Lightweight industrial sensor data stream integrity verification method based on physically unclonable functions
By adopting a lightweight verified data stream transmission method based on PUF in the SCADA system, combined with the HMAC algorithm, the problems of data transmission insecurity and device physical security are solved, and the integrity verification and device security of data streams are realized, reducing the stability of data streams. Computation and communication overhead.
Patent Information
- Application Number
- CN202211436657.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-16
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-11-16
AI Technical Summary
The existing SCADA system lacks an effective security mechanism in the industrial Internet of Things, which leads to insecure data transmission, is prone to malicious tampering and attacks, and lacks a node authentication mechanism, which cannot guarantee the physical security of the device.
Using a lightweight verifiable data streaming method based on physical non-cloneable function (PUF), combined with HMAC algorithm, bidirectional authentication and data stream integrity between PLC and sensor nodes are achieved through the stability of PUF and the security of HMAC. Verify, resist denial of service attacks, replay attacks and man-in-the-middle attacks, and ensure the physical security of sensor nodes through bit selection algorithms.
This method provides critical security properties, ensures the integrity of data transmission and the physical security of the device, reduces computing and communication overhead, is suitable for secure communication in SCADA systems, and overcomes the problems of security vulnerabilities and high overhead in existing solutions. .
Smart Images

Figure CN116094719B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of industrial Internet of Things and data security transmission protection, and in particular to a lightweight industrial sensor data stream integrity verification method based on a physical unclonable function. Background Art
[0002] Critical infrastructure and the booming Industry 4.0 are increasingly applying more advanced technologies and tools, such as computers, electrical and mechanical devices, to monitor the physical processes of industrial production. The network generated by intelligent computing integration for collecting and monitoring real-time data to ensure the automated operation, process control and monitoring of industrial infrastructure is called the Industrial Internet of Things. It is widely used in oil refining, chemical production, power and power grid control, water plant monitoring, transportation and water conservancy facility protection, etc. According to statistics, almost 80% or more of the automated infrastructure related to national economy and people's livelihood relies on the Industrial Internet of Things for process control production.
[0003] However, the current global cybersecurity threats and risks are becoming increasingly prominent. The frequency and complexity of cyber attacks against the Industrial Internet of Things are increasing. Supervisory control and data acquisition (SCADA) systems are a common type of industrial Internet of Things. The general SCADA network communication structure is as follows: Figure 1 As shown in Figure 1. It monitors and controls different kinds of industrial components by collecting field data and issuing commands to geographically remote field stations. Current SCADA systems lack reliable security mechanisms. Compared with smooth and reliable operation, general security and network security were not the main concerns of early SCADA systems. Therefore, this unguarded state leads to some security risks. For example, the PLC in the SCADA system collects perception data from sensor nodes. However, the adversary can maliciously intercept or tamper with the transmitted data stream, which may cause the PLC node to receive incorrect data and make wrong judgments. In addition, SCADA systems that lack authentication mechanisms have security vulnerabilities that can be exploited by adversaries to introduce malicious nodes, whose malicious behavior will lead to insecure data communication. On the other hand, SCADA systems generally monitor remote industrial production, and their equipment is usually far away from the security zone. Attackers may try to change the configuration of remote sensor nodes, target them for physical attacks, and obtain confidential information such as session keys from the nodes. For this, the existing state-of-the-art SCADA system data transmission schemes cannot guarantee security. Summary of the invention
[0004] The present invention is used to solve the problem of insecure data transmission in the industrial Internet of Things. Specifically, a lightweight verifiable data stream transmission method based on a physically unclonable function is proposed, which utilizes the lightweight properties of the HMAC (hash operation message authentication code) and the security characteristics of the PUF (physical unclonable function) mechanism.
