An encryption method of artificial intelligence data

Abstract

The present application relates to the field of data encryption, and more particularly to an artificial intelligence data encryption method, comprising: analyzing the action execution intensity and the perception data richness of the intelligent device to determine whether to perform encryption enhancement processing; in the encryption enhancement processing, determining the encryption block interval adjustment based on the action complexity or the number of environmental targets based on the corresponding space-time constraint reference value of the intelligent device; determining whether to perform encryption upload optimization according to the scene encryption pressure and the number of encryption anchor selection devices; in the encryption upload optimization, determining the order adjustment self-optimization of the upload sequence or recording the transmission channel as a to-be-analyzed optimization channel for cluster optimization based on the task coordination depth and the route intersection degree corresponding to the upload sequence of the transmission channel; the present application reduces the indirect leakage risk of data through the encryption security coordination of device data.

Description

Technical Field

[0001] This invention relates to the field of data encryption, and more particularly to a method for encrypting artificial intelligence data. Background Technology

[0002] With the widespread application of robots, smart mobile devices, and various artificial intelligence terminals, data encryption, as a key technology for ensuring system security and preventing the leakage of sensitive information, has become a research hotspot. Especially in multi-agent collaborative operation scenarios, such as unmanned workshops, warehousing and logistics, and emergency rescue, intelligent devices need to frequently collect environmental perception data, execute control commands, and periodically upload data to the cloud or central server for analysis and decision-making. Existing technologies for data encryption of intelligent devices mainly adopt a fixed-period full-data encryption upload mode. This means that regardless of whether the device is currently in a highly active task state or an idle standby state, it is encrypted and transmitted with the same encryption strength and the same data block interval. However, this approach suffers from uneven utilization of encryption resources and a lack of secure coordination in multi-device transmission scheduling. How to improve the adaptive capability of data encryption in artificial intelligence terminals to achieve improved encryption effectiveness is a technical problem that urgently needs to be solved by those skilled in the art.

[0003] Chinese Patent Publication No. CN120128398B discloses a method, system, and terminal for encrypted transmission of multimodal data from a composite robot. The method involves data transmission and includes: acquiring collection information from the composite robot; classifying and encrypting the collection information using a preset encryption algorithm to obtain encrypted information; acquiring data retrieval information from the composite robot; determining whether the data retrieval information is consistent with preset baseline retrieval information; if the data retrieval information is inconsistent with the baseline retrieval information, continuing to acquire data retrieval information; if the data retrieval information is consistent with the baseline retrieval information, encrypting the data retrieval information using a preset encryption algorithm to obtain encrypted retrieval information; decrypting the encrypted information based on the encrypted retrieval information to obtain target collection information and outputting it to a preset retrieval platform. It is evident that this solution has the following problems: in scenarios where multiple robots operate collaboratively, there may be multiple robots performing tasks or having related sensory data, such as two robots performing the same handling task. This high degree of correlation allows attackers to infer the collaborative relationship between different devices through traffic analysis, packet size comparison, and other means, leading to indirect leakage of related data. Summary of the Invention

[0004] Therefore, the present invention provides an encryption method for artificial intelligence data to overcome the problems faced by existing technologies in the encryption of data of artificial intelligence terminals, such as the inability to perform targeted analysis based on the associated data of multiple robots working together, and the risk of indirect data leakage caused by the lack of secure coordination in the transmission scheduling of encrypted data of devices.

[0005] To achieve the above objectives, the present invention provides a method for encrypting artificial intelligence data, comprising: Analyze the intensity of actions performed by smart devices and the richness of perceived data to determine whether to perform encryption enhancement processing; In the encryption enhancement process, the interval between encryption blocks is adjusted based on the spatiotemporal constraint reference value corresponding to the smart device, and based on the complexity of the action or the number of environmental targets. Whether to perform encrypted upload optimization depends on the encryption pressure of the scenario and the number of encryption anchor devices. In encrypted upload optimization, the order of upload sequences is adjusted based on the task collaboration depth and route intersection degree corresponding to the upload sequence of the transmission channel, or the transmission channel is recorded as the channel to be analyzed and optimized for cluster optimization. In cluster optimization, the number of associated channels corresponding to each channel to be analyzed and optimized is obtained, and the collaborative optimization processing or encrypted upload interval adjustment of the channel to be analyzed and optimized is determined based on the average number of associated channels and the difference in the number of associated channels.

