Cloud-based monitoring system based on artificial intelligence analysis

Abstract

The present invention discloses a cloud monitoring system based on artificial intelligence analysis, which specifically relates to the field of cloud monitoring. It is used to solve the problem of data integrity destruction caused by cross-channel sequence number conflicts in heterogeneous network switching scenarios. The system embeds the real-time status parameters of the transmission path into the fragmentation encapsulation process, and constructs a multi-dimensional verification system covering timing jitter and spatial motion characteristics, so that the detection and repair of conflicting fragments can accurately match the network fluctuation characteristics and the laws of physical scene changes; through the dual optimization of transmission window cardinality reverse compensation and priority self-feedback suppression, while eliminating multi-path sequence number misalignment, the conflict regeneration path of the high-jitter channel is actively blocked, forming a positive cycle of enhanced transmission stability and data integrity maintenance, and improving the spatiotemporal alignment accuracy of video streams and sensor data in complex network environments, providing a highly reliable real-time data foundation for intelligent security scenarios, and effectively avoiding the monitoring blind spots caused by mechanical verification rules of traditional solutions.

Description

Technical Field

[0001] The present invention relates to the field of cloud monitoring, and more specifically, to a cloud monitoring system based on artificial intelligence analysis. Background Art

[0002] To cope with complex network environments, modern cloud-based surveillance systems generally employ multi-path parallel transmission and adaptive compression technologies. They use the Z-standard algorithm to segment video streams into independently sequenced segments and utilize dynamic window adjustment mechanisms to improve transmission throughput. This system effectively reduces lag rates in cross-regional deployments and relies on continuous sequence number verification to ensure frame integrity in the cloud-based depacketization engine, enabling the AI ​​module to accurately analyze security events.

[0003] However, under conditions of multipath link fluctuations (such as handovers between satellite and 4G) and ACK loss, fragment retransmissions and window expansions can lead to cross-channel sequence number conflicts, triggering the depacketization engine's strong continuity check rules and causing key frames to be discarded. This issue directly disrupts the spatiotemporal alignment of the video stream and sensor data, rendering the cloud-based monitoring system's anomaly identification and compliance forensics capabilities ineffective, creating a monitoring blind spot.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a cloud monitoring system based on artificial intelligence analysis, which fundamentally solves the problem of data integrity destruction caused by cross-channel sequence number conflicts in heterogeneous network switching scenarios through dynamic perception and closed-loop coordination mechanism of multi-path transmission status. The real-time status parameters of the transmission path are embedded in the fragmentation encapsulation process, and a multi-dimensional verification system covering timing jitter and spatial motion characteristics is constructed, so that the detection and repair of conflicting fragments can accurately match the network fluctuation characteristics and the laws of physical scene changes; through the dual optimization of transmission window cardinality reverse compensation and priority self-feedback suppression, while eliminating multi-path sequence number misalignment, the conflict regeneration path of the high-jitter channel is actively blocked, forming a positive cycle of enhanced transmission stability and data integrity maintenance, improving the spatiotemporal alignment accuracy of video streams and sensor data in complex network environments, providing a highly reliable real-time data foundation for intelligent security scenarios, and effectively avoiding the monitoring blind spots caused by mechanical verification rules in traditional solutions, so as to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The cloud-based monitoring system based on artificial intelligence analysis includes: a path marking unit, a conflict perception unit, a spatiotemporal screening unit, a cross-path correction unit, and a window adjustment unit;

[0008] Path marking unit: When encapsulating fragments, it adds a channel identifier and the current transmission window base to each fragment, and generates a fragment transmission priority label based on the instantaneous delay jitter at the moment of path switching;

[0009] Conflict perception unit: When receiving a fragment, it extracts the sequence number offset history of the current channel based on the channel identifier, and dynamically adjusts the conflict judgment threshold based on the delay jitter. If the difference between the fragment sequence number and the offset exceeds the threshold, it is marked as a conflicting fragment and temporarily stored;

[0010] The spatiotemporal screening unit selects repairable fragments that meet both delay tolerance and spatial consistency by calculating the deviation between the actual transmission delay and path switching jitter of conflicting fragments and analyzing the continuity of the spatial motion vectors of the sensor data of adjacent fragments.

[0011] Cross-path correction unit: Based on the transmission window base difference of the repairable fragments, it reversely deduces the sequence number misalignment caused by multipath switching, dynamically corrects the fragment sequence number, verifies its continuity with the current depacketization window, and injects it into the depacketization queue after passing.

[0012] Window adjustment unit: Based on the fragment correction results, it freezes the window base expansion of the channel that has been successfully corrected, and rolls back the window base to the pre-conflict state for the channel that has failed to be corrected, and reduces the transmission priority of subsequent path switching.

[0013] In a preferred embodiment, the path marking unit includes the following contents:

[0014] During the fragment encapsulation stage, a channel identifier, a transmission window cardinality, and a priority tag are attached to each video fragment. The channel identifier is defined as the unique identifier of the transmission channel, the transmission window cardinality is defined as the starting sequence number of the current transmission window, and the priority tag is determined based on the comparison result of the instantaneous delay jitter with a preset threshold. The instantaneous delay jitter is defined as the absolute value of the difference in transmission delays between adjacent video fragments.

[0015] In a preferred embodiment, the conflict sensing unit includes the following:

[0016] The sequence offset history is queried according to the channel identifier, and the conflict judgment threshold is dynamically adjusted according to the priority tag. The conflict judgment difference between the sequence offset of the fragment sequence number and the expected sequence number, as well as the sequence offset and the historical offset average value in the sequence offset history are calculated. When the conflict judgment difference is greater than the conflict judgment threshold, the video fragment is marked as a conflicting fragment and temporarily stored. When the conflict judgment difference is not greater than the conflict judgment threshold, the video fragment is put into the depacketization queue.

[0017] In a preferred embodiment, the spatiotemporal screening unit includes the following:

[0018] By calculating the deviation between the actual transmission delay and the path switching jitter and analyzing the spatial motion vector continuity of the sensor data of adjacent slices, we screen out repairable slices that meet both delay tolerance and spatial consistency. When the deviation is less than the deviation threshold and the spatial motion vector continuity is less than the continuity measurement threshold, the conflicting slice is marked as repairable.

