Wireless communication system assisted by edge computing

Abstract

The invention discloses an edge computing assisted wireless communication system, which relates to the technical field of edge computing, and comprises an edge computing node cluster which is deployed at a base station and an access point and provides distributed computing and storage capability; the resource sensing module is used for monitoring computing resources, communication resources and task queue states of the edge nodes in real time; the dynamic scheduling controller is used for dynamically deciding a task unloading path based on reinforcement learning and game theory algorithms, wherein the task unloading path comprises local processing, edge node cooperative processing, cloud unloading and global time delay and energy consumption optimization; the cross-layer communication interface supports multi-protocol fusion of 5G NR and Wi-Fi 6, and realizes low-delay data interaction between edge nodes; and the security verification module is used for ensuring credibility and integrity of data between edge nodes through a lightweight block chain and a TEE (Trusted Execution Environment). According to the invention, through a hybrid decision framework of reinforcement learning and game theory, the problem that a single algorithm in a traditional scheme cannot give consideration to dynamic environment adaptability and multi-node benefit coordination is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of edge computing, and specifically relates to a wireless communication system assisted by edge computing. Background Art

[0002] With the popularization of a large number of mobile terminals such as smart phones and Internet of Things devices, mobile data traffic has increased exponentially. When dealing with such a huge amount of data, traditional cloud computing architectures face problems such as bandwidth bottlenecks and transmission delays, and it is difficult to meet application scenarios with high real-time requirements, such as autonomous driving and industrial automation control. The wide application of Internet of Things devices requires local real-time data processing and analysis. For example, sensors in smart homes need to process the collected data in a timely manner to achieve intelligent control of home appliances; production equipment in smart factories needs to monitor and analyze operation data in real time to improve production efficiency and product quality. Transmitting all this data to the cloud for processing will cause significant delays and cannot meet the real-time requirements of applications. The application of artificial intelligence algorithms in wireless communication is becoming more and more extensive, such as wireless resource management, signal detection and processing, etc. However, the training and inference of artificial intelligence models usually require a large amount of computing resources and data, and it is difficult to implement on terminal devices.

[0003] Edge computing can sink computing resources to the network edge, close to data sources and user terminals, providing strong support for the application of artificial intelligence technology in wireless communication. However, there are still problems to be solved in the existing technology: the problem of resource allocation and scheduling, and the limited computing power of edge nodes, including how to dynamically allocate computing and communication resources to optimize the global performance. Summary of the Invention

[0004] To solve the above technical problems, a wireless communication system assisted by edge computing is provided, and this technical solution solves the problem of the above-mentioned resource allocation and scheduling.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A wireless communication system assisted by edge computing, comprising: An edge computing node cluster: deployed at base stations and access points, providing distributed computing and storage capabilities; A resource awareness module: real-time monitoring of the computing resources, communication resources, and task queue status of edge nodes; A dynamic scheduling controller: based on reinforcement learning and game theory algorithms, dynamically determining task offloading paths including local processing, edge node collaborative processing, and cloud offloading, to optimize the global delay and energy consumption; A cross-layer communication interface: supporting multi-protocol fusion of 5G NR and Wi-Fi 6 to achieve low-latency data interaction between edge nodes; Security verification unit: Through lightweight blockchain and TEE, ensure the credibility and integrity of data among edge nodes.

[0006] Preferably, the edge computing node cluster specifically includes: The hardware composition includes a computing unit, a storage unit, and a communication unit; Computing unit: Heterogeneous computing architecture, including CPU + GPU / TPU for general computing and dedicated acceleration; FPGA / ASIC reconfigurable hardware; low-power chips, such as the ARM Cortex-M series MCUs for IoT devices; Storage unit: Hierarchical storage design, with high-speed cache NVMe SSD for storing hot data; persistent storage 3D NAND Flash; distributed storage engine using lightweight Ceph and EdgeFS for cross-node data synchronization; Communication unit: Multi-mode network interface, including 5G NR baseband unit, Wi-Fi 6 / 6E, and time-sensitive network interface.