[0006] Furthermore, for smart devices with an action execution density greater than the action execution density threshold or a perception data richness greater than the perception data richness threshold, encryption enhancement processing is determined to be performed.

[0007] Furthermore, the encryption enhancement process includes: Detect the spatiotemporal constraint reference values ​​corresponding to intelligent devices; If the spatiotemporal constraint reference value is less than the preset spatiotemporal constraint reference value, the encrypted block interval will be reduced based on the action complexity. If the spatiotemporal constraint reference value is greater than or equal to the preset spatiotemporal constraint reference value, the encryption block interval is reduced based on the number of environmental targets.

[0008] Furthermore, based on the condition that the scene encryption pressure is greater than the preset scene encryption pressure or the number of encryption anchor devices is greater than the preset number of encryption anchor devices, it is determined to perform encryption upload optimization.

[0009] Furthermore, encrypted upload optimizations include: Obtain the task collaboration depth and route intersection degree corresponding to the upload sequence of each transmission channel; For transmission channels where the task collaboration depth is less than the preset task collaboration depth and the route intersection degree is less than the preset route intersection degree, self-adjustment optimization is determined. For transmission channels whose task collaboration depth is greater than or equal to the preset task collaboration depth or whose route intersection degree is greater than or equal to the preset route intersection degree, the transmission channel is identified as the channel to be analyzed and optimized.

[0010] Furthermore, for encrypted anchor selection devices, these are intelligent devices with encryption strength greater than the preset encryption strength or command hit rate greater than the preset command hit rate.

[0011] Furthermore, for any transmission channel, the corresponding self-tuning optimization includes: Based on the cooperative interval target conditions, the order of the upload sequence corresponding to the transmission channel is adjusted; The target condition for the coordination interval is that the sequential coordination interference value corresponding to each encrypted data packet in the upload sequence is minimized.

[0012] Furthermore, if the average number of associated channels is less than the preset average number of associated channels or the difference in the number of associated channels is greater than or equal to the preset difference in the number of associated channels, then each channel to be analyzed and optimized is subjected to collaborative optimization processing based on the descending order of the number of associated channels. In the collaborative optimization process for any channel to be analyzed and optimized, the transmission channel with the lowest transmission pressure among the transmission channels that meet the collaborative optimization conditions is selected as the target insertion channel, and the risk equipment corresponding to the channel to be analyzed and optimized is assigned to the target insertion channel.

[0013] Furthermore, the collaborative optimization condition is: the smoothing insertion coefficient between the channel to be analyzed and optimized and the target insertion channel is greater than the preset smoothing insertion coefficient.

[0014] Furthermore, if the average number of associated channels is greater than or equal to the preset average number of associated channels and the difference in the number of associated channels is less than the preset difference in the number of associated channels, then the encrypted upload interval for each channel to be analyzed and optimized will be increased. The increase in the encrypted upload interval is positively correlated with both the depth of task collaboration and the degree of route overlap.

[0015] Compared with existing technologies, the beneficial effects of this invention are that it determines whether to perform encryption enhancement processing by analyzing the action execution density and perceptual data richness of intelligent devices. The action execution density and perceptual data richness reflect intelligent devices currently in a highly active task state, with large data collection volumes and drastically changing working environments, thus enabling targeted encryption enhancement processing. Compared to full encryption modes that perform encryption at fixed periods and uniform strength, this invention effectively avoids wasting encryption resources and improves the utilization efficiency of encryption resources.

[0016] Furthermore, in the encryption enhancement process of this invention, the spatiotemporal constraint reference value of the intelligent device is detected, and the encryption block interval is reduced based on either the complexity of the action or the number of environmental targets. When the activity space of the intelligent device is limited, the encryption granularity is adjusted primarily based on the fineness of its own actions; when the activity space coverage is large, the encryption is adjusted based on the number of surrounding environmental targets. This achieves adaptive adjustment of the block interval, enabling the encryption granularity to match the actual operating scenario of the device, ensuring data security in high-complexity actions or high-density environments while avoiding the performance overhead caused by excessive encryption.

