A method and system for multi-dimensional intelligent service scheduling of Internet of Things (IoT) devices

Abstract

This invention discloses a multi-dimensional intelligent service scheduling method and system for IoT devices, comprising the following steps: S1, request reception and feature extraction: receiving service scheduling requests and extracting scheduling feature information from the requests; S2, multi-dimensional scheduling rule matching: according to the preset priority order of the scheduling feature information, sequentially matching and querying the information in multiple dimension scheduling rule bases. Once a match is successfully found in a certain dimension, the matching process terminates and enters the service allocation stage. This invention has the advantages of accurate and efficient scheduling, solving the problem that existing IoT devices are not suitable for accurate and efficient scheduling of various different IoT devices during service scheduling, which easily leads to scheduling failures and inconsistencies.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, specifically to a multi-dimensional intelligent service scheduling method and system for IoT devices. Background Technology

[0002] Internet of Things (IoT) devices are physical entities capable of data collection, transmission, interaction, and control via the internet or other communication networks. They typically embed sensors, processors, communication modules, and software systems, enabling them to sense their environment, process information, and communicate online. These devices are widely used in smart homes, industrial automation, smart cities, connected vehicles, healthcare, and agricultural monitoring. Typical examples include smart cameras, smart locks, temperature and humidity sensors, wearable devices, smart meters, connected appliances, and various industrial controllers. IoT devices can connect to cloud platforms or edge computing nodes via Wi-Fi, Bluetooth, Zigbee, and 4G / 5G cellular networks to achieve remote monitoring, data analysis, automated decision-making, and collaborative operation, thereby building an intelligent ecosystem. With the rapid development of IoT technology, hundreds of millions of IoT devices urgently need to connect to cloud services. These devices are highly diverse in type, widely distributed geographically, and have significantly different access terminals, which places higher and more refined requirements on the backend service scheduling system. Currently, the commonly used scheduling schemes in the industry mainly include static configuration, DNS round-robin, and regional scheduling. However, these existing technologies have many limitations, such as overly coarse scheduling granularity, lack of session consistency guarantees, lagging fault response mechanisms, rigid configuration update mechanisms, and failure to fully consider terminal heterogeneity. Therefore, there is an urgent need for an intelligent service scheduling method and system that supports multi-dimensional, fine-grained, highly available, and dynamically adaptable intelligent service scheduling to address the challenges of large-scale, heterogeneous, and highly reliable service access in the current Internet of Things (IoT). Summary of the Invention