[0005] To achieve the above purpose, the technical solution adopted by the present invention is a lightweight verifiable data stream transmission method based on a physical unclonable function to solve the problems raised in the previous text. The method is suitable for communication and transmission between PLC nodes and sensor nodes in the SCADA system in the industrial Internet of Things, and aims to solve the problems of malicious tampering of information flow in the SCADA system, lack of node authentication mechanism, and inability to ensure the physical security of equipment. The innovation of the method lies in: 1. The present invention designs a new lightweight verifiable data stream transmission method based on a physical unclonable function, which can provide key security properties, including resistance to denial of service attacks, replay attacks, man-in-the-middle attacks, etc. At the same time, the present invention introduces an advanced bit selection algorithm to ensure the stability and efficiency of PUF use. 2. The method ensures the physical security of sensor nodes, and it is impossible to obtain confidential information through side channel attacks and bypass attacks, and to obtain confidential information by parsing the intrinsic configuration of the sensor. 3. The method performs continuous integrity verification on the data stream generated by the field sensor. Comparison of the proposed method with other existing methods shows that the method has a small computational overhead and communication overhead, which is crucial in the SCADA system. As shown in the figure, the implementation steps of the method are as follows:
[0006] Step (1) Registration phase: In order to ensure the stable use of PUF in this method, the present invention selects a PUF method based on SRAM (static random access memory). The most stable unit in SRAM is selected in combination with the data remanence algorithm;
[0007] Step (2) Setup phase: The method generates and stores the method parameters required for the authentication and data transmission phases. The specific flow of this step is shown in the figure;
[0008] Step (3) Authentication and transmission phase: In this phase, the method focuses on the two-way authentication between the smart sensor and the PLC and the use of the HMAC algorithm to verify the integrity of the data stream transmission. In addition, it also includes key negotiation, timestamp mechanism, identity and key update operations. The specific process of this step is shown in the figure;
[0009] Further, in step (1), the PUF is constructed using the most stable unit in the SRAM, specifically:
[0010] 1.1) Construct a PUF model. The concept of PUF comes from using the physically random and unpredictable hardware characteristics of existing physical entities at the microscopic scale as the source of key generation. Its most notable feature is the generation of a set of one-to-one corresponding challenge-response pairs (CRP), which takes a sequence as a challenge input and produces a string of bits called a response. This transformation is derived from physical randomness. The response R of PUF to the challenge C can be expressed by the formula:
[0011] R i =PUF(C i )
[0012] Among them, PUF represents an abstract function, which is a mapping of the function of a real physical device. It inputs a binary sequence and outputs a string of corresponding responses. Ci represents the challenge binary sequence input to the PUF function; Ri represents the corresponding response sequence of the i-th.
[0013] PUF is usually used in the form of integrated circuits (ICs). The uniqueness of the physical microstructure produced during the manufacturing process of the integrated circuit ensures that no two PUF structures are identical and cannot be tampered with. However, PUF noise caused by environmental conditions remains a limiting factor in PUF design, which may cause one or more output bits of the PUF to be incorrect in any challenge. A(d,n,l,l,∈) secure PUF needs to meet the following requirements:
[0014] 1. Any two PUFs, PUF i (·)and PUF j (·), for any input C∈{0,1} k , Pr[HD(PUF i (C),PUF j (C))>d]≥1-∈
[0015] PUF i (·) and PUF j (·) are two PUF functions with different internal physical microstructures; k represents an arbitrary natural number; C represents a subset of an arbitrary string consisting of 0 and 1; HD(a,b) represents the Hamming code distance between a and b; d represents the minimum Hamming distance between two PUF responses; Pr(·) represents the probability. ε is a probability number, usually set to 0.001. This formula represents that for two different PUFs, when any challenge C is input, the probability that the Hamming code distance between the responses is greater than the minimum Hamming distance d is greater than or equal to 1-ε.
[0016] 2. Any PUF, PUF i (·), for any input C1,,C n {0,1} k , Pr[HD(PUF i (C1),PUF i (C2))>d]≥1-∈.