[0017] Furthermore, this invention monitors the encryption pressure of the scene and the number of encryption anchor devices to determine whether to perform encryption upload optimization. When the encryption pressure of the scene is too high or there are too many encryption anchor devices, the encryption upload optimization process is triggered. This enables further encryption optimization for scenarios with high pressure or multiple high-risk devices, thereby improving the adaptability of the encryption method.

[0018] Furthermore, in the encrypted upload optimization process of this invention, the task collaboration depth and route intersection degree of each transmission channel are obtained, and the transmission channels are classified. For channels with low collaboration depth and low route intersection degree, self-adjustment optimization by adjusting the upload sequence order is performed; for channels with high collaboration depth or high route intersection degree, they are marked as channels to be analyzed and optimized, and enter cluster optimization. This effectively reduces the risk of traffic feature leakage caused by data from collaborative devices being too close in transmission time sequence or frequently crossing trajectories, and improves data transmission security in multi-agent collaborative scenarios. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method for encrypting artificial intelligence data according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating whether encryption enhancement processing is performed in an embodiment of the present invention; Figure 3 This is a flowchart illustrating whether encrypted upload optimization is performed in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0023] Please see Figure 1 The diagram shown is a schematic representation of an artificial intelligence data encryption method according to an embodiment of the present invention. The present invention provides an artificial intelligence data encryption method, comprising: Analyze the intensity of actions performed by smart devices and the richness of perceived data to determine whether to perform encryption enhancement processing; In the encryption enhancement process, the interval between encryption blocks is adjusted based on the spatiotemporal constraint reference value corresponding to the smart device, and based on the complexity of the action or the number of environmental targets. Whether to perform encrypted upload optimization depends on the encryption pressure of the scenario and the number of encryption anchor devices. In encrypted upload optimization, the order of upload sequences is adjusted based on the task collaboration depth and route intersection degree corresponding to the upload sequence of the transmission channel, or the transmission channel is recorded as the channel to be analyzed and optimized for cluster optimization. In cluster optimization, the number of associated channels corresponding to each channel to be analyzed and optimized is obtained, and the collaborative optimization processing or encrypted upload interval adjustment of the channel to be analyzed and optimized is determined based on the average number of associated channels and the difference in the number of associated channels.

[0024] This invention is applied to the encrypted uploading of data from artificial intelligence terminals, specifically, to the data encryption of artificial intelligence robots. Intelligent devices are robots with robotic arms and mobility capabilities. Application scenarios include, but are not limited to, smart warehousing where intelligent devices handle and stack goods. In these scenarios, the transmission channel is a logical data upload link established between edge communication devices and the cloud. Edge communication devices include, but are not limited to, edge gateways and field servers. Data collected by the intelligent device is uploaded to the edge communication device via a local communication network. Several transmission channels exist between the edge communication device and the cloud, and each channel can handle multiple data upload tasks.

[0025] Please see Figures 1 to 3 For smart devices where the action execution density is greater than the action execution density threshold or the perception data richness is greater than the perception data richness threshold, encryption enhancement processing is determined to be performed.

[0026] In this invention, the periodic determination of whether to perform encryption enhancement processing is performed at the end of each sampling period. In this embodiment, the sampling period is 5 minutes. It can be understood that the shorter the sampling period, the more timely the system responds to changes in the intensity of smart device actions and the richness of perceived data, that is, the more sensitive the encryption enhancement processing is to be triggered. Therefore, the greater the user's sensitivity to encryption response delay, the shorter the sampling period.

[0027] For a single smart device, the method for determining the corresponding action execution density is to obtain the total number of action commands executed by the smart device within the most recent sampling period, and divide it by the duration of the sampling period, with the unit being times per minute. It is understood that the smart device can receive control commands in real time and execute corresponding actions according to the command type. Action commands include, but are not limited to, joint movement commands of the robotic arm, translation and rotation commands of the mobile chassis, grasping and releasing commands, etc. The sources of control commands include those issued by the user or those issued by the smart device's own task decision module; this is content easily understood by those skilled in the art and will not be elaborated further.