[0003] The purpose of this invention is to provide a multi-dimensional intelligent service scheduling method and system for IoT devices, which has the advantages of accurate and efficient scheduling. It solves the problem that existing IoT devices are not easy to accurately and efficiently schedule multiple different IoT devices during service scheduling, which can easily lead to scheduling failures and inconsistencies. To achieve the above objectives, the present invention provides the following technical solution: a multi-dimensional intelligent service scheduling method for Internet of Things (IoT) devices, comprising the following steps: S1. Request reception and feature extraction: Receive service scheduling requests and extract scheduling feature information from the requests; S2. Multi-dimensional scheduling rule matching: According to the preset priority order of scheduling feature information, the information is matched and queried in the scheduling rule base of multiple dimensions in turn. Once a match is successfully found in a certain dimension, the matching process is terminated and the service allocation stage is entered. S3. Intelligent service node allocation: After a successful match, a two-level allocation algorithm is used to determine the specific service node. S4. Health status verification and failover: Before returning the allocation result, verify the health status of the service node. S5. Differentiated service configuration assembly: Based on the type of service scheduling request, assemble differentiated service configuration responses to respond to different types of service scheduling requests. S6. Response return: Return the assembled service configuration information to the service scheduling requester to complete one scheduling process. In a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices according to the present invention, the service scheduling request in S1 is initiated by the IoT device or user terminal, and the scheduling feature information extracted from the request includes the device's unique identifier, the SIM card integrated circuit identifier, the client's network address, and the geographical location information. In a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices according to the present invention, the multiple dimensions in S2 include a first dimension, a second dimension, a third dimension, and a fourth dimension. The first dimension has the highest priority, and the priorities of the second, third, and fourth dimensions decrease in that order. The first dimension is precise matching of the device's unique identifier, querying whether there is a dedicated scheduling rule for the unique identifier of the device. The second dimension is matching of the SIM card integrated circuit identifier, querying whether there is a dedicated scheduling rule for the SIM card integrated circuit identifier. The third dimension is matching of the client's network address, querying whether the client's network address IP falls into a preset scheduling rule. The fourth dimension is matching of the geographical region, matching the regional scheduling rule according to the geographical location information. As a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices of the present invention, the scheduling rules in S2 are stored in a relational database and cached in the memory of the service process in the form of key-value pairs to meet the high-concurrency scheduling requirements. The background daemon process periodically loads rules from the database, calculates the hash digest of the rule set, and compares it with the digest of the last loaded rule set. If the digests are the same, it means that the rules have not changed and the update is skipped. If the digests are different, a new cache structure is constructed and the cache is updated by atomic replacement. As a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices of the present invention, the first-level allocation algorithm in S3 is a weighted selection algorithm, which selects a group of nodes from the candidate service node group according to the weight ratio. The second-level allocation algorithm is a deterministic allocation algorithm, which uses the hash value of the device identifier to deterministically select a specific node within the group, ensuring that the same device is always assigned to the same node. As a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices of the present invention, when the weighted selection algorithm is used for calculation, it is assumed that there are n candidate service node groups, and each group i corresponds to a non-negative weight. Next, calculate the total weight of all groups. Then, a uniformly distributed random number r is generated within the interval. Finally, each group is traversed sequentially, and its weight value is accumulated. When the accumulated sum is greater than r for the first time, the current group is selected as the target service node group. When the deterministic allocation algorithm is calculated, a cryptographic hash function is first applied to the unique identifier D of the device to