An authentication method for verifying integrity of data individuals based on watermark technology

Abstract

The application discloses an authentication method for verifying data individual integrity based on a watermark technology, which independently verifies the integrity of each data item, and can reduce the probability of misjudgment of data and improve the performance of data integrity verification when verifying at the receiving end after data transmission through a wireless channel. Based on the watermark method of a single data item, a corresponding watermark value is generated for each data, and the generated watermark value is embedded into the carrier data. Therefore, the scheme can effectively detect tampering attacks, reduce the overall missed detection rate of tampering attacks, and avoid false alarms. Compared with other individual data verification schemes, the scheme has higher data tampering identification capability. In addition, the data individual does not need a flag bit and a buffer area, thereby reducing communication overhead and saving network resources.

Description

Technical Field

[0001] This invention belongs to the fields of information security and data authentication technology, and specifically relates to an authentication method for verifying the integrity of individual data based on watermarking technology. Background Technology

[0002] Wireless sensor networks (WSNs) are task-oriented networks, and their network security issues urgently need further resolution. Research on WSN network security will inevitably influence the development of the entire wireless sensor network industry. Data integrity, data source authentication, and data copyright protection can all be considered security requirements for data streams transmitted in sensor networks. Due to the unique environment in which sensor nodes operate, information is easily stolen and tampered with, making integrity authentication indispensable. Many current solutions are based on cryptography or complex computation, but this increases the amount of data transmitted over the network and is unsuitable for the limited computing power of wireless sensors. Existing digital watermarking schemes are suitable for this characteristic of wireless sensor networks, but they suffer from a high false alarm rate in integrity authentication. Since most data processing is based on group operations rather than individual point-to-point, research on independent individual authentication in wireless sensor networks based on digital watermarking is of great significance.

[0003] Data integrity authentication is a key issue that needs to be addressed in wireless sensor networks. Furthermore, digital watermarking can serve as one technique for verifying data tampering. The main advantages of digital watermarking are its small size, low storage requirements, and high imperceptibility. Currently, data integrity authentication in wireless sensor networks based on digital watermarking is mainly achieved through data packetization. However, packet authentication has certain limitations, such as packet synchronization and the transparency of packet markings, which increase memory consumption. For example, a chained packet authentication scheme uses synchronization points to group data for integrity authentication. Another work processes the data stream using chaotic sequences and introduces homomorphic encryption to prevent eavesdropping on valid data, but the encryption operation increases computational complexity.

[0004] In recent years, many papers have further explored the concept of individual identity verification. For example, watermarking is generated by randomly selecting data from a queue. However, due to the correlation between data, this scheme significantly reduces the false alarm rate while increasing the false alarm rate. To further improve this method, improvements have been made, reducing the overall error rate of data tampering detection. Since it does not involve grouping, the key advantage of individual integrity authentication is that it is unaffected by group marking. Individual schemes have a higher data individual identification rate, while eliminating the need for flag bits and buffers for data operations, reducing communication overhead and saving network resources.

[0005] Most existing watermarking authentication schemes for sensor networks are based on data packets, maintaining the transparency of the original data by sharing all data within the packet. However, watermarking authentication schemes still have the following problems: (1) If the packet length of a packet-based watermarking scheme is N, an attacker only needs to modify the data items at a frequency of one-N times to make all data items in the entire data stream verify as tampered, thus rendering them unusable. (2) If the packet flag is tampered with, the data stream receiver will be unable to synchronize with the sender, and the authentication of at least two adjacent data packets will fail. The packet flag data becomes a vulnerable link in the entire authentication scheme, affecting the imperceptibility of the watermark. Summary of the Invention

[0006] To address the aforementioned issues, this solution proposes a wireless sensor network data independent individual authentication method based on digital watermarking technology, aiming at the integrity verification of individual data items. This method performs integrity verification independently on each data item, reducing the probability of data misjudgment during verification at the receiving end after data transmission through the wireless channel, thus improving the performance of data integrity verification.

[0007] This invention presents a method for independent individual authentication of wireless sensor network data based on digital watermarking technology. This method independently verifies the integrity of each data item. First, a hash operation is performed on a data item, and the result is segmented. Then, each segment is folded into one bit, resulting in a bit string divided into two parts. Subsequently, two decimal numbers are calculated. The embedded method replaces the lowest decimal part of the data item with the two generated decimal numbers. Finally, the order of the decimal parts is randomly scrambled. Finally, the carrier data with the embedded watermark is generated.