[0028] For a single smart device, the method for confirming the richness of its corresponding perception data is as follows: obtain the values ​​collected by each perception sensor of the smart device in the most recent sampling period. The sensors and the corresponding collected data include, but are not limited to, the brightness value of the vision sensor, the distance value of the lidar, the torque value of the force sensor, and the acceleration value of the IMU. For each sensor, calculate the absolute value of the difference between the maximum and minimum values ​​in its collected values, i.e., the range value. Use the range value as the numerator to calculate the ratio between the range value and the maximum value, and record it as the reference sub-value. Take the average value of the reference sub-values ​​of all sensors to obtain the richness of perception data.

[0029] In this embodiment of the invention, the action execution density threshold is 6 times / minute. It can be understood that action execution density reflects the frequency of action command execution by the smart device per unit of time. A higher density indicates a more active task state for the device, generating more control command data, and thus a more urgent need for encryption enhancement processing. The greater the user's sensitivity to data security during a highly active device state, the lower the action execution density threshold.

[0030] In this embodiment of the invention, the perceptual data richness threshold is 0.5. It can be understood that perceptual data richness reflects the extent to which the range of changes in the values ​​perceived by each sensor within the sampling period is covered by the intelligent device. A higher richness indicates a more dynamic environment or more complex changes in the device's own state, resulting in more valuable and sensitive perceptual data, thus necessitating more urgent encryption and enhancement processing. The greater the user's sensitivity to data security under drastic changes in environmental perception, the lower the perceptual data richness threshold.

[0031] For smart devices with an action execution density less than or equal to the action execution density threshold and a perception data richness less than or equal to the perception data richness threshold, no encryption enhancement processing is required.

[0032] Specifically, the encryption enhancement process includes: Detect the spatiotemporal constraint reference values ​​corresponding to intelligent devices; If the spatiotemporal constraint reference value is less than the preset spatiotemporal constraint reference value, the encrypted block interval will be reduced based on the action complexity. If the spatiotemporal constraint reference value is greater than or equal to the preset spatiotemporal constraint reference value, the encryption block interval is reduced based on the number of environmental targets.

[0033] For a smart device, its movement trajectory within the most recent sampling period is detected, and the area of ​​the smallest rectangle that can encompass the movement trajectory is denoted as the spatiotemporal constraint reference value. In this embodiment of the invention, the preset spatiotemporal constraint reference value is 4.0 square meters. It can be understood that a smaller spatiotemporal constraint reference value indicates a more concentrated movement trajectory and a more restricted activity space for the smart device, meaning the device is confined to a small area. In this case, the complexity of the device's actions has a more significant impact on the encryption block interval, thus triggering a reduction in the encryption block interval based on action complexity. Conversely, a larger spatiotemporal constraint reference value indicates a wider movement range and more free activity space for the device. In this case, the environmental targets faced by the device are more diverse, thus triggering a reduction in the encryption block interval based on the number of environmental targets. The greater the user's sensitivity to action complexity, the larger the preset spatiotemporal constraint reference value.

[0034] The motion complexity is the average number of joint controls corresponding to each control command received by the intelligent device within the sampling period. For a single control command, the corresponding number of joint controls is the number of joints that need to be driven to execute the command. In this embodiment of the invention, the preset motion complexity is 0.5. It can be understood that the higher the motion complexity, the more joints the intelligent device needs to drive simultaneously for the control command executed, and the stronger the motion coupling. That is, the device is currently in a high-complexity motion execution state, and the control command data is more refined and sensitive. Therefore, the greater the user's sensitivity to data security in complex motion scenarios, the lower the preset motion complexity.

[0035] The number of environmental targets refers to the total number of target objects perceived by sensors with target recognition capabilities, such as visual sensors and LiDAR, of smart devices within the sampling period. Target objects include, but are not limited to, other smart devices, obstacles, shelves, personnel, and markers. The same target object appearing repeatedly is counted only once. In this embodiment of the invention, the preset number of environmental targets is 3. It can be understood that a larger number of environmental targets indicates a denser concentration of other smart devices, obstacles, or personnel around the smart device, resulting in richer and more sensitive collaborative relationships and location information in the perceived data. Therefore, the need to reduce the encryption block interval based on the number of environmental targets is more urgent. The greater the user's sensitivity to data security in densely populated environments, the smaller the preset number of environmental targets.