[0019] In a preferred embodiment, the spatiotemporal screening unit further includes the following:

[0020] The actual transmission delay is defined as the difference between the receiving timestamp and the sending timestamp. The path switching jitter is determined based on the priority tag. The deviation is calculated as the absolute value of the difference between the actual transmission delay and the path switching jitter divided by the path switching jitter. The spatial motion vector continuity is obtained by calculating the minimum relative change rate of the spatial motion vector of the conflicting slice and the adjacent slice.

[0021] In a preferred embodiment, the path correction unit includes the following:

[0022] For repairable fragmented data packets, the sequence number misalignment is derived by calculating the difference between the transmission window base and the current unpacking window base, combined with the nonlinear adjustment of the channel historical misalignment and the priority coefficient. The fragment sequence number is dynamically corrected according to the sequence number misalignment, and the continuity of the corrected fragment sequence number and the current unpacking window is verified. The fragmented data packets that pass the verification are injected into the unpacking queue.

[0023] In a preferred embodiment, the path correction unit further includes the following:

[0024] The processing process of the corrected fragment sequence number is as follows: extract the original fragment sequence number from the header of the repairable fragment data packet; then, subtract the sequence number misalignment amount from the original fragment sequence number to obtain the corrected fragment sequence number.

[0025] In a preferred embodiment, the path correction unit further includes the following:

[0026] The judgment process of continuity check is as follows: check whether the corrected fragment sequence number meets any of the following conditions: first, it falls within the depacketization window range; second, it is equal to the current depacketization window base minus one, that is, it is continuous with the lower boundary of the window; third, it is equal to the current depacketization window base plus the window size, that is, it is continuous with the upper boundary of the window; if any condition is met, it is considered to have passed the check; if none of the conditions are met, it is marked as unavailable and the corresponding fragment data packet is discarded.

[0027] In a preferred embodiment, the path correction unit further includes the following:

[0028] The priority coefficient is obtained by mapping the fragment transmission priority labels generated by the path marking unit.

[0029] In a preferred embodiment, the window adjustment unit includes the following contents:

[0030] According to the fragmentation correction results, the stability of the transmission channel is judged by counting the continuous correction success count. The transmission window base extension is frozen for the transmission channel whose continuous correction success count reaches the stability threshold. The transmission window base is rolled back to the pre-conflict window base for the transmission channel where fragmentation correction fails, and the transmission priority is adjusted through nonlinear calculation.

[0031] The technical effects and advantages of the cloud-based monitoring system based on artificial intelligence analysis of the present invention are:

[0032] The present invention fundamentally solves the problem of data integrity destruction caused by cross-channel sequence number conflicts in heterogeneous network switching scenarios through dynamic perception and closed-loop coordination mechanisms of multi-path transmission status. The real-time status parameters of the transmission path are embedded in the fragmentation encapsulation process, and a multi-dimensional verification system covering timing jitter and spatial motion characteristics is constructed, so that the detection and repair of conflicting fragments can accurately match the network fluctuation characteristics and the laws of physical scene changes; through the dual optimization of transmission window cardinality reverse compensation and priority self-feedback suppression, while eliminating multi-path sequence number misalignment, the conflict regeneration path of the high-jitter channel is actively blocked, forming a positive cycle of enhanced transmission stability and data integrity maintenance, improving the spatiotemporal alignment accuracy of video streams and sensor data in complex network environments, providing a highly reliable real-time data foundation for intelligent security scenarios, and effectively avoiding the monitoring blind spots caused by mechanical verification rules in traditional solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Schematic diagram of the process of the cloud monitoring system based on artificial intelligence analysis of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Example 1: Figure 1 The present invention provides a cloud monitoring system based on artificial intelligence analysis, including: a path marking unit, a conflict perception unit, a spatiotemporal screening unit, a cross-path correction unit, and a window adjustment unit;

[0036] Path marking unit: When encapsulating fragments, it adds a channel identifier and the current transmission window base to each fragment, and generates a fragment transmission priority label based on the instantaneous delay jitter at the moment of path switching.

[0037] Conflict perception unit: When receiving a fragment, it extracts the sequence number offset history of the current channel based on the channel identifier, and dynamically adjusts the conflict judgment threshold based on the delay jitter. If the difference between the fragment sequence number and the offset exceeds the threshold, it is marked as a conflicting fragment and temporarily stored.

[0038] Spatiotemporal screening unit: For conflicting shards, by calculating the deviation between their actual transmission delay and path switching jitter and analyzing the continuity of the spatial motion vectors of the sensor data of adjacent shards, it screens out repairable shards that meet both delay tolerance and spatial consistency.

[0039] Cross-path correction unit: Based on the transmission window cardinality difference of the repairable fragments, it reversely deduces the sequence number misalignment caused by multipath switching, dynamically corrects the fragment sequence number and verifies its continuity with the current depacketization window, and then injects it into the depacketization queue.

[0040] Window adjustment unit: Based on the fragment correction results, it freezes the window base expansion of the channel that has been successfully corrected, and rolls back the window base to the pre-conflict state for the channel that has failed to be corrected, and reduces the transmission priority of subsequent path switching.

[0041] In cloud-based monitoring systems, accurate transmission and spatiotemporal alignment of video streams and sensor data are the core guarantees for anomaly identification and compliance forensics. For multi-path parallel transmission scenarios in complex network environments (such as satellite and 4G switching), the video stream is divided into fragments with independent sequence numbers through the Z-standard algorithm, and the throughput is optimized using a dynamic window adjustment mechanism. However, multi-path link fluctuations and ACK loss can lead to fragment retransmission and window expansion, causing cross-channel sequence number conflicts, which in turn trigger the unpacking engine to mistakenly abandon key frames, destroying data integrity. To solve this problem, the present invention lays the foundation for subsequent conflict detection and repair by adding key metadata during the fragment encapsulation stage. This abstract focuses on the specific processing technology logic of the path marking unit, and ensures the accuracy and availability of metadata for the message processing of video fragments during encapsulation, so that the conflict perception unit can rely on these metadata to achieve dynamic correction of sequence number conflicts and data alignment.

