Multi-mode communication network dynamic scheduling system fusing AI computing power

Abstract

The invention provides a multimode communication network dynamic scheduling system fused with AI computing power, which belongs to the technical field of communication equipment and artificial intelligence hardware fusion, and is characterized in that based on a four-layer hardware architecture of'core control-computing power support-multimode communication-high-speed interconnection ', an interconnection layer is constructed by adopting high-speed interfaces such as PCIe / SGMII + and the like, and inter-module fault is realized; 10ms low-delay data transmission is realized; the AI computing power module integrated with the 8TOPS NPU is deeply coupled with the multi-mode communication module, and the AI computing power module is quantized and compressed through an LSTM / Transform model and then solidified to an NPU special storage to realize scene self-adaptive scheduling; a'acquisition-identification-decision-feedback 'hardware closed loop is established, the communication interruption rate is reduced in an emergency scene, the overall power consumption is reduced, and AI reasoning is delayed. Through breakthrough hardware architecture design, AI computing power and a multi-mode communication module are deeply integrated, an intelligent communication resource scheduling system is constructed, and dynamic optimal configuration of communication resources in multiple scenes can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of the integration of communication equipment and artificial intelligence hardware, and in particular to a multi-mode communication network dynamic scheduling system that integrates AI computing power. The system is suitable for hardware-level application scenarios such as IoT terminal clusters, ubiquitous connections in smart cities, and emergency rescue site communications that have high requirements for multi-scenario adaptability, communication reliability, and computing power coordination. Background Art

[0002] With the large-scale implementation of 5G, IoT, and AI technologies, multi-scenario communication needs have placed more stringent requirements on hardware equipment: high-bandwidth scenarios (such as 4K / 8K video backhaul and AI model training data transmission) require hardware-level support for Tbps rates; low-latency scenarios (such as autonomous driving control instructions and industrial PLC real-time communication) require hardware links to ensure responses within 5ms; low-power scenarios (such as farmland sensors and smart meters) require hardware modules to support long-term battery life. However, traditional multi-mode communication equipment has three major hardware bottlenecks:

[0003] Rigid hardware architecture: Communication modules (such as 5G, Wi-Fi, and Bluetooth) are stacked using non-standard interfaces, resulting in fragmented frequency band resources and difficulty in coordination.

[0004] Lack of computing power: The lack of dedicated AI computing hardware leads to reliance on external servers for decision-making, resulting in high scheduling delays (decision-making delays in traditional solutions > 100ms).

[0005] Insufficient scenario adaptation: Hardware configurations based on fixed rules (such as fixed frequency band allocation and fixed power mode) cannot dynamically match complex scenario requirements (such as the need to ensure communication continuity and low command latency during emergency rescue). For example, at emergency rescue sites, traditional equipment lacks hardware-level intelligent decision-making driven by AI computing power, which often leads to bandwidth resource congestion (critical instructions are stuck) or redundant module idleness (non-essential links continue to consume power), seriously affecting rescue efficiency. Therefore, there is an urgent need for a multi-mode communication device with integrated AI computing power at the hardware level to achieve dynamic scheduling through deep collaboration of hardware modules. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a multi-mode communication network dynamic scheduling system that integrates AI computing power, which successfully solves the problem of rigid traditional communication resource allocation mode. Through the dynamic optimization configuration mechanism, the communication scheduling system has stronger environmental adaptability and resource utilization efficiency, providing an innovative solution for the construction of intelligent communication networks in the 5G / 6G era.

[0007] The technical solution of the present invention is:

[0008] A multi-mode communication network dynamic scheduling system that integrates AI computing power. Through a breakthrough hardware architecture design, it deeply integrates AI computing power with multi-mode communication modules to build an intelligent communication resource scheduling system that can achieve dynamic optimization of communication resources in multiple scenarios.

[0009] The present invention adopts an innovative four-layer hardware architecture design:

[0010] Core control layer: As the brain of the system, responsible for overall scheduling decisions.

[0011] Computing power support layer: Integrates heterogeneous AI computing units to provide intelligent analysis capabilities.

[0012] Multi-mode communication layer: supports the collaborative operation of multiple communication protocols and frequency bands.

[0013] High-speed interconnection layer: realizes high-speed data interaction between modules through standardized interfaces.