convert it into a hash value of fixed length, and then the hash value is further interpreted as a non-negative integer H(D). Then, the integer is modulo the number of healthy and available nodes N in the current group, that is, the node_index=H(D)modN is calculated. The result is the 0-starting index of the target node in the group. Since the hash function is deterministic, as long as the number of nodes N in the group remains unchanged, any device identifier D will always be mapped to the same node index, thus satisfying the strong consistency requirement in the scheduling process. In a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices of the present invention, the health of the verification target node in S4 is directly returned and fed back. If the verification target node is unhealthy, a failover is triggered, and a replacement node is selected from other healthy nodes in the same group. The replacement node is then health-verified until a healthy node is matched. If all nodes are unhealthy, the initial default node is returned. As a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices of the present invention, in step S5, the device end and the user end are distinguished by the request source identifier, WiFi devices and cellular devices are distinguished by the validity of the SIM card integrated circuit identifier and the device type field, domestic requests and overseas requests are distinguished by the geographical location information, different distribution channels are distinguished by the application store identifier, the service configuration is dynamically assembled, and independent configuration and optimization are performed for specific terminal types. In a preferred embodiment of the multi-dimensional intelligent service scheduling method for IoT devices according to the present invention, after the scheduling is completed in S6, the service scheduling takes effect on the client. After the device receives the response, it establishes a long connection using the returned P2P address. All subsequent communications are directed to this node. Deterministic hashing is used to ensure that the next scheduling will still point to the same node. After the user receives the response, it calls the returned API address to obtain the device status and loads UI resources from the CDN. A multi-dimensional intelligent service scheduling system for Internet of Things (IoT) devices includes: The request access layer is responsible for receiving scheduling requests from IoT devices or user terminals, parsing parameters and identifying geographical locations, and serving as the unified entry point for the system to the outside world. The scheduling decision layer dynamically decides and allocates the optimal service node based on multi-dimensional rules and intelligent load balancing algorithms. The data persistence layer persistently stores scheduling rules, service node configurations, and terminal type mapping data, providing data support for scheduling decisions. The request access layer and the scheduling decision layer communicate using the gRPC protocol, which supports binary serialization and bidirectional streaming. The scheduling decision layer and the data persistence layer use a database connection pool to support high-concurrency reads. Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves precise multi-granularity scheduling from single devices to regions through a four-dimensional priority progressive scheduling decision mechanism. It solves the problems of coarse scheduling granularity and inability to customize services in traditional solutions. It supports the isolation of problematic devices, provides exclusive channels for VIP customers, and enables device-level canary releases. It also ensures that the time complexity of scheduling decisions is greatly reduced. At the same time, it uses a deterministic hash algorithm to stably map device identifiers to fixed service nodes, effectively ensuring session consistency and avoiding state loss and connection interruption caused by DNS round-robin. This makes P2P long connections more stable, eliminates the need for repeated authentication for users, and provides a smoother experience. 2. This invention combines proactive health detection with intelligent degradation strategies to achieve second-level fault detection and automatic failover, completely eliminating reliance on DNS cache timeouts, significantly shortening fault recovery time, and supporting automatic node regression. At the configuration level, a hot-update mechanism based on incremental hashing is introduced, enabling scheduling rule changes to take effect within seconds without requiring service restarts or waiting for cache refreshes, balancing agile operation with system stability. Furthermore, the system can sense terminal types and implement differentiated service strategies, dynamically matching the optimal connection solution for different devices, such as providing P2P services, HTTP DNS, CDN acceleration, or carrier-optimized channels, thereby optimizing overall connection quality and user experience. Attached Figure Description