[0007] Preferably, the edge computing node cluster specifically includes: The software architecture includes a resource virtualization layer, a task scheduling engine, and an edge service framework; Resource virtualization layer: Containerized platforms are Kubernetes Edge Edition and OpenYurt, supporting dynamic deployment of microservices, and lightweight virtual machines isolate tasks with high security requirements; Task scheduling engine: Real-time resource monitoring, with collected metrics including CPU utilization, memory occupancy, GPU video memory, and channel RTT; network traffic prediction based on LSTM; Edge service framework: Functional modules include local AI inference service, data preprocessing, and protocol conversion gateway.

[0008] Preferably, the resource awareness module specifically includes: Computing resource monitoring unit: Collect hardware metrics, including real-time utilization, temperature, and power consumption of CPU / GPU / TPU status, remaining memory capacity of memory and storage, Swap usage rate, and storage IOPS; virtualized resource management, including container / virtual machine metrics, Docker container CPU quota, and KVM virtual machine vCPU scheduling latency; Communication resource monitoring unit: Monitor the wireless channel status, according to 5G / Wi-Fi parameters, including RTT, packet loss rate, and channel quality; according to protocol stack performance, including TCP / UDP throughput and multi-path transmission status of the QUIC protocol stream; network topology awareness, neighbor node discovery, automatically identify adjacent edge nodes and inter-node link bandwidth based on mDNS and LLDP protocols.

[0009] Preferably, the resource awareness module specifically includes: Task queue status monitoring unit: Task attribute tracking, including real-time queue depth, monitoring the number of tasks to be processed, classifying them by priority, and the resource requirements of single tasks; Historical load analysis, including sliding window statistics and peak load prediction; Quality of service metrics, SLA violation risk, obtained through the task deadline default probability and the critical path task dependency graph.

[0010] Preferably, the dynamic scheduling controller specifically includes: State awareness unit: Environmental state input, including real-time resource data, CPU / GPU utilization, remaining bandwidth, and node geographical location from the resource awareness module; Task characteristics, obtained according to the amount of computation, data volume, deadline, and priority label; Network topology, obtained through the communication delay matrix between edge nodes; Algorithm engine unit: Reinforcement learning model, with algorithm selection as Proximal Policy Optimization and Soft Actor-Critic for high-dimensional continuous action spaces; State space design, including node load rate, channel quality index, and task queue length; The reward function is a multi-objective weighted sum; Game theory cooperation mechanism, non-cooperative game modeling, with edge nodes as players, cost function = local processing energy consumption + communication overhead; Nash equilibrium solution, converging to the optimal offloading strategy through a distributed iterative algorithm.

[0011] Preferably, the dynamic scheduling controller specifically includes: Action execution unit: Offloading decision, including local processing for lightweight tasks; Edge collaboration, sharing the computing load through D2D communication and edge Mesh network; Cloud offloading, only for compute-intensive non-real-time tasks; Resource reservation instructions, preallocating GPU video memory and bandwidth for high-priority tasks.

[0012] Preferably, the cross-layer communication interface specifically includes: Hardware infrastructure: Multi-mode radio frequency front end, which is a reconfigurable radio frequency chip, supporting a software-defined radio architecture for 5GNR and Wi-Fi 6, and a dynamic frequency band switching circuit; Intelligent antenna system, millimeter wave beamforming and Wi-Fi 6 MU-MIMO cooperative scheduling; Protocol acceleration hardware, using a dedicated baseband processor for 5G LDPC encoding and decoding and hardware acceleration for Wi-Fi 6 OFDMA resource unit allocation; Software protocol stack design: Cross-layer protocol fusion engine, physical layer adaptation, dynamically select modulation mode and switch in real time according to channel SNR; MAC layer scheduling, through a unified time slot allocation framework; network layer routing, based on SDN flow table rules, supporting IP-over-5G / Wi-Fi dual-stack transmission; low-latency optimization technology, data plane acceleration, enhanced by UDP protocol, multi-path transmission of QUIC protocol, automatically select the optimal link; through zero-copy technology, DPDK / SPDK user-mode drivers, reduce the data transfer overhead from kernel mode to user mode; control plane optimization, through the joint energy-saving scheduling of 5G URLLC and Wi-Fi 6 Target Wake Time.

[0013] Preferably, the cross-layer communication interface specifically includes: Key interaction process: Multi-protocol collaborative transmission, link selection phase, select 5G NR according to the delay budget, select Wi-Fi 6 for high-throughput tasks, and monitor the RTT in real time; data fragmentation and aggregation, fragment large files and transmit them concurrently through dual protocols, and the receiving end reorganizes based on the sequence number, with an out-of-order tolerance mechanism; seamless handover process, hard handover, when the 5G signal strength < -110dBm, trigger Wi-Fi 6 association; soft handover, 5G and Wi-Fi 6 dual connection, packet replication transmission.