[0014] The technology utilizes a closed-loop control process of "data collection - scenario identification - strategy decision-making - execution feedback," ensuring the real-time and reliability of the scheduling process through hardware-level collaborative working mechanisms. Its core innovative advantages lie in: deeply coupling AI computing power with communication hardware, breaking through the traditional static resource allocation model, significantly improving scheduling intelligence, optimizing resource utilization efficiency, and enhancing multi-scenario adaptability.

[0015] Further,

[0016] The core control layer is interconnected through standardized high-speed interfaces to ensure low latency in data transfer and high compatibility in module expansion.

[0017] The main controller of the core control layer, as the core hardware hub of the system, integrates a multi-protocol communication control unit; including:

[0018] Directly connected to the network controller via Gigabit Ethernet interface;

[0019] External 5G / Wi-Fi baseband module via two sets of PCIe 3.0 interfaces;

[0020] Connect to the Gigabit Ethernet switch chip through the SGMII+ interface and output 4 Gigabit network ports;

[0021] The network controller in the core control layer serves as the hardware hub for data interaction and implements multi-module data routing based on a dedicated routing chip. It includes:

[0022] Connect to the AI ​​computing module through the PCIe 2.0 interface to collect status data of the multi-mode communication module and terminal requirements in real time;

[0023] The scheduling instructions generated by the AI ​​computing module are mapped to the main controller at the hardware level to ensure that the instruction execution delay is less than 10ms.

[0024] Further,

[0025] The computing power support layer is dedicated computing power hardware that integrates AI acceleration chips and algorithm processing units, including:

[0026] Directly connected to the network controller via the PCIe 2.0 interface, it provides real-time intelligent decision-making computing power.

[0027] Provides high-speed interfaces of PCIe, USB 3.1, and DisplayPort;

[0028] Supports Wi-Fi 6 and Bluetooth;

[0029] Integrated multi-constellation GNSS receiver, supporting L1 band precise positioning;

[0030] CPU architecture: 1+3+4 triple cluster design;

[0031] GPU: Integrated Arm Mali-G57 quad-core GPU, running at 850MHz;

[0032] NPU computing power: comprehensive computing power reaches 8TOPS.

[0033] The computing power support layer is based on a hardware-accelerated machine learning model; it analyzes the status data and environmental parameters of the multi-mode communication layer, identifies the scenario type and generates corresponding scheduling strategies.

[0034] Further,

[0035] The multi-mode communication layer is a hardware-based communication unit that covers multiple scenarios and includes three types of dedicated modules:

[0036] High-bandwidth unit: 5G / Wi-Fi baseband module;

[0037] Low latency unit: 5G URLLC communication module;

[0038] Low power consumption unit: Bluetooth 5.4 low power consumption module.

[0039] The high-bandwidth unit is connected to the main controller via a PCIe 3.0 interface;

[0040] The low-latency unit and low-power unit are directly connected to the main controller through the UART+SPI composite interface.

[0041] The present invention achieves three major innovative breakthroughs through hardware-level collaborative design: First, an interconnection layer is constructed using high-speed interfaces such as PCIe / SGMII+ to achieve low-latency data transmission of <10ms between modules; second, the AI ​​computing power module integrated with the 8TOPS NPU is deeply coupled with the multi-mode communication module (5G URLLC <1ms / Bluetooth 5.4 <1mW / millimeter wave band (28 / 39GHz)), and the LSTM / Transformer model is quantized and compressed (such as INT8 precision) and then solidified into the NPU dedicated storage to achieve scene-adaptive scheduling; finally, a "collection-recognition-decision-feedback" hardware closed loop is established, which reduces the communication interruption rate by 80% and the overall power consumption by 35% in emergency scenarios, and the AI ​​inference delay is <5ms, which is significantly better than traditional software scheduling solutions. This design achieves a technological leap from fixed functions to intelligent adaptation through the collaborative optimization of hardware AI and communication protocols.

[0042] The beneficial effects of the present invention are

[0043] This invention innovatively integrates AI computing power modules with multi-mode communication hardware. Through a four-layer hardware architecture (core control - computing power support - multi-mode communication - high-speed interconnection), it achieves intelligent scene adaptation, efficient resource utilization and hardware-level reliability. It supports multi-mode communications and standardized expansion interfaces such as 5G / 4G / Wi-Fi / Bluetooth / Ethernet local networks, reduces the communication interruption rate in emergency scenarios, and provides an integrated "perception-decision-execution" intelligent communication solution for scenarios such as the industrial Internet and emergency rescue. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the overall framework of the present invention;

[0045] Figure 2 This is a working diagram of the core modules (main controller, AI computing power module) of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0047] Smart city ubiquitous IoT scenario: Core requirements: Millions of terminals (such as electricity meters and street light sensors) must balance low power consumption and burst traffic carrying capacity.

