Methods for Identifying and Preventing Malicious Payload Escape in IoT Protocols

By stripping physical layer header information in IoT boundary defense, eliminating the need for full stream reassembly, and using byte relative shift values ​​and time intervals to determine malicious payload escape, the problem of balancing resource consumption and identification accuracy in IoT boundary defense is solved, achieving high-precision malicious payload identification and real-time assurance.

CN123122927APending Publication Date: 2026-11-17BEIJING XINAN BORUI TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611325807.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

IoT boundary defense struggles to eliminate the resource overhead of full stream reassembly while maintaining the accuracy of malicious payload escape identification. Furthermore, occasional disorder in physical channels can easily lead to the false interception of legitimate control commands, and long-term, low-frequency, covert fragmented escapes create blind spots in security defense.

Method used

By stripping the physical layer header information from the network frames entering the boundary, the full stream reassembly is eliminated. The expected logical boundary termination line value is determined by the arithmetic sum of the absolute byte relative shift value and the length indicator. Combined with the state feature value of the application layer specific write control function code, a high-risk escape identification instruction is generated. The session is blocked by comparing the arrival time interval of the current adjacent messages with the industry benchmark jitter threshold. The dynamic attenuation operator is injected into the context state tracking flow table to relax the security tolerance boundary.

Benefits of technology

On resource-constrained IoT edge nodes, this approach balances the accuracy and real-time performance of malicious payload escape identification, reduces memory consumption, minimizes false interception of legitimate control commands, eliminates security blind spots, and meets the real-time requirements of industrial control networks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122927A_ABST
    Figure CN123122927A_ABST
Patent Text Reader

Abstract

This application provides a method for identifying and defending against malicious payload escape in IoT protocols, relating to the field of network communication security technology. It involves stripping the physical layer header information from network frames flowing into the boundary to obtain the data payload stream; eliminating full stream reassembly, and determining the expected logical boundary termination line value by the sum of the absolute byte relative shift value and the length indicator; obtaining the topology displacement boundary variance by subtracting the starting byte relative shift value of the next time-series network frame from the expected logical boundary termination line value of the previous frame; generating a high-risk escape identification command when the variance is less than zero and meets multiple conditions for joint judgment; blocking the current session based on a comparison of the arrival time interval of adjacent messages with an industrial benchmark jitter threshold, or by injecting a dynamic attenuation operator to relax the security tolerance boundary, blocking the current session when continuous anomalies occur within the observation window. Based on the above, it can improve the problem of difficulty in balancing identification accuracy with resource overhead and latency in existing technologies.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of network communication security technology, specifically to a method for identifying and defending against malicious payload escape in Internet of Things (IoT) protocols. Background Technology

[0002] The Internet of Things (IoT) communication networks are widely deployed in fields such as industrial automation control, smart warehousing, and smart city edge nodes. Industrial control bus protocols and smart device-specific protocols carry high-real-time, high-frequency control command flows. Edge firewalls or security gateways located at the boundaries of these networks constitute the first line of defense against unauthorized information penetration. Conventional boundary defense schemes rely on pre-set signature databases and fixed sliding window mechanisms to match incoming network packets one by one.

[0003] As attack methods evolve, attackers exploit inherent vulnerabilities in the fragmentation and reassembly rules of IoT protocols to carry out malicious payload escapes. They deliberately construct distorted message sequences containing out-of-order protocol fragments, overwritten overlapping segments, and abnormal message headers, hiding malicious injection code in multiple seemingly legitimate discrete frames in an attempt to bypass static feature matching. Existing IoT boundary defense solutions in the industry generally rely on a design paradigm of reassembly first, then auditing. That is, boundary defense nodes must allocate a full stream reassembly buffer in physical memory to completely restore all discrete fragmented messages into a continuous data stream before performing deep packet inspection and semantic auditing.

[0004] In IoT edge gateways and access nodes where hardware resources are highly limited and real-time control interactions are crucial, the aforementioned design paradigm directly leads to an adversarial relationship between security identification accuracy and edge engineering resource bottlenecks. Firstly, if the firewall forces a frame-by-frame unpacking and reassembly strategy, the extremely small memory capacity of IoT devices cannot support the state cache of high-concurrency sessions, causing a state cache explosion, resulting in firewall crashes or operational response delays exceeding 50 milliseconds, leading to industrial production line shutdowns. Secondly, if the firewall abandons long context auditing to meet high real-time control interactions, malicious payloads can successfully penetrate the boundary and enter the device's internal operations, resulting in control theft. Furthermore, in real industrial bus or wireless IoT environments, frequent natural packet loss and retransmission can easily lead to widespread false interception of legitimate control commands, or leave a security blind spot of protocol parsing asymmetry for maliciously constructed long-period, low-frequency, covert fragmentation escapes.

[0005] In other words, existing technologies have the problem that IoT boundary defenses cannot simultaneously eliminate the resource overhead of full stream reassembly while ensuring the accuracy of malicious payload escape identification. Furthermore, occasional disorder in physical channels can easily lead to the misinterpretation of legitimate control commands, and long-term, low-frequency, covert fragment escapes create blind spots in security defenses. Summary of the Invention

[0006] In view of this, the purpose of this application is to provide a method for identifying and defending against malicious payload escape in Internet of Things protocols, so as to improve the problem of difficulty in balancing security identification accuracy with resource consumption and latency in the prior art.