[0008] Specifically, the present invention proposes an authentication method for verifying the integrity of individual data based on watermarking technology, comprising the following steps:

[0009] The j-th original sensor data d j Let's denote the original data as A, and the data item received by the aggregation node as received data B;

[0010] Watermark generation includes:

[0011] sensor data d j Multiply by 100 and round down to get the integer m;

[0012] sensor data d j The decimal part is limited to 4, if the sensor data d j If the decimal part is less than 4, fill the remaining digits with 0. If the sensor data d j If the decimal part has more than 4 decimal places, delete the extra decimal parts. Sensor data d jThe third and fourth decimal parts of the four decimal parts are represented as decimal l1 and decimal l2, respectively;

[0013] A 128-bit binary number M is generated from an integer m using a hash operation;

[0014] Divide the binary number M into six equal parts and fold them into six bits, which are represented as b1, b2, b3, b4, b5, b6 respectively.

[0015] Divide the six bits into two equal segments, and convert the first three bits (bit segment one, b1, b2, b3) and the last three bits (bit segment two, b4, b5, b6) into decimal numbers B1 and B2 respectively as watermarks:

[0016] B k =Dec[b(k 2 :3k)](k=1,2)

[0017] B k This represents the k-th decimal digit used as a watermark, where b represents bits and Dec represents the decimal conversion function.

[0018] Watermark embedding:

[0019] Replace decimal l1 and decimal l2 with watermark B1 and watermark B2 respectively:

[0020] d′ j =Replace(B1,B2,l1,l2,d) j );

[0021] Where d j Let d′ represent the data from the j-th sensor. j This indicates that the j-th sensor data should be replaced with data in the decimal place, and Replace represents the replacement function;

[0022] After embedding, the order of the decimal parts of the received data B is randomly scrambled to obtain the final generated data items:

[0023] d″ j =RanScram(d′) j )

[0024] d″ j This represents the j-th final generated data item, and RanScram represents the random scrambling function;

[0025] Comparison: When the aggregation node receives a data item, it determines whether the data has been tampered with by comparing the watermark.

[0026] Furthermore, the formula for generating a 128-bit binary number M from an integer m using a hash operation is as follows:

[0027] M = Hash(m), where Hash represents the hash function.

[0028] Furthermore, the binary number M is divided into six equal segments and folded into six bits, represented as b1, b2, b3, b4, b5, b6 respectively:

[0029] X i =Slice(M)(i=1,2…6)

[0030] b i =Folding(X) i (i = 1, 2…6)

[0031] X i This represents the i-th segment of M divided into six equal segments, where b i This represents the i-th bit, Slice represents the piecewise function, and Folding represents the folding function;

[0032] Furthermore, the range of the two decimal digits B1 and B2 is 0≤B1, B2≤7.

[0033] Furthermore, when the aggregation node receives data items from received data B, it restores the fractional parts of the data items from received data B to their original order by reverse scrambling. Then, it calculates the watermark using the same method as in the sensor node watermark generation and compares it with the extracted watermark.

[0034] Furthermore, the calculation method for the currently collected data items is the same as that for the sensor nodes, resulting in two decimal numbers, denoted as B′1 and B′2; the lowest two decimal parts of the data item are extracted, denoted as l′1 and l′2; and the generated and extracted watermarks are compared to check for tampering.

[0035] Furthermore, if the generated watermarks B′1 and B′2 are different from the original watermark, it is determined that the non-watermarked part of the data item has been tampered with during transmission.

[0036] If the extracted watermark information l′1 and l′2 change, it is determined that the watermark part of the data item has been tampered with during transmission.

[0037] This solution aims to verify the integrity of individual data items and proposes a method for independent individual data authentication in wireless sensor networks based on digital watermarking technology. This method performs integrity verification independently on each data item. After the data is transmitted through the wireless channel, the verification at the receiving end can reduce the probability of data being misjudged and improve the performance of data integrity verification.

[0038] This paper studies a watermarking method based on individual data items. Each data item generates a corresponding watermark value, which is then embedded into the carrier data. Therefore, this scheme can effectively detect tampering attacks while reducing the overall false negative rate and avoiding false alarms. Compared with other individual data verification schemes, this scheme has a higher data tampering detection capability. Furthermore, individual data items do not require flag bits and buffers, reducing communication overhead and saving network resources. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the overall process of this invention.

[0040] Figure 2 This is a framework diagram of the watermarking research method of the present invention.

[0041] Figure 3 A flowchart for digital watermark generation and embedding.