[0042] Fragmentation encapsulation involves splitting an original data stream (such as a video stream) into multiple independent small data blocks (i.e., fragments) during data transmission. The necessary metadata and control information are then appended to each fragment to ensure that the receiving end can correctly reassemble and process the fragments. By breaking large data streams into smaller chunks, this approach optimizes data transmission efficiency and reliability, making it particularly useful in scenarios with unstable networks or limited bandwidth.

[0043] Each shard typically contains the following key elements:

[0044] Data content, i.e., a portion of the original data stream, such as a fragment of a video frame;

[0045] Sequence number, used to identify the order of the fragments in the original data stream to ensure correctness during reassembly;

[0046] Metadata, such as timestamps or channel identifiers, describing the properties of the shard and the transmission path;

[0047] and control information, such as checksums or priority tags, to ensure data integrity and optimize transmission strategies.

[0048] Through additional metadata and control information, fragment encapsulation can achieve intelligent transmission management, such as dynamically adjusting the transmission path or priority scheduling, thereby significantly improving the overall system performance and user experience.

[0049] The path marking unit includes the following:

[0050] Additional channel identifiers:

[0051] Each video segment is assigned a unique identifier for the transmission channel to which it belongs, called a channel identifier. The channel identifier is an integer value generated by the multipath transmission protocol during path allocation and is used to distinguish different transmission channels, such as satellite channels or mobile communication channels. The additional channel identifier is processed by reserving a fixed number of bytes, such as four bytes, in the header of the video segment and writing the channel identifier of the current transmission channel into this reserved field. The purpose of the additional channel identifier is to ensure that the receiving end can extract the sequence offset history of the corresponding transmission channel based on the channel identifier, thereby accurately identifying the source channel of the video segment.

[0052] Addition of the transfer window base:

[0053] The starting sequence number of the current transmission window is appended to each video fragment, which is called the transmission window base. The transmission window base represents the baseline value of the video fragment sequence number within the current transmission window and is updated in real time by the dynamic window adjustment mechanism based on the transmission status. The additional transmission window base is processed by appending a fixed number of bytes to the video fragment header, such as four bytes of space, and writing the transmission window base of the current transmission window into this additional field. The purpose of appending the transmission window base is to provide benchmark data for subsequent steps, which is used to reversely deduce the sequence number misalignment caused by multipath switching, thereby realizing dynamic correction of the video fragment sequence number.

[0054] Calculation of instantaneous delay jitter:

[0055] At the moment of transmission channel switching, the instantaneous delay jitter of the current transmission channel is calculated, which is called instantaneous delay jitter. Instantaneous delay jitter is defined as the absolute value of the difference in transmission delays between two adjacent video fragments. The specific calculation method is: record the time when the sender sends each video fragment and the time when the receiver returns the confirmation signal, calculate the transmission delay of the current video fragment, that is, the difference between the sending time and the time when the confirmation signal returns; then, subtract the transmission delay of the previous video fragment from the transmission delay of the current video fragment, and take the absolute value of the difference as the instantaneous delay jitter. The purpose of calculating instantaneous delay jitter is to quantify the impact of transmission channel switching on transmission stability and provide a basis for generating priority labels. The sender maintains a delay record table for each transmission channel and calculates the instantaneous delay jitter based on the delay records of the two most recent video fragments when the path switches.

[0056] Generation of priority labels:

[0057] Based on instantaneous delay jitter, a transmission priority tag, called a priority tag, is generated for each video segment. The priority tag generation rules are as follows: If the instantaneous delay jitter is less than the preset low delay jitter threshold, the priority tag is set to high, indicating high transmission channel stability; if the instantaneous delay jitter is greater than or equal to the low delay jitter threshold and less than the preset high delay jitter threshold, the priority tag is set to medium, indicating medium transmission channel stability; if the instantaneous delay jitter is greater than or equal to the high delay jitter threshold, the priority tag is set to low, indicating low transmission channel stability. The priority tag values ​​are high, medium, and low, respectively, represented by two binary digits: high is represented by zeros, medium by zeros and one, and low by one. Priority tags are generated by appending a two-digit field to the video segment header and writing the encoded value of the priority tag into this field. The purpose of generating priority tags is to provide a basis for dynamic adjustment in subsequent conflict determination steps and to provide a reference for transmission channel priority adjustments.

[0058] The metadata generated above, including the channel identifier, transmission window base, and priority tag, is integrated with the video segment data to generate a complete data packet. The data packet structure is defined as follows: the channel identifier is written first, occupying four bytes; the transmission window base is written next, occupying four bytes; then the priority tag is written, occupying two bytes; and finally, the video segment data is appended. This is done by sequentially writing the channel identifier, transmission window base, and priority tag into the packet header according to a predefined format, and appending the video segment data to form the encapsulated complete data packet.

[0059] In multi-path parallel transmission scenarios in complex network environments (such as switching between satellite and 4G networks), video streams are segmented into fragments with independent sequence numbers using a specific algorithm, and a dynamic window adjustment mechanism is used to optimize transmission throughput. However, multi-path link fluctuations and acknowledgment (ACK) loss can lead to fragment retransmissions and window expansions, causing cross-channel sequence number conflicts. This can cause the depacketization engine to erroneously discard key frames, compromising data integrity. To address this issue, we process the encapsulated video fragment packets to identify and mark conflicting fragments.

[0060] The conflict perception unit includes the following:

[0061] Extract the sequence offset history of the current channel based on the channel identifier:

[0062] The channel identifier is extracted from the header of the received encapsulated video fragment data packet. The channel identifier is an integer value that uniquely identifies the transmission channel to which the data packet belongs. The channel identifier is extracted by parsing the reserved field in the data packet header, such as the first four bytes, and reading the channel identifier value therein. For each transmission channel, a sequence offset history is maintained. The sequence offset history is a list that records the sequence offsets of the video fragments received by the channel over a period of time. The offset is defined as the difference between the video fragment sequence number and the expected sequence number and is used to reflect the sequence number change trend of the channel. Based on the extracted channel identifier, the corresponding sequence offset history is queried and extracted from the channel management module. The channel management module maintains a mapping table with the channel identifier as the key and the sequence offset history of the channel as the value. The sequence offset history is extracted by searching the mapping table using the channel identifier to obtain the sequence offset history list for the channel.