[0014] Preferably, the security verification module specifically includes: Lightweight blockchain: Optimize the consensus mechanism, a practical Byzantine fault tolerance variant, only trusted anchor nodes in the edge cluster participate in the consensus, compress the transaction verification time; directed acyclic graph structure, adopt the parallel verification mechanism of IOTA Tangle, support high-concurrency micro-transactions; data deposit design, key behaviors are uploaded to the chain, only store the task offloading decision hash, node identity certificate, resource usage audit log; off-chain data is stored through IPFS, only the fingerprint is stored on the chain; smart contract rules, automatically trigger exception handling; Trusted execution environment integration: Hardware-level security, TEE type selection includes Intel SGX, ARM TrustZone, and domestic solutions; secure enclave function, key management never leaves the TEE through the private key of the edge node, and sensitive computing directly executes on encrypted data for AI model inference; TEE-blockchain collaboration, off-chain calculation verification, including the TEE generating zero-knowledge proofs to prove the correctness of the calculation process, and the blockchain only verifies the proofs rather than the original data; Hybrid security architecture: When the node identity is forged, hardware identity binding based on TEE, each node has a unique key + blockchain CA; when the data is tampered with, through blockchain deposit + Merkle tree verification; when a man-in-the-middle attack occurs, end-to-end encryption implemented within the TEE; when a denial of service occurs, reputation-based traffic filtering, and the blockchain records the historical behavior scores of nodes.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes a distributed deployment of heterogeneous edge node clusters, combined with a security verification module of lightweight blockchain and TEE, to achieve the sinking of computing power and local data processing; the cross-layer communication interface supports multi-protocol intelligent switching of 5G / Wi-Fi 6, forming a seamless network architecture of "cloud-edge-end", significantly reducing transmission delay and backhaul load.

[0016] The dynamic scheduling controller innovatively integrates reinforcement learning and game theory to solve the problems of dynamic environment adaptation and multi-node interest balance in resource allocation. The reinforcement learning model optimizes the offloading strategy in real time through a high-dimensional state space, and the game theory mechanism realizes autonomous cooperation between nodes through virtual currency incentives. This hybrid algorithm supports millisecond-level decision-making and can dynamically adjust and optimize weights according to the task type, taking into account the multi-objective requirements of delay, energy consumption, and cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is an internal framework diagram of a wireless communication system assisted by edge computing; Figure 2 is an internal framework diagram of an edge computing node cluster; Figure 3 is an internal framework diagram of a resource awareness module Figure 4 is an internal framework diagram of a dynamic scheduling controller. DETAILED DESCRIPTION OF THE INVENTION

[0018] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0019] Referring to Figure 1 as shown, a wireless communication system assisted by edge computing includes: Edge computing node cluster: Deployed at base stations and access points, providing distributed computing and storage capabilities; Resource awareness module: Real-time monitoring of the computing resources, communication resources, and task queue status of edge nodes; Dynamic scheduling controller: Based on reinforcement learning and game theory algorithms, dynamically deciding task offloading paths including local processing, edge node collaborative processing, and cloud offloading, optimizing global delay and energy consumption; Cross-layer communication interface: Supporting multi-protocol fusion of 5G NR and Wi-Fi 6 to achieve low-latency data interaction between edge nodes; Security verification module: Ensuring the credibility and integrity of data between edge nodes through lightweight blockchain and TEE.

[0020] It should be noted that this solution forms a dynamic closed-loop control system with perception - decision - execution linkage: Resource perception module (collecting over 100,000 indicators per second) → Dynamic scheduling controller (generating strategies within 20 ms) → Cross-layer communication interface (protocol switching delay < 5 ms) to form a real-time closed loop.

[0021] Unified data bus design, implemented using the Apache Arrow in-memory format, enables efficient cross-module transmission of resource monitoring data (Float16 compressed), scheduling instructions (CBOR encoded), and security credentials (ASN.1 DER format).