[0017] This formula shows that for the same PUF function, the probability that the Hamming code distance between the responses obtained after inputting any two challenges C1 and C2 is greater than the minimum Hamming distance d is greater than or equal to 1-ε.
[0018] 3. Any two PUFs, PUF i (·), PUF i* (·), for any input C1,,C n {0,1} k , Pr[H ∞ PUF i (C k ),PUF m (C j )) 1≤j,k≤n,i≠m,j≠k >λ]≥1-∈
[0019] H ∞ Represents minimum entropy, which is used to represent the probability distribution of PUF response; PUF i (·) and PUF m (·) represents two different PUFs; j, k, n, i, m are the subscripts of the parameters. λ is generally a real number between 0.999 and 0.9999. This condition means that when using multiple inputs to evaluate different PUFs, the minimum entropy of the PUF output must be greater than λ with a high probability, and the probability of this event occurring is greater than or equal to 1-ε.
[0020] 1.2) Use the data remanence algorithm to select the SRAM storage unit with the most stable state characteristics. SRMA refers to static random access memory. Among them, the memory address of the SRAM stable unit is used as the challenge C in the PUF mechanism, and the value (0 / 1) stored in the SRAM stable unit corresponding to the memory address is used as the response R in the PUF mechanism.
[0021] The data remanence algorithm finds the most stable cells in the SRAM. First, a value ("0" or "1") is written to the entire memory. Next, the power is briefly turned off, at which point some cells are flipped ("1" becomes "0" or "0" becomes "1"). The most stable cells are the ones that can flip easily and quickly after the opposite data is written. For example, a strong "1" cell is a bit that flips quickly when a "0" is written to its memory. Conversely, if a "1" is written to a memory location and flipped quickly, this means it is a strong "0" cell. The method selects the cell that flips its value in the shortest time to build the PUF.
[0022] Furthermore, step (2) generates and stores method parameters required for the authentication and data transmission phases, specifically:
[0023] 2.1)PLC generates a challenge, which is a sequence of memory addresses selected from the stable bits stored in the registration phase.PLC transmits the generated challenge to the smart sensor.
[0024] 2.2) The smart sensor receives the challenge C and uses the PUF to generate a response R. It sends its own ID and response R to the PLC. The ID is a number set by the scheme for the smart sensor to indicate its identity.
[0025] 2.3) PLC generates SID through SM3-Hash algorithm according to the received sensor ID. This is a virtual identity designed for smart sensors to achieve anonymous communication of sensors. In addition, we use the fuzzy extractor algorithm and BCH code as the error correction code to calculate (K, hd) = FE.Gen (R). Finally, the virtual identity SID of the smart sensor, the challenge C, and the key K are stored in the PLC. {SID, C, hd} is sent to the smart sensor. Among them, the SM3-Hash algorithm takes the ID as input and outputs a 256-bit SID as the virtual identity of this sensor; the fuzzy extractor algorithm takes the response R as input, and inputs the key K and auxiliary data hd. Among them. The SM3-Hash algorithm is a cryptographic hashing (hash, hash) algorithm independently developed by my country. BCH code is a message error correction code. FE.Gen (.) is the key generation sub-algorithm in the fuzzy extractor algorithm.
[0026] 2.4) The smart sensor receives the {SID, C, hd} data sent from the PLC.
[0027] Furthermore, in step (3), the authentication and transmission phase steps are executed to complete the two-way authentication between the smart sensor and the PLC, and use the HMAC algorithm to verify the integrity of the data stream. The specific steps in step (3) are:
[0028] 3.1) The smart sensor node sends its SID to the PLC to inform the PLC that communication is about to begin.