[0036] When reducing the encrypted block interval based on action complexity, the adjusted encrypted block interval = basic encrypted block interval × max(α × preset action complexity / action complexity, minimum adjustment coefficient). The encryption block interval is reduced based on the number of environmental targets. The adjusted encryption block interval = basic encryption block interval × max(β × preset number of environmental targets / number of environmental targets, minimum adjustment coefficient). Where α is the first block adjustment value and β is the second block adjustment value, in this embodiment of the invention, the first block adjustment value α is 1.2 and the second block adjustment value β is 0.9. It can be understood that the larger α is, the more sensitive the adjusted interval is to changes in action complexity when adjusting the encryption block interval based on action complexity; that is, the greater the reduction in the encryption block interval under the same action complexity. The smaller α is, the smoother the adjustment effect. Therefore, the greater the user's sensitivity to increased encryption strength in action-complex scenarios, the larger the first block adjustment value α should be. Similarly, the larger β is, the more sensitive the adjusted interval is to changes in the number of environmental targets when adjusting the encryption block interval based on the number of environmental targets; that is, the greater the reduction in the encryption block interval under the same number of environmental targets. The smaller β is, the smoother the adjustment effect. Therefore, the greater the user's sensitivity to increased encryption strength in dense environmental scenarios, the larger the second block adjustment value β should be.

[0037] In this invention, the data collected by the smart device during the sampling period is encrypted in blocks. This involves dividing the data into several encrypted data blocks of equal size, and then encrypting each block to obtain an encrypted data packet. The block interval is the number of bytes contained in each encrypted data block. A smaller block interval results in finer data segmentation and more granular encryption, making it more difficult for an attacker to recover complete information from a single data block. However, this also increases the computational and transmission overhead. Conversely, a larger block interval leads to higher encryption efficiency but weakens the fine-grained protection capability. The block interval typically ranges from 64 bytes to 1024 bytes. In this embodiment, the basic block interval is set to 256 bytes. It is understood that the greater the user's requirement for encryption granularity, the smaller the basic block interval.

[0038] Specifically, encryption upload optimization is performed based on the condition that the scene encryption pressure is greater than the preset scene encryption pressure or the number of encryption anchor devices is greater than the preset number of encryption anchor devices.

[0039] The scene encryption pressure is the average of the current transmission pressure values ​​of all transmission channels. For a single transmission channel, the transmission pressure value is the current pending data queue length divided by the preset allowed pending data queue length. The pending data queue length is measured in units of the number of encrypted data packets. The preset allowed pending data queue length is the maximum allowed pending data queue length for a single transmission channel. The larger the actual pending data queue length, the more severe the backlog in the transmission channel and the higher the risk of data congestion, leading to increased transmission pressure and scene encryption pressure. Therefore, the need for encryption upload optimization is more urgent. The greater the user's sensitivity to transmission congestion, the smaller the preset allowed pending data queue length.

[0040] In this embodiment of the invention, the preset allowed queue length for data to be uploaded is 100 encrypted data packets. It is understood that the lower the user's tolerance for queue length, the smaller the preset allowed queue length for data to be uploaded.

[0041] Specifically, for encrypted anchor selection devices, these are smart devices with encryption strength greater than the preset encryption strength or command hit rate greater than the preset command hit rate.

[0042] For any smart device, the corresponding encryption strength is the number of encryption enhancement processes in the most recent sampling periods / the number of sampling periods. The command hit rate is the average of the number of control commands received by the smart device from the user in the most recent sampling periods / the number of control commands received by all smart devices from the user in the most recent sampling periods. The number of sampling periods is set based on user needs. It can be understood that the more sensitive the user is to the timeliness of the encryption strength determination, the smaller the number of sampling periods. Preferably, the number of sampling periods is 5.

