduling algorithm based on asynchronous advantage updates, integrating terminal distribution, network status, and historical feedback to construct a state-enhanced task graph. This enables joint optimization of time windows, push order, and terminal resources, improving the reachability of broadcast tasks and the overall system load balancing capability.

[0160] In this invention, the abnormal interference suppression module combines residual convolutional neural networks and recurrent neural networks to identify multiple interference sources in the signal and perform adaptive repair, ensuring stable transmission of broadcast content in complex electromagnetic environments. The terminal collaborative feedback module constructs a federated reinforcement learning architecture, combining local playback status, user feedback, and environmental parameters to optimize the reception strategy while protecting data privacy. Through global model aggregation, the strategy achieves self-learning and continuous optimization on the terminal side.

[0161] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A multifunctional integrated digital network broadcasting system based on intelligent algorithms, characterized in that, It includes a multimodal data acquisition module, a broadcast content generation module, a broadcast scheduling decision module, an abnormal interference suppression module, and a terminal collaborative feedback module that are connected in sequence. The multimodal data acquisition module is used to collect multi-source information within the broadcast coverage area in real time; The broadcast content generation module is used to perform cluster analysis on the semantic features of the content using the sparse tensor semantic clustering algorithm based on the multi-source information, and to construct a content-context tag mapping relationship by combining historical playback data to generate structured broadcast content. The broadcast scheduling decision module is used to construct a broadcast task flow graph with time window constraints based on the structured broadcast content, combined with terminal location, regional congestion status and historical broadcast response data, and to jointly optimize the scheduling strategies of broadcast time window, content priority and terminal scheduling order by adopting a graph attention scheduling algorithm based on asynchronous advantage update, and generate broadcast scheduling decision instructions. The abnormal interference suppression module is used to identify potential interference sources during broadcast signal transmission using a residual convolutional neural network algorithm, and to reconstruct the disturbed broadcast segment through a signal repair mechanism driven by a recurrent neural network; the potential interference sources include channel conflict, signal attenuation, and malicious interference; The terminal collaborative feedback module is used to receive the broadcast signal controlled by the broadcast scheduling instruction, perform playback operation on the terminal, and collect terminal playback status data, user interaction feedback data and environmental response data at the same time. It also dynamically adjusts the terminal receiving parameters and content push strategy through a federated reinforcement learning algorithm. The operation of the abnormal interference suppression module includes the following steps: The system performs real-time acquisition of multi-channel signal streams in the broadcast signal transmission link and uses a multi-scale decomposition method to extract multi-dimensional characteristic parameters such as signal energy spectrum distribution, instantaneous frequency characteristics, channel utilization, and interference index. Based on the residual convolutional neural network algorithm, the above feature parameters are modeled in time and fused in space. Combined with channel history data and statistical thresholds, the system can automatically detect abnormal signal segments and classify the interference type. For the detected abnormal segments, an adaptive signal repair mechanism driven by a recurrent neural network is adopted to reconstruct the content of the disturbed signal based on the temporal pattern and amplitude characteristics of historical reference segments. By comparing the signals before and after reconstruction, and combining the current link dynamics with the target terminal status, an interference suppression control vector is generated, including strategy parameters at the levels of forwarding delay adjustment, link remapping, frequency band avoidance, and content reconstruction. The abnormal suppression processing results are then synchronously fed back to the terminal collaborative feedback module.

2. The multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 1, characterized in that, The multi-source information includes environmental sound signals, real-time voice signals collected by microphones, video surveillance image data, regional meteorological parameters, electromagnetic interference signal detection data, terminal device geographical location and motion status data, user interaction behavior data, and network link status parameters.

3. The multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 1, characterized in that, The operation of the broadcast content generation module includes the following steps: The multi-source information is standardized and preprocessed, and a multi-order feature tensor containing semantic, temporal, and modal dimensions is constructed. Based on the multi-order feature tensor, the sparse tensor semantic clustering algorithm is used to compress the semantic units and cluster them into topics. The semantic core units in multi-source content are identified by low-rank representation and semantic sparsity constraints to form a set of candidate broadcast semantic segments. Based on the candidate broadcast semantic fragment set, combined with historical broadcast playback data and terminal response records, a content-context tag mapping relationship is constructed using a tag migration modeling method, and a content-context matching matrix is ​​generated using a multi-scenario adaptation model based on tag weight fitting. Based on the candidate broadcast semantic fragment set and content-context matching matrix, a broadcast structure generation mechanism based on constraint satisfaction problem solving is adopted to reconstruct the order and optimize the format of the candidate content, generating structured broadcast content including time series logic, context adaptability and multimodal fusion features.