[0042] Figure 4 This is a flowchart of the digital watermark extraction and comparison process.

[0043] Figure 5 This is the test result for the first modified attack scenario.

[0044] Figure 6 This is the test result for the second modified attack scenario.

[0045] Figure 7 This is the test result for the third modified attack scenario.

[0046] Figure 8 The results of the CVR test are for insert attack testing.

[0047] Figure 9 The results of the CVR test were used to test the deletion attack.

[0048] Figure 10 To modify the CVR test results of the attack test.

[0049] Figure 11 Compare the FNR and FPR of the various schemes.

[0050] Figure 12 Comparison of CVR among the various schemes.

[0051] Figure 13 This is a comparison of the computation time between the various schemes.

[0052] Figure 14 This is a portion of the actual sensor data from Intel Berkeley Research Lab. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] The overall flow of the implementation scenario of the present invention is as follows: Figure 1 As shown, the first step is to establish an individual authentication model, collect data from sensor nodes, and manage the data effectively. Based on this, research on digital watermarking technology and data individual integrity verification will be conducted. Finally, the research results will be improved through model verification, algorithm verification, and optimization verification.

[0055] First, the data is observed and analyzed, with the primary goal of studying digital watermarks embedded in individual data items. Second, the method of embedding them into individual data items is considered. Typically, the embedding of digital watermarks does not affect the normal use of the carrier data, and the watermark information is not easily detected, modified, or destroyed, thus possessing high security. As a mature and well-researched information hiding technology, digital watermarking has wide applications in the fields of copyright protection and integrity authentication of multimedia information. Digital watermarks are small in size and low in capacity, and their embedding into the carrier does not affect the normal use of the source data. The research methods and technical routes for digital watermarking are as follows... Figure 2 As shown.

[0056] Verifying the integrity of individual data items requires direct targeting to each individual item. Researching individual data items necessitates considering how to generate special watermarks to ensure each data item can be authenticated and addressing the issue of how to embed the watermark into the carrier. Digital watermarks are embedded in data items and extracted at the aggregation node; this process is reversible, and the embedded watermark cannot be easily identified. Currently, the most common watermark embedding method is the LSB (least significant bit) replacement method. Generally, the least significant bit of data is meaningless, but in special cases such as military and medical environments, the embedded watermark must be reversible to ensure data accuracy. Therefore, the embedding method must be pseudo-random to prevent attackers from maliciously destroying the least significant bit; otherwise, it will cause a significant number of missed detections during data verification. The process of digital watermarking and generation is as follows... Figure 3 As shown.

[0057] The process of watermark extraction and comparison is as follows: Figure 4As shown, to reduce the overall false positive rate, some digital watermarking research schemes use traditional comparison methods, that is, comparing the generated watermark and the extracted watermark. The correctness of the comparison result mainly depends on the correctness of the watermark. Therefore, further improving the performance of individual data authentication while maintaining low overhead and low latency is a problem worthy of further research. The main problem with watermark comparison is that the aggregation node compares the generated watermark and the extracted watermark to determine whether the data has been tampered with. That is, the data reaching the user needs to be watermarked using the same method as the watermark generated by the sensor node.

[0058] Watermark generation and embedding: First, we will use sensor data d j Multiply by 100, then round down to the nearest integer m. Then, d j Limited to 4 decimal places, either fill with 0s or remove the decimal part, where the third and fourth decimals are represented as l1 and l2 respectively. Typically, d j The first two decimal parts are considered meaningful. The watermark generation algorithm generates a 128-bit binary number M using the hash operation in formula (1) and m, and divides M into six equal segments using the Slice function in formula (2). Accordingly, the resulting six segments are folded into six bits b1, b2, ..., b6, as shown in formula (3). For example, suppose d j =19.7336, then m=1973, and l1 and l2 are 3 and 6 respectively.

[0059] M = Hash(m) (1)

[0060] X i =Slice(M)(i=1,2…6) (2)

[0061] b i =Folding(X) i (i = 1, 2…6) (3)

[0062] Finally, these six bits are divided into two equal segments (three bits in each segment). The first three bits (b1, b2, b3) and the last three bits (b4, b5, b6) are converted into decimal numbers B1 and B2 respectively as watermarks, as shown in equation (4).

[0063] B k =Dec[b(k 2 :3k)](k=1,2) (4)

[0064] Watermark embedding is achieved by replacing the two smallest decimal parts l1 and l2 with B1 and B2 respectively, as shown in equation (5). Finally, after embedding, the order of the decimal parts of the data item is randomly scrambled, as shown in equation (6). Therefore, randomly changing the position of the watermark can intentionally prevent the watermark from being identified and attacked. The final generated data item is represented as d″. j .