[0007] To achieve the above objectives, this application adopts the following technical solution:

[0008] Methods for identifying and preventing malicious payload escape in IoT protocols include:

[0009] The physical layer header information is stripped from the current network frame flowing into the boundary to obtain the data payload stream;

[0010] To avoid full stream reassembly, the arithmetic sum of the absolute byte relative shift value of the current network frame and the length indicator is used to determine the expected logical boundary termination line value, which is then written into the context state tracking flow table along with the state feature value of the application layer specific write control function code.

[0011] The topology displacement boundary variance is obtained by subtracting the relative shift value of the starting byte of the next time-series network frame from the value of the expected logical boundary termination line of the previous frame. When the topology displacement boundary variance is less than zero and its absolute value is greater than the safety tolerance boundary, the historical overlap frequency does not fluctuate monotonically within a consecutive preset number of message periods, and the state feature value is activated, a high-risk escape identification instruction is generated.

[0012] The current session is blocked based on a comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, or the high-risk escape identification command is suspended and the dynamic attenuation operator is injected in reverse into the context state tracking flow table to relax the security tolerance boundary. The current session is blocked when the variance of the topological displacement boundary is continuously less than zero and the state feature value is continuously active within the extended observation window.

[0013] In a preferred embodiment of this application, in the aforementioned method for identifying and defending against malicious payload escape in IoT protocols, the step of determining the expected logical boundary termination line value by the arithmetic sum of the absolute byte relative shift value of the current network frame and the length indicator further includes:

[0014] Align the network layer protocol fragmentation boundary with the application layer specific write control function code features using syntax tree topology alignment, and construct a sequence of relative offset matrix matrices for protocol fields.

[0015] The state feature value is used to characterize the write / overwrite action of the current industrial protocol control plane for registers or sensitive configuration parameters. The state feature value transitions from an inactive state to an active state when the application layer packet header of the current network frame carries the register write / overwrite opcode.

[0016] In a preferred embodiment of this application, after the step of subtracting the relative shift value of the start byte of the next time-series network frame from the value of the expected logical boundary termination line of the preceding frame to obtain the topology displacement boundary variance in the above-mentioned method for identifying and defending against malicious payload escape in IoT protocols, the method further includes:

[0017] When the variance of the topology displacement boundary is equal to zero, it is determined that the current session is in a linearly increasing steady-state flow condition, the abnormal counter of the current session is cleared, and the next time-series network frame is forwarded normally.

[0018] When the variance of the topological displacement boundary is greater than zero, it is determined that there is a fragmentation hole in the physical transmission timing of the current protocol stream. The current protocol stream is then attached to the timing monitoring queue, waiting for subsequent messages to cover and fill the fragmentation hole.

[0019] In a preferred embodiment of this application, in the above-mentioned method for identifying and protecting against malicious payload escape in IoT protocols, the security tolerance boundary is 16 bytes and the preset quantity is 5.

[0020] The historical overlap frequency is counted as follows: the number of times the variance of the topological displacement boundary is less than zero in the current session is counted in terms of message period. When the value sequence of the historical overlap frequency is neither monotonically increasing nor monotonically decreasing in 5 consecutive message periods, it is determined that the historical overlap frequency exhibits non-monotonic fluctuation.

[0021] In a preferred embodiment of this application, in the aforementioned method for identifying and protecting against malicious payload escape in IoT protocols, the step of blocking the current session or suspending the high-risk escape identification command based on a comparison between the arrival time interval of current adjacent packets and an industrial benchmark jitter threshold includes:

[0022] When the arrival time interval of the current adjacent messages is within the range of the industrial reference jitter threshold, a blocking instruction is issued according to the high-risk escape identification instruction to cut off the transmission control protocol connection of the current session and shield the corresponding industrial bus logical node.

[0023] When the arrival time interval between adjacent messages is greater than the industrial benchmark jitter threshold, it is determined that the current network topology has caused an occasional large-scale disorder caused by physical channel degradation. The high-risk escape identification instruction is suspended, the current network frame is temporarily stored in the temporary storage queue, and the remaining part of the current network frame after stripping the application layer semantics is forwarded to the internal industrial automation network. The observation window is extended to the next 3 consecutive message circulation cycles.

[0024] In a preferred embodiment of this application, in the aforementioned method for identifying and protecting against malicious payload escape in IoT protocols, the dynamic attenuation operator is generated as follows:

[0025] The ratio of the industrial benchmark jitter threshold to the current adjacent message arrival time interval is determined as the dynamic attenuation operator;

[0026] The dynamic attenuation operator is injected into the context state tracking flow table in an inverse relationship, so that the security tolerance boundary is monotonically widened as the arrival time interval of the current adjacent message increases;

[0027] When the current adjacent message arrival time interval falls back to within the industrial reference jitter threshold range, the injection of the dynamic attenuation operator is canceled, and the safety tolerance boundary is returned to its original value.

[0028] In a preferred embodiment of this application, in the above-mentioned method for identifying and defending against malicious payload escape in IoT protocols, the current adjacent message arrival time interval is obtained by collecting the absolute timestamps of two adjacent fragmented messages arriving at the access node and subtracting them using the timestamp calibration unit built into the communication network card. The unit of the current adjacent message arrival time interval is milliseconds, and the industrial benchmark jitter threshold is 120 milliseconds.

