Low-power communication optimization method, system, electronic device and storage medium for IoT devices based on dynamic adaptive scheduling

Abstract

The present invention relates to the technical field of the Internet of Things, and in particular to a low-power communication optimization method, system, electronic device and storage medium for Internet of Things devices based on dynamic adaptive scheduling. The method includes the following steps: the device layer cyclically collects environmental data to generate an original data set; the device layer calculates the variance of the environmental data in the original data set based on a sliding window, and adjusts the device mode by comparing the variance and the fluctuation threshold; in active mode, the time interval for reporting data; in sleep mode, the time interval for reporting data is extended; the gateway forwards compressed data packets and network load status, and broadcasts the network load status to the device layer; the cloud optimizes dynamic parameters through a prediction model, and broadcasts the optimized dynamic parameters to the device layer through the gateway, adjusting the delay of reporting data under different modes of the device layer. The present invention can reduce device energy consumption and improve network throughput by combining a dual-mode communication mechanism with a dynamic adjustment strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of the Internet of Things, and in particular to a method, system, electronic device and storage medium for optimizing low-power communication of Internet of Things devices based on dynamic adaptive scheduling. Background Art

[0002] The Internet of Things (IoT) refers to the connection of any object to a network through information sensing devices and agreed-upon protocols. Objects exchange and communicate information through information media to achieve intelligent identification, positioning, tracking, and monitoring. In short, interconnected sensor devices report collected data to designated servers according to pre-set instructions.

[0003] However, sensor devices typically collect and report data periodically, based on a preset, fixed time period or triggered by a simple event. First, when sensors use a fixed time period, even if the environment in which they operate remains normal, they will still report a large amount of collected data to the server, resulting in the transmission of a large amount of redundant data. Second, whether using a fixed time period or a simple event trigger, a large number of sensors may report data simultaneously, triggering intense channel contention and significantly increasing packet loss rates. Third, sensors can only report data based on established communication mechanisms, making them difficult to adapt to complex and dynamic scenarios. For example, a water level sensor does not need to report water level data frequently when the water level is safe. However, when the water level approaches or has reached a dangerous level, it needs to report water level data frequently, relying on manually set parameters. Summary of the Invention

[0004] In order to adaptively adjust the data transmission cycle and communication mode and realize the joint optimization of device energy consumption and network load, the present invention provides a low-power communication optimization method, system, electronic device and storage medium for Internet of Things devices based on dynamic adaptive scheduling.

[0005] The present invention provides a low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling, which adopts the following technical solutions:

[0006] A method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling includes the following steps:

[0007] The device layer cyclically collects environmental data and generates raw data sets;

[0008] The device layer calculates the variance of the environmental data in the original data set based on the sliding window. If the variance exceeds the preset fluctuation threshold, it enters the active mode. If the variance is less than or equal to the fluctuation threshold, it enters the sleep mode.

[0009] In active mode, the device layer compresses the data in the original data set twice using different encoding methods to generate a data compression package;

[0010] In sleep mode, the device layer generates random delays to extend the time interval for reporting data;

[0011] The gateway forwards compressed data packets and network load status according to the network load, and broadcasts the network load status to the device layer;

[0012] The cloud obtains the compressed data packets and network load status forwarded by the gateway, optimizes the dynamic parameters through the prediction model, and broadcasts the optimized dynamic parameters to the device layer through the gateway to adjust the delay of reporting data in different modes of the device layer.

[0013] In a specific implementation plan, the fluctuation threshold is determined based on the historical average and the network load factor, and the calculation formula is:

[0014]

[0015] In the above formula, is the fluctuation threshold;

[0016] for environmental sensitivity;

[0017] is the historical mean;

[0018] is the network weight;

[0019] is the network load factor.

[0020] In a specific possible implementation scheme, in active mode, the device layer compresses the data in the original data set twice using different encoding methods, using difference encoding and Huffman encoding.