[0063] Dynamically adjust the conflict determination threshold based on delay jitter:

[0064] The priority tag is extracted from the header of the received encapsulated video fragment data packet. The priority tag is generated by the sender in the path tag unit based on the instantaneous delay jitter at the moment of path switching. The value is high, medium, or low, respectively indicating the real-time stability of the transmission channel. The priority tag is extracted by parsing the reserved fields in the packet header, such as the two bytes after the channel identifier and the transmission window base, and reading the priority tag encoded value. Based on the extracted priority tag, the conflict determination threshold is dynamically adjusted to adapt to the network conditions of the transmission channel. The conflict determination threshold is a numerical value used to determine whether the offset of the video fragment sequence number is abnormal. For example, the dynamic adjustment method may be: if the priority tag is high, the conflict determination threshold is set to 1.0 times the basic conflict determination threshold; if the priority tag is medium, the conflict determination threshold is set to 1.5 times the basic conflict determination threshold; if the priority tag is low, the conflict determination threshold is set to 2.0 times the basic conflict determination threshold. The basic conflict determination threshold is a preset constant, such as 5, which represents the tolerance for sequence number offset under ideal network conditions. The dynamic adjustment process is to select the corresponding multiplication factor based on the priority label, multiply it by the basic conflict determination threshold, and obtain the current conflict determination threshold.

[0065] Determine whether the difference between the shard sequence number and the offset exceeds the threshold:

[0066] Extract the fragment sequence number from the header of the received encapsulated video fragment packet. The fragment sequence number is an integer value that indicates the sequence number of the video fragment in the video stream. The fragment sequence number is added by the sender in the path tag unit. Extracting the fragment sequence number involves parsing the reserved field in the packet header, such as the four bytes after the priority tag, and reading the fragment sequence number value. Based on the starting sequence number of the current depacketization window and the status of the received video fragments, the current expected video fragment sequence number, called the expected sequence number, is calculated. The expected sequence number reflects the sequence number of the next video fragment that the receiver expects to receive. It is calculated by determining the sequence number of the next video fragment to be received based on the current state of the depacketization window. The sequence offset of the current video fragment is calculated. The sequence offset is defined as the difference between the video fragment sequence number and the expected sequence number. This offset is calculated by subtracting the calculated expected sequence number from the extracted fragment sequence number. The sequence offset of the current video fragment is compared with the historical offset in the sequence offset history of the channel to calculate the difference in conflict determination.

[0067] The conflict determination difference is defined as the absolute difference between the current sequence number offset and the average value of the historical offsets. It is calculated as follows: first, the average value of all historical offsets in the sequence number offset history is calculated, then the average value is subtracted from the current sequence number offset, and the absolute value is taken to obtain the conflict determination difference. The conflict determination difference is compared with the dynamically adjusted conflict determination threshold: if the conflict determination difference is greater than the conflict determination threshold, the video segment is determined to be a conflict segment, marked as a conflict, and temporarily stored in the conflict segment buffer; if the conflict determination difference is less than or equal to the conflict determination threshold, the video segment is determined to be non-conflicting and is directly placed in the unpacking queue for subsequent unpacking processing.

[0068] During the fragmentation phase, the system extracts the channel identifier, priority tag, and fragment sequence number for each encapsulated video fragment data packet, extracts the sequence number offset history based on the channel identifier, dynamically adjusts the conflict determination threshold based on the priority tag, calculates the sequence number offset and conflict determination difference, and ultimately determines whether the video fragment is a conflict fragment and performs corresponding processing. The channel identifier ensures the targeted extraction of historical data, the dynamically adjusted conflict determination threshold enhances the ability to adapt to network fluctuations, and the determination of sequence number differences accurately locates abnormal video fragments. These processing results provide the spatiotemporal screening unit with data packets marked as conflict fragments and temporarily stored, ensuring that the subsequent repair process can perform targeted analysis and processing based on the tag information of the conflict perception unit, thereby maintaining the integrity and spatiotemporal alignment of video streams and sensor data in complex network environments in the cloud-based monitoring system.

[0069] Video streams are segmented into fragments with independent sequence numbers and transmitted over multiple paths with dynamically adjusted windows to optimize performance. However, network link fluctuations and lost acknowledgments can trigger fragment retransmissions or window expansions, leading to cross-channel sequence number conflicts and the depacketization engine's inadvertent loss of key frames, compromising data integrity. To address this issue, the spatiotemporal screening unit focuses on filtering and repairing conflicting fragments. By analyzing transmission delays and the spatial consistency of sensor data, it selects repairable fragments, providing reliable input for subsequent sequence number correction and depacketization queue injection, thereby ensuring data transmission integrity and spatiotemporal alignment.

[0070] The spatiotemporal screening unit includes the following:

[0071] Get the deviation between the actual transmission delay and the path switching jitter:

[0072] For each conflicting fragmented data packet, its transmission delay characteristics are evaluated to determine whether the delay anomaly is within the tolerable range.

[0073] Calculation of actual transmission delay:

[0074] The actual transmission delay is defined as the time difference between when a conflicting fragment packet is sent from the sender and when it is received by the receiver. The actual transmission delay is calculated by extracting the send and receive timestamps from the header of the conflicting fragment packet. The send timestamp is the time the fragment was sent when the sender encapsulated the fragment, and the receive timestamp is the time the fragment was received when the receiver received it. The actual transmission delay is calculated by subtracting the send timestamp from the receive timestamp.