[0029] This application also provides an IoT protocol malicious payload escape identification and security defense device, including: a data acquisition module, used to strip the physical layer header information of the current network frame flowing into the boundary to obtain the data payload stream; a protocol parsing module, used to avoid full stream reassembly, determine the expected logical boundary termination line value by arithmetic sum of the absolute byte relative shift value and the length indicator of the current network frame, and write the state feature value of the application layer specific write control function code into the context state tracking flow table; and a state auditing module, used to obtain the topology displacement boundary by subtracting the starting byte relative shift value of the next time-series network frame from the expected logical boundary termination line value of the previous time-series frame. When the variance of the topological displacement boundary is less than zero and its absolute value is greater than the safety tolerance boundary, the frequency of historical overlap does not fluctuate monotonically within a consecutive preset number of message periods, and the state feature value is activated, a high-risk escape identification instruction is generated; the safety execution module is used to block the current session based on the comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, or to suspend the high-risk escape identification instruction and inject the dynamic attenuation operator in reverse into the context state tracking flow table to relax the safety tolerance boundary, and block the current session when the variance of the topological displacement boundary is continuously less than zero and the state feature value is continuously activated within the extended observation window.

[0030] Based on the above, this application also provides an electronic device, including: a memory for storing a computer program; and a processor connected to the memory for executing the computer program stored in the memory to implement the above-mentioned method for identifying and defending against malicious payload escape in Internet of Things protocols.

[0031] Based on the above, this application also provides a computer-readable storage medium storing a computer program, which executes the various steps of the above-described method for identifying and defending against malicious payload escape in Internet of Things protocols when it is run.

[0032] The above-described embodiments of this disclosure have the following beneficial effects:

[0033] The IoT protocol malicious payload escape identification and security defense method provided in this application firstly strips the physical layer header information from the current network frame flowing into the boundary to obtain the data payload stream, allowing protocol parsing to directly target the standard network layer and transport layer data payloads without touching the physical layer overhead. Secondly, it eliminates the need for full stream reassembly, directly determining the expected logical boundary termination line value by arithmetic summing the absolute byte relative shift value and the length indicator, reducing high-dimensional spatial feature matching to one-dimensional linear temporal differential judgment, without reserving linearly growing physical memory space for concurrent sessions. Then, it triggers the corresponding judgment path based on the polarity of the topological displacement boundary variance, and jointly judges the overlap amplitude, historical overlap frequency, and application layer specific write control function code state feature value in the overlap direction to establish a cross-layer criterion for network layer protocol fragmentation boundary and application layer sensitive write control semantics. Finally, it selects direct blocking or logical backoff based on the comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, and injects the dynamic attenuation operator in reverse into the context state tracking flow table to relax the security tolerance boundary, forming a feedback collaborative closed loop with bidirectional dependence between steps. Based on the above, since the physical overhead of full stream reassembly buffer is eliminated and only instantaneous field offset differential comparison is performed, boundary forwarding does not need to wait for fragment reassembly. Thus, the risk of state cache explosion can be eliminated and boundary forwarding latency can be reduced. In this way, it is possible to balance the accuracy and real-time performance of malicious payload escape identification on resource-constrained IoT edge nodes, thereby improving the problem of difficulty in balancing identification accuracy, resource overhead, and latency in existing technologies.

[0034] Meanwhile, because the dynamic attenuation operator monotonically relaxes the security tolerance boundary as the time interval between the arrival of adjacent messages increases, and cancels the injection and returns to the original value after the time interval falls back, the judgment threshold and the physical channel jitter amplitude form an inverse coupling. In this way, the judgment divergence caused by physical channel noise can be suppressed, and the misinterpretation of legitimate control commands caused by natural packet loss and retransmission due to strong electromagnetic interference in industrial sites can be reduced. Furthermore, when interleaved overwriting continues to exist within the extended observation window and the write control state continues to be active, the backoff is terminated and the blocking is implemented, which prevents long-period, low-frequency covert fragmentation from accumulating and penetrating through the sliding window mechanism, thus eliminating the security defense blind spot of protocol parsing asymmetry.

[0035] Furthermore, in mainstream industrial gateway environments, the physical memory space occupied per thousand concurrent sessions is reduced from 128MB based on full stream reassembly in existing technologies to 4.2MB, resulting in a reduction of over 96% in physical memory consumption; the latency of edge forwarding is reduced from 45 to 120 milliseconds in existing technologies to 8 to 15 microseconds, meeting the 50-millisecond real-time requirement of industrial control networks; and in high-noise industrial environments, the accuracy of malicious payload escape identification is improved from 62% in existing technologies to 99.4%. Attached Figure Description

[0036] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0037] Figure 1 A flowchart illustrating the method for identifying and protecting against malicious payload escape in IoT protocols provided in this application embodiment;

[0038] Figure 2 A schematic diagram of the structure of the IoT protocol malicious payload escape identification and security defense device provided in the embodiments of this application;

[0039] Figure 3 A structural block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0041] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0042] The IoT protocol malicious payload escape identification and security defense method provided in this application operates on the protocol parsing plane of an IoT edge firewall or security gateway, reconstructing the firewall's boundary flow processing rules without changing standard network hardware. The method is deployed on an IoT edge firewall node, which is physically connected in series between an industrial fieldbus and a remote management platform. The edge firewall node has built-in data acquisition, protocol parsing, status auditing, and security execution modules. IoT protocol data packet sequences flow from the remote management platform into the edge firewall node via a wireless IoT link, are audited, and then forwarded to the internal industrial automation network. The method is applicable to scenarios including malicious code injection in industrial control bus protocols such as Modbus and protocols specific to smart devices.