4. The multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 3, characterized in that, The formula for the sparse tensor semantic clustering algorithm is as follows: in, Let S represent the set of optimal semantic clustering results; S represents the set of all clustering partitioning schemes. Represents multi-order feature tensor elements; This represents the content-context label weight matrix; i represents the semantic unit number; j represents the time window number; k represents the modality number; This represents the set of data indices within the c-th cluster; This represents the number of multi-source data entries in the c-th cluster; This represents the cluster indicator vector for the c-th cluster; The c-th topic cluster contains the number of feature entries; C represents the final number of topic categories. This represents the sparsity adjustment parameter.

5. A multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 1, characterized in that, The operation of the broadcast scheduling decision module includes the following steps: Based on the structured broadcast content, combined with the terminal spatial distribution, congestion status within the region, and historical broadcast response data, a broadcast task flow graph with time window constraints is constructed, where nodes represent specific broadcast content, edges represent the temporal or logical dependencies between content, and time windows are used to limit the executable time range of each task node. Collect information on the link accessibility, reception capability, interaction behavior characteristics and content preference of each terminal, construct a terminal state tensor, and fuse it with the broadcast task graph to generate a state-enhanced scheduling graph. Based on the state-enhanced scheduling graph, a graph attention scheduling algorithm based on an asynchronous advantage update mechanism is adopted to dynamically calculate the scheduling value function of each broadcast task node. Based on the structural correlation between nodes, content priority and terminal feedback potential, attention weights are allocated, and the broadcast time window allocation, content delivery priority ranking and target terminal scheduling order are jointly optimized through asynchronous strategy. Based on the optimization results, broadcast scheduling decision instructions are generated, including content identifiers, target terminal ID lists, predetermined time windows, and frequency resource allocation information. These instructions are then encapsulated and issued through the heterogeneous link control module.

6. A multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 5, characterized in that, The asynchronous advantage update mechanism specifically includes the following steps: Based on the state-enhanced scheduling graph, for each broadcast task node, the node state transition sequence is asynchronously sampled based on the node's historical scheduling state, current environmental characteristics, and terminal feedback data, and the scheduling value function of the task node is dynamically updated. By performing weighted difference on the scheduling value function at different time steps, the advantage estimation function is calculated, and combined with the context attention weight of each node in the graph structure, local sensitivity adjustment of the scheduling priority of broadcast content and the selection of target terminals is achieved. In scenarios involving multi-task concurrency and resource contention, a distributed parallel computing approach is adopted to independently optimize the scheduling parameters of each node and update the global scheduling strategy parameters in an asynchronous manner. The global scheduling strategy parameter update based on the asynchronous advantage update mechanism integrates the latest scheduling priority and resource allocation suggestions of the nodes into the final broadcast scheduling decision output, including content distribution order, target terminal grouping and time slot resource allocation information, and sends scheduling instructions to the heterogeneous link control module.

7. A multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 5, characterized in that, The formula for the graph attention scheduling algorithm is as follows: in, This represents the final scheduling priority score of broadcast task node i; This represents the content importance score corresponding to the i-th task node in the structured broadcast content; This represents the user activity index within the target area of ​​task i; This represents the historical feedback score for content i; This indicates the current link bandwidth utilization rate of the geographical area where task i is located; Indicates the maximum link bandwidth capacity; This represents the congestion risk coefficient of task i in the state-enhanced scheduling graph; Indicates the remaining schedulable time window for task i; , and This represents the weighting factor for the score; The exponential amplification factor representing the response feedback score; and These represent the slope of the Sigmoid function and the center position parameter in the time window urgency score, respectively.

8. A multifunctional integrated digital network broadcasting system based on intelligent algorithms according to claim 1, characterized in that, The terminal collaborative feedback module constructs a policy adaptation mechanism based on a federated reinforcement learning architecture. It periodically optimizes the receiving policy using playback status data, interaction feedback data, and environmental awareness parameters collected locally on the terminal, and achieves cross-terminal policy updates through global policy aggregation.

Images