[0065] d′ j =Replace(B1,B2,l1,l2,d) j (5)

[0066] d″ j =RanScram(d′) j (6)

[0067] Since the watermark is generated by folding into 6 digits, the range of the two decimal numbers B1 and B2 generated from this segment is 0≤B1,B2≤7.

[0068] Watermark extraction and comparison: When the sink node receives a data item, it first restores the fractional parts of the sensor data item to their original order by reverse scrambling. Next, it calculates the watermark using the same method as in the sensor node watermark generation and compares it with the extracted watermark.

[0069] The data items collected are calculated using the same method as those described for the sensor nodes, resulting in two decimal numbers, denoted as B′1 and B′2. Simultaneously, the two lowest decimal parts of each data item are extracted, denoted as l′1 and l′2. Then, the generated and extracted watermarks are compared to check for any possible malicious tampering.

[0070] If the non-watermarked portion of a data item is tampered with during transmission, the generated watermarks B′1 and B′2 will differ from the original watermark. Similarly, tampering with the watermarked portion of a data item will cause the extracted watermark information l′1 and l′2 to change compared to the original content. Therefore, the proposed scheme detects data tampering in both of these ways.

[0071] Data integrity authentication: A missed detection occurs when a tampered data item is not identified as such at the sink node. Although the hash value calculated for the tampered data item changes, bit flipping keeps the generated watermark value unchanged. The folding operation causes the generated watermark information bits to shift from 0 to 1 and vice versa, with a probability of 1 / 2 for each shift. It is easy to know that the flipped bits K follow a binomial distribution, where the probability distribution P of “K” can be calculated using formula (7).

[0072]

[0073] Where P represents the probability of watermark bit flipping, K represents watermark bit flipping, k belongs to the element in the set of watermark bit flipping counts {0,1,2…N}, and N represents the total number of watermark bits.

[0074] There are three scenarios for data modification. The first scenario corresponds to modifying only the data used to generate the watermark. In this case, the six digits calculated from the tampered data are flipped during the folding process, and the resulting decimal number is still equal to the number of digits related to the extracted watermark. The false negative rate (FNR) in this scenario is shown in formula (8). Similarly, the second scenario involves only modifying the watermark portion. Since the watermark information is a decimal number with a value range of 0 to 7, the changed watermark information will not be equal to the generated watermark, so the FNR is 0 in this case. The last scenario refers to modifying both the data used to generate the watermark and the watermark portion. Since both of these parts of the data are modified, the FNR in this scenario is lower than in the first scenario.

[0075]

[0076] Where k represents the number of watermark bits flipped, and N represents the total number of watermark bits.

[0077] The insertion operation is similar to the last data modification scenario. Since the inserted data item does not embed watermark information, the FNR in this case is lower than that of data modification. Furthermore, when a data item is deleted, it does not affect the authentication of other data items. Therefore, the deletion operation will not result in missed detections of other data items, and the FNR is 0.

[0078] A false alarm occurs when an unaltered data item is detected as a tampered item during extraction and comparison. For an unaltered data item, the generated watermark is only related to the item itself. Therefore, the watermark extracted from the data item will be equal to the generated watermark, meaning that a correct data item will not be mistakenly identified as a tampered item. According to the analysis, all three tampering operations only affect the value of the current data item. If the data item is not associated, no false alarm will occur. Therefore, the false alarm rate (FPR) calculation formula is shown in (9), which is 0 under all three tampering methods.

[0079]

[0080] Where n is the number of all non-tampered data items that were mistakenly identified as tampered, and Truenum is the number of all truly non-tampered items.

[0081] Figure 5-7 Three modification attack cases were tested for FNR and FPR. Figure 5 This is the result of the first test that modified the attack scenario. Figure 6 This is the test result for the second type of modified attack scenario. Figure 7This is the test result for the third type of modified attack scenario.

[0082] Figure 5-7 Experimental results for three different modification attack scenarios are presented. Modifications to different parts of the data are explained in three cases. In the first case, only the data used to generate the watermark was modified. For example... Figure 5 As shown, when the modification rate increases, the FPR remains at 0, and the FNR is close to 0.02. In the second case, only the watermark portion of the data is modified. Figure 6 The display shows that FNR and FPR remain at 0. In the third case, both the data used to generate the watermark and the watermark portion are modified. In this case, the generation and extraction of the watermark from the partial data are consistent. From Figure 7 It can be seen that the value of FNR lies between the first and second cases.