[0021] In a specific embodiment, the difference encoding includes the following steps:

[0022] Calculate the difference between the current environmental data and the environmental data collected last time ,like , then discard the current data, only record the timestamp, and generate a compressed data packet; if , then retain , and record and timestamp to generate compressed data packets.

[0023] In a specific implementation scheme, the gateway calculates the channel utilization before forwarding the compressed data packet based on the network load and network load status, and follows the forwarding principle:

[0024] like , allowing the device layer to report immediately;

[0025] like , random delay 0-5 seconds;

[0026] like , only high-priority devices are allowed to report.

[0027] In a specific feasible implementation plan, the prediction model is a lightweight LSTM model, which is trained with the goal of optimizing global energy consumption and network efficiency and outputs dynamic parameters.

[0028] In a specific possible implementation scheme, in sleep mode, the device layer adopts a discontinuous reception mechanism.

[0029] The present invention also provides a low-power communication optimization system for IoT devices based on dynamic adaptive scheduling, which adopts the following technical solutions:

[0030] A low-power communication optimization system for IoT devices based on dynamic adaptive scheduling, used to execute the above-mentioned low-power communication optimization method for IoT devices based on dynamic adaptive scheduling, further comprising:

[0031] Perception module, used to collect environmental data and generate raw data sets;

[0032] The dynamic adjustment module obtains the original data set of the perception module, calculates the data volatility and compares it with the volatility threshold, and adjusts the entry into sleep mode or active mode; adjusts the delay of the reported data compression package according to the dynamic parameters;

[0033] The gateway calculates channel utilization, broadcasts network load status, forwards data compression packages to the cloud, and forwards dynamic parameters to the dynamic adjustment module;

[0034] In the cloud, the data compression package forwarded by the gateway is obtained, the dynamic parameters are optimized through the prediction model, and the dynamic parameters are sent to the gateway.

[0035] The present invention also provides an electronic device, which adopts the following technical solution:

[0036] An electronic device comprises at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling.

[0037] The present invention also provides a non-transitory computer-readable storage medium storing computer instructions, which adopts the following technical solution:

[0038] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the above-mentioned low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling.

[0039] In summary, the present invention has the following beneficial effects:

[0040] 1. Environmental awareness is achieved through the device layer, network complexity is perceived through the gateway, and multi-dimensional prediction and optimization are performed using cloud-based predictive models, avoiding performance bottlenecks caused by single-dimensional optimization. Data compression in both sleep and active modes effectively reduces device energy consumption by 30%-50%. Global optimization based on network load conditions increases network throughput by nearly 20%, with the ability to dynamically adapt to the environment.

[0041] 2. The device layer dynamically adjusts modules to automatically switch between active and dormant modes, adjust the delay in reporting compressed data packets, and globally optimize the device layer and network load status through the cloud, achieving decentralized collaboration and reducing communication dependencies. An embedded-friendly simplified LSTM version reduces resource usage.

[0042] 3. Covers multi-dimensional performance indicators such as energy consumption, network, and storage; achieving significant optimization results in multiple fields such as agriculture, industry, and home. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is the flow chart of dynamic scheduling algorithm.

[0044] Figure 2 This is a schematic diagram of switching between active mode and sleep mode.

[0045] Figure 3 It is a data compression flow chart.

[0046] Figure 4 It is a system architecture diagram.

[0047] Figure 5 This is the communication sequence diagram between the device and the gateway.

[0048] Figure 6 It is the core class diagram of the system. DETAILED DESCRIPTION

[0049] The following combination Figures 1-6 The present invention is described in further detail.

[0050] This method uses the NS-3 simulation platform (simulating large-scale deployments of 1,000 nodes) for network simulation. The Keysight N6705B DC power analyzer is used for energy analysis. Python and TensorFlow are used for model training, and AWS IoT Core is used for cloud computing. The following steps are also included:

[0051] S1, the device layer cyclically collects environmental data and generates raw data sets.

[0052] Environmental data is collected through sensors, terminal devices and other hardware devices according to the preset cycle logic, and the collected environmental data is saved in the original data set according to the time sequence. express Environmental data collected at all times.