[0075] Determination of path switching jitter:

[0076] Path switching jitter indicates the degree of change in channel delay at the moment of path switching. The receiving end determines path switching jitter based on the priority tag extracted from the header of the conflicting fragment packet. The priority tag is generated by the sending end in the path marking unit based on the instantaneous delay jitter at the moment of path switching, and can be high, medium, or low. Depending on the priority tag, the path switching jitter is set to a predefined low jitter value, medium jitter value, or high jitter value, corresponding to high, medium, and low priority tags, respectively. For example, the low jitter value can be set to 10 milliseconds, the medium jitter value to 30 milliseconds, and the high jitter value to 50 milliseconds.

[0077] The priority label reflects the network status of the sender at the moment of path switching. By associating it with the path switching jitter, the receiver can be provided with an expectation of network fluctuations when transmitting conflicting fragmented packets.

[0078] Calculation of deviation:

[0079] Skewness is defined as the difference between actual transmission delay and path switching jitter. Skewness is calculated by first calculating the absolute value of the difference between actual transmission delay and path switching jitter, then dividing this absolute difference by the path switching jitter to obtain a dimensionless ratio, the skewness. A smaller skewness indicates a closer match between actual transmission delay and path switching jitter, and the delay characteristics of conflicting fragmented packets conform to the expected fluctuations during path switching.

[0080] Analyze the spatial motion vector continuity of adjacent slice sensor data:

[0081] The spatial consistency of the conflicting fragmented packets is evaluated using sensor data to ensure that their motion characteristics match the video stream context.

[0082] Extraction of spatial motion vector:

[0083] Each collision fragment packet contains sensor data associated with a video frame, recording spatial information. The sensor data in the collision fragment packet is extracted to calculate the spatial motion vector. The spatial motion vector is a three-dimensional vector that represents the speed of an object in space, with components corresponding to the x, y, and z directions. The sensor data is encapsulated with the video fragment at the transmitter and retrieved by parsing the collision fragment packet at the receiver.

[0084] The spatial motion vector reflects the motion state of an object in a video frame and can be used to evaluate the continuity between conflicting fragment packets and adjacent fragment packets in the spatial context.

[0085] Identification of adjacent fragmented packets:

[0086] The preceding and following adjacent fragment packets are identified based on the fragment sequence number extracted from the header of the conflicting fragment packet. The fragment sequence number is added by the sender to the path marker unit and indicates the order of the fragments in the video stream. The preceding fragment packet is the fragment packet with a sequence number one less than the conflicting fragment packet, and the following fragment packet is the fragment packet with a sequence number one greater than the conflicting fragment packet. If the preceding or following fragment packet has not been received, the conflicting fragment packet is not included in the continuity analysis.

[0087] The spatial motion vectors of the previous fragment data packet and the next fragment data packet can provide context information of the conflicting fragment data packet in the video stream, which is used to evaluate their spatial consistency.

[0088] Calculation of continuity measures:

[0089] The continuity metric is defined as the relative rate of change between the spatial motion vector of the colliding fragment and the spatial motion vector of the adjacent fragment. The continuity metric is calculated by first calculating the modulus of the difference between the spatial motion vector of the colliding fragment and the spatial motion vector of the preceding fragment, then dividing the difference by the modulus of the spatial motion vector of the preceding fragment to obtain a relative rate of change. Similarly, the relative rate of change between the spatial motion vector of the colliding fragment and the spatial motion vector of the succeeding fragment is calculated. Finally, the minimum of these two relative rates is taken as the continuity metric. The modulus of the spatial motion vector is calculated by taking the square root of the sum of the squares of its three components. The continuity metric is a dimensionless ratio; smaller values ​​indicate greater spatial continuity between the colliding fragment and the adjacent fragment.

[0090] By calculating the continuity metric, we can quantify the similarity in motion characteristics between the conflicting fragment packets and the adjacent fragment packets, ensuring that the spatial context of the conflicting fragment packets in the video stream is consistent.

[0091] Filter for repairable fragmented packets:

[0092] Based on the above calculation results, filter conditions are set to filter out repairable fragmented data packets.

[0093] Delay tolerance condition judgment:

[0094] Determine whether the deviation is less than a preset deviation threshold, such as 0.2. If the deviation is less than the deviation threshold, the difference between the actual transmission delay of the conflicting fragmented packet and the path switching jitter is within an acceptable range, meeting the delay tolerance requirement. The delay tolerance condition ensures that the transmission delay of the conflicting fragmented packet is within a controllable range, preventing conflicting fragmented packets with excessive or insufficient delay from being mistakenly identified as repairable fragmented packets.

[0095] Judgment of spatial consistency conditions:

[0096] Determine whether the continuity metric is less than a preset continuity threshold, such as 0.1. If the continuity metric is less than the continuity threshold, it indicates that the conflicting fragment packet has a high degree of continuity with the adjacent fragment packets in terms of spatial motion vectors, meeting the spatial consistency requirement. The spatial consistency condition ensures that the content of the conflicting fragment packet in the video stream matches that of the adjacent fragment packets, avoiding the inclusion of conflicting fragment packets with discontinuous content in the repair process.

[0097] Application of filtering rules:

[0098] If the conflicting fragmented data packet satisfies both the delay tolerance condition and the spatial consistency condition, that is, the deviation is less than the deviation threshold and the continuity metric is less than the continuity metric threshold, then the conflicting fragmented data packet is marked as a repairable fragmented data packet and passed to the path correction unit for sequence number correction; otherwise, the conflicting fragmented data packet is marked as an unrepairable fragmented data packet and discarded.

[0099] Taking transmission delay and spatial consistency into consideration, the repairability of conflicting fragmented data packets can be comprehensively evaluated to ensure that the selected conflicting fragmented data packets are both acceptable in terms of transmission characteristics and match the video stream in terms of content.

[0100] By calculating the deviation between the actual transmission delay and path switching jitter, as well as the spatial motion vector continuity of the sensor data of adjacent fragmented packets in the spatiotemporal screening unit, repairable fragmented packets that meet both delay tolerance and spatial consistency are screened out. The deviation metric quantifies the degree of consistency between the transmission delay of the conflicting fragmented packets and the expected path switching, while the continuity metric ensures the spatial context continuity of the conflicting fragmented packets in the video stream. These screening results provide the path correction unit with reliable repairable fragmented packets, ensuring that subsequent sequence number correction based on the transmission window cardinality difference can accurately restore the fragment order, thereby maintaining the integrity and spatiotemporal alignment of video streams and sensor data in complex network environments in cloud-based monitoring systems.