[0043] In the initial state, the edge firewall node allocates a context state tracking flow table in physical memory for the current session. The expected logical boundary termination line value field in the context state tracking flow table is initially zero, the exception counter is initially zero, the application layer specific write control function code status feature value is initially inactive, and the security tolerance boundary retains its original 16-byte value.

[0044] S1, strip the physical layer header information from the current network frame flowing into the boundary to obtain the data payload stream.

[0045] In this embodiment, the data acquisition module captures the sequence of IoT protocol data packets flowing into the boundary in real time, extracts the currently arriving network frames, strips the physical layer header information, and obtains the corresponding standard network layer and transport layer data payload streams as the original data source for subsequent logical analysis. In other words, the data payload stream is the standard network layer and transport layer data payload stream.

[0046] It should be noted that the source and form of the data payload stream are not limited and can be selected according to the type of IoT protocol actually deployed. For example, the data acquisition module receives incoming network frames through the communication network card of the edge firewall node. The network frames can be industrial control bus protocol messages such as Modbus, or smart device-specific protocol messages.

[0047] S2, exempt from full stream reassembly, determine the expected logical boundary termination line value by arithmetic sum of the absolute byte relative shift value of the current network frame and the length indicator, and write it into the context state tracking flow table along with the state feature value of the application layer specific write control function code.

[0048] In this embodiment, the protocol parsing module receives the data payload stream, performs dephysicalization processing on the data payload stream, skips the full stream reassembly buffer, and does not store the payload data in the full stream reassembly. The protocol parsing module reads the absolute byte relative shift value of the current network frame in the current session data stream as a first feature parameter, and extracts the length indicator from the control header of the current network frame as a constraint rule. Both the absolute byte relative shift value and the length indicator are in bytes. The full stream reassembly includes the reassembly of network layer fragmented packets and the stream reassembly of transport layer session packets.

[0049] Based on this, the protocol parsing module uses the arithmetic sum of the absolute byte relative shift value and the length indicator to map and determine the expected logical boundary termination line value of the current network frame, which is used as the second feature parameter.

[0050] Simultaneously, the protocol parsing module performs syntax tree topology alignment between the network layer protocol fragmentation boundary and the application layer specific write control function code features, constructs a sequence of relative offset matrices for protocol fields, and synchronously writes the state feature values ​​of the application layer specific write control function codes into the context state tracking flow table. These state feature values ​​characterize the write / overwrite actions of the current industrial protocol control plane for registers or sensitive configuration parameters. For example, the state feature value transitions from an inactive state to an active state when the application layer header of the current network frame carries a register write / overwrite opcode, and remains inactive when the application layer header does not carry a register write / overwrite opcode.

[0051] Regarding S2, it should be noted that the specific method of syntax tree topology alignment is not limited and can be selected according to actual needs. For example, in an alternative implementation, the protocol parsing module first performs field-level syntax tree parsing on the network layer protocol fragmentation boundary to obtain the fragmentation offset node sequence, and then performs function code-level syntax tree parsing on the application layer message header to obtain the write control function code node sequence. Then, the two node sequences are topologically aligned according to byte offset positions. Fragmentation boundary nodes and write control function code nodes in the same byte interval occupy the same matrix row in the protocol field relative offset matrix sequence. In this way, the fragmentation boundary information of the network layer and the sensitive write control semantics of the application layer form a one-to-one cross-layer mapping relationship in the byte dimension.

[0052] S3, the topology displacement boundary variance is obtained by subtracting the relative shift value of the starting byte of the next time-series network frame from the value of the expected logical boundary termination line of the previous sequence. When the topology displacement boundary variance is less than zero and its absolute value is greater than the safety tolerance boundary, the historical overlap frequency does not fluctuate monotonically within a consecutive preset number of message periods, and the state feature value is activated, a high-risk escape identification instruction is generated.

[0053] In this embodiment, when the next time-series network frame arrives, the state audit module extracts the relative shift value of the starting byte of the next time-series network frame in real time, and performs a numerical logic comparison between the relative shift value of the starting byte and the value of the preceding expected logical boundary termination line stored in the context state tracking flow table to construct a one-dimensional linear state shift trajectory sequence and obtain the topological displacement boundary variance. The topological displacement boundary variance is obtained by subtracting the value of the preceding expected logical boundary termination line from the relative shift value of the starting byte.