[0029] 3.2) PLC searches for SID in the database. After successfully finding it, it reads (C, K) of the smart sensor node accordingly. After that, PLC generates a timestamp T1 = time.now() based on the current time. This is to resist DoS attacks. Next, PLC calculates the encrypted timestamp The benefit is that it greatly saves the computational overhead of encryption and decryption. Then, the PLC calculates the HMAC value H1 = Hmac(K, TS). This step has two purposes. First, the PLC proves its identity to the smart sensor, because currently only the authentic and reliable PLC has the stored key K. Second, H1 also verifies the authenticity of T1, preventing it from being intercepted and tampered by the adversary during transmission. Finally, {T1 * ,H1} is sent to the smart sensor. Among them, T1 = time.now() is the function to generate the current time of the system; middle It is an XOR operation. T1 and hash(K) are both 256-bit binary strings, so they can be XORed. H1=Hmac(K,TS), where Hmac is a message authentication code algorithm based on the SM3 hash algorithm, which outputs a 256-bit binary number for message integrity authentication and source identity authentication.
[0030] 3.3) The smart sensor receives {T1 * ,H1}, and generates a PUF response based on the challenge C transmitted and stored in the previous stage Due to the microscopic physical properties of the SRAM that constitutes the PUF, the response generated at this time may be noisy and cannot be directly used for calculations. Afterwards, the smart sensor selects the auxiliary data hd from its memory and calculates K=FE.Rec(R * ,hd) to reconstruct the key K. Next, the smart sensor calculates To decrypt T1. Then calculate the verification value H 11 =Hmac(K,T1). After completing the above steps, the smart sensor verifies H 11 Is it the same as H1? If the verification is successful, the sensing data can be read from the smart sensor memory and H2 = Hmac (K, data) can be calculated. Finally, {data, H2} is sent to the PLC. Among them, FE.Rec (.) is the key reconstruction sub-algorithm in the fuzzy extractor algorithm
[0031] 3.4) PLC receives data and H2 from the smart sensor. Then it starts to verify the identity and data. First, PLC calculates the current timestamp T2 = time.now() and calculates H 22 =Hmac(K,data). If the verification time of the program is within the threshold t we set, and the integrity of the transmitted data is guaranteed, it means that the data is safe. The threshold t is a time value, which is determined according to the processor performance of the current system, the complexity of the processing flow, etc. For example, the experimental setting in the present invention is 0.24s.
[0032] 3.5) At this point, a round of data transmission and verification is completed. Next, the smart meter and PLC begin to synchronously update the key and identity. The smart meter and PLC use the hash strategy to calculate the updated key K new = hash(K) and the updated virtual identity
[0033] This method adopts a lightweight verifiable data stream transmission method based on physical unclonable functions to solve the problems mentioned above. This method is suitable for communication transmission between PLC nodes and sensor nodes in the SCADA system in the industrial Internet of Things, aiming to solve the problems of malicious tampering of information flow in the SCADA system, lack of node authentication mechanism, and inability to ensure the physical security of equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 SCADA system architecture diagram
[0035] Figure 2 Diagram of data transmission from sensor to PLC
[0036] Figure 3 Setting the stage for an efficient PUF-based authenticated secure transmission scheme
[0037] Figure 4 The authentication and data transmission phases of the PUF-based efficient authentication and security transmission scheme
[0038] Figure 5 Smart sensor architecture diagram DETAILED DESCRIPTION
[0039] The purpose of the present invention is to propose a lightweight verifiable data stream transmission method based on a physically unclonable function, which utilizes the lightweight properties of HMAC (hash operation message authentication code) and the security characteristics of PUF (physical unclonable function) mechanism to ensure the secure transmission of perception data between PLC and field sensors.
[0040] The technical solution provided by the present invention will be described in detail below in conjunction with specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention.