[0053] S2, the device layer calculates data volatility based on the sliding window.

[0054] Data volatility is expressed by variance, according to the preset sliding window calculate Variance at time ,in, Indicates the The sliding window length of different hardware devices is different. The sliding window is expressed as .variance The calculation formula is:

[0055]

[0056] In the above formula, represents the data in the sliding window, ;

[0057] is the average value of all data in the sliding window.

[0058] The variance With the preset fluctuation threshold For comparison, if , indicating that the data is highly volatile and there may be abnormalities. It is necessary to enter active mode and the device layer needs to report data to the gateway more frequently. If , indicating that the data volatility is small and the situation is normal. It is necessary to enter or maintain sleep mode to reduce the frequency of data reporting from the device layer to the gateway.

[0059] Fluctuation threshold Determined by the historical average and network load factor, the calculation formula is:

[0060]

[0061] In the above formula, is the environmental sensitivity, which is initially set to 0.3 in this embodiment;

[0062] is the historical mean;

[0063] is the network weight, which is initially set to 0.7 in this embodiment;

[0064] is the network load factor.

[0065] In S3, active mode, the device layer compresses the environmental data in the original data set through difference coding and Huffman coding.

[0066] Calculate the difference between the current environmental data and the environmental data collected last time, that is:

[0067]

[0068] In the above formula, That is the difference between the two environmental data, if , then discard the current data, only record the timestamp, and generate a compressed data packet; if , then retain , and record The device layer then generates a compressed data packet with the data and timestamp. This data packet is further compressed using Huffman coding. The device layer then sends the compressed data packet to the gateway. Because the data packet has undergone two compressions, it is smaller and reported faster, allowing the device layer to report more data packets to the gateway in the same amount of time. Furthermore, the device layer can report the compressed data packet immediately, further shortening the device layer's communication cycle.

[0069] In S4, sleep mode, the device layer uses the discontinuous reception (DRX) mechanism and only listens to gateway broadcasts in a specific time window.

[0070] Discontinuous Reception (DRX) is a technology used to optimize the energy consumption of terminal devices in mobile communication systems. Its core idea is to balance the real-time communication needs of devices with power consumption control through a periodic sleep and wake-up mechanism, thereby extending battery life and improving system resource utilization.

[0071] The device layer generates a random delay time to extend the time interval for the device layer to report data, so that the frequency of the device layer reporting to the gateway is reduced. That is, within the same time, the number of data packets reported by the device layer to the gateway is reduced.

[0072] It is understandable that no matter in active mode or sleep mode, the device layer will report the data compression package to the gateway. The difference is that in active mode, in order to improve data efficiency, two compressions are used to make the compressed data packet smaller, while in sleep mode, only one compression is used.

[0073] S5, the gateway monitors the network load status.

[0074] Calculating channel utilization According to the pre-selected forwarding principle, the gateway will forward the compressed data packet and network load status to the cloud. The forwarding principle is as follows:

[0075] like , allowing the device layer to report immediately;

[0076] like , random delay of 0-5 seconds, that is, based on the original delay of reporting data compression package at the device layer, it is further extended by 0-5 seconds;

[0077] like Only high-priority devices are allowed to report, including fire alarms and earthquake alarms. The gateway sends the compressed data package and network load status to the cloud based on the channel utilization.

[0078] S6, predictive model training is performed in the cloud and network efficiency is optimized.

[0079] In this embodiment, the prediction model is a lightweight LSTM model, which is an embedded-friendly simplified version of the LSTM and uses less than 5% of the resources. The model architecture is shown in Table 1.

[0080] Table 1

[0081]

[0082] The LSTM model is trained to optimize global energy consumption and network efficiency based on historical data, compressed packets forwarded by the gateway, and network load. It outputs dynamic parameters used to update loop logic, environmental sensitivity, and network weights. The cloud sends these dynamic parameters to the device layer via over-the-air (OTA). Upon receiving these dynamic parameters, the device layer automatically updates loop logic and fluctuation thresholds.