[0054] Specifically, when the topology displacement boundary variance is equal to zero, the current session is determined to be in a linearly increasing steady-state flow condition. The state audit module clears the abnormal counter of the current session and performs normal forwarding for the next time-series network frame. When the topology displacement boundary variance is greater than zero, the current protocol stream is determined to have fragmentation holes in the physical transmission timing. The state audit module attaches the current protocol stream to the timing monitoring queue, waiting for subsequent packets to cover and fill the fragmentation holes. When the topology displacement boundary variance is less than zero, the current session is determined to have forward interleaving or unordered overwriting of payloads. The state audit module takes the absolute value of the topology displacement boundary variance to obtain the value of the current overlapping bytes.

[0055] Furthermore, the state audit module retrieves the historical overlap frequency of the current session and performs a multi-condition joint determination based on the value of the overlapping byte, the historical overlap frequency, and the state feature value. If all three conditions are met simultaneously—the value of the overlapping byte is greater than the security tolerance boundary, the historical overlap frequency exhibits non-monotonic fluctuations within a preset number of consecutive message periods, and the state feature value is in an active state—it is determined that there is intentionally constructed malicious payload escape injection behavior in the current data stream. The state audit module then generates a high-risk escape identification command and sends it to the security execution module.

[0056] In this embodiment, the security tolerance boundary is 16 bytes, and the preset quantity is 5. The historical overlap frequency is counted as follows: the number of times the variance of the topology displacement boundary is less than zero in the current session is counted in terms of message period. When the value sequence of the historical overlap frequency is neither monotonically increasing nor monotonically decreasing within 5 consecutive message periods, it is determined that the historical overlap frequency exhibits non-monotonic fluctuation. It should be noted that overlap caused by natural packet loss and retransmission usually exhibits a regular sequence of monotonically increasing or monotonically decreasing within adjacent message periods, while artificially constructed escape injection, as a boundary detection logic, will alternately change the overlap amplitude within consecutive message periods, exhibiting non-monotonic fluctuation. Based on this difference, occasional physical overlap can be distinguished from intentionally constructed distortion overwrite.

[0057] In an alternative implementation, the safety tolerance boundary and the preset number can be adjusted according to the message length distribution and control command frequency of the deployed industrial protocol. For example, the safety tolerance boundary can be selected between 8 bytes and 32 bytes, and the preset number can be selected between 3 and 8. The specific values ​​need to be calibrated in conjunction with the on-site working conditions.

[0058] S4, based on the comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, block the current session, or suspend the high-risk escape identification instruction and inject the dynamic attenuation operator in reverse into the context state tracking flow table to relax the security tolerance boundary. If the variance of the topological displacement boundary is continuously less than zero and the state feature value is continuously active within the extended observation window, block the current session.

[0059] In this embodiment, the security execution module receives the high-risk escape identification command and initiates a dynamic sliding window mechanism to eliminate the risk of false alarms caused by natural packet loss and retransmission due to strong electromagnetic interference in the industrial environment. The security execution module retrieves the current adjacent packet arrival time interval recorded at the firewall interface for the current session. This current adjacent packet arrival time interval is obtained by subtracting the absolute timestamps of two adjacent sets of fragmented packets arriving at the access node, collected by the timestamp calibration unit built into the communication network card. The unit of the current adjacent packet arrival time interval is milliseconds. The industrial benchmark jitter threshold is 120 milliseconds.

[0060] Specifically, when the arrival time interval between current adjacent messages is within the industrial reference jitter threshold range, the security execution module directly issues a blocking command based on the high-risk escape identification command, severing the transmission control protocol connection of the current session and shielding the corresponding industrial bus logical node. The full name of the transmission control protocol is Transmission Control Protocol, abbreviated as TCP.

[0061] When the arrival time interval of the current adjacent packets exceeds the industrial benchmark jitter threshold, it is determined that the current network topology has experienced sporadic large-scale out-of-order packets caused by physical channel degradation. The security execution module initiates a logical backoff procedure and performs the following three actions in sequence: S41, suspends the high-risk escape identification instruction, performs conditional release on the current network frame, temporarily stores the current network frame in the temporary storage queue of the edge gateway to maintain the session's liveness, and forwards the remaining part of the current network frame after stripping application layer semantics to the internal industrial automation network; S42, extends the observation window of the current session to the next 3 consecutive packet flow cycles; S43, generates a dynamic attenuation operator based on the ratio of the industrial benchmark jitter threshold to the arrival time interval of the current adjacent packets, and injects the dynamic attenuation operator in reverse in place into the context state tracking flow table according to the inverse proportional relationship, so that the security tolerance boundary monotonically widens as the arrival time interval of the current adjacent packets increases, widens the logical boundary elastic threshold of subsequent packets, and suppresses the judgment divergence caused by physical channel noise.

[0062] Within the extended observation window, if the relative shift value of the starting byte of subsequent network frames is consistently less than the value of the preceding expected logical boundary termination line, the variance of the topology displacement boundary remains less than zero, and the state characteristic value in the current session remains active, then it is determined that the current interleaving overwrite exhibits a deliberately constructed high-density burst malicious payload escape injection characteristic. The logical backoff procedure terminates, the security execution module directly issues a blocking command, cuts off the transmission control protocol connection of the current session, and shields the corresponding industrial bus logical node.

[0063] Furthermore, when the arrival time interval of the current adjacent message falls back to within the range of the industrial reference jitter threshold, the security execution module cancels the injection of the dynamic attenuation operator and returns the security tolerance boundary to its original value, i.e., 16 bytes. The current session returns to the linearly increasing steady-state flow condition, and the entire dynamic closed loop is reset.