[0041] The present invention proposes a lightweight industrial sensor data stream integrity verification scheme based on PUF to solve the problems mentioned above. The invention is applicable to the communication transmission between PLC nodes and sensor nodes in the SCADA system, and aims to solve the problems of malicious tampering of information flow in the SCADA system, lack of node authentication mechanism, and inability to ensure the physical security of equipment. At the same time, problems such as high communication and computing overhead in secure communication schemes are also solved. The lightweight industrial sensor data stream integrity verification scheme based on PUF proposed in the present invention is mainly divided into three stages, namely, registration stage, setting stage, and verification and data transmission stage.
[0042] Step (1) Registration phase: The present invention uses an SRAM-based PUF algorithm, whose input challenge is actually a set of memory addresses of the SRAM, and whose output response is the bias value ("0" or "1") of the SRAM cell corresponding to the input address. If the SRAM is a component of the PUF, it must have a stable output. Due to temperature, voltage, aging and other noise effects, the bias values of some SRAM cells may be unstable. Therefore, the present invention uses the data remanence method as a bit selection algorithm to find the most stable cell in the SRAM. First, the value ("0" or "1") is written to the entire memory. Second, the power is turned off briefly, at which time some cells are flipped ("1" becomes "0" or "0" becomes "1"). The most stable cells are those that can be easily and quickly flipped after the opposite data is written. For example, a strong "1" cell is a bit that can flip quickly when "0" is written to its memory. Conversely, if a "1" is flipped quickly, it means that it is a strong "0" cell. We select the cell whose value flips in the shortest time for the PUF challenge. The PLC has enough storage space to store these selected memory addresses. Since the total capacity of the SRAM is 256KB, the present invention has 256KB of memory bits that can be used for selection by the data remanence algorithm. The present invention selects the most stable 4662 memory bits from the 256KB memory. These stable bits will be disrupted as the source of the challenge sequence of the PUF.
[0043] Step (2) Setup phase: In this phase, the sensor and PLC begin to interact, and both ends generate and store the required scheme parameters for the authentication and data transmission phases through calculation and transmission. The present invention randomly selects 2331 stable bits from the challenge source generated in the registration phase as a challenge, and the PUF embedded in the smart meter generates a 2331-bit response after inputting the challenge. Among them, each 63-bit response is encrypted by the BCH encryption algorithm to generate a 7-bit key K and 56-bit auxiliary data. The present invention uses the SM3 cryptographic hash algorithm as a hash algorithm. SM3 is mainly used for digital signature and verification, message authentication code generation and verification, random number generation, etc., and its algorithm is public. This algorithm generates a hash value of 256 bits for the input bit message after padding and iterative compression, wherein XOR, modulo, modulo addition, shift, and, or, and not operations are used, and it is composed of padding, iterative process, message expansion and compression function. After these operations, the algorithm generates a value of 256 bits in length. The lengths of C, R, ID, SID, K, and hd are 2331b, 2331b, 256b, 256b, 256b, and 2072b respectively. The specific steps are as follows:
[0044] Step (2-1): PLC reads all challenge data bits from the storage space, a total of 4662 bits. Then randomly selects 2331 bits as the input challenge Generate: C, which is essentially a sequence of memory addresses. PLC transmits the generated challenge to the smart sensor.
[0045] Step (2-2): The smart sensor receives the challenge C from the PLC and uses the PUF mechanism to generate a response of 2331b by measuring the internal physical microstructure of the linked SRAM In addition, the smart sensor uses the system's own clock to generate a random number, and then uses the SM3 cryptographic hash algorithm to use this random number as input to input a 256-bit string. As an ID representing its own identity. Finally, it sends its own ID and response R to the PLC.
[0046] Step (2-3): PLC generates a 256-bit SID based on the received sensor ID using the SM3 cryptographic hash algorithm. This is a virtual identity designed specifically for smart sensors. All subsequent communications are performed with the virtual identity. In addition, the present invention uses an auxiliary data algorithm with BCH code as the error correction code to calculate (K, hd) = FE.Gen (R). Among them, each 63-bit response is encrypted by the BCH encryption algorithm to generate a 7-bit key K and 56 bits of auxiliary data. Finally, the virtual identity SID of the smart sensor, the challenge C, and the key K are stored in the PLC. is sent to the smart sensor.