[0004] Figure 1 This is a flowchart of the scheduling method of the present invention; Figure 2 This is a framework diagram of the scheduling system of the present invention; Figure 3 This is a flowchart of the multi-dimensional scheduling strategy of the present invention; Figure 4 This is a flowchart of the intelligent load balancing process of the present invention; Figure 5 A hot update flowchart is configured for this invention. Detailed Implementation

[0005] Example 1 Please see Figures 1-5A multi-dimensional intelligent service scheduling method for IoT devices includes the following steps: S1. Request reception and feature extraction: Receive service scheduling requests and extract scheduling feature information from the requests; S2. Multi-dimensional scheduling rule matching: According to the preset priority order of scheduling feature information, the information is matched and queried in the scheduling rule base of multiple dimensions in turn. Once a match is successfully found in a certain dimension, the matching process is terminated and the service allocation stage is entered. S3. Intelligent service node allocation: After a successful match, a two-level allocation algorithm is used to determine the specific service node. S4. Health status verification and failover: Before returning the allocation result, verify the health status of the service node. S5. Differentiated service configuration assembly: Based on the type of service scheduling request, assemble differentiated service configuration responses to respond to different types of service scheduling requests. S6. Response return: Return the assembled service configuration information to the service scheduling requester to complete one scheduling process. Furthermore, the service scheduling request in S1 is initiated by the IoT device or user terminal, and the scheduling feature information extracted from the request includes the device's unique identifier, SIM card integrated circuit identifier, client network address, and geographical location information. Furthermore, S2 includes multiple dimensions, including the first, second, third, and fourth dimensions. The first dimension has the highest priority, while the priorities of the second, third, and fourth dimensions decrease in that order. The first dimension is a precise match of the device's unique identifier, which checks whether there is a dedicated scheduling rule for that unique device identifier. The second dimension is a match of the SIM card's integrated circuit identifier, which checks whether there is a dedicated scheduling rule for that SIM card's integrated circuit identifier. The third dimension is a match of the client's network address, which checks whether the client's network address IP falls into a preset scheduling rule. The fourth dimension is a match of the geographical region, which matches the regional scheduling rule based on the geographical location information. Furthermore, the scheduling rules in S2 are stored in a relational database and cached in the service process memory as key-value pairs to meet the high-concurrency scheduling requirements. The background daemon periodically loads rules from the database, calculates the hash digest of the rule set, and compares it with the digest of the last loaded rule set. If the digests are the same, it means that the rules have not changed and the update is skipped. If the digests are different, a new cache structure is built and the cache is updated by atomic replacement. Furthermore, the first-level allocation algorithm in S3 is a weighted selection algorithm, which selects a group of nodes from the candidate service node group according to the weight ratio. The second-level allocation algorithm is a deterministic allocation algorithm, which uses the hash value of the device identifier to deterministically select a specific node within the group, ensuring that the same device is always assigned to the same node. Furthermore, when the weighted selection algorithm performs the calculation, it assumes there are n candidate service node groups, and each group i corresponds to a non-negative weight. Next, calculate the total weight of all groups. Then, a uniformly distributed random number r is generated within the interval. Finally, each group is traversed sequentially, and its weight value is accumulated. When the accumulated sum is greater than r for the first time, the current group is selected as the target service node group. When the deterministic allocation algorithm is calculated, a cryptographic hash function is first applied to the unique identifier D of the device to convert it into a hash value of fixed length, and then the hash value is further interpreted as a non-negative integer H(D). Then, the integer is modulo the number of healthy and available nodes N in the current group, that is, the node_index=H(D)modN is calculated. The result is the 0-starting index of the target node in the group. Since the hash function is deterministic, as long as the number of nodes N in the group remains unchanged, any device identifier D will always be mapped to the same node index, thus satisfying the strong consistency requirement in the scheduling process. Furthermore, in S4, if the target node is healthy, it will directly return and report the node's health. If the target node is unhealthy, a failover will be triggered, and a replacement node will be selected from other healthy nodes in the same group. The replacement node will then undergo health checks until a healthy node is found. If all nodes are unhealthy, the initial default node will be returned. Furthermore, S5 distinguishes between device and user ends by request source identifier, distinguishes between WiFi and cellular devices by SIM card integrated circuit identifier validity and device type field, distinguishes between domestic and overseas requests by geolocation information, distinguishes between different distribution channels by app store identifier, dynamically assembles service configurations, and independently configures and optimizes for specific terminal types. Furthermore, after scheduling is completed in S6, the service scheduling takes effect on the client. After the device receives the response, it establishes a long connection using the returned P2P address. All subsequent communication is directed to this node. Deterministic hashing is used to ensure that the next scheduling will still point to the same node. After the user receives the response, it calls the returned API address to obtain the device status and loads UI resources from the CDN. By constructing a multi-dimensional, priority-based, and rule-driven intelligent scheduling mechanism, the accuracy and flexibility of IoT service scheduling are significantly improved. During the scheduling process, rules are matched sequentially based on four dimensions: unique device identifier, SIM card identifier, IP address, and geographical region. This ensures that high-priority dedicated policies take effect first. At the same time, combined with a two-level node allocation algorithm, weighted selection achieves on-demand load distribution, and hash modulo ensures session consistency. This approach balances the utilization of cluster resources while avoiding connection jitter and state loss caused by frequent switching of service nodes by devices. In addition, the scheduling rules adopt memory caching, database persistence, and incremental update mechanisms, which support high-concurrency scheduling while