[0022] Joint modeling of 5G KPIs and computing loads, establishing a cross-layer mathematical model of channel quality (CQI) → computing task shard size → encryption time consumption, to achieve:

[0023] In the formula, is the computing offloading ratio, reflecting the distribution of computing tasks among different computing resources; is the encryption strength level, used to measure the security and resource consumption of encryption operations; is the task latency, which refers to the time required from the start to the completion of a task, including the time overheads of computing, transmission, encryption, and decryption processes; is the task energy consumption, which refers to the total energy consumed during the completion of a task, including the energy consumption of computing, communication, and security mechanisms; is the security level, used to measure whether the security of the system meets the requirements of the service level agreement; is the security level 3, which is a specific security level standard used to ensure that the system meets certain security performance requirements.

[0024] Refer to Figure 2 As shown, the edge computing node cluster specifically includes: The hardware composition includes a computing unit, a storage unit, and a communication unit; Computing unit: Heterogeneous computing architecture, including CPU + GPU / TPU for general computing and dedicated acceleration; FPGA / ASIC reconfigurable hardware; low-power chips, such as the ARM Cortex-M series MCUs for IoT devices; Storage unit: Hierarchical storage design, with high-speed cache NVMe SSD for storing hot data; persistent storage 3D NANDFlash; distributed storage engines using lightweight Ceph and EdgeFS for cross-node data synchronization; Communication unit: Multi-mode network interface, including 5G NR baseband unit, Wi-Fi 6 / 6E, and time-sensitive network interface; The software architecture includes a resource virtualization layer, a task scheduling engine, and an edge service framework; Resource virtualization layer: The containerization platform is the edge version of Kubernetes and OpenYurt, supporting the dynamic deployment of microservices and isolating high-security requirement tasks with lightweight virtual machines; Task scheduling engine: Real-time resource monitoring, and the collected metrics include CPU utilization, memory occupancy, GPU video memory, and channel RTT; Network traffic prediction based on LSTM; Edge service framework: The functional modules include local AI inference service, data preprocessing, and protocol conversion gateway.

[0025] It should be noted that the computing units include: Scene adaptive acceleration: For AI tasks, the GPU (NVIDIA H100) processes matrix operations, and the TPU (Google Coral) optimizes the inference latency; For signal processing, the FPGA (Xilinx Versal) is reconfigured in real time as a 5G LDPC decoder or OFDM modulator; In the energy efficiency priority scenario, the ARM Cortex-M55 MCU (+Ethos-U55 NPU) realizes the sensor data analysis with μW-level power consumption.

[0026] Storage-communication joint optimization, near-storage computing, the NVMe SSD has a built-in AI accelerator, such as the Samsung SmartSSD, and directly performs data filtering at the storage end; The 3D NAND Flash cooperates with 5G URLLC: The hot data is pre-loaded into the SSD cache to meet the 1ms-level access latency; Hardware-level integration of multi-mode communication, baseband chip innovation, the Qualcomm X75 baseband, the world's first 5G Advanced-Ready+Wi-Fi 7 integrated solution, supporting joint channel sensing; The hardware timestamp accuracy of the time-sensitive network reaches 10ns.

[0027] Security isolation of the resource virtualization layer: Refer to Figure 3 As shown, the resource perception module specifically includes: Computing resource monitoring unit: Collecting hardware metrics, including the real-time utilization rate, temperature, and power consumption of the CPU / GPU / TPU status, the remaining memory capacity of the memory and storage, the Swap usage rate, and the storage IOPS; Virtualized resource management, including container / virtual machine metrics, the CPU quota of Docker containers, and the vCPU scheduling latency of KVM virtual machines; Communication Resource Monitoring Unit: Monitor the wireless channel status, according to 5G / Wi-Fi parameters, including RTT, packet loss rate, channel quality; according to protocol stack performance, including TCP / UDP throughput and multi-path transmission status of the QUIC protocol stream; network topology awareness, neighbor node discovery, automatically identify adjacent edge nodes and inter-node link bandwidth based on mDNS and LLDP protocols; Task Queue Status Monitoring Unit: Task attribute tracking, including real-time queue depth, monitor the number of tasks to be processed, classify by priority, and single-task resource requirements; historical load analysis, including sliding window statistics and peak load prediction; quality of service metrics, SLA violation risk, obtained through task deadline default probability and critical path task dependency graph.