[0043] In this embodiment of the invention, the preset encryption strength is 0.6. It can be understood that a higher encryption strength indicates a higher frequency of encryption enhancement processing triggered on the smart device within the most recent sampling periods, meaning a stronger data protection need and a more active encryption state. Therefore, it should be marked as an encryption anchor device as a reference benchmark for upload optimization. The greater the user's emphasis on the device's encryption activity, the lower the preset encryption strength. In this embodiment of the invention, the preset command hit rate is 1.0. It can be understood that a higher command hit rate indicates that the smart device receives more user commands relative to the scene average, meaning more frequent device task interactions. Therefore, it should be marked as an encryption anchor device as a reference benchmark for upload optimization. The greater the user's sensitivity to the criticality of device tasks, the lower the preset command hit rate.

[0044] In this embodiment of the invention, the preset number of encrypted anchor devices is 3. It can be understood that a larger number of encrypted anchor devices indicates more intelligent devices with high encryption strength or high command hit rate in the current scenario, meaning a higher density of overall security-sensitive devices and a higher risk of indirect leakage of associated data during collaborative uploads. Therefore, the need for encrypted upload optimization is more urgent. The greater the user's sensitivity to the risks caused by the number of security-sensitive devices, the smaller the preset number of encrypted anchor devices.

[0045] Specifically, the encrypted upload optimizations include: Obtain the task collaboration depth and route intersection degree corresponding to the upload sequence of each transmission channel; For transmission channels where the task collaboration depth is less than the preset task collaboration depth and the route intersection degree is less than the preset route intersection degree, self-adjustment optimization is determined. For transmission channels whose task collaboration depth is greater than or equal to the preset task collaboration depth or whose route intersection degree is greater than or equal to the preset route intersection degree, the transmission channel is identified as the channel to be analyzed and optimized.

[0046] For a single transmission channel, the corresponding task collaboration depth is the average number of collaborative tasks of each smart device in the upload sequence of that transmission channel. For a single smart device, the corresponding number of collaborative tasks is the number of other smart devices in the upload sequence that have a collaborative working relationship with that smart device in the most recent sampling period. For any two smart devices, if there are cargo handling or palletizing tasks for the same cargo area in the most recent sampling period, then the two smart devices have a collaborative working relationship.

[0047] For a single transmission channel, the corresponding route intersection degree is determined by recording the movement trajectory of each smart device in the most recent sampling period of the transmission channel's upload sequence, counting the intersection points between all movement trajectories, and the route intersection degree is the total number of intersection points.

[0048] In this embodiment of the invention, the preset task collaboration depth is 2. The greater the task collaboration depth, the denser the collaborative working relationship between the intelligent devices in the upload sequence of the transmission channel, and the stronger the business correlation between the uploaded data. Therefore, simple self-adjustment optimization is not possible, and it is selected as the channel to be analyzed and optimized for cluster optimization. The greater the user's sensitivity to the risk of collaboration association, the smaller the preset task collaboration depth.

[0049] In this embodiment of the invention, the preset route intersection degree is 8. The larger the route intersection degree, the more frequently the movement trajectories of each intelligent device in the uploaded sequence intertwine in space, and the easier it is for environmental perception-related data to be correlated. This increases the risk of data packet transmission being attacked by traffic analysis or time-series correlation attacks, making simple self-adjustment impossible. Therefore, it is selected as a channel to be analyzed and optimized for cluster optimization. The greater the user's sensitivity to trajectory correlation risks, the smaller the preset route intersection degree.

[0050] In addition, each transmission channel corresponds to several smart devices, and the upload sequence is obtained by randomly arranging the encrypted data packets generated by the smart devices in the most recent sampling period; in the initial scenario, that is, when the smart devices are first put into use, they are assigned to the transmission channel with the lowest transmission pressure value.

[0051] Specifically, for any transmission channel, the corresponding self-tuning optimization includes: Based on the cooperative interval target conditions, the order of the upload sequence corresponding to the transmission channel is adjusted; The target condition for the coordination interval is that the sequential coordination interference value corresponding to each encrypted data packet in the upload sequence is minimized.

[0052] The method for confirming the sequential collaborative interference value is as follows: traverse two adjacent encrypted data packets in the upload sequence, determine whether there is a collaborative working relationship between the corresponding smart devices, and contribute one interference count if a collaborative relationship exists; otherwise, contribute 0. Summate the interference counts of all adjacent data packet pairs to obtain the sequential collaborative interference value.