ensuring the real-time nature of configuration changes and system stability. Furthermore, this solution enables dynamic assembly and precise delivery of differentiated service configurations for heterogeneous terminals. By identifying the request source, network type, geographical location, and application distribution channel characteristics, the system can provide highly customized service addresses for device terminals, user terminals, domestic and international users, or users from different operators. This optimizes connection quality, reduces latency, and improves user experience. Combined with health checks and automatic fault transfer mechanisms, even in the event of some node anomalies, it can quickly switch to available nodes to ensure service continuity. Overall, this scheduling method not only improves the service elasticity and operational efficiency of the IoT platform but also provides solid technical support for large-scale, multi-scenario, and highly reliable device access and interaction. Example 2 Please see Figure 2 A multi-dimensional intelligent service scheduling system for Internet of Things (IoT) devices, comprising: The request access layer is responsible for receiving scheduling requests from IoT devices or user terminals, parsing parameters and identifying geographical locations, and serving as the unified entry point for the system to the outside world. The scheduling decision layer dynamically decides and allocates the optimal service node based on multi-dimensional rules and intelligent load balancing algorithms. The data persistence layer persistently stores scheduling rules, service node configurations, and terminal type mapping data, providing data support for scheduling decisions. The request access layer and the scheduling decision layer communicate using the gRPC protocol, which supports binary serialization and bidirectional streaming. The scheduling decision layer and the data persistence layer use a database connection pool to support high-concurrency reads. Through a layered architecture design, the IoT device scheduling system achieves high cohesion, low coupling, and high-performance collaboration. The request access layer serves as a unified entry point, efficiently processing scheduling requests from massive heterogeneous terminals and completing parameter parsing and geographic location identification. The scheduling decision layer relies on a multi-dimensional rule engine and intelligent load balancing algorithm to complete precise node allocation within milliseconds. The data persistence layer provides a stable and scalable configuration data foundation for the entire scheduling logic. The three layers communicate using the gRPC protocol and a database connection pool, fully leveraging the advantages of gRPC in binary serialization, low-latency calls, and bidirectional streaming. At the same time, the connection pool improves database access efficiency, effectively supporting the real-time scheduling requirements of high concurrency and low latency, and significantly enhancing the system's throughput and response speed. Furthermore, this architecture boasts excellent maintainability and scalability. With clearly defined responsibilities and interfaces at each layer, changes to business rules only require configuration updates at the scheduling decision layer or data persistence layer, without altering the access logic. The strong-typed contracts of gRPC also reduce the risk of errors in cross-service calls. Simultaneously, memory caching and incremental synchronization mechanisms can be seamlessly integrated with this layered structure, further ensuring hot configuration updates and system stability. Overall, this system not only meets the core requirements of IoT scenarios for precise device-level scheduling, session consistency, and fault self-healing, but also provides a flexible and reliable architectural foundation for supporting more terminal types, network environments, and business scenarios in the future. The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional intelligent service scheduling method for Internet of Things (IoT) devices, characterized in that, Includes the following steps: S1. Request reception and feature extraction: Receive service scheduling requests and extract scheduling feature information from the requests; S2. Multi-dimensional scheduling rule matching: According to the preset priority order of scheduling feature information, the information is matched and queried in the scheduling rule base of multiple dimensions in turn. Once a match is successfully found in a certain dimension, the matching process is terminated and the service allocation stage is entered. S3. Intelligent service node allocation: After a successful match, a two-level allocation algorithm is used to determine the specific service node. S4. Health status verification and failover: Before returning the allocation result, verify the health status of the service node. S5. Differentiated service configuration assembly: Based on the type of service scheduling request, assemble differentiated service configuration responses to respond to different types of service scheduling requests. S6. Response return: The assembled service configuration information is returned to the service scheduling requester, completing one scheduling process. The service scheduling request in S1 is initiated by an IoT device or user terminal. The scheduling feature information extracted from the request includes the device's unique identifier, SIM card integrated circuit identifier, client network address, and geographic location information. The multiple dimensions in S2 include a first dimension, a second dimension, a third dimension, and a fourth dimension. The first dimension has the highest priority, and the priorities of the second, third, and fourth dimensions decrease in that order. The first dimension is a precise match of the device's unique identifier, which checks whether there is a dedicated scheduling rule for that unique device identifier. The second dimension is a match of the SIM card's integrated circuit identifier, which checks whether there is a dedicated scheduling rule for that SIM card's integrated circuit identifier. The third dimension is a match of the client's network address, which checks whether the client's network address IP falls into a preset scheduling rule. The fourth dimension is a match of the geographical region, which matches the regional scheduling rule based on the geographical location information. The first-level allocation algorithm in S3 is a weighted selection algorithm, which selects a group of nodes from the candidate service node group according to the weight ratio. The second-level allocation algorithm is a deterministic allocation algorithm, which uses the hash value of the device identifier to deterministically select a specific node within the group, ensuring that the same device is always assigned to the same node. In S5, the device end and the user end are distinguished by the request source identifier, WiFi devices and cellular devices are distinguished by the validity of the SIM card integrated circuit identifier and the device type field, domestic requests and overseas requests are distinguished by the geographical location information, and different distribution channels are distinguished by the application store identifier. Service configuration is dynamically assembled and independently configured and optimized for specific terminal types.