[0083] Figure 8-10 CVR test results for different tampering attacks. Figure 8 Insertion attack test, Figure 9 Deletion attack test, Figure 10 Modify the attack test.

[0084] from Figure 8-10 It can be seen that for all different insertion, deletion, and modification rates, the CVR gradually decreases as the modification and insertion rates increase. Meanwhile, the accuracy of verifying the integrity of tampered data exceeds 0.99. If a data item is deleted during transmission, the receiving end cannot verify its integrity; it can only inspect the received data item.

[0085] When data is subjected to tampering attacks during transmission, the proposed scheme inspects each data item. CVR means verifying the integrity of all received data. During verification, since no data is discarded, this scheme has a significant advantage in verifying individual data. Furthermore, it can be seen that the proposed scheme performs well in verifying the integrity of all test data.

[0086] X. Shi, “A Statistical Integrity Authentication Scheme without Grouping for Streaming Data,” 2014 Ninth International Conference on P2P, Parallel, Grid, Cloud and Internet Computing, 2014, pp. 548-552. This paper presents a method for generating watermarks by randomly selecting data from a queue, referred to here as the grouped random selection queue method. Y. Xiao and G. Gao, “Digital Watermark-Based Independent Individual Certification Scheme in WSNs,” IEEE Access, vol. 7, pp. 145516-145523, 2019. This method improves upon the random selection queue method and is referred to here as the non-grouped random selection queue method.

[0087] Figure 11-13 The comparison between the various options is shown. Figure 11 Compare the FNR and FPR of the various schemes. Figure 12 Comparison of CVR among the various schemes. Figure 13 The computation time of each scheme is compared. In the figure,

[17] represents the grouped random selection queue method, and

[18] represents the non-grouped random selection queue method.

[0088] Figure 11 The reported FPR of the grouped random selection queue method and the non-grouped random selection queue method exceeds 0.8 and approaches 1 with increasing modification rate. The FNR of the grouped random selection queue method is approximately 0, while the FNR of the non-grouped random selection queue method is between 0 and 0.2. However, compared with the other two methods, the proposed method has a smaller and more stable FNR. Furthermore, in the two methods mentioned above, the generated watermarks are consistent for related data items. Therefore, when one data item is tampered with, the authentication of other related, untampered data items fails. Thus, authentication unrelated to data has a higher tamper detection rate compared to authentication related to data. Additionally, in the grouped random selection queue method, caching data in the queue consumes more storage space. In contrast, the proposed method caches only one data item, thus having lower resource consumption. In summary, from... Figure 11 It is clear that, in terms of data integrity verification, the scheme proposed in this work is superior to the comparative schemes in the literature.

[0089] from Figure 12It can be seen that the CVR of schemes

[17] and

[18] increases with the increase of the modification rate. As the number of modified data increases, the total number of correct data decreases. The trend of the false alarm rate is relatively stable, so the number of misjudged data will decrease significantly. It can be seen that the CVR increases with the increase of the modification rate. The CVR of the proposed scheme is stable and close to 1 under the change of modification rate. Since the proposed scheme has no false alarms, the trend of CVR change is only related to false alarms.

[0090] In our experiments, tampering attacks were categorized into three types: data modification, insertion, and deletion. The performance of the proposed scheme was tested using MATLAB simulations. The raw data stream used in the experiments came from a real wireless sensor network deployed at Intel Berkeley Research Lab (IntelLab Data, http: / / db.lcs.mit.edu / labdata / labdata.html). The experimental data consisted of 10,000 records collected by the wireless sensors, each containing several data items such as air temperature, voltage, humidity, and time. For ease of description, the experimental results presented in this paper only pertain to one data item (temperature).

[0091] The performance of the proposed independent individual integrity authentication scheme is evaluated through tampering experiments and comparisons with other individual schemes. The experimental framework tests the ability of each scheme to authenticate data integrity using the correlation values ​​of FNR, FPR, and CVR.

[0092] Each sensor data point not only generates corresponding watermark information but also serves as the carrier for embedding the watermark. Therefore, the information embedding method must meet the requirements of security and small capacity. Considering the contrast of watermarks in the scheme, we redesigned the position of the embedded watermark in the data. The watermark at the receiving end does not involve data calculation; it is placed in the meaningless decimal places of the sensor data and then randomly shuffled. Figure 14 This shows some actual sensor data from Intel Berkeley Research Lab.