[0053] Specifically, if the average number of associated channels is less than the preset average number of associated channels or the difference in the number of associated channels is greater than or equal to the preset difference in the number of associated channels, then each channel to be analyzed and optimized will be collaboratively optimized based on the descending order of the number of associated channels. For a single channel to be analyzed and optimized, the number of associated channels is the number of other channels to be analyzed and optimized that are associated with that channel. If there are at least a preset number of smart devices in two channels to be analyzed and optimized that have a collaborative relationship with other smart devices in another channel to be analyzed and optimized, then the two channels to be analyzed and optimized are considered to be associated. In this embodiment of the invention, the preset number of associated devices is 50% of the sum of the smart devices in the two channels to be analyzed and optimized. The preset number of associated devices is rounded up. The more sensitive the user is to the data association risk of the channel to be analyzed and optimized, the smaller the preset number of associated devices.

[0054] The mean number of associated channels is the average number of associated channels corresponding to each channel to be analyzed and optimized, and the difference in the number of associated channels is the variance of the number of associated channels corresponding to each channel to be analyzed and optimized.

[0055] In the collaborative optimization process for any channel to be analyzed and optimized, the transmission channel with the lowest transmission pressure among those satisfying the collaborative optimization conditions is selected and designated as the target insertion channel. The risky device corresponding to the channel to be analyzed and optimized is then assigned to the target insertion channel. The risky device is the intelligent device with the largest number of collaborative tasks in the channel to be analyzed and optimized. When a risky device is inserted into the target insertion channel, the encrypted data packet corresponding to the risky device is deleted from the channel to be analyzed and optimized and inserted into the upload sequence corresponding to the target insertion channel. Furthermore, a self-adjustment optimization is performed once for the upload sequences corresponding to both the channel to be analyzed and optimized and the target insertion channel.

[0056] Specifically, the collaborative optimization condition is: the smoothing insertion coefficient between the channel to be analyzed and optimized and the target insertion channel is greater than the preset smoothing insertion coefficient.

[0057] For two channels to be analyzed and optimized, the corresponding smoothing insertion coefficient is the reciprocal of the number of intelligent devices that have a collaborative relationship between the two channels. The preset smoothing insertion coefficient is 0.2 in this embodiment of the invention. It can be understood that the larger the smoothing insertion coefficient, the fewer intelligent devices that have a collaborative relationship between the two channels, that is, the weaker the collaborative association between the channels and the lower the data coupling. The less transmission interference caused by migrating risky devices from one channel to another, the easier it is to meet the collaborative optimization. The greater the user's sensitivity to the migration risk caused by collaborative association, the larger the preset smoothing insertion coefficient.

[0058] Specifically, if the average number of associated channels is greater than or equal to the preset average number of associated channels and the difference in the number of associated channels is less than the preset difference in the number of associated channels, then the encrypted upload interval for each channel to be analyzed and optimized will be increased. The increase in the encrypted upload interval is positively correlated with both the depth of task collaboration and the degree of route overlap.

[0059] For a single channel to be analyzed and optimized, the corresponding encrypted upload interval is the interval between each adjacent encrypted data packet sent to the cloud within a single upload sequence. The adjusted encrypted upload interval = basic encrypted upload interval × (1 + duration adjustment coefficient × impact value). The impact value = (task collaboration depth normalization value + route intersection degree normalization value) / 2. The task collaboration depth normalization value and route intersection degree normalization value are obtained after normalizing the task collaboration depth and route intersection degree respectively. The preferred value for the duration adjustment coefficient is 1.1. The greater the user's sensitivity to collaboration-related risks, the greater the duration adjustment coefficient. The basic encrypted upload interval is 15 seconds. It can be understood that the basic encrypted upload interval determines the baseline sparsity of data transmission. The longer the interval, the fewer data packets are uploaded per unit time, the lower the transmission bandwidth usage, and the lower the possibility of risk-related leakage. However, the risk of data backlog and delay increases. The less sensitive the user is to data real-time performance and the greater the sensitivity to data-related leakage risks, the greater the basic encrypted upload interval.