[0048] Device Applications: Intelligent sleep scheduling: The AI ​​module automatically switches to Bluetooth 5.4 low-power mode based on terminal behavior patterns (e.g., a daily meter reading), reducing static power consumption to 1mW. Burst traffic response: High-bandwidth demand events (e.g., sudden facial recognition on a security camera) trigger the activation of the 5G / Wi-Fi module within milliseconds. Containerized load balancing, enabled by Nebula Intelligent Computing network technology, achieves nearly 100% link utilization, supporting daily terabyte-level data exchange. Energy-saving results: Overall energy consumption is reduced by 35% compared to traditional solutions, and terminal battery life is increased by three times.

[0049] The main technologies used in this article are as follows:

[0050] Internet of Things (IoT): Originating in the media industry, the IoT represents the third revolution in the information technology industry. The IoT connects any object to a network through information sensing devices and agreed-upon protocols. Objects exchange and communicate via information media, enabling intelligent identification, positioning, tracking, and monitoring.

[0051] Artificial Intelligence (AI) is a new technical discipline that studies and develops theories, methods, techniques, and application systems for simulating, extending, and expanding human intelligence. A branch of computer science, AI seeks to understand the essence of intelligence and develop new intelligent machines that can respond in a manner similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems.

[0052] Computing Technology: Computing technology is the ability to efficiently process and compute massive amounts of data through the collaborative work of computer hardware (such as CPUs, GPUs, and TPUs) and software systems. Its core metrics include computing speed (FLOPS), energy efficiency, and parallel processing capabilities. Current mainstream technologies encompass general-purpose computing chips (such as the x86 architecture), specialized acceleration chips (such as NPUs for AI training), distributed computing frameworks (such as Hadoop / Spark), and cutting-edge technologies such as quantum computing. These technologies are widely used in AI training (such as large models with hundreds of billions of parameters), scientific simulations (such as climate prediction), and blockchain verification. China plans to achieve a computing power of over 300 EFlops (300 petaflops) by 2025, accounting for approximately 30% of the global total, making it the core infrastructure of the digital economy.

[0053] Dynamic Scheduling: Dynamic scheduling is typically performed when the scheduling environment and tasks are subject to unpredictable fluctuations. Compared to static scheduling, dynamic scheduling can produce more actionable decision-making solutions tailored to the actual production site conditions.

[0054] Multimode communication technology: Multimode communication technology is a comprehensive solution that integrates multiple communication modes (such as 5G / 4G / Wi-Fi / Bluetooth / local area networks) to achieve intelligent switching and collaborative operation. Its core advantage lies in overcoming the distance and stability limitations of a single communication mode. It is widely used in fields such as marine communications, the Industrial Internet of Things, and 6G evolution. Leveraging dynamic mode switching, heterogeneous network convergence, and edge computing enhancements, this technology ensures reliable connections in complex environments.

[0055] The present invention provides a multi-mode communication dynamic scheduling system that integrates AI computing power (referred to as "this system"), which realizes hardware-level dynamic intelligent scheduling of communication resources in multiple scenarios through deep integration of hardware architecture and AI computing power modules, combined with the collaborative design of multi-mode communication hardware.

[0056] 1. Hardware Architecture Design

[0057] This system adopts a four-layer hardware architecture: core control - computing power support - multi-mode communication - high-speed interconnection. The core modules are interconnected through standardized high-speed interfaces to ensure low latency in data flow and high compatibility of module expansion. The specific structure is described as follows:

[0058] 1. Main controller: As the core hardware hub of the system, it integrates a multi-protocol communication control unit (supporting hardware stack analysis of 5G, Wi-Fi, Bluetooth and other protocols).

[0059] Directly connect to the network controller via the Gigabit Ethernet interface to achieve high-speed data transfer.

[0060] External 5G / Wi-Fi baseband module is connected through two sets of PCIe 3.0 interfaces (supporting hardware adaptation of multi-mode communication protocols).

[0061] Connect to the Gigabit Ethernet switch chip through the SGMII+ interface and output 4 Gigabit network ports (used to support high-speed interconnection of servers, industrial terminals and other equipment).