[0064] The following example illustrates the operation flow of this application embodiment using a complete message cycle sequence.

[0065] Assuming the current session is in a linearly increasing steady-state flow condition during the 99th message cycle, the preceding expected logical boundary termination line value stored in the context state tracking flow table is 1024 bytes, the exception counter is zero, the state characteristic value is inactive, and the security tolerance boundary retains its original 16-byte value.

[0066] In the 100th message cycle, the next time-series network frame arrives. The protocol parsing module reads the absolute byte relative shift value of the next time-series network frame as 1024 bytes and the length indicator as 512 bytes. After arithmetic summation of the two, the expected logical boundary termination line value is updated to 1536 bytes and written to the context state tracking flow table. The state auditing module extracts the start byte relative shift value of the next time-series network frame, which is measured to be 1024 bytes. The difference between this and the previous expected logical boundary termination line value of 1024 bytes is zero bytes, indicating a linear steady state. The state auditing module clears the anomaly counter and performs normal forwarding of the next time-series network frame.

[0067] In the 101st message cycle, the measured relative shift value of the starting byte of the next time-series network frame is 1500 bytes, and the expected logical boundary termination value is 1536 bytes. The difference yields a topology displacement boundary variance of -36 bytes. The state audit module takes the absolute value of this topology displacement boundary variance, finding the current overlapping byte value to be 36 bytes, exceeding the original 16-byte security tolerance boundary. The state audit module retrieves the historical overlap frequency of the current session, which exhibits non-monotonic fluctuations over five consecutive message cycles from the 97th to the 101st message cycle. Simultaneously, the protocol parsing module detects that the application layer header of the next time-series network frame carries a register write-overwrite opcode, and the state characteristic value transitions from inactive to active. When all three conditions are met, the state audit module generates the high-risk escape identification instruction and sends it to the security execution module.

[0068] While performing the 101st message cycle, the security execution module retrieves the arrival time interval of the current adjacent message. The difference between the absolute timestamp of the next time-series network frame collected by the timestamp calibration unit and the absolute timestamp of the previous network frame is measured to be 180 milliseconds, which is greater than the 120-millisecond industrial benchmark jitter threshold. Based on this, the security execution module determines that the current network topology has experienced sporadic large-scale out-of-order events caused by physical channel degradation, and initiates a logical backoff procedure: suspending the high-risk escape identification command; temporarily storing the next time-series network frame in the temporary storage queue of the edge gateway to maintain session liveness; conditionally allowing the remaining portion of the next time-series network frame after stripping application layer semantics to the internal industrial automation network; extending the observation window to the 102nd, 103rd, and 104th consecutive message flow cycles; generating the dynamic attenuation operator based on the ratio of the 120-millisecond industrial benchmark jitter threshold to the current adjacent message arrival time interval of 180 milliseconds; and injecting the dynamic attenuation operator in reverse proportion into the context state tracking flow table, so that the security tolerance boundary is widened in the same direction according to the ratio of 180 milliseconds to 120 milliseconds, widening from the original 16 bytes to 24 bytes.

[0069] In the 102nd message cycle, the next time-series network frame arrives. The measured relative shift value of the starting byte remains consistently less than the expected logical boundary termination value of the preceding sequence. The measured variance of the topology displacement boundary is -28 bytes, and its absolute value of 28 bytes exceeds the currently temporarily effective 24-byte security tolerance boundary. The state characteristic value remains active, and the abnormal state has not been eliminated. In the 103rd message cycle, the measured variance of the topology displacement boundary is -31 bytes, and the state characteristic value remains active, indicating that the abnormal state persists. In the 104th message cycle, the measured variance of the topology displacement boundary is -33 bytes, and the state characteristic value remains active. The extended observation window has been traversed completely. The topology displacement boundary variance remains less than zero throughout, and the state characteristic value remains active throughout. The security execution module determines that the interleaved overwrite exhibits characteristics of deliberately constructed high-density burst malicious payload escape injection, and the logical backoff procedure terminates. The security execution module directly issues a blocking command to cut off the transmission control protocol connection of the current session and shield the corresponding industrial bus logic node, thereby performing in-situ precise interception of the malicious payload escape injection behavior.

[0070] Entering the 105th message cycle, the edge firewall node receives a new round of message sequences legitimately retransmitted by the remote management platform. The arrival time interval between current adjacent messages drops back to 80 milliseconds, falling within the 120-millisecond industrial benchmark jitter threshold range. The reverse injection path of the dynamic attenuation operator is synchronously withdrawn, and the security tolerance boundary returns from 24 bytes to its original 16-byte value. The protocol parsing module remaps the expected logical boundary termination line value based on the absolute byte relative shift value and the length indicator. The state audit module finds that the variance of the topology displacement boundary is zero bytes. The current session returns to a linearly increasing steady-state flow condition, and the entire dynamic closed loop is reset.

[0071] Combination Figure 2 This application also provides an IoT protocol malicious payload escape identification and security defense device. This device may include a data acquisition module, a protocol parsing module, a status auditing module, and a security execution module.