[0083] For ease of understanding, further explanation is provided below through Examples 1-3.

[0084] Example 1:

[0085] Smart Agriculture - Temperature and Humidity Monitoring Scenario

[0086] Hardware equipment:

[0087] Sensor: SHT30 temperature and humidity sensor (sampling accuracy ±0.2°C, ±2%RH);

[0088] Communication module: LoRaWAN terminal (Semtech SX1276, operating frequency band 868MHz);

[0089] Gateway: MultiTech Conduit AP 868.

[0090] Software Configuration:

[0091] Dynamic scheduling algorithm parameters: sliding window length = 10 minutes, fluctuation threshold = 0.5°C.

[0092] Data compression: difference coding + Huffman coding.

[0093] Experimental results:

[0094] Table 2

[0095]

[0096] Data source:

[0097] Test period: June 1-June 30, 2023 (greenhouse environment, average daily temperature fluctuation of 1.5°C).

[0098] Data collection tools: Wireshark packet capture analysis + device-side energy consumption monitoring chip (TIINA219).

[0099] In Table 2, ↓ indicates a decrease. Analysis of the data in Table 2 shows that by adopting the above-mentioned low-power communication optimization method for IoT devices based on dynamic adaptive scheduling, the average daily energy consumption, average daily data transmission volume and network packet loss rate are effectively reduced, and the system performance is effectively improved.

[0100] Example 2:

[0101] Industrial Internet of Things - Motor Vibration Monitoring Scenario

[0102] Hardware equipment:

[0103] Sensor: ADI ADXL356 vibration sensor (sampling rate 1kHz);

[0104] Edge computing unit: Raspberry Pi 4B (running a lightweight LSTM model);

[0105] Communication protocol: MQTT over Wi-Fi (priority channel allocation).

[0106] Dynamic scheduling strategy:

[0107] Active mode: When the vibration amplitude > 5g is detected, the original waveform data will be reported immediately;

[0108] Sleep mode: When the vibration amplitude is less than 2g, the statistical value (mean + variance) is reported every 30 minutes.

[0109] Experimental results:

[0110] Table 3

[0111]

[0112] Test method:

[0113] Fault simulation: Inject wear particles into motor bearings and record the delay from abnormal vibration to cloud-based alarm.

[0114] Model accuracy: LSTM predicts vibration trends for the next 5 minutes with a RMSE (root mean square error) of 0.18g.

[0115] By analyzing the data in Table 3, it can be seen that after adopting the above-mentioned low-power communication optimization method for IoT devices based on dynamic adaptive scheduling, the fault detection delay, false alarm rate and average monthly communication traffic have all been significantly reduced, and less resources can be consumed to monitor motor vibration more accurately.

[0116] Example 3:

[0117] Smart Home – Security Camera Network

[0118] Hardware equipment:

[0119] Equipment: Hikvision DS-2CD2043G0-I camera (1080P, motion detection function);

[0120] Gateway: Huawei HiLink Smart Gateway (supports Wi-Fi 6).

[0121] Dynamic scheduling strategy:

[0122] Active mode: When motion is detected, it immediately switches to 1080P real-time video streaming;

[0123] Sleep mode: When no one is moving, one low-resolution image (240P) is uploaded every 10 minutes.

[0124] Experimental results:

[0125] Table 4

[0126]

[0127] Data Validation:

[0128] Test scenario: simulates a home environment, triggering motion events 20 times per day;

[0129] Compression algorithm: H.265 encoding + dynamic frame rate adjustment, bandwidth usage reduced by 45%.

[0130] Analysis of the data in Table 4 shows that the combination of active and dormant modes significantly reduces monthly storage usage. Less data reporting results in lower network latency during emergencies, allowing users to be informed more promptly. Furthermore, lower daily device heat generation extends device lifespan.