[0101] The path correction unit solves the sequence number misalignment problem caused by multi-path switching for the repairable fragmented data packets screened out by the time-space screening unit.

[0102] The path correction unit includes the following:

[0103] Step 1. Calculate the transmission window cardinality difference:

[0104] For each repairable fragmented packet, the difference between its transmission window cardinality and the current depacketization window cardinality is calculated to assess the degree of fragment sequence number misalignment. The current depacketization window cardinality is defined as the starting sequence number of the window currently being processed by the depacketization engine. It is maintained by the depacketization engine in real time and reflects the depacketization status of the receiving end. The transmission window cardinality difference is calculated by extracting the transmission window cardinality from the header of the repairable fragmented packet. This value is added by the sender in the path marker unit during fragment encapsulation and represents the fragment window position of the sender. Then, the current depacketization window cardinality is subtracted from the extracted transmission window cardinality to obtain the transmission window cardinality difference.

[0105] The current unpacking window base serves as a benchmark for the receiver's processing progress. Comparing this with the transmission window base of the repairable fragmented data packet quantifies the window offset between the two. The transmission window base difference reflects the relative position of the window to which the fragment belongs and the current unpacking window. A positive value indicates that the fragment comes from a future window, while a negative value indicates that it comes from a past window, intuitively demonstrating the direction and degree of sequence number misalignment.

[0106] Step 2. Reversely deduce the sequence number misalignment caused by multipath switching:

[0107] Based on the transmission window cardinality difference and channel history information, the amount of sequence number misalignment caused by multipath switching is derived and calculated using a nonlinear adjustment strategy. The channel history misalignment is maintained by the channel management module, recording the channel's accumulated sequence number offsets from past multipath switching events. This is updated using the sequence number offset history extracted from the conflict awareness unit. The sequence number misalignment calculation process is as follows: First, the priority tag is extracted from the header of the repairable fragment packet. The priority coefficient is determined based on the priority tag. If the priority tag is high, the priority coefficient is twice the baseline value; if it is medium, it is the baseline value; and if it is low, it is half the baseline value. Next, the absolute value of the transmission window cardinality difference and the absolute value of the channel history misalignment are calculated. The transmission window cardinality difference is then multiplied by the priority coefficient and then divided by the absolute value of the transmission window cardinality difference plus one to obtain the first partial result. The channel history misalignment is multiplied by the priority coefficient and divided by the absolute value of the channel history misalignment plus one to obtain the second partial result. Finally, the first and second partial results are added to obtain the sequence number misalignment.

[0108] For example, the calculation method of the sequence number misalignment can be as follows:

[0109]

[0110] in:

[0111] PT coef : Priority coefficient, fragment transmission priority label (PT) mapping generated by the path label unit:

[0112] PT = high, TP coef =2.0;

[0113] PT = medium, TP coef =1.0;

[0114] PT = low, PT coef =0.5.

[0115] This coefficient reflects the impact of the channel's transmission priority on the misalignment calculation.

[0116] |WBD| and |HDM|: are the transmission window base difference and the absolute value of the channel history misalignment, respectively, used for nonlinear normalization to avoid overfitting caused by a single weight factor. and Nonlinear attenuation is introduced to balance the contribution of current window difference and historical dislocation, ensuring that high-priority channels rely more on real-time data and low-priority channels rely more on historical trends.

[0117] Step 3. Dynamically modify the shard number:

[0118] Based on the calculated sequence number misalignment, the fragment sequence number of the repairable fragment packet is corrected. The corrected fragment sequence number is calculated by extracting the original fragment sequence number from the repairable fragment packet header. This sequence number is encapsulated by the sender in the path tag unit and represents the order of the fragments in the video stream. Then, the sequence number misalignment is subtracted from the original fragment sequence number to obtain the corrected fragment sequence number.

[0119] By subtracting the sequence number misalignment amount and adjusting the fragment sequence number to its correct position in the video stream, the sequence number misalignment caused by multipath switching can be eliminated, ensuring that the fragment sequence number is consistent with the actual transmission order.

[0120] Step 4. Verify the continuity with the current unpacking window:

[0121] Verify that the corrected fragment sequence number is continuous with the current depacketization window to determine whether it should be injected into the depacketization queue. The depacketization window range is defined as the interval from the current depacketization window base to the current depacketization window base plus the window size minus one, with the window size preset by the depacketization engine. The continuity check is performed by checking whether the corrected fragment sequence number meets any of the following conditions: first, it falls within the depacketization window range; second, it is equal to the current depacketization window base minus one, meaning it is continuous with the lower boundary of the window; third, it is equal to the current depacketization window base plus the window size, meaning it is continuous with the upper boundary of the window. If any of these conditions are met, the verification is considered passed; if none of them are met, the fragmented packet is marked as unavailable and discarded.

[0122] The depacketization window range specifies the range of fragment sequence numbers currently acceptable to the receiver. The continuity check ensures that the corrected fragment sequence numbers match the progress of the current depacketization window, avoiding depacketization errors caused by the injection of discontinuous fragments.

[0123] Step 5. Inject unpacking queue:

[0124] For fragmented data packets that pass the continuity check, they are inserted into the correct position in the depacketization queue according to the corrected fragment sequence number, ensuring that the depacketization engine processes them in order. The insertion process involves finding the corresponding sequential position in the depacketization queue based on the corrected fragment sequence number and placing the fragmented data packet there, maintaining sequence number continuity in the queue. Inserting the fragmented data into the depacketization queue in the corrected fragment sequence number order ensures that the fragmented data received by the depacketization engine is in the correct order, maintaining the temporal and spatial consistency of the video stream and sensor data.