[0062] 2. Network controller: Serves as the data interaction hardware hub, implementing multi-module data routing based on a dedicated routing chip.

[0063] Connect to the AI ​​computing power module through the PCIe 2.0 interface (bandwidth ≥ 5Gbps) to collect status data of the multi-mode communication module (such as bandwidth occupancy, latency, power consumption) and terminal requirements (such as transmission type and priority) in real time.

[0064] Map the scheduling instructions generated by the AI ​​computing module to the main controller at the hardware level to ensure that the instruction execution delay is less than 10ms

[0065] 3. AI computing power module: It is dedicated computing power hardware that integrates AI acceleration chips (such as GPU / NPU) and algorithm processing units.

[0066] Directly connected to the network controller via the PCIe 2.0 interface, it provides real-time intelligent decision-making computing power.

[0067] Provides high-speed interfaces such as PCIe, USB 3.1, and DisplayPort

[0068] Supports Wi-Fi 6 (802.11ac) and Bluetooth.

[0069] Integrated with multi-constellation GNSS receiver (GPS / GLONASS / BDS / Galileo), it supports L1 band precise positioning.

[0070] CPU architecture: 1+3+4 triple-cluster design (1×2.7GHz Cortex-A76+3×2.3GHz Cortex-A76+4×2.1GHz Cortex-A55).

[0071] GPU: Integrated Arm Mali-G57 quad-core GPU, running at 850MHz.

[0072] NPU computing power: comprehensive computing power reaches 8TOPS (focused on AI task processing)

[0073] Core functions: Based on hardware-accelerated machine learning models (such as LSTM / Transformer), it analyzes the status data and environmental parameters (such as interference intensity and device movement speed) of multi-mode communication modules, identifies scenario types, and generates corresponding scheduling strategies.

[0074] 4. Multi-mode communication module: This is a hardware communication unit that covers multiple scenarios and includes three types of dedicated modules:

[0075] High-bandwidth unit: 5G / Wi-Fi baseband module (supporting millimeter wave frequency bands (such as 28 / 39GHz).

[0076] Low-latency unit: 5G URLLC communication module (supports ultra-reliable low-latency communication, air interface delay <1ms).

[0077] Low power unit: Bluetooth 5.4 low power module (static power consumption <1mW).

[0078] The high-bandwidth unit is connected to the main controller via a PCIe 3.0 interface (supporting high-band hardware adaptation).

[0079] The low-latency unit and low-power unit are directly connected to the main controller through the UART+SPI composite interface (to ensure fast hardware-level response to control instructions).

[0080] 2. Dynamic Scheduling Method (Hardware-Level Closed-Loop Implementation)

[0081] This invention is based on the hardware-level closed-loop process of "data collection-scene recognition-strategy decision-execution feedback" and realizes dynamic scheduling through the hardware collaboration of each module. The specific steps are as follows:

[0082] 1. Data acquisition layer (hardware real-time perception)

[0083] The multi-mode communication module reports status data (current bandwidth, air interface delay, module power consumption) in real time through dedicated hardware interfaces (such as 5G modules and Bluetooth modules). The network controller synchronously collects terminal requirements (transmission type: video / command / sensor data; priority: high / medium / low) and environmental parameters (signal interference intensity, device movement speed) through the hardware routing chip, and transmits them to the AI ​​computing power module through the PCIe 2.0 interface (hardware-level high-speed channel), ensuring that the data collection delay is less than 2ms.

[0084] 2. Scene recognition layer (AI hardware accelerated reasoning)

[0085] The AI ​​computing module uses a hardware-accelerated machine learning model (pre-trained on multi-scenario datasets) to perform real-time reasoning based on collected multi-dimensional data (communication status, terminal requirements, and environmental parameters), identifying the current scenario type at the hardware level.

[0086] High-bandwidth scenarios (such as 4K live streaming): trigger conditions are bandwidth requirements > 10Gbps and latency tolerance > 10ms;

[0087] Low-latency scenarios (such as industrial control): Trigger conditions are latency requirements <5ms and packet loss rate <0.1%;

[0088] Low-power scenarios (such as sensor networks): Trigger conditions are terminal battery life > 72 hours and data transmission frequency < 1 time / minute;

[0089] Emergency scenarios (such as earthquake rescue): The trigger conditions are public network interruption and device movement speed > 50km / h, and communication continuity must be guaranteed.