[0047] Step (2-4): The smart sensor receives the data sent from the PLC. The data is stored and is ready for the authentication and data transmission phases.
[0048] Step (3) The focus of this stage is the two-way authentication between the smart sensor and the PLC and the use of HMAC for data stream integrity transmission. In addition, it also includes operations such as key negotiation, timestamp mechanism, identity and key update. In the authentication and data transmission stage. The smart sensor end uses the system's own millis() function to generate an execution timestamp in milliseconds. The present invention uses HMAC based on the SM3 cryptographic hash algorithm as a hash algorithm to generate a 256b Hmac output value. The FE.Rec() function uses the BCH decryption algorithm, takes 56b of auxiliary data and 63b of PUF response as input, and outputs a 7b key K. In addition, T1,T * ,H1,R * ,H 11 ,H2,T2,K new ,SIDnew The lengths of are 256b, 256b, 256b, 2331b, 256b, 256b, 256b, 256b, 256b, and 256b respectively. The specific steps are as follows:
[0049] Step (3-1): The smart sensor node sends its SID and a communication request signal to the PLC to inform the PLC that the authentication and data transmission phase communication is about to begin.
[0050] Step (3-2): After receiving the data, the PLC searches for the SID in its own storage space and reads the (C, K) of the smart sensor node accordingly. After that, the PLC uses the system's internal clock to generate a timestamp T1 = time.now(). This step is to defend against DoS attacks. Next, the PLC calculates T1 and hash(K) are both binary strings of size 256b, so XOR operation can be used. This can play the role of encrypting TS, and its benefit is that it greatly saves the computational overhead of encryption and decryption. Then, PLC calculates H1=Hmac(K,T1). This step has two purposes. First, PLC proves its identity to the smart sensor, because currently only the authentic and reliable PLC has the stored key. Secondly, H1 also verifies the authenticity of T1 to prevent it from being intercepted and tampered by the opponent during transmission. Finally, PLC queues the message {T1 * ,H1} is sent to the smart sensor.
[0051] Step (3-3): The smart sensor receives {T1 * ,H1}. Afterwards, the smart sensor uses the embedded PUF mechanism to generate a PUF response At this point, the response generated by the PUF may be noisy, and we cannot use it directly for calculations. Afterwards, the smart sensor selects the hd stored in the setup phase from its storage space and uses the key recovery algorithm of the fuzzy extractor to recover the noisy response R * and the corresponding auxiliary data hd as input, the recovery key K = PUF (R * ,hd). Next, the smart sensor calculates To decrypt T1. Then calculate H 11 =Hmac(K,T1). After completing the above steps, the smart sensor verifies H1,H 11 If the verification is successful, the sensing data can be read from the smart sensor memory and H2 = Hmac (K, data) can be calculated. Finally, {data, H2} is sent to the PLC. In the experimental link of the present invention, the sensor uses a temperature sensor and sends temperature data every 2 seconds. For example, the physical quantity measured by the data in this experiment is 25.6°
[0052] Step (3-4): PLC receives data data and H2 from the smart sensor. The next step is to verify the data and identity. First, PLC calculates the current timestamp T2 = time.now() and calculates H 22 =Hmac(K,data). If the verification time of the program is within the threshold t we set, and the integrity of the transmitted data is guaranteed, it means that the data is safe. The threshold t is a time value, which is determined according to the processor performance of the current system, the complexity of the processing flow, etc. For example, the experimental setting in the present invention is 0.24s.
[0053] Step (3-5): A round of data transmission and verification has been completed. Next, the smart meter and PLC start to synchronously update the key and identity. The smart meter and PLC calculate K new = hash(K) and After that, the smart sensor and PLC can communicate directly using this key and identity.