[0072] The data acquisition module is used to strip the physical layer header information from the current network frame flowing into the boundary to obtain the data payload stream. In this embodiment, the data acquisition module can be used to perform... Figure 1 The relevant content regarding the data acquisition module shown in S1 can be found in the previous description of S1.

[0073] The protocol parsing module is used to eliminate full stream reassembly, determine the expected logical boundary termination line value by arithmetically adding the absolute byte relative shift value of the current network frame and the length indicator, and write it along with the state feature value of the application layer specific write control function code into the context state tracking flow table. In this embodiment, the protocol parsing module can be used to execute... Figure 1 The details of S2 shown above, and the relevant content regarding the protocol parsing module, can be found in the previous description of S2.

[0074] The state audit module is used to obtain the topology displacement boundary variance by subtracting the relative shift value of the starting byte of the next time-series network frame from the value of the expected logical boundary termination line of the preceding frame. When the topology displacement boundary variance is less than zero and its absolute value is greater than the safety tolerance boundary, the historical overlap frequency does not fluctuate monotonically within a consecutive preset number of message periods, and the state feature value is activated, a high-risk escape identification instruction is generated. In this embodiment, the state audit module can be used to execute... Figure 1 The relevant content regarding the status audit module shown in S3 can be found in the previous description of S3.

[0075] The security execution module is used to block the current session based on a comparison between the arrival time interval of current adjacent messages and an industrial benchmark jitter threshold, or to suspend the high-risk escape identification command and inject the dynamic attenuation operator in reverse into the context state tracking flow table to relax the security tolerance boundary. The current session is blocked when the variance of the topological displacement boundary remains less than zero and the state feature value remains active within the extended observation window. In this embodiment, the security execution module can be used to execute... Figure 1 The relevant content regarding the security execution module shown in S4 can be found in the previous description of S4.

[0076] Combination Figure 3 This application also provides an electronic device, which includes a memory and a processor connected to the memory. The memory stores a computer program, and the processor executes the computer program stored in the memory to implement the aforementioned method for identifying and defending against malicious payload escape in Internet of Things protocols.

[0077] Optionally, the memory may be, but is not limited to, random access memory, read-only memory, programmable read-only memory, erasable read-only memory, electrically erasable read-only memory, etc. Among them, random access memory is abbreviated as RAM; read-only memory is abbreviated as ROM.

[0078] Optionally, the processor may include at least one of a central processing unit (CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices. The CPU is the abbreviation for Central Processing Unit.

[0079] It is understood that the structure of the above-described electronic device is for illustrative purposes only. The electronic device may also include more or fewer components than described above, or have a different configuration than described above. For example, it may also include a communication interface for communicating with industrial fieldbus and remote management platform.

[0080] In this embodiment of the application, corresponding to the above-described method for identifying and defending against malicious payload escape in Internet of Things (IoT) protocols, a computer-readable storage medium is also provided. The computer-readable storage medium stores a computer program, which, when executed, performs each step of the method for identifying and defending against malicious payload escape in IoT protocols. The specific steps executed by the computer program during runtime are not detailed here, but can be found in the preceding explanation of the method for identifying and defending against malicious payload escape in IoT protocols.

[0081] Based on the above embodiments, the IoT protocol malicious payload escape identification and security defense method provided in this application firstly obtains the data payload stream by stripping the physical layer header information of the current network frame flowing into the boundary, so that the protocol parsing directly faces the standard network layer and transport layer data payload; secondly, without the method of full stream reassembly, the expected logical boundary termination line value is directly determined by the arithmetic sum of the absolute byte relative shift value and the length indicator, reducing the high-dimensional spatial feature matching to one-dimensional linear temporal differential judgment; then, the polarity of the topological displacement boundary variance triggers three independent judgment paths, and performs multi-condition joint judgment in the overlapping direction based on the overlapping byte value, historical overlapping frequency and the state feature value of the application layer specific write control function code; finally, based on the comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, it selects direct blocking or logical backoff, and injects the context state tracking flow table in reverse with a dynamic attenuation operator to form a feedback collaborative closed loop with bidirectional dependence between steps. By eliminating the physical overhead of full stream reassembly buffer and performing only instantaneous field offset differential comparison, edge forwarding does not need to wait for fragment reassembly. This eliminates the risk of state cache explosion and reduces edge forwarding latency, thereby enabling the balance between malicious payload escape identification accuracy and real-time performance on resource-constrained IoT edge nodes. This improves the problem of difficulty in balancing identification accuracy with resource overhead and latency in existing technologies.

[0082] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, and the division of modules is only a logical functional division. In actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment according to actual needs.

[0084] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the steps of the methods described in the various embodiments of this application.

[0085] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for identifying and protecting against malicious payload escape in Internet of Things (IoT) protocols, characterized in that, include: The physical layer header information is stripped from the current network frame flowing into the boundary to obtain the data payload stream; To avoid full stream reassembly, the arithmetic sum of the absolute byte relative shift value of the current network frame and the length indicator is used to determine the expected logical boundary termination line value, which is then written into the context state tracking flow table along with the state feature value of the application layer specific write control function code. The topology displacement boundary variance is obtained by subtracting the relative shift value of the starting byte of the next time-series network frame from the value of the expected logical boundary termination line of the previous frame. When the topology displacement boundary variance is less than zero and its absolute value is greater than the safety tolerance boundary, the historical overlap frequency does not fluctuate monotonically within a consecutive preset number of message periods, and the state feature value is activated, a high-risk escape identification instruction is generated. The current session is blocked based on a comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, or the high-risk escape identification command is suspended and the dynamic attenuation operator is injected in reverse into the context state tracking flow table to relax the security tolerance boundary. The current session is blocked when the variance of the topological displacement boundary is continuously less than zero and the state feature value is continuously active within the extended observation window.

2. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, The step of determining the expected logical boundary termination line value by arithmetic summation of the absolute byte relative shift value of the current network frame and the length indicator further includes: Align the network layer protocol fragmentation boundary with the application layer specific write control function code features using syntax tree topology alignment, and construct a sequence of relative offset matrix matrices for protocol fields. The state feature value is used to characterize the write / overwrite action of the current industrial protocol control plane for registers or sensitive configuration parameters. The state feature value transitions from an inactive state to an active state when the application layer packet header of the current network frame carries the register write / overwrite opcode.

3. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, After the step of subtracting the relative shift value of the start byte of the next time-series network frame from the value of the expected logical boundary termination line of the preceding frame to obtain the topology displacement boundary variance, the method further includes: When the variance of the topology displacement boundary is equal to zero, it is determined that the current session is in a linearly increasing steady-state flow condition, the abnormal counter of the current session is cleared, and the next time-series network frame is forwarded normally. When the variance of the topological displacement boundary is greater than zero, it is determined that there is a fragmentation hole in the physical transmission timing of the current protocol stream. The current protocol stream is then attached to the timing monitoring queue, waiting for subsequent messages to cover and fill the fragmentation hole.

4. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, The security tolerance boundary is 16 bytes, and the preset quantity is 5. The historical overlap frequency is counted as follows: the number of times the variance of the topological displacement boundary is less than zero in the current session is counted in terms of message period. When the value sequence of the historical overlap frequency is neither monotonically increasing nor monotonically decreasing in 5 consecutive message periods, it is determined that the historical overlap frequency exhibits non-monotonic fluctuation.

5. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, The step of blocking the current session or suspending the high-risk escape identification command based on the comparison between the current adjacent message arrival time interval and the industrial benchmark jitter threshold includes: When the arrival time interval of the current adjacent messages is within the range of the industrial reference jitter threshold, a blocking instruction is issued according to the high-risk escape identification instruction to cut off the transmission control protocol connection of the current session and shield the corresponding industrial bus logical node. When the arrival time interval between adjacent messages is greater than the industrial benchmark jitter threshold, it is determined that the current network topology has caused an occasional large-scale disorder caused by physical channel degradation. The high-risk escape identification instruction is suspended, the current network frame is temporarily stored in the temporary storage queue, and the remaining part of the current network frame after stripping the application layer semantics is forwarded to the internal industrial automation network. The observation window is extended to the next 3 consecutive message circulation cycles.

6. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, The dynamic decay operator is generated as follows: The ratio of the industrial benchmark jitter threshold to the current adjacent message arrival time interval is determined as the dynamic attenuation operator; The dynamic attenuation operator is injected into the context state tracking flow table in an inverse relationship, so that the security tolerance boundary is monotonically widened as the arrival time interval of the current adjacent message increases; When the current adjacent message arrival time interval falls back to within the industrial reference jitter threshold range, the injection of the dynamic attenuation operator is canceled, and the safety tolerance boundary is returned to its original value.

7. The method for identifying and protecting against malicious payload escape in IoT protocols according to claim 1, characterized in that, The current adjacent message arrival time interval is obtained by collecting the absolute timestamps of two adjacent fragmented messages arriving at the access node and subtracting them using the timestamp calibration unit built into the communication network card. The unit of the current adjacent message arrival time interval is milliseconds, and the industrial reference jitter threshold is 120 milliseconds.

8. A device for identifying and protecting against malicious payload escape in Internet of Things (IoT) protocols, characterized in that: include: The data acquisition module is used to strip the physical layer header information from the current network frame flowing into the boundary to obtain the data payload stream; The protocol parsing module is used to eliminate full stream reassembly. It determines the expected logical boundary termination line value by arithmetically adding the absolute byte relative shift value of the current network frame and the length indicator, and writes it into the context state tracking flow table along with the state feature value of the application layer specific write control function code. The status audit module is used to obtain the topology displacement boundary variance by subtracting the relative shift value of the starting byte of the next time-series network frame from the value of the expected logical boundary termination line of the previous frame. When the topology displacement boundary variance is less than zero and its absolute value is greater than the safety tolerance boundary, the historical overlap frequency does not fluctuate monotonically within a consecutive preset number of message periods, and the status feature value is activated, a high-risk escape identification instruction is generated. The security execution module is used to block the current session based on a comparison between the arrival time interval of the current adjacent messages and the industrial benchmark jitter threshold, or to suspend the high-risk escape identification instruction and inject the dynamic attenuation operator in reverse into the context state tracking flow table to relax the security tolerance boundary. The current session is blocked when the variance of the topological displacement boundary is continuously less than zero and the state feature value is continuously active within the extended observation window.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the IoT protocol malicious payload escape identification and security defense method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed, performs the IoT protocol malicious payload escape identification and security defense method according to any one of claims 1 to 7.