[0060] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0061] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for encrypting artificial intelligence data, characterized in that, include: Analyze the intensity of actions performed by smart devices and the richness of perceived data to determine whether to perform encryption enhancement processing; In the encryption enhancement process, the interval between encryption blocks is adjusted based on the spatiotemporal constraint reference value corresponding to the smart device, and based on the complexity of the action or the number of environmental targets. Whether to perform encrypted upload optimization depends on the encryption pressure of the scenario and the number of encryption anchor devices. In encrypted upload optimization, the order of upload sequences is adjusted based on the task collaboration depth and route intersection degree corresponding to the upload sequence of the transmission channel, or the transmission channel is recorded as the channel to be analyzed and optimized for cluster optimization. In cluster optimization, the number of associated channels corresponding to each channel to be analyzed and optimized is obtained, and the collaborative optimization processing or encrypted upload interval adjustment of the channel to be analyzed and optimized is determined based on the average number of associated channels and the difference in the number of associated channels.

2. The method for encrypting artificial intelligence data according to claim 1, characterized in that, For smart devices with an action execution density greater than the action execution density threshold or a perception data richness greater than the perception data richness threshold, encryption enhancement processing is determined to be performed.

3. The method for encrypting artificial intelligence data according to claim 2, characterized in that, Encryption enhancement processes include: Detect the spatiotemporal constraint reference values ​​corresponding to intelligent devices; If the spatiotemporal constraint reference value is less than the preset spatiotemporal constraint reference value, the encrypted block interval will be reduced based on the action complexity. If the spatiotemporal constraint reference value is greater than or equal to the preset spatiotemporal constraint reference value, the encryption block interval is reduced based on the number of environmental targets.

4. The method for encrypting artificial intelligence data according to claim 2, characterized in that, Based on the condition that the scene encryption pressure is greater than the preset scene encryption pressure or the number of encryption anchor devices is greater than the preset number of encryption anchor devices, it is determined to perform encryption upload optimization.

5. The method for encrypting artificial intelligence data according to claim 4, characterized in that, Encrypted upload optimizations include: Obtain the task collaboration depth and route intersection degree corresponding to the upload sequence of each transmission channel; For transmission channels where the task collaboration depth is less than the preset task collaboration depth and the route intersection degree is less than the preset route intersection degree, self-adjustment optimization is determined. For transmission channels whose task collaboration depth is greater than or equal to the preset task collaboration depth or whose route intersection degree is greater than or equal to the preset route intersection degree, the transmission channel is identified as the channel to be analyzed and optimized.

6. The method for encrypting artificial intelligence data according to claim 5, characterized in that, For encrypted anchor selection devices, these are intelligent devices with encryption strength greater than the preset encryption strength or command hit rate greater than the preset command hit rate.

7. The method for encrypting artificial intelligence data according to claim 6, characterized in that, For any transmission channel, the corresponding self-tuning optimizations include: Based on the cooperative interval target conditions, the order of the upload sequence corresponding to the transmission channel is adjusted; The target condition for the coordination interval is that the sequential coordination interference value corresponding to each encrypted data packet in the upload sequence is minimized.

8. The method for encrypting artificial intelligence data according to claim 7, characterized in that, If the average number of associated channels is less than the preset average number of associated channels or the difference in the number of associated channels is greater than or equal to the preset difference in the number of associated channels, then the channels to be analyzed and optimized are processed collaboratively based on the descending order of the number of associated channels. In the collaborative optimization process for any channel to be analyzed and optimized, the transmission channel with the lowest transmission pressure among the transmission channels that meet the collaborative optimization conditions is selected as the target insertion channel, and the risk equipment corresponding to the channel to be analyzed and optimized is assigned to the target insertion channel.

9. The method for encrypting artificial intelligence data according to claim 8, characterized in that, The collaborative optimization condition is that the smoothing insertion coefficient between the channel to be analyzed and the target insertion channel is greater than the preset smoothing insertion coefficient.

10. The method for encrypting artificial intelligence data according to claim 9, characterized in that, If the average number of associated channels is greater than or equal to the preset average number of associated channels and the difference in the number of associated channels is less than the preset difference in the number of associated channels, then the encrypted upload interval for each channel to be analyzed and optimized will be increased. The increase in the encrypted upload interval is positively correlated with both the depth of task collaboration and the degree of route overlap.

Images