[0054] This paper proposes a lightweight PUF-based industrial sensor data stream integrity verification scheme for secure communication between field sensors and PLC in SCADA systems. The proposed scheme allows field sensors to interact with PLCs securely and effectively using a common negotiated key. The lightweight nature of the scheme and the security characteristics of PUF are of great help in overcoming security vulnerabilities. Compared with existing schemes, the proposed scheme has less computational and communication overhead.
[0055] In the embodiments provided in this application, it should be understood that the disclosed method can be implemented in other ways without exceeding the spirit and scope of this application. The current embodiment is only an illustrative example and should not be used as a limitation. The specific content given should not limit the purpose of this application.
[0056] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A lightweight industrial sensor data stream integrity verification method based on physical unclonable functions, characterized in that: The implementation steps are as follows: Step (1) Registration phase: Select the SRAM-based PUF method; and select the most stable unit in the SRAM in combination with the data remanence algorithm; Step (2) Setup phase: The method generates and stores the method parameters required for the authentication and data transfer phases; Step (3) authentication and transmission phase; Step (1) uses the most stable unit in SRAM to build PUF, specifically: 1.1) Construct PUF model; PUF response R to challenge C is expressed by formula: R i =PUF(C i ) Among them, PUF represents an abstract function, which is a mapping of the function of a real physical device. It inputs a binary sequence and outputs a string of corresponding responses; C i represents the challenge binary sequence input to the PUF function; R i represents the response sequence corresponding to the i-th one; A(d,n,l,λ,ε) secure PUF needs to meet the following requirements: 1) Any two PUFs, PUF i (·) and PUF j (·), for any input C∈{0,1} k , Pr[HD(PUF i (C1),PUF i (C2))>d]≥1-ε; PUF i (·) and PUF j (·) are two PUF functions with different internal physical microstructures; k represents an arbitrary natural number; C represents a subset of an arbitrary string consisting of 0 and 1; HD(a,b) represents the Hamming code distance between a and b; d represents the minimum Hamming distance between two PUF responses; Pr(·) represents the probability; ε is a probability number, set to 0.001; this formula represents that for two different PUFs, when any challenge C is input, the probability that the Hamming code distance between the responses is greater than the minimum Hamming distance d is greater than or equal to 1-ε; 2) Any PUF, PUF i (·), for any input C1,…,C n ∈{0,1} k , Pr[HD(PUF i (C1),PUF i (C2))>d]≥1-ε; This formula shows that for the same PUF function, the probability that the Hamming code distance between the responses obtained after inputting any two challenges C1 and C2 is greater than the minimum Hamming distance d is greater than or equal to 1-ε; 3) Any two PUFs, PUF i (·), PUF i* (·), for any input C1,…,C n ∈{0,1} k , Pr[H ∞ PUF i (C k ),PUF m (C j )) i≤j,k≤n,i≠m,j≠k >λ]≥1-ε; H ∞ Represents minimum entropy, which is used to represent the probability distribution of PUF response; PUF i (·) and PUF m (·) represents two different PUFs; j, k, n, i, m are the subscripts of the parameters; λ is a real number between 0.999 and 0.9999; this condition means that when using multiple inputs to evaluate different PUFs, the minimum entropy of the PUF output must be greater than λ with a high probability, and the probability of this event occurring is greater than or equal to 1-ε; 1.2) Use the data remanence algorithm to select the SRAM storage unit with the most stable state characteristics; SRMA refers to static random access memory; the memory address of the SRAM stable unit is used as the challenge C in the PUF mechanism, and the value 0 / 1 stored in the SRAM stable unit corresponding to the memory address is used as the response R in the PUF mechanism; First, the value 0 / 1 is written to the entire memory. Second, the power is turned off briefly, at which point some cells are flipped, i.e., "1" becomes "0" or "0" becomes "1". The cells whose values flip in the shortest