[0125] The path correction unit uses the repairable fragmented data packets output by the spatiotemporal screening unit as a basis. By calculating the transmission window cardinality difference and combining the historical channel misalignment with the nonlinear adjustment of the priority coefficient, it derives the sequence number misalignment. It then dynamically corrects the fragment sequence number, verifies its continuity with the current depacketization window, and finally injects the fragmented data packet into the depacketization queue. This process addresses the sequence number misalignment problem of multi-path transmission in cloud-based monitoring systems, integrating real-time and historical data to ensure the integrity and spatiotemporal alignment of video streams and sensor data.

[0126] In cloud-based monitoring systems, the reliable transmission of video streams and sensor data is a core requirement for anomaly identification and compliance forensics, and it is necessary to cope with multi-path transmission challenges in complex network environments, such as cross-channel sequence number conflicts caused by network fluctuations and fragment retransmissions. The aforementioned path marking unit to the cross-path correction unit has completed fragment encapsulation, conflict detection, repairable fragment screening and sequence number correction, and successfully injected the corrected fragment data packets into the depacketization queue, ensuring the initial continuity of the video stream and sensor data. However, continuous fluctuations in the network environment may cause changes in the stability of the transmission channel, and some channels may affect the overall transmission performance due to correction failures. To this end, the window adjustment unit further optimizes the configuration of the transmission channel based on the correction results of the path correction unit, and improves transmission stability and data integrity by dynamically adjusting the window base and transmission priority, providing reliable guarantees for subsequent data processing.

[0127] The window adjustment unit includes the following:

[0128] Statistics of channel stability with successful continuous corrections:

[0129] For each transmission channel, the stability index of its continuous correction success is calculated based on the correction results of the fragments in the path correction unit. The specific processing logic is to maintain a continuous correction success count for the identifier of each transmission channel, and its initial value is set to zero. When the fragment correction of a certain transmission channel is successful, the continuous correction success count of the transmission channel is increased by one; when the fragment correction of the transmission channel in the path correction unit fails, the continuous correction success count of the transmission channel is reset to zero. At the same time, a stability threshold is pre-set as the basis for judging whether the transmission channel is stable. The stability threshold is a fixed value used to compare with the continuous correction success count to determine the stability status of the transmission channel.

[0130] Freeze Window Cardinality Extension:

[0131] For transmission channels that have been continuously corrected successfully, a freeze operation is performed to extend the transmission window cardinality to maintain stability. First, the continuous correction success count for each transmission channel is checked to see if it is greater than or equal to the preset stability threshold. If the continuous correction success count for a transmission channel reaches or exceeds the stability threshold, the transmission window cardinality for that transmission channel is locked to the current value, and any further increases to the transmission window cardinality are prohibited for a preset freeze duration. The freeze duration is a predetermined period of time that ensures that the transmission window cardinality remains constant over time.

[0132] Roll back the window base to the state before the conflict:

[0133] For transmission channels where fragmentation correction fails, the transmission window cardinality is rolled back to its pre-conflict stable state. A pre-conflict window cardinality is maintained for each transmission channel, recording the transmission window cardinality value at the time of the last successful fragmentation correction for that transmission channel. Each time fragmentation correction succeeds for a transmission channel, the pre-conflict window cardinality for that transmission channel is updated to the current transmission window cardinality. If fragmentation correction fails, the pre-conflict window cardinality remains unchanged. If fragmentation correction fails for a transmission channel, the current transmission window cardinality for that transmission channel is adjusted to the recorded pre-conflict window cardinality.

[0134] The pre-collision window base reflects the last stable state of the transmission channel before a collision. Rolling back the transmission window base to the pre-collision window base eliminates the sequence number misalignment caused by expanding the window base, thereby restoring the stable transmission capacity of the transmission channel.

[0135] Lower the transmission priority for subsequent path switching:

[0136] For transmission channels where fragment correction fails, the transmission priority is adjusted to reduce traffic allocation to unstable transmission channels. A transmission priority is maintained for each transmission channel, with its initial value set to a baseline value. When fragment correction fails for a transmission channel, the transmission priority of that transmission channel is lowered. The specific adjustment method is to multiply the current transmission priority by an attenuation factor. This attenuation factor is determined through a nonlinear calculation method based on the size of the continuous correction success count, ensuring that the reduction in transmission priority is related to the degree of instability of the transmission channel. When fragment correction is successful for a transmission channel, its transmission priority is slowly restored. The specific adjustment method is to add a recovery factor to the current transmission priority. This recovery factor is calculated based on the size of the continuous correction success count, ensuring that the transmission priority is gradually increased only when the transmission channel remains stable. In subsequent path switching, the traffic ratio is allocated according to the transmission priority of each transmission channel, and transmission channels with lower transmission priorities receive less traffic allocation.

[0137] For example, the transmission priority can be dynamically adjusted in the following ways:

[0138] A transmission priority TP is maintained for each channel, with an initial value of 1.0, indicating the default priority.

[0139] When the fragmentation correction fails, reduce TP. The calculation formula is:

[0140]

[0141] Parameter explanation: CSC is the continuous correction success count of the current channel (0 if failed);

[0142] in the formula The priority of unstable channels is quickly attenuated using an exponential function.

[0143] When the fragmentation is corrected successfully, TP is slowly restored. The calculation formula is:

[0144]

[0145] TP new Indicates the transmission priority of a transmission channel at the current moment, TP current Indicates the new value obtained after adjusting the transmission priority of a transmission channel, that is, the updated transmission priority.

[0146] in the formula is the recovery factor.

[0147] The recovery process is smooth and controlled by the CSC, ensuring that priority is gradually increased only when the channel remains stable.

[0148] By lowering the transmission priority of transmission channels where fragment correction fails, we can reduce traffic allocation to unstable transmission channels, thereby prioritizing stable transmission channels with higher transmission priorities. This dynamic adjustment optimizes the overall transmission strategy and improves data transmission reliability. The mechanism of slowly restoring transmission priority avoids frequent adjustments caused by short-term fluctuations.

[0149] Based on the fragment correction results from the path correction unit, the window adjustment unit dynamically optimizes the transmission channel by freezing the transmission window base extension for transmission channels that have successfully corrected fragments, and rolling back the transmission window base to the pre-conflict state and lowering the transmission priority of transmission channels that have failed fragment correction. This ensures stable transmission and spatiotemporal alignment of video streams and sensor data in complex network environments.