[0028] It should be noted that the Computing Resource Monitoring Unit: Heterogeneous chip monitoring, GPU / TPU status, collect SM unit utilization through NVIDIA DCGM 4.0, combined with the power consumption model of AMD ROCm (accuracy ±3W); low-power MCU, use Arm Energy Probe technology to achieve μA-level current monitoring; Virtualization layer perspective, capture the vCPU stealing time of KVM virtual machines through eBPF hooks to identify physical CPU overload.

[0029] Communication Resource Monitoring Unit: 5G / Wi-Fi joint probe, CSI feedback compression of 5G NR, based on JPEG 2000 encoding, reducing the reported data volume by 90%; OFDMA sub-carrier level monitoring of Wi-Fi 6, obtained through the Intel AX210 network card driver API; topology discovery optimization, extend the LLDP protocol frame to carry the computing power label of edge nodes.

[0030] Task Queue Status Monitoring Unit: Real-time priority classifier, dynamically adjust task priorities based on the XGBoost model (input: task type / historical execution time / SLA level); Load prediction model, use the Temporal Fusion Transformer algorithm to fuse sliding window statistics and external events, such as sudden increase in sports event traffic.

[0031] Refer to Figure 4 As shown, the dynamic scheduling controller specifically includes: Status Awareness Unit: Environmental status input, including real-time resource data, CPU / GPU utilization, remaining bandwidth, node geographical location from the resource awareness module; task characteristics, obtained according to the amount of computation, data volume, deadline, priority label; network topology, obtained through the communication delay matrix between edge nodes; Algorithm Engine Unit: Reinforcement learning models with proximal policy optimization and flexible Actor-Critic algorithms selected for high-dimensional continuous action spaces; state space design including node load rate, channel quality index, and task queue length; reward function as a multi-objective weighted sum; Game theory collaboration mechanism, non-cooperative game modeling with edge nodes as players, cost function = local processing energy consumption + communication overhead; Nash equilibrium solution converging to the optimal offloading strategy through a distributed iterative algorithm; Action Execution Unit: Offloading decisions include local processing for lightweight tasks; edge collaboration to share computational load through D2D communication and edge Mesh networks; cloud offloading only for compute-intensive non-real-time tasks; resource reservation instructions to pre-allocate GPU memory and bandwidth for high-priority tasks.

[0032] It should be noted that the State Awareness Unit: 5G-A enhanced delay matrix, obtaining the delay between edge nodes in real time through the Xhaul Awareness Protocol (XAP) defined in 3GPP Release 19, supporting dynamic topology mapping in the millimeter wave band; Dynamic labeling of task characteristics, automatically generating task characteristic labels using the Gemini-Nano edge model, such as mapping "automatic driving emergency braking" to URLLC L3-level priority; Quantum random number injection, providing true random seeds for the game theory mechanism, implemented through the Guodun Quantum QRNG chip to prevent strategy prediction attacks.

[0033] Hardware acceleration of the algorithm engine: Algorithm Engine Unit: Scene adaptive improvement of reinforcement learning, enhanced attention mechanism, embedding the Transformer-XL module in the Critic network of PPO to achieve long-term dependence modeling of node load (memory span up to 1 hour); Multi-objective dynamic weights, automatically adjusting the weights of the reward function according to the task type:

[0034] Wherein, is the delay weight, used to measure the importance of delay in the reward function, which is automatically adjusted according to the task type; is a ultra-reliable low-latency communication task, which is a task type with extremely high requirements for delay; is a massive machine-type communication task, which is a task type with relatively low requirements for delay; Incentive design of game theory mechanism, virtual currency economic model, nodes contribute computing power to obtain "Edge Coins", which can be used to purchase bandwidth reservation rights (automatically settled by smart contracts), anti-Sybil attack design, and each TEE hardware identity is bound to a unique wallet address; Extension of incomplete information game, using Harsanyi transformation to process node private information, such as remaining battery power, and achieving incentive compatibility through Bayesian game.

[0035] In summary, the advantages of the present invention are as follows: An edge computing system with full-stack collaboration of "perception - decision - execution" is constructed. Through distributed deployment of heterogeneous node clusters, the computing power is sunk and data is processed locally, breaking through the limitations of the traditional cloud computing centralized architecture; innovatively integrating reinforcement learning and game theory algorithms, supporting millisecond-level decision optimization, and being able to dynamically adjust weights such as latency and energy consumption according to different task types such as URLLC / mMTC, realizing multi-objective collaborative optimization; for high-demand scenarios such as industrial control and autonomous driving, ensuring a response at the 10ms level through edge collaborative computing and TSN network, and combining TEE security verification to ensure zero loss of key instructions, significantly improving reliability; adopting a hybrid security architecture of lightweight blockchain + TEE, reducing computing power overhead while ensuring data credibility, and being compatible with domestic and international standard hardware to avoid the risk of technology monopoly.