[0093] Table 1 Comparison of other grouping schemes with the scheme of this invention

[0094]

[0095] In the table above, the label

[15] is the scheme described by H. Guo, Y. Li, and S. Jajodia, “Chaining watermarks for detecting malicious modifications to streaming data,” Information Sciences, vol. 177, no. 1, pp. 281-298, 2007. It proposes a chain-based group authentication scheme that uses synchronization points to group data to achieve data integrity authentication.

[0096] The label

[16] indicates the scheme described by Jiang Wenxian, Zhang Zhenxing, and Wu Jingjing, “A wireless sensor network data fusion protocol based on reversible digital watermark authentication”, Journal of Communications, 39(03):118-127, 2018. It processes the data stream through chaotic sequences and introduces homomorphic encryption to prevent valid data from being eavesdropped, but the encryption operation increases the amount of computation.

[0097] The label

[19] refers to the scheme described in X. Shi, D. Xiao, “A reversible watermarking authentication scheme for wireless sensor networks,” Information Sciences, 240 (Complete): 173-183, 2013.

[0098] In the grouping scheme, if one node is attacked, the data packets associated with that node will be dropped, leading to packet loss and misidentification. Furthermore, the proposed scheme can identify all data items. As shown in Table 1, the proposed scheme has a higher data tampering detection rate than the grouping scheme.

[0099] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. An authentication method for verifying the integrity of individual data based on watermarking technology, characterized in that, Includes the following steps: The j-th original sensor data d j Let's denote the original data as A, and the data item received by the aggregation node as received data B; Watermark generation includes: sensor data d j Multiply by 100 and round down to get the integer m; sensor data d j The decimal part is limited to 4 digits if the sensor data d j If the decimal part is less than 4 digits, fill the remaining digits with 0. If the sensor data d j If the decimal part is greater than 4 digits, remove the extra decimal parts. (Sensor data d) j The third and fourth decimal places of the four-digit fractional part are represented as decimal l1 and decimal l2, respectively. A 128-bit binary number M is generated from an integer m using a hash operation; Divide the binary number M into six equal parts and fold them into six bits, which are represented as b1, b2, b3, b4, b5, b6 respectively. Divide the six bits into two equal segments, and convert the first three bits (bit segment one) and the last three bits (bit segment two) into decimal numbers B1 and B2 respectively as watermarks: This represents the k-th decimal digit used as a watermark, where b represents bits and Dec represents the decimal conversion function. Watermark embedding: Replace decimal l1 and decimal l2 with watermark B1 and watermark B2 respectively: ； Where d j This represents the data from the j-th sensor. This indicates that the data from the j-th sensor should be replaced with the data in the decimal place, and Replace represents the replacement function; After embedding, the order of the decimal parts of the received data B is randomly scrambled to obtain the final generated data items: This represents the j-th final generated data item, and RanScram represents the random scrambling function; Comparison: When the aggregation node receives a data item, it determines whether the data has been tampered with by comparing the watermark.

2. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 1, characterized in that: From integers m A 128-bit binary number is generated through hash operation. M The formula is: Here, Hash represents the hash function.

3. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 2, characterized in that: binary number M After being divided into six equal segments, they are folded into six bits, which are represented as follows: b 1, b 2, b 3, b 4, b 5, b The formula for 6 is: Indicates will M The i-th segment, which is divided into six equal parts. Let represent the i-th bit, Slice represent the segmentation function, and Folding represent the folding function.

4. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 3, characterized in that: Two decimal digits B 1 and B The range of 2 is 0≤ B 1, B 2≤7.

5. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 4, characterized in that: When the aggregation node receives a data item from received data B, it restores the fractional part of the data item to its original order by reverse scrambling, and then calculates the watermark using the same method as in the sensor node watermark generation and compares it with the extracted watermark.

6. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 5, characterized in that: The calculation method for the currently collected data items is the same as that used to describe the sensor nodes, resulting in two decimal numbers, represented as follows: and Extract the two lowest decimal parts of the data item, and represent them as follows: and Compare the generated and extracted watermarks to check for tampering.

7. The authentication method for verifying the integrity of individual data based on watermarking technology according to claim 6, characterized in that: If the generated watermark and If the watermark is different from the original watermark, it is determined that the non-watermark portion of the data item has been tampered with during transmission; if the extracted watermark information... and If a change is found, it is determined that the watermark portion of the data item has been tampered with during transmission.

Images