[0131] The data of Examples 1-3 can be combined to obtain Table 5:

[0132] Table 5

[0133]

[0134] The present invention also discloses a low-power communication optimization system for Internet of Things devices based on dynamic adaptive scheduling, comprising:

[0135] It includes a perception module, which is used to collect environmental data through sensors and terminals and generate raw data sets;

[0136] The dynamic scheduling module obtains the original data set of the perception module, calculates the data volatility and compares it with the volatility threshold, and adjusts the mode to enter the sleep mode or active mode;

[0137] The gateway is used to broadcast network load status and calculate channel utilization, obtain compressed data packets and network load status from the dynamic scheduling module, and report them to the cloud;

[0138] In the cloud, the compressed data packets and network load status reported by the gateway are obtained, and data analysis and model training are performed. The dynamic parameters after model training are output to the dynamic adjustment module to adjust the reporting cycle and compressed data packets of sensors and terminals.

[0139] The present invention also discloses an electronic device, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so as to enable the at least one processor to execute the above-mentioned low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling.

[0140] The present invention also discloses a non-transient computer-readable storage medium storing computer instructions, which are used to enable a computer to execute the above-mentioned low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling.

[0141] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A low-power communication optimization method for IoT devices based on dynamic adaptive scheduling, characterized by: The steps include: The device layer cyclically collects environmental data and generates raw data sets; The device layer calculates the variance of the environmental data in the original data set based on the sliding window. If the variance exceeds the preset fluctuation threshold, it enters the active mode. If the variance is less than or equal to the fluctuation threshold, it enters the sleep mode. In active mode, the device layer compresses the data in the original data set twice using different encoding methods to generate a data compression package; In sleep mode, the device layer generates random delays to extend the time interval for reporting data; The gateway forwards compressed data packets and network load status according to the network load, and broadcasts the network load status to the device layer; The cloud obtains the compressed data packets and network load status forwarded by the gateway, optimizes the dynamic parameters through the prediction model, and broadcasts the optimized dynamic parameters to the device layer through the gateway to adjust the delay of reporting data in different modes of the device layer.

2. The method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling according to claim 1, characterized in that: In active mode, the device layer compresses the data in the original data set twice through different coding methods, using difference coding and Huffman coding.

3. The method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling according to claim 2, characterized in that: The difference encoding includes the following steps: Calculate the difference between the current environmental data and the environmental data collected last time ,like , then discard the current data, only record the timestamp, and generate a compressed data packet; like , then retain , and record and timestamp to generate compressed data packets.

4. The method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling according to claim 1, characterized in that: Before forwarding compressed data packets based on network load and network load status, the gateway calculates channel utilization and follows the forwarding principles: like , allowing the device layer to report immediately; like , random delay 0-5 seconds; like , only high-priority devices are allowed to report.

5. The method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling according to claim 1, characterized in that: The prediction model is a lightweight LSTM model, which is trained with the goal of optimizing global energy consumption and network efficiency and outputs dynamic parameters.

6. The method for optimizing low-power communication of IoT devices based on dynamic adaptive scheduling according to claim 1, characterized in that: In sleep mode, the device layer adopts a discontinuous reception mechanism.

7. A low-power communication optimization system for IoT devices based on dynamic adaptive scheduling, characterized by: The method for optimizing low-power communication of an IoT device based on dynamic adaptive scheduling according to any one of claims 1 to 6 further includes: Perception module, used to collect environmental data and generate raw data sets; The dynamic adjustment module obtains the original data set of the perception module, calculates the data volatility and compares it with the volatility threshold, and adjusts the entry into sleep mode or active mode; adjusts the delay of the reported data compression package according to the dynamic parameters; The gateway calculates channel utilization, broadcasts network load status, forwards data compression packages to the cloud, and forwards dynamic parameters to the dynamic adjustment module; In the cloud, the data compression package forwarded by the gateway is obtained, the dynamic parameters are optimized through the prediction model, and the dynamic parameters are sent to the gateway.

8. An electronic device, characterized in that: The invention comprises at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by at least one of the processors, and the instructions are executed by at least one of the processors to enable at least one of the processors to execute the low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute the low-power communication optimization method for Internet of Things devices based on dynamic adaptive scheduling as described in any one of claims 1-6.

Images