[0090] 3. Policy decision layer (hardware-level policy generation)

[0091] The AI ​​computing module generates hardware-executable scheduling policies based on the scenario type and transmits them to the network controller via the PCIe 2.0 interface. Specific policies include:

[0092] High-bandwidth scenarios: The hardware switches to the 5G / Wi-Fi baseband module, allocates the 6 GHz unlicensed frequency band (dedicated hardware band), and disables redundant links (such as Bluetooth modules) to reduce interference.

[0093] Low-latency scenario: The hardware switches to the 5G URLLC module, QoS priority queuing (hardware-level queue management) is enabled, and 1 Gbps of dedicated bandwidth is reserved.

[0094] Low-power scenario: The hardware switches to Bluetooth low-power mode and reduces the transmit power (hardware register configuration: 20dBm → 10dBm);

[0095] Emergency scenario: The hardware activates multi-link redundancy (5G public network + satellite communication module) and automatically switches to the narrowband private network (if deployed, triggered by a hardware switch).

[0096] 4. Execution feedback layer (hardware status adjustment and optimization)

[0097] The main controller executes the scheduling strategy through the multi-protocol control unit (hardware-level protocol stack) and adjusts the hardware status of the multi-mode communication module (such as switching frequency bands and opening / closing redundant links). The network controller collects the execution effect (actual bandwidth, latency, power consumption) in real time through the hardware monitoring interface, and feeds it back to the AI ​​computing power module, iteratively optimizing the machine learning model parameters (hardware-level parameter updates) to improve the accuracy of subsequent scheduling strategies.

[0098] 3. Hardware-level innovation advantages

[0099] This system achieves the following core advantages through the deep integration of AI computing modules and multi-mode communication hardware:

[0100] Hardware-level intelligent scene adaptation

[0101] As a dedicated hardware unit, the AI ​​computing module supports real-time scenario reasoning (single-cycle latency <5ms) and dynamically adjusts the hardware configuration of the multi-mode communication module (such as frequency band, power, and link), overcoming the hardware limitations of traditional equipment's "fixed-rule scheduling." In measured emergency scenarios, the communication interruption rate was reduced by 80% compared to traditional solutions (based on a multi-link redundant hardware design).

[0102] Efficient use of hardware resources

[0103] Through hardware-level intelligent switching between high-bandwidth, low-latency, and low-power modules, performance and energy efficiency are balanced: high-speed modules are activated in high-bandwidth scenarios, while redundant hardware is hibernated in low-power scenarios. This results in a 35% reduction in overall device power consumption compared to traditional solutions (due to the dynamic hibernation mechanism of hardware modules), significantly improving the PUE of network equipment.

[0104] Hardware scalability and compatibility

[0105] It adopts standardized high-speed interfaces such as PCIe 3.0 and SGMII+, supports the expansion of new communication modules such as satellite communication and UWB ultra-wideband (hardware interface is plug-and-play), and adapts to the future integrated air-space-ground communication needs; it is designed based on international specifications such as 3GPP 5G standards and Bluetooth SIG protocols to ensure hardware-level interoperability with mainstream terminals (such as mobile phones, industrial PLCs, and sensors).

[0106] Hardware reliability assurance

[0107] The core modules (main controller, AI computing power module) use industrial-grade chips (operating temperature -40℃ ~ 85℃), and the multi-mode communication module supports hardware-level anti-interference design (such as the beamforming hardware algorithm of the 5G module), ensuring stable operation in complex environments (such as high temperature and strong electromagnetic interference), meeting the high reliability requirements of scenarios such as the Industrial Internet and emergency rescue.

[0108] Through the deep integration of hardware architecture and AI computing power, this invention provides an integrated solution of "hardware perception-intelligent decision-making-real-time execution" for multi-scenario communications, promoting the evolution of communication equipment towards "intelligent, efficient, and ubiquitous".

[0109] The present invention adopts a four-layer hardware architecture (core control - computing power support - multi-mode communication - high-speed interconnection), which realizes efficient data transmission through the high-speed interconnection layer and reduces delays.

[0110] The hardware embedding of AI computing modules improves AI reasoning speed and supports real-time decision-making; it reduces CPU load and improves overall system efficiency.

[0111] The specialized design of the multi-mode communication module (supporting 5G / 4G / Wi-Fi / Bluetooth / local network) enables seamless switching between multiple networks and ensures communication continuity.

[0112] Dynamically schedule the hardware-level closed-loop process (data collection - scenario recognition - policy decision - execution feedback) to achieve end-to-end latency response of <10ms, meeting low-latency requirements; reduce software scheduling overhead through hardware collaboration; closed-loop feedback ensures continuous policy optimization.