[0036] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A wireless communication system assisted by edge computing, characterized in that Including: Edge computing node cluster: Deployed at base stations and access points to provide distributed computing and storage capabilities; Resource awareness module: Real-time monitor the computing resources, communication resources and task queue status of edge nodes; Dynamic scheduling controller: Based on reinforcement learning and game theory algorithms, dynamically decide the task offloading paths including local processing, edge node collaborative processing, and cloud offloading, and optimize the global delay and energy consumption; Cross-layer communication interface: Support multi-protocol fusion of 5G NR and Wi-Fi 6 to achieve low-latency data interaction between edge nodes; Security verification module: Ensure the credibility and integrity of data between edge nodes through lightweight blockchain and TEE.

2. The wireless communication system assisted by edge computing according to claim 1, characterized in that The edge computing node cluster specifically includes: The hardware composition includes a computing unit, a storage unit, and a communication unit; Computing unit: Heterogeneous computing architecture, including CPU+GPU / TPU for general computing and dedicated acceleration; FPGA / ASIC reconfigurable hardware; low-power chips, such as the ARM Cortex-M series MCUs for IoT devices; Storage unit: Hierarchical storage design, with high-speed cache NVMe SSD for storing hot data; persistent storage 3D NAND Flash; distributed storage engine using lightweight Ceph and EdgeFS for cross-node data synchronization; Communication unit: Multi-mode network interface, including 5G NR baseband unit, Wi-Fi 6 / 6E, and time-sensitive network interface.

3. The wireless communication system assisted by edge computing according to claim 2, wherein, The edge computing node cluster specifically includes: The software architecture includes a resource virtualization layer, a task scheduling engine, and an edge service framework; Resource virtualization layer: The containerization platform is the edge version of Kubernetes and OpenYurt, which supports the dynamic deployment of microservices and lightweight virtual machine isolation for high-security requirement tasks; Task scheduling engine: Real-time resource monitoring, and the collected metrics include CPU utilization, memory occupancy, GPU video memory, and channel RTT; network traffic prediction based on LSTM; Edge service framework: The functional modules include local AI inference service, data preprocessing, and protocol conversion gateway.

4. A wireless communication system assisted by edge computing according to claim 3, wherein The resource awareness module specifically includes: Computing resource monitoring unit: Collect hardware metrics, including the real-time utilization rate, temperature and power consumption of CPU / GPU / TPU status, the remaining memory capacity of memory and storage, Swap usage rate, and storage IOPS; virtualized resource management, including container / virtual machine metrics, Docker container CPU quota, and KVM virtual machine vCPU scheduling delay; Communication resource monitoring unit: Monitor the wireless channel status, according to 5G / Wi-Fi parameters, including RTT, packet loss rate, and channel quality; according to the protocol stack performance, including TCP / UDP throughput and the multi-path transmission status of the QUIC protocol stream; network topology awareness, neighbor node discovery, automatically identify neighboring edge nodes and the link bandwidth between nodes based on the mDNS and LLDP protocols.

5. The wireless communication system assisted by edge computing according to claim 4, wherein The resource awareness module specifically includes: Task Queue Status Monitoring Unit: Task attribute tracking, including real-time queue depth, monitoring the number of tasks to be processed, classifying by priority, and single-task resource requirements; historical load analysis, including sliding window statistics and peak load prediction; quality of service metrics, SLA violation risk, obtained through task deadline default probability and critical path task dependency graph.