2. The method for multi-dimensional intelligent service scheduling of IoT devices according to claim 1, characterized in that: The scheduling rules in S2 are stored in a relational database and cached in the service process memory in the form of key-value pairs to meet the high-concurrency scheduling requirements. The background daemon periodically loads rules from the database, calculates the hash digest of the rule set, and compares it with the digest of the last loaded rule set. If the digests are the same, it means that the rules have not changed and the update is skipped. If the digests are different, a new cache structure is constructed and the cache is updated by atomic replacement.

3. The method for multi-dimensional intelligent service scheduling of IoT devices according to claim 1, characterized in that: When the weighted selection algorithm is used for calculation, it is assumed that there are n candidate service node groups, and each group i corresponds to a non-negative weight. Next, calculate the total weight of all groups. Then, a uniformly distributed random number r is generated within the interval. Finally, each group is traversed sequentially, and its weight value is accumulated. When the accumulated sum is greater than r for the first time, the current group is selected as the target service node group. When the deterministic allocation algorithm is calculated, a cryptographic hash function is first applied to the unique identifier D of the device to convert it into a hash value of fixed length, and then the hash value is further interpreted as a non-negative integer H(D). Then, the integer is modulo the number of healthy and available nodes N in the current group, that is, the node_index=H(D)modN is calculated. The result is the 0-starting index of the target node in the group. Since the hash function is deterministic, as long as the number of nodes N in the group remains unchanged, any device identifier D will always be mapped to the same node index, thus satisfying the strong consistency requirement in the scheduling process.

4. The method for multi-dimensional intelligent service scheduling of IoT devices according to claim 1, characterized in that: If the target node being verified in S4 is healthy, the health status of the node is directly returned and fed back. If the target node being verified is unhealthy, a failover is triggered. A replacement node is selected from other healthy nodes in the same group, and the health of the replacement node is verified until a healthy node is matched. If all nodes are unhealthy, the initial default node is returned.

5. The method for multi-dimensional intelligent service scheduling of IoT devices according to claim 1, characterized in that: After the scheduling in S6 is completed, the service scheduling takes effect on the client. After the device receives the response, it establishes a long connection using the returned P2P address. All subsequent communications are directed to this node. Deterministic hashing is used to ensure that the next scheduling will still point to the same node. After the user receives the response, it calls the returned API address to obtain the device status and loads UI resources from the CDN.

6. A multi-dimensional intelligent service scheduling system for Internet of Things (IoT) devices, applicable to the multi-dimensional intelligent service scheduling method for IoT devices as described in any one of claims 1-5, characterized in that, include: The request access layer is responsible for receiving scheduling requests from IoT devices or user terminals, parsing parameters and identifying geographical locations, and serving as the unified entry point for the system to the outside world. The scheduling decision layer dynamically decides and allocates the optimal service node based on multi-dimensional rules and intelligent load balancing algorithms. The data persistence layer persistently stores scheduling rules, service node configurations, and terminal type mapping data, providing data support for scheduling decisions. The request access layer and the scheduling decision layer communicate using the gRPC protocol, which supports binary serialization and bidirectional streaming. The scheduling decision layer and the data persistence layer use a database connection pool to support high-concurrency reads.

Images