[0113] Multi-scenario adaptation, automatic identification of network environments, and dynamic parameter adjustment; reduce manual configuration and lower operation and maintenance costs; and improve QoS guarantees in different scenarios.

[0114] Efficient use of hardware resources, maximizing computing resource utilization through intelligent scheduling; reducing idle resources and lowering energy consumption; supporting stable operation under high-load scenarios; and reducing static power consumption through hardware-level module hibernation mechanisms.

[0115] The present invention implements communication resource scheduling through hardware-level closed-loop control, including:

[0116] (1) Real-time collection of communication status, terminal requirements and environmental parameters;

[0117] (2) AI hardware-accelerated scene recognition: Identify high-bandwidth / low-latency / low-power / emergency scenarios based on machine learning models, with a latency of <5ms;

[0118] (3) Generate and execute strategies:

[0119] -High-bandwidth scenarios: Enable the 5G / Wi-Fi module and allocate a dedicated frequency band;

[0120] - Low latency scenario: Switch to 5G URLLC module and reserve QoS channel;

[0121] - Low power scenario: Activate Bluetooth low power mode and reduce transmit power;

[0122] -Emergency scenario: start multi-link redundant communication;

[0123] (4) Feedback optimization: Collect execution effect data and iteratively update AI model parameters.

[0124] The above description is only a preferred embodiment of the present invention and is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention are included in the scope of protection of the present invention.

Claims

1. A multi-mode communication network dynamic scheduling system integrating AI computing power, characterized by: It includes four layers of hardware architecture: Core control layer: As the brain of the system, responsible for overall scheduling decisions; Computing power support layer: Integrates heterogeneous AI computing units to provide intelligent analysis capabilities; Multi-mode communication layer: supports the collaborative operation of several communication protocols and frequency bands; High-speed interconnection layer: realizes high-speed data interaction between modules through standardized interfaces.

2. The system according to claim 1, wherein: The core control layer is interconnected through standardized high-speed interfaces to ensure low latency in data transfer and high compatibility in module expansion.

3. The system according to claim 2, characterized in that The main controller of the core control layer, as the core hardware hub of the system, integrates a multi-protocol communication control unit; including: Directly connected to the network controller via Gigabit Ethernet interface; External 5G / Wi-Fi baseband module via two sets of PCIe 3.0 interfaces; Connect to the Gigabit Ethernet switch chip through the SGMII+ interface and output 4 Gigabit network ports.

4. The system according to claim 2, wherein: The network controller in the core control layer serves as the hardware hub for data interaction and implements multi-module data routing based on a dedicated routing chip. It includes: Connect to the AI ​​computing module through the PCIe 2.0 interface to collect status data of the multi-mode communication module and terminal requirements in real time; The scheduling instructions generated by the AI ​​computing module are mapped to the main controller at the hardware level to ensure that the instruction execution delay is less than 10ms.

5. The system according to claim 1, wherein: The computing power support layer is dedicated computing power hardware that integrates AI acceleration chips and algorithm processing units, including: Directly connected to the network controller via the PCIe 2.0 interface, providing real-time intelligent decision-making computing power; Provides high-speed interfaces of PCIe, USB 3.1, and DisplayPort; Supports Wi-Fi 6 and Bluetooth; Integrated multi-constellation GNSS receiver, supporting L1 band precise positioning; CPU architecture: 1+3+4 triple cluster design; GPU: Integrated Arm Mali-G57 quad-core GPU, running at 850MHz; NPU computing power: comprehensive computing power reaches 8TOPS.

6. The system according to claim 5, characterized in that The computing power support layer is based on a hardware-accelerated machine learning model; it analyzes the status data and environmental parameters of the multi-mode communication layer, identifies the scenario type and generates corresponding scheduling strategies.

7. The system according to claim 1, wherein: The multi-mode communication layer is a hardware-based communication unit that covers multiple scenarios and includes three types of dedicated modules: High-bandwidth unit: 5G / Wi-Fi baseband module; Low latency unit: 5G URLLC communication module; Low power consumption unit: Bluetooth 5.4 low power consumption module.

8. The system according to claim 7, characterized in that The high-bandwidth unit is connected to the main controller via a PCIe 3.0 interface; The low-latency unit and low-power unit are directly connected to the main controller through the UART+SPI composite interface.

Images