time are selected to construct the PUF. Furthermore, step (2) generates and stores method parameters required for the authentication and data transmission phases, specifically: 2.1) PLC generates a challenge, which is a sequence of memory addresses selected from the stable bits stored in the registration phase; PLC transmits the generated challenge to the smart sensor; 2.2) The smart sensor receives the challenge C and generates a response R using the PUF; it sends its ID and the response R to the PLC; the ID is a number set by the smart sensor to indicate its identity; 2.3) PLC generates SID through SM3-Hash algorithm according to the received sensor ID, which is a virtual identity designed for smart sensors to achieve anonymous communication of sensors; using fuzzy extractor algorithm, BCH code is used as error correction code to calculate (K, hd) = FE.Gen (R); finally, the virtual identity SID of the smart sensor, challenge C, and key K are stored in PLC; {SID, C, hd} is sent to the smart sensor; where, The SM3-Hash algorithm takes the ID as input and outputs a 256-bit SID as the virtual identity of this sensor; the fuzzy extractor algorithm takes the response R as input and outputs the key K and auxiliary data hd; FE.Gen(·) is the key generation sub-algorithm in the fuzzy extractor algorithm; 2.4) The smart sensor receives {SID, C, hd} data sent from PLC; Furthermore, in step (3), the authentication and transmission phase steps are executed to complete the two-way authentication between the smart sensor and the PLC, and use the HMAC algorithm to verify the integrity of the data stream transmission; the specific steps in step (3) are: 3.1) The smart sensor node sends its SID to the PLC to inform the PLC that communication is about to begin; 3.2) PLC searches for SID in the database. After successfully finding it, it reads (C, K) of the smart sensor node accordingly. Then, PLC generates a timestamp T1 = time.now() based on the current time to resist DoS attacks. Next, PLC calculates the encrypted timestamp PLC calculates HMAC value H1=Hmac(K,T1); First, the PLC proves its identity to the smart sensor, because only the authentic and reliable PLC has the stored key K; second, H1 also verifies the authenticity of T1 to prevent it from being intercepted and tampered by the adversary during transmission; finally, {T1 * ,H1} is sent to the smart sensor; where, T1=time.now() is a function to generate the current time of the system; middle It is an XOR operation. Both T1 and hash(K) are 256-bit binary strings, so they can be XORed. H1=Hmac(K,T1), where Hmac is a message authentication code algorithm based on the SM3 hash algorithm, which outputs a 256-bit binary number and is used for message integrity authentication and source identity authentication. 3.3) The smart sensor receives {T1 * ,H1}, and generates a PUF response based on the challenge C transmitted and stored in the previous stage Due to the microscopic physical properties of the SRAM that constitutes the PUF, the response generated at this time may be noisy and cannot be directly used for calculations; after that, the smart sensor selects the auxiliary data hd from its memory and calculates K=FE.Rec(R * ,hd) to reconstruct the key K; Next, the smart sensor calculates To decrypt T1; then calculate the verification value H 11 =Hmac(K,T1); After completing the above steps, the smart sensor verifies H 11 Is it the same as H1? If the verification is successful, read the sensing data from the smart sensor memory and calculate H2 = Hmac (K, data); finally, {data, H2} is sent to the PLC; where FE.Rec (·) is the key reconstruction sub-algorithm in the fuzzy extractor algorithm; 3.4) PLC receives data and H2 from the smart sensor; then it starts to verify the identity and data; first, PLC calculates the current timestamp T2 = time.now(), and calculates H 22 =Hmac(K,data); If the verification time of the program is within the threshold t we set and the integrity of the transmitted data is guaranteed, it means that the data is safe; where the threshold t is a time value set to 0.24s; 3.5) At this point, a round of data transmission and verification is completed; next, the smart sensor and PLC begin to synchronously update the key and identity; the smart sensor and PLC use the hash strategy to calculate the updated key K new = hash(K) and the updated virtual identity