6. The edge-computing-assisted wireless communication system according to claim 5, wherein The dynamic scheduling controller specifically includes: State Awareness Unit: Environmental state input, including real-time resource data, CPU / GPU utilization, remaining bandwidth, and node geographical location from the resource awareness module; task characteristics, obtained according to the amount of computation, data volume, deadline, and priority label; network topology, obtained through the communication delay matrix between edge nodes. Algorithm Engine Unit: Reinforcement learning model, with algorithm selection of Proximal Policy Optimization and Soft Actor-Critic for high-dimensional continuous action space; state space design, including node load rate, channel quality index, and task queue length; the reward function is a multi-objective weighted sum. Game theory cooperation mechanism, non-cooperative game modeling, with edge nodes as players, cost function = local processing energy consumption + communication overhead; Nash equilibrium solution, converging to the optimal offloading strategy through a distributed iterative algorithm.

7. The edge-computing-assisted wireless communication system according to claim 6, wherein The dynamic scheduling controller specifically includes: Action Execution Unit: Offloading decision, including local processing for lightweight tasks; edge collaboration to share the computing load through D2D communication and edge Mesh network; cloud offloading, only for compute-intensive non-real-time tasks; resource reservation instructions to pre-allocate GPU video memory and bandwidth for high-priority tasks.

8. A wireless communication system assisted by edge computing according to claim 7, characterized in that The cross-layer communication interface specifically includes: Hardware infrastructure: Multi-mode RF front-end, a reconfigurable RF chip, supporting a software-defined radio architecture for 5GNR and Wi-Fi 6, with a dynamic frequency band switching circuit; intelligent antenna system, millimeter wave beamforming and Wi-Fi 6 MU-MIMO cooperative scheduling; protocol acceleration hardware, using a dedicated baseband processor for 5G LDPC encoding and decoding and hardware acceleration for Wi-Fi 6 OFDMA resource unit allocation. Software protocol stack design: Cross-layer protocol fusion engine, physical layer adaptation, by dynamically selecting the modulation method and switching in real-time according to the channel SNR; MAC layer scheduling, through a unified time slot allocation framework; network layer routing, based on SDN flow table rules, supporting IP-over-5G / Wi-Fi dual-stack transmission; low-latency optimization technology, data plane acceleration, enhanced through the UDP protocol, multi-path transmission of the QUIC protocol, and automatically selecting the optimal link; through zero-copy technology, DPDK / SPDK user-mode drivers to reduce the data transfer overhead from the kernel mode to the user mode; control plane optimization, through the joint energy-saving scheduling of 5G URLLC and Wi-Fi 6 Target Wake Time.

9. The edge-computing-assisted wireless communication system according to claim 8, wherein The cross-layer communication interface specifically includes: Key interaction process: Multi-protocol collaborative transmission, link selection phase, select 5G NR according to latency budget, select Wi-Fi 6 for high-throughput tasks, and monitor the RTT in real time; data fragmentation and aggregation, large files are fragmented and transmitted concurrently through dual protocols, and the receiver reorganizes based on sequence numbers, with an out-of-order tolerance mechanism; seamless handover process, hard handover, when the 5G signal strength < -110 dBm, trigger Wi-Fi 6 association; soft handover, dual connection of 5G and Wi-Fi 6, and packet replication transmission.

10. The edge computing-assisted wireless communication system according to claim 9, wherein The security verification module specifically includes: Lightweight blockchain: Optimized consensus mechanism, practical Byzantine fault tolerance variant, only trusted anchor nodes in the edge cluster participate in consensus, compressing transaction verification time; directed acyclic graph structure, adopting the parallel verification mechanism of IOTA Tangle, supporting high-concurrency microtransactions; data deposit design, key behaviors are uploaded to the chain, only storing task offloading decision hashes, node identity certificates, and resource usage audit logs; off-chain data is stored through IPFS, and only fingerprints are stored on the chain; smart contract rules, automatically trigger exception handling; Trusted execution environment integration: Hardware-level security, TEE type selection includes Intel SGX, ARM TrustZone, and domestic solutions; secure enclave function, key management never leaves the TEE through the private key of the edge node, and sensitive computing directly executes the AI model inference on encrypted data; TEE-blockchain collaboration, off-chain calculation verification, including the TEE generating zero-knowledge proofs to prove the correctness of the calculation process, and the blockchain only verifies the proofs rather than the original data; Hybrid security architecture: When node identity is forged, hardware identity binding based on TEE, each node has a unique key + blockchain CA; when data is tampered with, blockchain deposit + Merkle tree verification; when man-in-the-middle attack occurs, end-to-end encryption implemented within the TEE; when denial of service occurs, reputation-based traffic filtering, and the blockchain records the historical behavior scores of nodes.

Images