[0150] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0151] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0152] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0153] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article or device. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0154] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. Cloud monitoring system based on artificial intelligence analysis, characterized by: Including steps: Path marking unit, conflict perception unit, spatiotemporal screening unit, cross-path correction unit, and window adjustment unit; Path marking unit: When encapsulating fragments, it adds a channel identifier and the current transmission window base to each fragment, and generates a fragment transmission priority label based on the instantaneous delay jitter at the moment of path switching; The transmission window base is defined as the starting sequence number of the current transmission window; Conflict perception unit: When receiving a fragment, it extracts the sequence number offset history of the current channel based on the channel identifier, and dynamically adjusts the conflict judgment threshold based on the delay jitter. If the difference between the fragment sequence number and the offset exceeds the threshold, it is marked as a conflicting fragment and temporarily stored; The spatiotemporal screening unit selects repairable fragments that meet both delay tolerance and spatial consistency by calculating the deviation between the actual transmission delay and path switching jitter of conflicting fragments and analyzing the continuity of the spatial motion vectors of the sensor data of adjacent fragments. Cross-path correction unit: Based on the transmission window base difference of the repairable fragments, it reversely deduces the sequence number misalignment caused by multipath switching, dynamically corrects the fragment sequence number, verifies its continuity with the current depacketization window, and injects it into the depacketization queue after passing. The cross-path correction unit includes the following: For repairable fragmented packets, the difference between the transmission window base and the current depacketization window base is calculated, combined with the channel's historical misalignment and nonlinear adjustment of the priority coefficient to derive the sequence number misalignment. The fragment sequence number is dynamically corrected based on the sequence number misalignment. The corrected fragment sequence number is verified for continuity with the current depacketization window, and the fragmented packets that pass the verification are injected into the depacketization queue. The channel management module maintains the channel's historical misalignment, recording the channel's accumulated sequence number offsets from past multipath switching. This is updated using the sequence number offset history extracted by the conflict sensing unit. Window adjustment unit: Based on the fragment correction results, it freezes the window base expansion of the channel that has been successfully corrected, and rolls back the window base to the pre-conflict state for the channel that has failed to be corrected, and reduces the transmission priority of subsequent path switching.

2. The cloud-based monitoring system based on artificial intelligence analysis according to claim 1, characterized in that: The path marking unit includes the following: During the fragment encapsulation stage, a channel identifier, a transmission window cardinality, and a priority tag are attached to each video fragment. The channel identifier is defined as the unique identifier of the transmission channel, the transmission window cardinality is defined as the starting sequence number of the current transmission window, and the priority tag is determined based on the comparison result of the instantaneous delay jitter with a preset threshold. The instantaneous delay jitter is defined as the absolute value of the difference in transmission delays between adjacent video fragments.

3. The cloud-based monitoring system based on artificial intelligence analysis according to claim 2, characterized in that: The conflict perception unit includes the following: The sequence offset history is queried according to the channel identifier, and the conflict judgment threshold is dynamically adjusted according to the priority tag. The conflict judgment difference between the sequence offset of the fragment sequence number and the expected sequence number, as well as the sequence offset and the historical offset average value in the sequence offset history are calculated. When the conflict judgment difference is greater than the conflict judgment threshold, the video fragment is marked as a conflicting fragment and temporarily stored. When the conflict judgment difference is not greater than the conflict judgment threshold, the video fragment is put into the depacketization queue.

4. The cloud-based monitoring system based on artificial intelligence analysis according to claim 3 is characterized in that: The spatiotemporal screening unit includes the following: By calculating the deviation between the actual transmission delay and the path switching jitter and analyzing the spatial motion vector continuity of the sensor data of adjacent slices, we screen out repairable slices that meet both delay tolerance and spatial consistency. When the deviation is less than the deviation threshold and the spatial motion vector continuity is less than the continuity measurement threshold, the conflicting slice is marked as repairable.

5. The cloud-based monitoring system based on artificial intelligence analysis according to claim 4 is characterized in that: The spatiotemporal screening unit also includes the following: The actual transmission delay is defined as the difference between the receiving timestamp and the sending timestamp. The path switching jitter is determined based on the priority tag. The deviation is calculated as the absolute value of the difference between the actual transmission delay and the path switching jitter divided by the path switching jitter. The spatial motion vector continuity is obtained by calculating the minimum relative change rate of the spatial motion vector of the conflicting slice and the adjacent slice.

6. The cloud monitoring system based on artificial intelligence analysis according to claim 1, characterized in that: The cross-path correction unit also includes the following: The processing process of the corrected fragment sequence number is as follows: extract the original fragment sequence number from the header of the repairable fragment data packet; then, subtract the sequence number misalignment amount from the original fragment sequence number to obtain the corrected fragment sequence number.

7. The cloud-based monitoring system based on artificial intelligence analysis according to claim 6, characterized in that: The cross-path correction unit also includes the following: The judgment process of continuity check is as follows: check whether the corrected fragment sequence number meets any of the following conditions: first, it falls within the depacketization window range; second, it is equal to the current depacketization window base minus one, that is, it is continuous with the lower boundary of the window; third, it is equal to the current depacketization window base plus the window size, that is, it is continuous with the upper boundary of the window; if any condition is met, it is considered to have passed the check; if none of the conditions are met, it is marked as unavailable and the corresponding fragment data packet is discarded.

8. The cloud-based monitoring system based on artificial intelligence analysis according to claim 7, characterized in that: The cross-path correction unit also includes the following: The priority coefficient is obtained by mapping the fragment transmission priority labels generated by the path marking unit.

9. The cloud-based monitoring system based on artificial intelligence analysis according to claim 1, characterized in that: The window adjustment unit includes the following: According to the fragmentation correction results, the stability of the transmission channel is judged by counting the continuous correction success count. The transmission window base extension is frozen for the transmission channel whose continuous correction success count reaches the stability threshold. The transmission window base is rolled back to the pre-conflict window base for the transmission channel where fragmentation correction fails, and the transmission priority is adjusted through nonlinear calculation.

Images